An efficient management method and system for smart home device data

Abstract

The present invention relates to the technical field of device data processing in smart homes, and particularly relates to an efficient management method and system for smart home device data. The present invention obtains smart home device data; obtains the initial time window and preset initial threshold of a reference device; then classifies the data to obtain valid device data and regular device data; then obtains a user behavior matrix, and obtains a behavior pattern intensity index according to the similarity and time distribution between column vectors; then obtains the final time window; obtains a linkage coefficient between different devices according to the row vector similarity between different devices; then screens out strongly correlated devices; obtains a correction threshold by using the data change of strongly correlated devices in each final time window; and filters all device data according to the correction threshold. The present invention can dynamically adapt to the daily living habits of users to obtain accurate filtering results of smart home device data, so as to efficiently manage smart homes.

Description

Technical Field

[0001] The present invention relates to the technical field of device data processing in smart homes, and particularly to an efficient management method and system for smart home device data. Background Art

[0002] Smart home is an automated home control system that takes a residence as a platform and integrates a variety of advanced technologies to enhance the security, convenience, and comfort of the home. There are a large number of devices in the smart home system, and most of them require real-time data collection, so a huge amount of device data will be generated. These device data are usually stored in the cloud to avoid the storage limitations of local devices. In addition, many functions of smart homes, such as remote operation, device linkage, and federated learning, rely on the powerful performance of cloud computing servers. Therefore, how to combine smart homes with cloud computing and perform efficient data analysis and processing is an extremely important topic.

[0003] Since the number of smart home devices is increasing and their functions are becoming more and more perfect, the amount of data generated by the devices is also becoming more and more huge. If this data is directly transmitted to the cloud for processing without filtering, it will bring a great burden to the cloud, occupying too much network resources and storage resources. In the current smart home system, some simple edge filtering algorithms are basically used to preliminarily process the data at the home device side to reduce the cloud transmission pressure and improve the response efficiency. The most common one is the filtering method based on rule engines and event-driven. This method filters regular data according to preset rules and only uploads abnormal events. However, this method is limited by fixed rules and is difficult to adapt to dynamic scenarios, suffering from problems such as rule rigidity and lack of flexibility. For example, when the behavior habits of family members change, if the rules are not adjusted in time, it is possible to mis-filter valid data, resulting in a decrease in data utilization efficiency and failing to achieve the expected effect. Summary of the Invention

[0004] In order to solve the technical problems that the conventional filtering method is difficult to adapt to the dynamic scenarios of smart homes, the filtering rules are rigid and lack flexibility, resulting in possible mis-filtering of smart home device data, a decrease in data utilization efficiency, failure to achieve the expected filtering effect, and ultimately the inability to efficiently manage smart homes, the purpose of the present invention is to provide an efficient management method and system for smart home device data, and the specific technical solutions adopted are as follows:

[0005] An efficient management method for smart home device data, the method includes:

[0006] Obtain the device data of each smart home device at different sampling times;

[0007] Select the device data of any one smart home device as the reference device data of the reference device; evenly divide all the sampling moments of the reference device to obtain the initial time window during the operation of the reference device; obtain the preset initial threshold for each initial time window of the reference device; classify the reference device data of each time window according to the preset initial threshold to obtain effective device data and regular device data; obtain the user behavior matrix of the reference device according to the distribution characteristics and data differences of all the effective device data and regular device data; obtain the behavior rule intensity index of each column vector according to the similarity characteristics and time distribution characteristics between different column vectors in the user behavior matrix; obtain the final time window during the operation of the reference device according to the behavior rule intensity index; obtain the linkage coefficient between the reference device and each other device according to the similarity characteristics between the corresponding row vectors in the user behavior matrix of the reference device and each other device;

[0008] Screen out all the strongly related devices of the reference device according to the linkage coefficient; obtain the correction threshold for each final time window according to the device data change characteristics of each strongly related device in each final time window; filter all the device data according to the correction threshold.

[0009] Further, the method for obtaining the effective device data and the regular device data includes:

[0010] When the data difference between every two adjacent device data in each initial time window is greater than the preset initial threshold, take the latter device data of every two adjacent device data as the effective device data; when the data difference between every two adjacent device data in each initial time window is less than the preset initial threshold, take the latter device data of every two adjacent device data as the regular device data.

[0011] Further, the method for obtaining the user behavior matrix includes:

[0012] Take the time period from when the user operates on the reference device until the device data of the reference device stops changing as each user behavior period, calculate the average of the continuous regular device data of each user behavior period of the reference device to obtain the regular interval data of each user behavior period in the reference device, and use the sequence composed of all the regular interval data of each user behavior period as the interval sequence;

[0013] Take the sequence composed of all the effective device data of each user behavior period of the reference device as the effective sequence;

[0014] Statistically calculate the effective sequence length, effective sequence difference, interval sequence length, and interval sequence mean within each user behavior period;

[0015] The effective sequence length, effective sequence difference, interval sequence length, and interval sequence mean within each user behavior period are combined into column vectors, and all column vectors are combined into a user behavior matrix.

[0016] Furthermore, the method for obtaining the behavior pattern intensity index includes:

[0017] Cluster all column vectors in the user behavior matrix to obtain all column vector clustering clusters;

[0018] Take the column vector clustering clusters with more than one column vector as the clustering clusters to be analyzed;

[0019] Obtain the behavior pattern intensity index according to the behavior pattern intensity index calculation formula, and the behavior pattern intensity index calculation formula is as follows:

[0020] ;

[0021] In the formula, represents the behavior pattern intensity index of each column vector; represents the number of column vectors in the clustering cluster to be analyzed; represents the preset time period included in the user behavior matrix; represents the time interval between every two adjacent column vectors in the clustering cluster to be analyzed; represents the standard deviation of the time interval; represents the normalization function; represents the floor function; represents the absolute value function.

[0022] Furthermore, the method for obtaining the final time window includes:

[0023] Take the column vectors with behavior pattern intensity index greater than the preset first threshold as valid column vectors; take the user behavior periods corresponding to all valid column vectors in the user behavior matrix of the reference device as each final time window.

[0024] Furthermore, the method for obtaining the linkage coefficient includes:

[0025] Each element in the first row vector of the user behavior matrix is the effective sequence length of each user behavior period, each element in the second row vector is the effective sequence difference of each user behavior period, each element in the third row vector is the interval sequence length of each user behavior period, and each element in the fourth row vector is the interval sequence mean of each user behavior period;

[0026] Obtain the linkage coefficient according to the linkage coefficient calculation formula, and the linkage coefficient calculation formula is as follows:

[0027] ;

[0028] Wherein, represents the linkage coefficient between the reference device and each other device; represents the time length corresponding to each valid sequence of the reference device; represents the time length corresponding to each valid sequence of each other device; represents the difference of each valid sequence of the reference device; represents the difference of each valid sequence of each other device; represents the standard deviation function; represents the first row vector of the user behavior matrix of the reference device; represents the first row vector of the user behavior matrix of each other device; represents the second row vector of the user behavior matrix of the reference device; represents the second row vector of the user behavior matrix of each other device; represents the third row vector of the user behavior matrix of the reference device; represents the third row vector of the user behavior matrix of each other device; represents the fourth row vector of the user behavior matrix of the reference device; represents the fourth row vector of the user behavior matrix of each other device; represents the dynamic time warping function.

[0029] Further, the method for obtaining all strongly correlated devices of the reference device includes:

[0030] Taking each other device with a linkage coefficient greater than a preset second threshold as a device to be analyzed;

[0031] Sorting all devices to be analyzed in descending order of the linkage coefficient, and selecting a preset number of other devices from front to back as the strongly correlated devices of the reference device.

[0032] Further, the method for obtaining the correction threshold includes:

[0033] Obtaining the correction threshold according to the correction threshold calculation formula, and the correction threshold calculation formula is as follows:

[0034] , ;

[0035] Wherein, represents the correction threshold of each final time window; represents the preset initial threshold of each final time window; is the number of strongly correlated devices of the reference device; represents the effective sequence length of the th strongly correlated device within the time range; represents the effective sequence difference of the th strongly correlated device within the time range; represents the time range; represents the start time node of each final time window; represents the time difference between the start time node of each final time window of the reference device and the start time node of the final time window of the th closest strongly correlated device; represents the normalization function.

[0036] An efficient management system for smart home device data, the system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the above-mentioned efficient management method for smart home device data.

[0037] A computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the above-mentioned efficient management method for smart home device data.

[0038] The present invention has the following beneficial effects:

[0039] The present invention obtains device data of each smart home device at different sampling moments to facilitate subsequent screening and filtering operations on the device data. In common edge filtering methods, in order to reduce the amount of data and the transmission pressure on the cloud, a threshold-triggered method is usually used to filter data. Therefore, an initial time window is first set for the device data obtained during the operation of the reference device, and a preset initial threshold in each initial time window is set to preliminarily filter the device data of the reference device. The device data of the reference device in each time window is classified according to the initial threshold to obtain valid device data and regular device data. Since the change of the device data of smart home is closely related to the user's behavior, the change of the device data can reflect the user's daily living habits. Therefore, a user behavior matrix of the reference device is obtained. By using the similarity features and time distribution features of the device data under different user behaviors, the regularity of the user behavior is analyzed to obtain a regularity intensity index. According to the behavior regularity intensity index, the final time window during the operation of the reference device is obtained. Since the preset initial threshold does not adaptively adjust with the change of the user's behavior habits, when the user changes habits, some valid device data may be ignored in the part where the original preset initial threshold is large, and multi-device fusion analysis can more accurately analyze the data change pattern to adjust the preset initial threshold. Therefore, the linkage coefficient between the reference device and each other device is analyzed, and then the strongly correlated devices of the reference device are screened out. In actual situations, when the strongly correlated devices of the reference device show a stable state, the filtering threshold of the final time window of the reference device should be adjusted larger to reduce the upload of regular device data. When the strongly correlated devices of the reference device show an active state, the filtering threshold of the final time window of the reference device should be adjusted smaller to improve the capture accuracy of valid device data. Therefore, according to the device data change characteristics of each strongly correlated device in each final time window, the correction threshold of each final time window is obtained. All device data is filtered according to the correction threshold. The present invention can dynamically adapt to the user's daily living habits to obtain accurate filtering results of smart home device data, so as to efficiently manage smart home. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0041] Figure 1 It is a flowchart of an efficient management method for smart home device data provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, in conjunction with the accompanying drawings and preferred embodiments, a method and system for efficient management of smart home device data proposed according to the present invention, including its specific implementation manner, structure, features, and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0044] The following specifically describes, in conjunction with the accompanying drawings, the specific solution of a method and system for efficient management of smart home device data provided by the present invention.

[0045] Please refer to Figure 1 , which shows a method for efficient management of smart home device data provided by an embodiment of the present invention. The method includes:

[0046] Step S1: Obtain the device data of each smart home device at different sampling times.

[0047] The embodiment of the present invention is mainly applied to the scenario of filtering device data when a smart home uploads an abnormal event. Since the amount of device data corresponding to a smart home abnormal event is small, while the amount of device data generated usually is extremely large, the embodiment of the present invention needs to first obtain the device data of each smart home device at different sampling times and perform subsequent screening and filtering operations on the device data.

[0048] In one embodiment of the present invention, the sampling time is set to 1 minute. It should be noted that in other embodiments of the present invention, the sampling time can be set by itself and is not limited herein.

[0049] Step S2: Arbitrarily select the device data of a smart home device as the reference device data of the reference device; evenly divide all the sampling moments of the reference device to obtain the initial time window during the operation of the reference device; obtain the preset initial threshold of the reference device in each initial time window; classify the reference device data of each time window according to the initial threshold to obtain valid device data and regular device data; obtain the user behavior matrix of the reference device according to the distribution characteristics and data differences of all the valid device data and regular device data; obtain the behavior rule intensity index of each column vector according to the similarity characteristics and time distribution characteristics between different column vectors in the user behavior matrix; obtain the final time window during the operation of the reference device according to the behavior rule intensity index; obtain the linkage coefficient between the reference device and each other device according to the similarity characteristics between the corresponding row vectors in the user behavior matrices of the reference device and each other device.

[0050] In the device layer of the smart home system, the functions and forms of home devices are diverse, and there are many devices that need to collect data in real time and at high frequency, such as indoor temperature control devices, air quality monitoring devices, smart refrigerators, etc. The data generated by these devices is large and is the main data source of the smart home system. In the common edge filtering methods, in order to reduce the data volume and the transmission pressure on the cloud, the threshold trigger method is usually used to filter data. This method will set a fixed threshold range in the rule in advance, and only when the data change exceeds the set range will it be uploaded as abnormal data, otherwise it will be regarded as regular device data for filtering. Therefore, in the embodiment of the present invention, an initial time window is first set for the device data obtained during the operation of the reference device, and a preset initial threshold in each initial time window is set to preliminarily filter the device data of the reference device.

[0051] In an embodiment of the present invention, each initial time window contains the device data of 100 sampling moments, that is, each initial time window is 1 hour and 40 minutes. It should be noted that in other embodiments of the present invention, the initial time window can be set by itself and is not limited here.

[0052] In an embodiment of the present invention, the preset initial threshold of each initial time window is set by using environmental factors or the parameters when the device leaves the factory. In an embodiment of the present invention, a process for setting the preset initial threshold of the device by using environmental factors is provided as follows:

[0053] Since the temperature changes significantly during sunrise in the morning and sunset in the evening, the normal change range of the indoor temperature per minute is approximately between 0.1°C and 0.5°C at this time. Therefore, the filtering threshold of the initial time window for the indoor temperature measurement device in these two periods is set to 0.5. While during other ordinary periods, the normal change range of the temperature per minute is approximately between 0.1°C and 0.3°C, so the filtering threshold of the initial time window in these periods is set to 0.3.

[0054] The preset initial threshold for each initial time window of other devices is set according to the above method, and it is required that the setting of the preset initial threshold conforms to the actual situation, which will not be elaborated here.

[0055] Classify the reference device data for each time window according to the initial threshold to obtain valid device data and regular device data. Preferably, in an embodiment of the present invention, the method for obtaining valid device data and regular device data includes:

[0056] When the data difference between every two adjacent device data in each initial time window is greater than the preset initial threshold, the latter device data of every two adjacent device data is used as valid device data; when the data difference between every two adjacent device data in each initial time window is less than the preset initial threshold, the latter device data of every two adjacent device data is used as regular device data.

[0057] Since the change of the device data of the smart home is closely related to the user's behavior, the change of the device data can reflect the user's daily living habits. To avoid the filtering accuracy from decreasing due to the change of the user's habits, it is necessary to capture the change of the user's behavior habits according to the data characteristics within the initial time window and adjust the length of the initial time window and the size of the preset initial threshold. Therefore, in an embodiment of the present invention, according to the distribution characteristics and data differences of all valid device data and regular device data, a user behavior matrix of the reference device is obtained.

[0058] Preferably, in an embodiment of the present invention, the method for obtaining the user behavior matrix includes:

[0059] Since the smart home device has a delay, that is, after the user performs an operation, the reference device will generate a change in device data after a certain period of time. Therefore, the period from when the user operates on the reference device until the device data of the reference device ends changing is used as each user behavior period. The average of the continuous regular device data of each user behavior period of the reference device is obtained to get the regular interval data of each user behavior period of the reference device, and the sequence composed of all the regular interval data of each user behavior period is used as the interval sequence.

[0060] The sequence composed of all the valid device data of each user behavior period of the reference device is used as the valid sequence.

[0061] Statistically analyze the effective sequence length, effective sequence difference, interval sequence length, and interval sequence mean within each user behavior period;

[0062] Form a column vector with the effective sequence length, effective sequence difference, interval sequence length, and interval sequence mean within each user behavior period, and form a user behavior matrix with all column vectors.

[0063] In an embodiment of the present invention, an example of a user behavior matrix is provided, which is specifically as follows:

[0064]

[0065] Since the data of smart home devices is closely related to people's daily routines, the embodiments of the present invention utilize the similar characteristics and time distribution characteristics of device data under different user behaviors to analyze the regularity of user behaviors.

[0066] Preferably, in an embodiment of the present invention, the method for obtaining the regularity strength index includes:

[0067] Cluster all column vectors in the user behavior matrix to obtain all column vector clustering clusters; use the column vector clustering clusters with more than one column vector as the clustering clusters to be analyzed.

[0068] Obtain the behavior regularity strength index according to the behavior regularity strength index calculation formula, and the behavior regularity strength index calculation formula is as follows:

[0069] ;

[0070] In the formula, represents the behavior regularity strength index of each column vector; represents the number of column vectors in the clustering cluster to be analyzed; represents the preset time period included in the user behavior matrix; represents the time interval between every two adjacent column vectors in the clustering cluster to be analyzed; represents the standard deviation of the time interval; represents the normalization function; represents the floor function; represents the absolute value function.

[0071] In an embodiment of the present invention, since a user may generate the same behavior multiple times at the same time node on the same day, the preset time period is set to 1 day, and the intensity regularity of each column vector is analyzed by the number of times the user generates the same behavior within each day.

[0072] In the behavior regularity strength index calculation formula, represents the fractional part. The closer the fractional part is to 1, that is the closer it is to 0, it indicates that the number of user behaviors corresponding to similar column vectors is closer to a multiple of the preset time period. This multiple is the number of times the user performs the same behavior within the preset time period, and the smaller it is, the higher the intensity of the user behavior pattern, and the larger the behavior pattern intensity index of the column vector. The smaller the standard deviation of the time intervals between every two adjacent column vectors in the clustering cluster to be analyzed, the more similar the interval times when the user performs the same behavior. At this time, the higher the intensity of the user behavior pattern, and the larger the behavior pattern intensity index of the column vector.

[0073] Preferably, in an embodiment of the present invention, obtaining the final time window during the operation of the reference device according to the behavior pattern intensity index includes:

[0074] Regarding the column vectors with a behavior pattern intensity index greater than a preset first threshold as valid column vectors; regarding the user behavior periods corresponding to all valid column vectors in the user behavior matrix of the reference device as each final time window. In an embodiment of the present invention, the preset first threshold is set to 0.7, and the preset first threshold can be set by itself and is not limited herein.

[0075] After determining the final time window, it is also necessary to adjust the size of the preset initial threshold to further enhance the filtering accuracy of the device data. In conventional algorithms, the preset initial threshold does not adaptively adjust with the change of the user behavior habit. When the user changes the habit, some parts with a relatively large original preset initial threshold may ignore some valid device data, resulting in over-filtering; while some parts with a relatively small original preset initial threshold may misjudge normal device data as valid device data, resulting in incomplete filtering. In the analysis of the previous steps, the relationship between the user behavior and the device data has been roughly identified. However, these rules are the analysis results obtained from the data of a single smart home device. Applying this result to the adjustment of the time window can improve the robustness of the algorithm, but when adjusting the preset initial threshold, it is necessary to more precisely analyze the data change pattern, otherwise the problem of inaccurate preset initial threshold still cannot be solved. Therefore, in the embodiment of the present invention, a fusion analysis is performed by combining the linkage relationship of multiple devices to obtain a filtering rule that more conforms to the user behavior characteristics. Since the devices with a strong correlation relationship with the reference device will also change their device data simultaneously when the data of the reference device changes, but there will be a lag, and the higher the intensity of the user behavior, the smaller the time delay of the lag. Therefore, according to this relationship, the linkage coefficient between the reference device and other devices is obtained, and then the strongly correlated devices of the reference device are screened out.

[0076] Preferably, in an embodiment of the present invention, the method for obtaining the linkage coefficient includes:

[0077] Each element in the first row vector of the user behavior matrix is the effective sequence length of each user behavior period, each element in the second row vector is the effective sequence difference of each user behavior period, each element in the third row vector is the interval sequence length of each user behavior period, and each element in the fourth row vector is the interval sequence mean of each user behavior period.

[0078] Obtain the linkage coefficient according to the linkage coefficient calculation formula, and the linkage coefficient calculation formula is as follows:

[0079] ;

[0080] In the formula, represents the linkage coefficient between the reference device and each other device; represents the time length corresponding to each effective sequence of the reference device; represents the time length corresponding to each effective sequence of each other device; represents the difference of each effective sequence of the reference device; represents the difference of each effective sequence of each other device; represents the standard deviation function; represents the first row vector of the user behavior matrix of the reference device; represents the first row vector of the user behavior matrix of each other device; represents the second row vector of the user behavior matrix of the reference device; represents the second row vector of the user behavior matrix of each other device; represents the third row vector of the user behavior matrix of the reference device; represents the third row vector of the user behavior matrix of each other device; represents the fourth row vector of the user behavior matrix of the reference device; represents the fourth row vector of the user behavior matrix of each other device; represents the dynamic time warping function.

[0081] In the linkage coefficient calculation formula, analyze each effective sequence of the reference device and each other device. When the difference in the time lengths corresponding to all the effective sequences of the reference device is smaller, and the difference in the effective sequence differences is smaller, and at the same time, the difference in the time lengths corresponding to all the effective sequences of the other device is smaller, and the difference in the effective sequence differences is smaller, it indicates that the lag time between the reference device and the other device is smaller, and at this time, the user behavior intensity is greater, that is The smaller it is, the greater the inverse trend of the lag time between the reference device and other devices and the user behavior intensity, the greater the linkage relationship between the reference device and the other device, and at this time, the greater the linkage coefficient between the reference device and the other device; use the dynamic time warping function to compare the similarity between each row vector in the user behavior matrix of the reference device and each other device, and the smaller it is, the higher the similarity between the reference device and each other device row vector, the greater the linkage relationship between the reference device and the other device, and at this time, the greater the linkage coefficient between the reference device and the other device.

[0082] Step S3: Screen out all strongly correlated devices of the reference device according to the linkage coefficient; obtain the correction threshold for each final time window according to the device data change characteristics of each strongly correlated device in each final time window; filter all device data according to the correction threshold.

[0083] According to the above steps, the linkage coefficient between the reference device and each other device can be obtained, and based on this, the strongly correlated devices of the reference device can be screened out.

[0084] Preferably, in an embodiment of the present invention, the method for obtaining all strongly correlated devices of the reference device includes:

[0085] Take each other device with a linkage coefficient greater than the preset second threshold as a device to be analyzed; in an embodiment of the present invention, the preset second threshold is set to 0.5. It should be noted that the preset second threshold can be set by itself and is not limited here.

[0086] Sort all devices to be analyzed in descending order according to the linkage coefficient, and select a preset number of other devices from front to back as the strongly correlated devices of the reference device. In an embodiment of the present invention, the preset number is set to 3. It should be noted that the preset number can be set by itself and is not limited here.

[0087] In actual situations, when the strongly correlated devices of the reference device show a stable state, the filtering threshold of the final time window of the reference device should be adjusted to be larger to reduce the upload of conventional device data; while when the strongly correlated devices of the reference device show an active state, the filtering threshold of the final time window of the reference device should be adjusted to be smaller to improve the capture accuracy of effective device data. Therefore, in the embodiment of the present invention, the correction threshold for each final time window is obtained according to the device data change characteristics of each strongly correlated device in each final time window.

[0088] Preferably, in an embodiment of the present invention, the method for obtaining the correction threshold includes:

[0089] Obtain the correction threshold according to the correction threshold calculation formula, and the correction threshold calculation formula is as follows:

[0090] ;

[0091] In the formula, represents the correction threshold for each final time window; represents the preset initial threshold for each final time window; is the number of strongly correlated devices of the reference device; represents the th strongly correlated device's effective sequence length within the time range; represents the th strongly correlated device's effective sequence difference within the time range; represents the time range; represents the start time node of each final time window; represents the time difference between the start time node of each final time window of the reference device and the start time node of the final time window of the th closest strongly correlated device; represents the normalization function.

[0092] In the correction threshold calculation formula, due to the weak lag of device data changes between strongly correlated devices, a smaller time range is specified to compare the degree of device data changes between two devices within the smaller time range, so as to facilitate the correction of the filtering threshold for each final time window. Therefore, in the embodiments of the present invention, is used as the time range for comparing two devices; the greater the effective sequence length and the effective sequence difference of each strongly correlated device within the time range, the higher the degree of device data changes and the longer the change time within the time range. At this time, the state of this strongly correlated device is more active and less suitable as a reference for adjusting the threshold of the reference device. At this time, the correction value for the preset initial threshold is smaller. Analyze all the strongly correlated devices of the reference device to obtain to correct the preset initial threshold; the 5 in the denominator is to control the scaling degree of the correction value. In other embodiments of the present invention, other values can be used, which are not limited herein.

[0093] Thus, the correction threshold for each final time window is obtained.

[0094] Filter the device data of each device through the division of each device's final time window and the correction threshold of each final time window, and upload the filtered data to the cloud.

[0095] In summary, obtain the device data of each smart home device at different sampling times; select the device data of any one smart home device as the reference device data of the reference device; evenly divide all the sampling times of the reference device to obtain the initial time window during the operation of the reference device; obtain the preset initial threshold of the reference device in each initial time window; classify the reference device data in each time window according to the preset initial threshold to obtain valid device data and regular device data; obtain the user behavior matrix of the reference device according to the distribution characteristics and data differences of all the valid device data and regular device data; obtain the behavior rule strength index of each column vector according to the similarity characteristics and time distribution characteristics between different column vectors in the user behavior matrix; obtain the final time window during the operation of the reference device according to the behavior rule strength index; obtain the linkage coefficient between the reference device and each other device according to the similarity characteristics between the corresponding row vectors in the user behavior matrix of the reference device and each other device; screen out all the strongly correlated devices of the reference device according to the linkage coefficient; obtain the correction threshold of each final time window according to the device data change characteristics of each strongly correlated device in each final time window; filter all the device data according to the correction threshold.

[0096] The second object of the embodiment of the present invention is to provide an efficient management system for smart home device data, the system includes a memory, a processor and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs in the processor, it can implement the method described in steps S1-S3.

[0097] The third object of the embodiment of the present invention is to provide a computer device, including a memory, a processor and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, it implements the steps of the method described in the above steps S1-S3.

[0098] It should be noted that: the above sequence of the embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0099] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. The key point of each embodiment is to illustrate the differences from other embodiments.

Claims

1. An efficient management method for smart home device data, characterized in that, The method includes: Obtaining device data of each smart home device at different sampling moments; Optionally selecting the device data of one smart home device as the reference device data of the reference device; evenly dividing all sampling moments of the reference device to obtain the initial time window during the operation of the reference device; obtaining the preset initial threshold of the reference device in each initial time window; classifying the reference device data of each time window according to the preset initial threshold to obtain valid device data and regular device data; obtaining the user behavior matrix of the reference device according to the distribution characteristics and data differences of all valid device data and regular device data; obtaining the behavior pattern intensity index of each column vector according to the similarity characteristics and time distribution characteristics between different column vectors in the user behavior matrix; obtaining the final time window during the operation of the reference device according to the behavior pattern intensity index; obtaining the linkage coefficient between the reference device and each other device according to the similarity characteristics between the corresponding row vectors in the user behavior matrices of the reference device and each other device; Screening out all strongly correlated devices of the reference device according to the linkage coefficient; obtaining the correction threshold of each final time window according to the device data change characteristics of each strongly correlated device in each final time window; filtering all device data according to the correction threshold; The method for obtaining valid device data and regular device data includes: When the data difference between every two adjacent device data in each initial time window is greater than the preset initial threshold, taking the latter device data of every two adjacent device data as valid device data; when the data difference between every two adjacent device data in each initial time window is less than the preset initial threshold, taking the latter device data of every two adjacent device data as regular device data; The method for obtaining the user behavior matrix includes: Taking the time period from when the user operates on the reference device until the device data of the reference device stops changing as each user behavior period, averaging the continuous regular device data of each user behavior period of the reference device to obtain the regular interval data of each user behavior period in the reference device, and taking the sequence composed of all regular interval data of each user behavior period as the interval sequence; Taking the sequence composed of all valid device data of each user behavior period of the reference device as the valid sequence; Counting the valid sequence length, valid sequence difference, interval sequence length, and interval sequence mean within each user behavior period; Taking the valid sequence length, valid sequence difference, interval sequence length, and interval sequence mean within each user behavior period to form a column vector, and forming a user behavior matrix with all column vectors.

2. The efficient management method for data of a smart home device according to claim 1, characterized in that, The method for obtaining the behavior pattern intensity index includes: Clustering all column vectors in the user behavior matrix to obtain all column vector clustering clusters; Taking the column vector clustering clusters with more than one column vector as the clustering clusters to be analyzed; Obtaining the behavior pattern intensity index according to the behavior pattern intensity index calculation formula, and the behavior pattern intensity index calculation formula is as follows: ； In the formula, represents the behavior pattern intensity index of each column vector; represents the number of column vectors for the clustering cluster to be analyzed; represents the preset time period included in the user behavior matrix; represents the time interval between every two adjacent column vectors in the clustering cluster to be analyzed; represents the standard deviation of the time interval; represents the normalization function; represents the floor function; represents the absolute value function.

3. The efficient management method for data of a smart home device according to claim 1, characterized in that, The method for obtaining the final time window includes: Take the column vectors with the behavior pattern intensity index greater than the preset first threshold as valid column vectors; take the user behavior periods corresponding to all the valid column vectors in the user behavior matrix of the reference device as each final time window.

4. The efficient management method for smart home device data according to claim 1, characterized in that The method for obtaining the linkage coefficient includes: Each element in the first row vector of the user behavior matrix is the effective sequence length of each user behavior period, each element in the second row vector is the effective sequence difference of each user behavior period, each element in the third row vector is the interval sequence length of each user behavior period, and each element in the fourth row vector is the interval sequence mean of each user behavior period; Obtain the linkage coefficient according to the linkage coefficient calculation formula, and the linkage coefficient calculation formula is as follows: ； Wherein, represents the linkage coefficient between the reference device and each other device; represents the time length corresponding to each valid sequence of the reference device; represents the time length corresponding to each valid sequence of each other device; represents the difference of each valid sequence of the reference device; represents the difference of each valid sequence of each other device; represents the standard deviation function; represents the first row vector of the user behavior matrix of the reference device; represents the first row vector of the user behavior matrix of each other device; represents the second row vector of the user behavior matrix of the reference device; represents the second row vector of the user behavior matrix of each other device; represents the third row vector of the user behavior matrix of the reference device; represents the third row vector of the user behavior matrix of each other device; represents the fourth row vector of the user behavior matrix of the reference device; represents the fourth row vector of the user behavior matrix of each other device; represents the dynamic time warping function.

5. The efficient management method for smart home device data according to claim 1, characterized in that, The method for obtaining all strongly correlated devices of the reference device includes: Take each other device with a linkage coefficient greater than the preset second threshold as a device to be analyzed; Sort all the devices to be analyzed in descending order of the linkage coefficient, and select a preset number of other devices from front to back as the strongly correlated devices of the reference device.

6. The efficient management method for smart home device data according to claim 1, characterized in that The method for obtaining the correction threshold includes: Obtain the correction threshold according to the correction threshold calculation formula, and the correction threshold calculation formula is as follows: ， ； Wherein, represents the correction threshold for each final time window; represents the preset initial threshold for each final time window; is the number of strongly correlated devices of the reference device; represents the th valid sequence length of the strongly correlated device within the time range; represents the th valid sequence difference of the strongly correlated device within the time range; represents the time range; represents the start time node of each final time window; represents the time difference between the start time node of each final time window of the reference device and the start time node of the final time window of the th closest strongly correlated device; represents the normalization function.

7. An efficient management system for smart home device data, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the efficient management method for smart home device data according to any one of claims 1 to 6.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the efficient management method for smart home device data according to any one of claims 1 to 6.

Images