Smart home humidification control system based on big data

Abstract

The invention discloses a smart home humidification control system based on big data, which comprises a data vectorization module, a score calculation module, a candidate set generation module, a current operation vector generation module and an instruction generation module, and relates to the technical field of big data and smart home. Based on the timeliness, the user operation frequency and the environment matching degree, a comprehensive score is generated for each operation vector, dynamic scoring is achieved, the timeliness of the operation vectors is independently attenuated along with the passing of the current time, the operation frequency is converted through a logarithmic function in the user operation frequency, and the user operation frequency is converted into the environment matching degree. The difference between high-frequency and low-frequency operation vector scores can be effectively reduced, the problem of overlarge frequency difference can be relieved, the importance degree of the operation frequency can be more stably reflected, in the environment matching degree, historical operation scores with similar environments are higher, personalized requirements in different scenes are met, and the method is suitable for popularization and application. Therefore, the comprehensive score of the operation vector can more accurately capture the user demand change.

Description

Technical Field

[0001] The present invention relates to the field of big data and smart home technologies, and in particular to a smart home humidification control system based on big data. Background Art

[0002] Humidification control refers to the process of adjusting and managing the operating status and related parameters of the humidification equipment to achieve specific humidification effects and environmental requirements.

[0003] Publication number CN112213976B discloses a smart home humidification control method and system based on big data. Multiple operation vectors are extracted from an operation vector collection constructed based on historical operation records, and then an identification operation element for characterizing the user's established operating habits is determined from a sequence set determined by the multiple operation vectors. Then, a preliminary identification operation element for characterizing the user's currently changing operating habits is determined from the sequence set. The operation classification model obtained by training the identification operation element and the preliminary identification operation element can self-optimize with each change in the user's operating habits, and output an execution result instruction for the humidifier that is more in line with the user's actual operating habits.

[0004] However, the above application still has the following problems: the above application relies on fixed-weight statistics of historical operations, such as "number of occurrences", and does not consider the timeliness of the operation. It is difficult to adapt to the fast-paced changes in user behavior, resulting in delayed response. The above application handles the operation frequency and time factors in isolation, and does not build a comprehensive scoring system, which reduces the recommendation accuracy. Summary of the Invention

[0005] In order to solve the technical problems existing in the background technology, the present invention proposes a smart home humidification control system based on big data.

[0006] The present invention proposes a smart home humidification control system based on big data, comprising:

[0007] Data vectorization module: This module obtains the user's historical operation records of using the humidifier, extracts the operation type, operation time, and ambient humidity of each historical operation record, vectorizes the operation type, operation time, and ambient humidity, and combines them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set.

[0008] Rating calculation module: Generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching;

[0009] Candidate set generation module: sets a threshold for the comprehensive score, and generates candidate sets for operation vectors with a comprehensive score higher than the threshold;

[0010] Current operation vector generation module: Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation;

[0011] The Graph Attention Network algorithm is a graph neural network algorithm based on the attention mechanism in the existing technology. It aims to efficiently process graph structured data by dynamically calculating the attention weights between nodes and selectively aggregating neighbor information.

[0012] Instruction generation module: According to the operation vector of the current operation obtained, the operation vector of the current operation is analyzed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier control instruction executes the corresponding action.

[0013] Preferably, in the score calculation module, a comprehensive score is generated for each operation vector as follows:

[0014] Timeliness Weight time The calculation formula is:

[0015]

[0016] Where t is the current time, t0 is the operation time of the operation vector, e is the base of the natural exponential function, and γ is the time sensitivity coefficient;

[0017] User operation frequency Weight freq The calculation formula is:

[0018] Weight freq =log(N+1);

[0019] Where N is the number of times the operation type of the operation vector appears in the historical operation record, and log is the logarithmic function;

[0020] Environmental Matching Weight env The calculation formula is:

[0021]

[0022] E current is the vector of current ambient humidity, E history is the vector of ambient humidity in the operation vector, k is the environmental similarity coefficient, and k is obtained by calculating the standard deviation of the historical ambient humidity data in the historical operation records;

[0023] The calculation formula of the operation vector comprehensive score CS is:

[0024] CS=α·Weight time +β·Weight freq+∈·Weight env ;

[0025] Among them, α, β and ∈ are weight coefficients, and the values of α, β and ∈ are automatically obtained through the cross-validation algorithm.

[0026] Preferably, the timeliness Weight time The calculation formula of γ is:

[0027]

[0028] Where γ0 is the initial value of γ, ΔE is the absolute value of the difference between the current ambient humidity and the ambient humidity in the operation vector, σ is the standard deviation of the ambient humidity fluctuation, and exp is the natural exponential function;

[0029] σ is obtained by statistical calculation of historical environmental humidity data in historical operation records.

[0030] Preferably, in the candidate set generation module, the threshold of the comprehensive score is set as follows:

[0031] Comprehensive score threshold T 评分阈值 for:

[0032] T 评分阈值 =B 初始阈值 ·(1-β·ΔH);

[0033] When ΔH=0, T 评分阈值 =B 初始阈值 ;

[0034] Among them, B 初始阈值 is the initial threshold, ΔH is the difference between the target humidity and the current ambient humidity, and the target humidity is the ideal ambient humidity value preset by the user.

[0035] Preferably, it also includes a real-time abnormal operation detection module: used to detect in real time whether the current operation is an abnormal operation, and if it is determined to be an abnormal operation, trigger the start of the emergency humidification mode.

[0036] Preferably, in the real-time abnormal operation detection module, whether the current operation is an abnormal operation is detected in real time as follows:

[0037] According to the operation vector set, a normal distribution model of operation habits is constructed, and the mean vector μ and covariance matrix ∑ of the operation vector set are calculated;

[0038] Calculate the square value D of the Mahalanobis distance between the current operation and μ 2 :

[0039] D 2 =(x-μ) T Σ -1 (x-μ);

[0040] Among them, x is the operation vector of the current operation, ∑ -1 is the inverse matrix of the covariance matrix ∑, and T is the transpose operation;

[0041] If D 2 If the preset threshold is exceeded, the current operation is determined to be an abnormal operation, triggering the start of the emergency humidification mode.

[0042] Preferably, in the real-time abnormal operation detection module, the emergency humidification mode is started as follows:

[0043]

[0044] Where ΔH is the difference between the target humidity and the current ambient humidity;

[0045] ΔH max is the maximum allowable humidity difference;

[0046] Power is the current humidification power;

[0047] Max_Power is the maximum power of the humidifier.

[0048] A smart home humidification control method based on big data, comprising the following steps:

[0049] S1. Obtain historical operation records of the user using the humidifier, extract the operation type, operation time, and ambient humidity of each historical operation record, vectorize the operation type, operation time, and ambient humidity, and combine them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set;

[0050] S2. Generate a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching;

[0051] S3. Set a threshold for the comprehensive score, and generate a candidate set for the operation vectors whose comprehensive scores are higher than the threshold;

[0052] S4. Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation;

[0053] S5. Real-time detection of whether the current operation is an abnormal operation. If it is determined to be an abnormal operation, triggering the start of the emergency humidification mode;

[0054] S6. According to the obtained operation vector of the current operation, the operation vector of the current operation is parsed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier controls the instruction to execute the corresponding action.

[0055] The big data-based smart home humidification control system proposed in the present invention has the following beneficial technical effects:

[0056] 1. The scoring calculation module generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environmental matching, achieving dynamic scoring. The timeliness of the operation vector decays independently with the passage of time. The user operation frequency is converted using a logarithmic function, which can effectively narrow the gap between the scores of high-frequency and low-frequency operation vectors, alleviate the problem of excessive frequency differences, and more stably reflect the importance of operation frequency. For environmental matching, historical operations with similar environments receive higher scores, adapting to personalized needs in different scenarios. This allows the comprehensive score of the operation vector to more accurately capture changes in user needs.

[0057] 2. The setting of the real-time abnormal operation detection module introduces the Mahalanobis distance to detect abnormal operations, determines that the current operation is an abnormal operation, triggers the start of the emergency humidification mode, and enhances the fault tolerance of the system. Even in the event of abnormal operation, the system can continue to operate in the emergency humidification mode, avoiding the paralysis of the entire humidification control system due to a single fault or abnormality, improving the overall reliability of the system and ensuring equipment safety.

[0058] 3. This application generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching to capture changes in user habits;

[0059] A threshold for the comprehensive score is set, and operation vectors with a comprehensive score higher than the threshold are generated into a candidate set. This focuses on high-quality data and generates operation vectors that better meet actual needs for the current operation. By reducing the amount of calculation, a faster response can be made.

[0060] According to the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation. The graph attention network algorithm can selectively aggregate neighbor information by dynamically calculating the attention weights between nodes, thereby mining the complex relationship between operation vectors. By mining the relationship between operation vectors, the graph attention network algorithm can generate a more reasonable operation vector for the current operation, providing more accurate information for the instruction generation module, so that the control instructions generated by the instruction generation module based on these operation vectors can better meet user needs and improve the intelligence level of the smart home humidification control system.

[0061] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a principle block diagram of the system of the present invention;

[0063] Figure 2 Flowchart of the method of the present invention. DETAILED DESCRIPTION

[0064] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention, and are not to be construed as limiting the present invention.

[0065] like Figure 1 The smart home humidification control system based on big data shown in FIG. 1 includes:

[0066] Data vectorization module: This module obtains the user's historical operation records of using the humidifier, extracts the operation type, operation time, and ambient humidity of each historical operation record, vectorizes the operation type, operation time, and ambient humidity, and combines them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set.

[0067] Rating calculation module: Generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching;

[0068] Candidate set generation module: sets a threshold for the comprehensive score, and generates candidate sets for operation vectors with a comprehensive score higher than the threshold;

[0069] Current operation vector generation module: Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation;

[0070] The Graph Attention Network algorithm is a graph neural network algorithm based on the attention mechanism in the existing technology. It aims to efficiently process graph structured data by dynamically calculating the attention weights between nodes and selectively aggregating neighbor information.

[0071] Instruction generation module: according to the obtained operation vector of the current operation, the operation vector of the current operation is analyzed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier control instruction executes the corresponding action;

[0072] This application generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching to capture changes in user habits;

[0073] A threshold for the comprehensive score is set, and operation vectors with a comprehensive score higher than the threshold are generated into a candidate set. This focuses on high-quality data and generates operation vectors that better meet actual needs for the current operation. By reducing the amount of calculation, a faster response can be made.

[0074] According to the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation. The graph attention network algorithm can selectively aggregate neighbor information by dynamically calculating the attention weights between nodes, thereby mining the complex relationship between operation vectors. By mining the relationship between operation vectors, the graph attention network algorithm can generate a more reasonable operation vector for the current operation, providing more accurate information for the instruction generation module, so that the control instructions generated by the instruction generation module based on these operation vectors can better meet user needs and improve the intelligence level of the smart home humidification control system.

[0075] In an optional embodiment, the score calculation module generates a comprehensive score for each operation vector as follows:

[0076] Timeliness of the operation vector Weight time The calculation formula is:

[0077]

[0078] Among them, t is the current time, t0 is the operation time of the operation vector, e is the base of the natural exponential function, and γ is the time sensitivity coefficient; thus making the timeliness of the operation vector Weight time decays independently with the current time t;

[0079] By independently calculating the t-t0 of the action vector, we ensure that the comprehensive score CS decays naturally over time, thereby accurately capturing the evolution of user habits.

[0080] In an optional embodiment, the calculation formula of γ is:

[0081]

[0082] Where γ0 is the initial value of γ, ΔE is the absolute value of the difference between the current ambient humidity and the ambient humidity in the operation vector, σ is the standard deviation of the ambient humidity fluctuation, and exp is the natural exponential function;

[0083] σ is obtained by statistical calculation of historical environmental humidity data in historical operation records;

[0084] Dynamic settings of γ can adapt to scene changes in real time and are highly flexible;

[0085] User operation frequencyWeigh freq The calculation formula is:

[0086] Weight freq =log(N+1);

[0087] Where N is the number of times the operation type of the operation vector appears in the historical operation record, and log is the logarithmic function;

[0088] Weight freq The calculation formula is designed to address the fact that the number of times different operation vectors appear in smart home operation records can vary significantly. If the score were measured directly based on the raw number of operation occurrences, high-frequency operation vectors would receive scores significantly higher than low-frequency operation vectors, severely weakening or even negligible the impact of low-frequency operation vectors on the overall score (CS). Using a logarithmic function to transform the operation frequency effectively narrows the gap between high- and low-frequency operation vector scores, alleviating the problem of excessive frequency discrepancies. The logarithmically transformed score can more consistently reflect the importance of operation frequency across different user operation data, improving the stability and generalization of the subsequent overall score (CS).

[0089] Environmental Matching Weight env The calculation formula is:

[0090]

[0091] E current is the current environment vector, E history is the historical operating environment vector, k is the environmental similarity coefficient, and k is obtained by calculating the standard deviation of the historical environmental humidity data in the historical operating records;

[0092] E current The value of can be obtained through sensor collection;

[0093] k is dynamically calculated based on the standard deviation of historical environmental data rather than a fixed threshold, adapting to environmental differences among different users;

[0094] Environmental Matching Weight env The calculation formula is to give higher scores to historical operations with similar environments, adapting to personalized needs in different scenarios.

[0095] The calculation formula of the operation vector comprehensive score CS is:

[0096] CS=α·Weight time +β·Weight freq +∈·Weight env ;

[0097] Among them, α, β and ∈ are weight coefficients, and the values of α, β and ∈ are automatically obtained through the cross-validation algorithm.

[0098] The scoring calculation module is set up to generate a comprehensive score for each operation vector based on timeliness, user operation frequency, and environmental matching, realizing dynamic scoring. The timeliness of the operation vector decays independently with the passage of time. The operation frequency is converted using a logarithmic function in the user operation frequency, which can effectively narrow the gap between the scores of high-frequency and low-frequency operation vectors, alleviate the problem of excessive frequency differences, and more stably reflect the importance of operation frequency. In the environmental matching, historical operations with similar environments have higher scores, which adapts to personalized needs in different scenarios, so that the comprehensive score of the operation vector can more accurately capture changes in user needs.

[0099] In an optional embodiment, in the candidate set generation module, the threshold of the comprehensive score is set as follows:

[0100] Comprehensive score threshold T 评分阈值 for:

[0101] T 评分阈值 =B 初始阈值 ·(1-β·ΔH);

[0102] When ΔH=0, T 评分阈值 =B 初始阈值 ;

[0103] Among them, B 初始阈值 is the initial threshold, ΔH is the difference between the target humidity and the current ambient humidity, and the target humidity is the ideal ambient humidity value preset by the user;

[0104] Comprehensive score threshold T 评分阈值 In the formula:

[0105] Lower the threshold in dry environments, expand the candidate set, and quickly respond to humidification needs;

[0106] In humid environments, the threshold is raised to reduce the candidate set and avoid over-humidification.

[0107] Real-time abnormal operation detection module: used to detect in real time whether the current operation is an abnormal operation. If it is determined to be an abnormal operation, it triggers the start of the emergency humidification mode;

[0108] By calculating the Mahalanobis distance with the historical mean, real-time anomaly detection is achieved during operation execution. This design ensures the security and response speed of the system.

[0109] Current operation refers to the real-time operation of the humidifier being performed.

[0110] In an optional embodiment, in the real-time abnormal operation detection module, whether the current operation is an abnormal operation is detected in real time as follows:

[0111] According to the operation vector set, a normal distribution model of operation habits is constructed, and the mean vector μ and covariance matrix ∑ of the operation vector set are calculated;

[0112] Calculate the square value D of the Mahalanobis distance between the current operation and μ 2 :

[0113] D 2 =(x-μ) T Σ -1 (x-μ);

[0114] Among them, x is the operation vector of the current operation, ∑ -1 is the inverse matrix of the covariance matrix ∑, and T is the transpose operation;

[0115] If D 2 If the preset threshold is exceeded, the current operation is determined to be an abnormal operation, triggering the start of the emergency humidification mode.

[0116] In an optional embodiment, in the real-time abnormal operation detection module, starting the emergency humidification mode is:

[0117]

[0118] Where ΔH is the difference between the target humidity and the current ambient humidity;

[0119] ΔH max is the maximum allowable humidity difference; ΔH max The value of can be dynamically adjusted through reinforcement learning algorithms;

[0120] Power is the current humidification power;

[0121] Max_Power is the maximum power of the humidifier;

[0122] The humidification power is positively correlated with the humidity difference ΔH, that is, the greater the humidity difference, the higher the power is, which is used to quickly make up for the humidity deficiency;

[0123] When the current ambient humidity is lower than the target humidity, the closer the current ambient humidity is to the target humidity, the smaller ΔH is, the power is reduced, and energy waste is reduced.

[0124] The setting of the real-time abnormal operation detection module introduces the Mahalanobis distance to detect abnormal operations, determines that the current operation is an abnormal operation, triggers the start of the emergency humidification mode, and enhances the fault tolerance of the system. Even in the event of abnormal operation, the system can continue to operate in the emergency humidification mode, avoiding the paralysis of the entire humidification control system due to a single fault or abnormality, improving the overall reliability of the system and ensuring equipment safety.

[0125] like Figure 2A smart home humidification control method based on big data is shown, comprising the following steps:

[0126] S1. Obtain historical operation records of the user using the humidifier, extract the operation type, operation time, and ambient humidity of each historical operation record, vectorize the operation type, operation time, and ambient humidity, and combine them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set;

[0127] S2. Generate a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching;

[0128] S3. Set a threshold for the comprehensive score, and generate a candidate set for the operation vectors whose comprehensive scores are higher than the threshold;

[0129] S4. Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation;

[0130] S5. Real-time detection of whether the current operation is an abnormal operation. If it is determined to be an abnormal operation, trigger the start of the emergency humidification mode; otherwise, enter S6;

[0131] S6. According to the obtained operation vector of the current operation, the operation vector of the current operation is parsed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier controls the instruction to execute the corresponding action.

[0132] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0133] In the embodiments provided by the present invention, it should be understood that the disclosed systems or methods can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative. For example, the division of modules is only a logical function division, and other division methods may be used in actual implementation.

[0134] Modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of these modules may be selected to achieve the objectives of this embodiment based on actual needs.

[0135] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing module, each module may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or hardware plus software functional modules.

[0136] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic characteristics of the present invention.

[0137] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A smart home humidification control system based on big data, characterized in that: include: Data vectorization module: This module obtains the user's historical operation records of using the humidifier, extracts the operation type, operation time, and ambient humidity of each historical operation record, vectorizes the operation type, operation time, and ambient humidity, and combines them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set. Rating calculation module: Generates a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching; Candidate set generation module: sets a threshold for the comprehensive score, and generates candidate sets for operation vectors with a comprehensive score higher than the threshold; Current operation vector generation module: Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation; Instruction generation module: According to the operation vector of the current operation obtained, the operation vector of the current operation is analyzed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier control instruction executes the corresponding action.

2. The smart home humidification control system based on big data according to claim 1 is characterized in that: In the score calculation module, a comprehensive score is generated for each operation vector as follows: Timeliness Weight time The calculation formula is: Where t is the current time, t0 is the operation time of the operation vector, e is the base of the natural exponential function, and γ is the time sensitivity coefficient; User operation frequency Weight freq The calculation formula is: Weight freq =log(N+1); Where N is the number of times the operation type of the operation vector appears in the historical operation record, and log is the logarithmic function; Environmental Matching Weight env The calculation formula is: E current is the vector of current ambient humidity, E history is the vector of ambient humidity in the operation vector, k is the environmental similarity coefficient, and k is obtained by calculating the standard deviation of the historical ambient humidity data in the historical operation records; The calculation formula of the operation vector comprehensive score CS is: CS=α·Weight time +β·Weight freq +∈·Weight env ; Among them, α, β and ∈ are weight coefficients, and the values of α, β and ∈ are automatically obtained through the cross-validation algorithm.

3. The smart home humidification control system based on big data according to claim 2 is characterized in that: Timeliness Weight time The calculation formula of γ is: Where γ0 is the initial value of γ, ΔE is the absolute value of the difference between the current ambient humidity and the ambient humidity in the operation vector, σ is the standard deviation of the ambient humidity fluctuation, and exp is the natural exponential function; σ is obtained by statistical calculation of historical environmental humidity data in historical operation records.

4. The smart home humidification control system based on big data according to claim 1, characterized in that: In the candidate set generation module, the threshold of the comprehensive score is set as follows: Comprehensive score threshold T 评分阈值 for: T 评分阈值 =B 初始阈值 ·(1-β·ΔH); When ΔH = 0, T 评分阈值 = B 初始阈值 ; Among them, B 初始阈值 is the initial threshold, ΔH is the difference between the target humidity and the current ambient humidity, and the target humidity is the ideal ambient humidity value preset by the user.

5. The smart home humidification control system based on big data according to claim 1, characterized in that: It also includes a real-time abnormal operation detection module: used to detect in real time whether the current operation is an abnormal operation. If it is determined to be an abnormal operation, it triggers the start of the emergency humidification mode.

6. The smart home humidification control system based on big data according to claim 5, characterized in that: In the real-time abnormal operation detection module, whether the current operation is an abnormal operation is detected in real time as follows: According to the operation vector set, a normal distribution model of operation habits is constructed, and the mean vector μ and covariance matrix ∑ of the operation vector set are calculated; Calculate the square value D of the Mahalanobis distance between the current operation and μ 2 : D 2 =(x-μ) T ∑ -1 (x-μ); Among them, x is the operation vector of the current operation, ∑ -1 is the inverse matrix of the covariance matrix ∑, and T is the transpose operation; If D 2 If the preset threshold is exceeded, the current operation is determined to be an abnormal operation, triggering the start of the emergency humidification mode.

7. The smart home humidification control system based on big data according to claim 5 or 6, characterized in that: In the real-time abnormal operation detection module, the emergency humidification mode is started as follows: Where ΔH is the difference between the target humidity and the current ambient humidity; ΔH max is the maximum allowable humidity difference; Power is the current humidification power; Max_Power is the maximum power of the humidifier.

8. The method for controlling humidification in a smart home based on big data according to any one of claims 1 to 7, characterized in that: The following steps are involved: S1. Obtain historical operation records of the user using the humidifier, extract the operation type, operation time, and ambient humidity of each historical operation record, vectorize the operation type, operation time, and ambient humidity, and combine them to form an operation vector. Multiple operation vectors are constructed to form an operation vector set; S2. Generate a comprehensive score for each operation vector based on timeliness, user operation frequency, and environment matching; S3. Set a threshold for the comprehensive score, and generate a candidate set for the operation vectors whose comprehensive scores are higher than the threshold; S4. Based on the operation vectors in the candidate set, the graph attention network algorithm is used to calculate the operation vector of the current operation; S5. Real-time detection of whether the current operation is an abnormal operation. If it is determined to be an abnormal operation, triggering the start of the emergency humidification mode; S6. According to the obtained operation vector of the current operation, the operation vector of the current operation is parsed and decomposed into various operation elements. Each operation element generates a corresponding control instruction through a preset mapping table, and the control instruction is sent to the humidifier. The humidifier controls the instruction to execute the corresponding action.

Images