Multidimensional data monitoring method and device for wearable devices

Abstract

The present invention discloses a multi-dimensional data monitoring method and device for wearable devices, and relates to the field of data processing technology. The method includes: a patch layout array of an interactive target wearable device is used to obtain K monitoring data sets; multiple recognition tasks are obtained, and mapping associations are performed in combination with K patch monitoring functions to obtain multiple task-patch set mapping relationships; the K monitoring data sets are subjected to data concentration fluctuation interval endpoint screening to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients; the recognition task monitoring step size is configured to obtain multiple first monitoring step sizes and multiple second monitoring step sizes; multiple task monitors are constructed; and the monitoring data of the K layout patches are subjected to long-term extraction and analysis to obtain multiple monitoring results. The technical problem of insufficient accuracy of multi-dimensional data monitoring of wearable devices in the prior art is solved, and the technical effect of improving data monitoring accuracy is achieved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a multidimensional data monitoring method and device for wearable devices. Background Art

[0002] With the widespread use of wearable devices, users' demands for device functions have gradually diversified. Especially in multi-task application scenarios such as health monitoring and sports tracking, devices need to collect and process multi-dimensional physiological data in real time in order to provide users with personalized services and suggestions. However, there are still some urgent problems to be solved in the existing wearable device data monitoring technology. Traditional monitoring methods often use a fixed monitoring step size, which fails to make reasonable adjustments based on the specific needs of the monitoring task and the dynamic changes of sensor data, resulting in low data utilization efficiency and inaccurate monitoring results. Summary of the invention

[0003] The present application provides a multi-dimensional data monitoring method and apparatus for wearable devices, which solves the technical problem of insufficient accuracy of multi-dimensional data monitoring of wearable devices in the prior art.

[0004] In view of the above problems, the present application provides a multi-dimensional data monitoring method and apparatus for wearable devices.

[0005] In a first aspect of the present application, a multi-dimensional data monitoring method for a wearable device is provided, the method comprising:

[0006] Interact with the patch layout array of the target wearable device, obtain K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions; obtain multiple recognition tasks of the target wearable device, map and associate them with the K patch monitoring functions of the K layout patches, and obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to one recognition task; perform data concentration fluctuation interval endpoint screening on the K monitoring data sets, determine K monitoring data medians and K monitoring data fluctuation coefficients; perform recognition task monitoring step configuration based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients, and obtain multiple first monitoring steps and multiple second monitoring steps; construct multiple task monitors based on the multiple first monitoring steps and the multiple second monitoring steps; perform long-term extraction and analysis on the monitoring data of the K layout patches according to the multiple task monitors and the multiple task-patch set mapping relationships, and obtain multiple monitoring results.

[0007] A second aspect of the present application provides a multi-dimensional data monitoring device for a wearable device, the device comprising:

[0008] A data acquisition module, the data acquisition module is used to interact with the patch layout array of the target wearable device, and obtain K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions; a mapping module, the mapping module is used to obtain multiple recognition tasks of the target wearable device, and map and associate the K patch monitoring functions of the K layout patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to an recognition task; a screening module, the screening module is used to perform data concentration fluctuation interval end on the K monitoring data sets Point screening, determining the median values ​​of K monitoring data sets and the K monitoring data fluctuation coefficients; a configuration module, the configuration module is used to identify the task monitoring step configuration based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients, and obtain multiple first monitoring steps and multiple second monitoring steps; a construction module, the construction module is used to construct multiple task monitors based on the multiple first monitoring steps and the multiple second monitoring steps; an analysis module, the analysis module is used to perform long-term extraction and analysis on the monitoring data of the K deployed patches according to the multiple task monitors and the multiple task-patch set mapping relationships, and obtain multiple monitoring results.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] First, the patch layout array of the interactive target wearable device is used to obtain K monitoring data sets of K layout patches of the patch layout array within the preset monitoring window, wherein the K layout patches include K layout positions. Next, multiple recognition tasks of the target wearable device are obtained, and mapping associations are performed in combination with the K patch monitoring functions of the K layout patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to an recognition task. Then, the K monitoring data sets are screened for endpoints of the data center fluctuation interval to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients. Further, the recognition task monitoring step size is configured based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients to obtain multiple first monitoring step sizes and multiple second monitoring step sizes; and multiple task monitors are constructed based on the multiple first monitoring step sizes and multiple second monitoring step sizes. Finally, the monitoring data of the K layout patches are subjected to long-term extraction and analysis according to the multiple task monitors and the multiple task-patch set mapping relationships to obtain multiple monitoring results. The invention solves the technical problem of insufficient precision of multi-dimensional data monitoring of wearable devices in the prior art, and achieves the technical effect of improving the precision of data monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic diagram of a multi-dimensional data monitoring method for a wearable device provided in an embodiment of the present application;

[0013] Figure 2 A schematic diagram of the structure of a multi-dimensional data monitoring device for a wearable device provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: data acquisition module 11 , mapping module 12 , screening module 13 , configuration module 14 , construction module 15 , analysis module 16 . DETAILED DESCRIPTION

[0015] The present application solves the technical problem of insufficient accuracy of multi-dimensional data monitoring of wearable devices in the prior art by providing a multi-dimensional data monitoring method and device for wearable devices.

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides a multi-dimensional data monitoring method for a wearable device, wherein the method comprises:

[0019] The patch layout array of the interactive target wearable device is used to obtain K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions.

[0020] Communicate with the target wearable device in an interactive manner to obtain the patch array deployed on the target wearable device; within a pre-set monitoring time window, obtain a set of monitoring data of K patches from the patch array, each patch corresponds to a specific deployment position, and a total of K patches participate in data collection.

[0021] Acquire multiple recognition tasks of the target wearable device, and map and associate them with the K patch monitoring functions of the K deployed patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to one recognition task.

[0022] The recognition task refers to the target wearable device being able to perform multiple functional tasks, including heart rate monitoring, step counting, sleep quality analysis, motion pattern recognition, body temperature monitoring, etc. The K patches deployed on the wearable device each have a specific monitoring function. For example, some patches may be used to monitor heart rate, some may be used to monitor acceleration (for step counting or motion pattern recognition), and some may be used to monitor body temperature, etc. According to the correspondence between the recognition task and the patch monitoring function, the monitoring functions of these patches are matched with the recognition task to obtain multiple task-patch set mapping relationships. The task-patch set mapping relationship refers to the association relationship between each recognition task and a set of patches responsible for the task. Through mapping, each task will be assigned to the corresponding monitoring patch set. For example, the heart rate monitoring task will be mapped to the patch set related to monitoring heart rate, and the motion tracking task will be mapped to the patch set related to acceleration and position monitoring.

[0023] The K monitoring data sets are subjected to data center fluctuation interval endpoint screening to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients.

[0024] For each monitoring data set, the endpoints of the fluctuation range are screened to identify the maximum and minimum values ​​in each monitoring data set, and then the fluctuation range of each monitoring data set is obtained; the monitoring data median and monitoring data fluctuation coefficient of each monitoring data set are calculated within the fluctuation range. The monitoring data median represents the central trend of the data within the monitoring time period, and the monitoring data fluctuation coefficient reflects the degree of discreteness or volatility of the data.

[0025] Furthermore, the K monitoring data sets are subjected to data center fluctuation interval endpoint screening to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients, including:

[0026] A first monitoring data set is randomly extracted from the K monitoring data sets, and the mean of the first monitoring data set is calculated to obtain a first monitoring comprehensive concentration value; the first monitoring comprehensive concentration value is used as the midpoint of the first fluctuation interval, and the left and right ends of the midpoint of the first fluctuation interval are respectively iterated according to the preset screening bandwidth to obtain a first fluctuation left-ward interval and a first fluctuation right-ward interval; the left endpoint of the first fluctuation left-ward interval is iterated to the left according to the preset screening bandwidth to obtain a first iterative fluctuation left-ward interval, and the right endpoint of the first fluctuation right-ward interval is iterated to the right according to the preset screening bandwidth to obtain a first iterative fluctuation right-ward interval; respectively determine whether the concentration density of the first iterative fluctuation left-ward interval is greater than or equal to the concentration density of the first fluctuation left-ward interval, and determine whether the concentration density of the first iterative fluctuation right-ward interval is greater than or equal to the concentration density of the first fluctuation left-ward interval. Whether the concentration density of the first iterative fluctuation rightward interval is greater than or equal to the concentration density of the first fluctuation rightward interval, if so, obtain the left-to-right iteration authentication pass result and the right-to-left iteration authentication pass result; based on the left iteration of the left endpoint of the first iterative fluctuation leftward interval and the right iteration of the left endpoint of the first iterative fluctuation rightward interval according to the preset screening bandwidth, until the preset iteration stop condition is met, obtain the first data center fluctuation interval; calculate the mean of the first data center fluctuation interval to obtain the first monitoring data median, and calculate the fluctuation variance of the first data center fluctuation interval to obtain the first monitoring data fluctuation coefficient; screen the data center fluctuation interval endpoints of the K monitoring data sets according to the preset screening bandwidth to obtain the K monitoring data medians and K monitoring data fluctuation coefficients.

[0027] Specifically, one is randomly selected from K monitoring data sets as the first monitoring data set, and the mean of the first monitoring data set is calculated to obtain a first monitoring comprehensive concentrated value, which represents the preliminary average level of this group of monitoring data and serves as the basis for subsequent fluctuation interval calculations; the first monitoring comprehensive concentrated value is taken as the midpoint, and it is iterated to the left and right ends according to the preset filtering bandwidth to form the first fluctuation left interval and the first fluctuation right interval respectively; after obtaining the first fluctuation left interval and the first fluctuation right interval, it is iterated to the left and right again according to the preset filtering bandwidth to obtain the first iteration fluctuation left interval and the first iteration fluctuation right interval respectively; for the fluctuation interval after each iteration, the first iteration fluctuation left interval and the first iteration fluctuation right interval are judged respectively. Whether the concentration density of the right-hand fluctuation interval is greater than or equal to the concentration density of the original fluctuation interval; if the interval density after iteration is higher, it indicates that the data in the interval is more representative, thereby obtaining the left-hand iteration authentication pass result and the right-hand iteration authentication pass result; continue to iterate the left-hand and right-hand iteration intervals according to the screening bandwidth until the preset iteration stop condition is met; finally, through iteration, obtain the fluctuation interval in the first data center; perform mean calculation again on the fluctuation interval in the first data center obtained finally, obtain the median value of the first monitoring data set, that is, the most accurate concentration value obtained after the entire screening process; at the same time, calculate the fluctuation variance of the fluctuation interval in the first data center, obtain the first monitoring data fluctuation coefficient, and the first monitoring data fluctuation coefficient reflects the degree of fluctuation of the data in the interval. According to the same preset screening bandwidth and iteration steps, similar processing is performed on all K monitoring data sets, and finally the concentration value and fluctuation coefficient of each monitoring data set are obtained, that is, the median value of K monitoring data sets and the fluctuation coefficient of K monitoring data.

[0028] Furthermore, the preset iteration stopping condition is that the concentration density difference between two adjacent iterations is less than or equal to the preset concentration density difference.

[0029] The condition for stopping the iteration is based on the concentration density difference. In each iteration, the concentration density difference between two adjacent intervals is calculated. When the concentration density difference between two adjacent iterations is less than or equal to the preset concentration density difference, the iteration process will stop.

[0030] Furthermore, the amount of monitoring data in the first iterative fluctuation leftward interval is counted, and the calculated result is divided by the interval length of the first iterative fluctuation leftward interval to obtain the concentration density of the first iterative fluctuation leftward interval.

[0031] The amount of monitoring data in the first iteration fluctuation left-ward interval refers to how many monitoring data points are collected in the first iteration fluctuation left-ward interval. For example, if there are 50 data points in the range of this interval, then the amount of monitoring data is 50. The interval length of the first iteration fluctuation left-ward interval refers to the numerical range of the first iteration fluctuation left-ward interval. For example, assuming that the left endpoint of the first iteration fluctuation left-ward interval is 70 and the right endpoint is 80, then the interval length is 80-70=10. Divide the amount of monitoring data in the first iteration fluctuation left-ward interval by the interval length of the first iteration fluctuation left-ward interval to obtain the concentration density of the first iteration fluctuation left-ward interval.

[0032] Based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients, identification task monitoring step configuration is performed to obtain multiple first monitoring step lengths and multiple second monitoring step lengths.

[0033] The monitoring step refers to the frequency or time interval of data collection during the execution of a task; the shorter the monitoring step, the more frequent the data collection and the higher the monitoring accuracy; the longer the step, the lower the monitoring frequency, but the consumption of computing resources can be reduced. The monitoring step is dynamically adjusted according to the mapping relationship between each task and patch and the corresponding data fluctuation. Specifically, the larger the fluctuation coefficient, the more dispersed the values ​​in the data set, and more frequent monitoring may be required to capture data changes; the smaller the fluctuation coefficient, the more concentrated the values ​​in the data set, and the monitoring frequency can be appropriately reduced. When configuring the monitoring step, the monitoring step can be divided into the first monitoring step and the second monitoring step. The first monitoring step may be used for tasks with large data fluctuations that require more frequent monitoring; while the second monitoring step may be used for tasks with small data fluctuations that can be monitored less frequently.

[0034] Further, based on the plurality of task-patch set mapping relationships and the K monitoring data fluctuation coefficients, the identification task monitoring step configuration is performed to obtain a plurality of first monitoring step lengths and a plurality of second monitoring step lengths, including:

[0035] Taking the multiple recognition tasks as indexes, the K monitoring data fluctuation coefficients are extracted in combination with the multiple task-patch set mapping relationships to obtain multiple monitoring data fluctuation coefficient sets; taking the K monitoring data fluctuation coefficients as indexes, the mapping quantity statistics are performed in combination with the multiple task-patch set mapping relationships to obtain K mapping quantities; taking the ratios of the K mapping quantities to the sum of the K mapping quantities as K mapping coefficients; performing weighted calculations based on the K mapping coefficients and the multiple monitoring data fluctuation coefficient sets to obtain multiple task monitoring coefficients corresponding to the multiple recognition tasks; multiplying the multiple task monitoring coefficients by a preset first monitoring step and a preset second monitoring step, respectively, to obtain the multiple first monitoring steps and the multiple second monitoring steps.

[0036] Specifically, taking multiple recognition tasks as indexes and combining the task-patch set mapping relationship, the fluctuation coefficient related to each task is extracted from the K monitoring data fluctuation coefficients to form multiple monitoring data fluctuation coefficient sets, each set contains the monitoring data fluctuation coefficients of all patches related to a specific task; taking the K monitoring data fluctuation coefficients as indexes and combining the task-patch set mapping relationship, the number of tasks or the number of patches corresponding to each fluctuation coefficient is counted to form K mapping quantities, and the mapping quantity reflects the importance or coverage of each fluctuation coefficient in the task set; the K mapping quantities are divided by the sum of the K mapping quantities to obtain K mapping coefficients, and the mapping coefficients represent the weight of each fluctuation coefficient in the whole; weighted calculation is performed based on the K mapping coefficients and multiple monitoring data fluctuation coefficient sets to obtain the task monitoring coefficient corresponding to each recognition task, and the task monitoring coefficient comprehensively considers the relationship between the task and the patch and the volatility of the data, and reflects the task's demand for monitoring frequency; the multiple task monitoring coefficients are multiplied by the preset first monitoring step and the preset second monitoring step to obtain multiple first monitoring steps and multiple second monitoring steps.

[0037] A plurality of task monitors are constructed based on the plurality of first monitoring steps and the plurality of second monitoring steps.

[0038] A plurality of task monitors are constructed according to a plurality of first monitoring steps and a plurality of second monitoring steps. The task monitors are used to continuously monitor each task, collect data at a certain frequency according to the allocated monitoring step, and perform data processing and analysis.

[0039] Furthermore, constructing a plurality of task monitors based on the plurality of first monitoring steps and the plurality of second monitoring steps includes:

[0040] Pre-construct an initial task monitor, wherein the initial task monitor includes a preset first monitoring branch, a preset second monitoring branch and a monitoring fusion fully connected layer, the preset first monitoring branch includes a preset first monitoring step, and the preset second monitoring branch includes a preset second monitoring step; respectively input the multiple first monitoring steps and the multiple second monitoring steps into the initial task monitor to obtain the multiple task monitors.

[0041] Specifically, the initial task monitor is the basic framework of the task monitor, which includes two monitoring branches (the first monitoring branch and the second monitoring branch), and a monitoring fusion fully connected layer for processing and integrating the results of these branches; the preset first monitoring branch and the preset second monitoring branch correspond to different monitoring frequencies and step sizes, respectively, and these branches perform data collection tasks of different frequencies through their respective monitoring step sizes; wherein, the preset first monitoring branch includes a preset first monitoring step size, and the preset second monitoring branch includes a preset second monitoring step size; the monitoring fusion fully connected layer is a layer for integrating the data results from the first monitoring branch and the second monitoring branch, which fuses the monitoring results under different step sizes to generate the final monitoring output; multiple first monitoring step sizes and second monitoring step sizes are input into the initial task monitor, and multiple task monitors are generated after being processed by the preset branches and the fusion layer.

[0042] Further, including:

[0043] Acquire multiple sample monitoring data sets, multiple sample first extraction data sets, multiple sample second extraction data sets, and multiple sample monitoring result sets as training data sets; perform supervised training using multiple sample monitoring data sets and multiple sample first extraction data sets in the training data set in combination with the preset first monitoring step length to obtain a preset first monitoring branch; perform supervised training using multiple sample monitoring data sets and multiple sample second extraction data sets in the training data set in combination with the preset second monitoring step length to obtain a preset second monitoring branch;

[0044] Perform supervised training using a plurality of sample first extraction data sets, a plurality of sample second extraction data sets, and a plurality of sample monitoring result sets to obtain a monitoring fusion fully connected layer;

[0045] The preset first monitoring branch and the preset second monitoring branch are connected in parallel, and the output layers of the preset first monitoring branch and the preset second monitoring branch are connected to the input layer of the monitoring fusion fully connected layer to obtain the initial task monitor.

[0046] Specifically, multiple sample monitoring data sets, multiple sample first extraction data sets, multiple sample second extraction data sets, and multiple sample monitoring result sets are collected, and these data sets will be used as training data sets to train various parts of the initial task monitor; supervised training is performed using multiple sample monitoring data sets and multiple sample first extraction data sets in the training data set, combined with a preset first monitoring step, so that the preset first monitoring branch can learn how to predict or monitor data changes based on the first extraction data set and the preset first monitoring step; similarly, supervised training is performed using multiple sample monitoring data sets and multiple sample second extraction data sets in the training data set, combined with a preset second monitoring step, Thereby, the preset second monitoring branch can learn how to predict or monitor data changes based on the second extracted data set and the preset second monitoring step; supervised training is performed using multiple sample first extracted data sets, multiple sample second extracted data sets and multiple sample monitoring result sets, so that the monitoring fusion fully connected layer can learn how to fuse the outputs of the two monitoring branches and generate the final monitoring results; the preset first monitoring branch and the preset second monitoring branch are connected in parallel, and their output layers are connected to the input layer of the monitoring fusion fully connected layer, thereby constructing a complete initial task monitor, which can generate monitoring results based on the input monitoring data set and the preset monitoring step.

[0047] By using multiple sample data for supervised training, the first and second monitoring branches are trained separately, and the outputs of the two branches are integrated through the monitoring fusion layer, and finally an initial task monitor that can flexibly adjust the monitoring frequency and monitoring tasks is constructed. The initial task monitor can select appropriate monitoring branches and output the best monitoring results according to different task requirements and data conditions, thereby improving the accuracy and efficiency of the monitoring task.

[0048] According to the multiple task monitors and the multiple task-patch set mapping relationships, long-term time series extraction and analysis are performed on the monitoring data of the K deployed patches to obtain multiple monitoring results.

[0049] According to the task-patch set mapping relationship, determine the task monitor corresponding to each deployed patch (K in total); for each deployed patch, extract the long time series data sequence from its monitoring data; input the extracted long time series data sequence into the corresponding task monitor, and each task monitor will process the input data according to its internal monitoring branch and monitoring step size, and generate corresponding monitoring results.

[0050] By utilizing multiple task monitors and task-patch set mapping relationships, long-term time series extraction and analysis are performed on the monitoring data of K deployed patches, which can capture the long-term trends and dynamic changes of the data and generate multiple monitoring results.

[0051] In summary, the embodiments of the present application have at least the following technical effects:

[0052] First, the patch layout array of the interactive target wearable device is used to obtain K monitoring data sets of K layout patches of the patch layout array within the preset monitoring window, wherein the K layout patches include K layout positions. Next, multiple recognition tasks of the target wearable device are obtained, and mapping associations are performed in combination with the K patch monitoring functions of the K layout patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to an recognition task. Then, the K monitoring data sets are screened for endpoints of the data center fluctuation interval to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients. Further, the recognition task monitoring step size is configured based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients to obtain multiple first monitoring step sizes and multiple second monitoring step sizes; and multiple task monitors are constructed based on the multiple first monitoring step sizes and multiple second monitoring step sizes. Finally, the monitoring data of the K layout patches are subjected to long-term extraction and analysis according to the multiple task monitors and the multiple task-patch set mapping relationships to obtain multiple monitoring results. The invention solves the technical problem of insufficient precision of multi-dimensional data monitoring of wearable devices in the prior art, and achieves the technical effect of improving the precision of data monitoring.

[0053] Embodiment 2, based on the same inventive concept as the multi-dimensional data monitoring method for wearable devices in the above embodiment, Figure 2 As shown, the present application provides a multi-dimensional data monitoring device for a wearable device, wherein the device comprises:

[0054] A data acquisition module 11, the data acquisition module 11 is used to interact with the patch layout array of the target wearable device, and obtain K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions; a mapping module 12, the mapping module 12 is used to obtain multiple recognition tasks of the target wearable device, and map and associate the K patch monitoring functions of the K layout patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to an recognition task; a screening module 13, the screening module 13 is used to perform data concentration fluctuation interval end on the K monitoring data sets Point screening, determining the median values ​​of K monitoring data sets and the K monitoring data fluctuation coefficients; a configuration module 14, the configuration module 14 is used to identify the task monitoring step configuration based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients, and obtain multiple first monitoring steps and multiple second monitoring steps; a construction module 15, the construction module 15 is used to construct multiple task monitors based on the multiple first monitoring steps and the multiple second monitoring steps; an analysis module 16, the analysis module 16 is used to perform long-term extraction and analysis on the monitoring data of the K deployed patches according to the multiple task monitors and the multiple task-patch set mapping relationships, and obtain multiple monitoring results.

[0055] Furthermore, the screening module 13 is used to perform the following method:

[0056] A first monitoring data set is randomly extracted from the K monitoring data sets, and the mean of the first monitoring data set is calculated to obtain a first monitoring comprehensive concentration value; the first monitoring comprehensive concentration value is used as the midpoint of the first fluctuation interval, and the left and right ends of the midpoint of the first fluctuation interval are respectively iterated according to the preset screening bandwidth to obtain a first fluctuation left-ward interval and a first fluctuation right-ward interval; the left endpoint of the first fluctuation left-ward interval is iterated to the left according to the preset screening bandwidth to obtain a first iterative fluctuation left-ward interval, and the right endpoint of the first fluctuation right-ward interval is iterated to the right according to the preset screening bandwidth to obtain a first iterative fluctuation right-ward interval; respectively determine whether the concentration density of the first iterative fluctuation left-ward interval is greater than or equal to the concentration density of the first fluctuation left-ward interval, and determine whether the concentration density of the first iterative fluctuation right-ward interval is greater than or equal to the concentration density of the first fluctuation left-ward interval. Whether the concentration density of the first iterative fluctuation rightward interval is greater than or equal to the concentration density of the first fluctuation rightward interval, if so, obtain the left-to-right iteration authentication pass result and the right-to-left iteration authentication pass result; based on the left iteration of the left endpoint of the first iterative fluctuation leftward interval and the right iteration of the left endpoint of the first iterative fluctuation rightward interval according to the preset screening bandwidth, until the preset iteration stop condition is met, obtain the first data center fluctuation interval; calculate the mean of the first data center fluctuation interval to obtain the first monitoring data median, and calculate the fluctuation variance of the first data center fluctuation interval to obtain the first monitoring data fluctuation coefficient; screen the data center fluctuation interval endpoints of the K monitoring data sets according to the preset screening bandwidth to obtain the K monitoring data medians and K monitoring data fluctuation coefficients.

[0057] Furthermore, the screening module 13 is used to perform the following method:

[0058] The preset iteration stopping condition is that the concentration density difference between two adjacent iterations is less than or equal to the preset concentration density difference.

[0059] Furthermore, the screening module 13 is used to perform the following method:

[0060] The amount of monitoring data in the first iterative fluctuation leftward interval is counted, and the calculated result is divided by the interval length of the first iterative fluctuation leftward interval to obtain the concentration density of the first iterative fluctuation leftward interval.

[0061] Furthermore, the configuration module 14 is used to execute the following method:

[0062] Taking the multiple recognition tasks as indexes, the K monitoring data fluctuation coefficients are extracted in combination with the multiple task-patch set mapping relationships to obtain multiple monitoring data fluctuation coefficient sets; taking the K monitoring data fluctuation coefficients as indexes, the mapping quantity statistics are performed in combination with the multiple task-patch set mapping relationships to obtain K mapping quantities; taking the ratios of the K mapping quantities to the sum of the K mapping quantities as K mapping coefficients; performing weighted calculations based on the K mapping coefficients and the multiple monitoring data fluctuation coefficient sets to obtain multiple task monitoring coefficients corresponding to the multiple recognition tasks; multiplying the multiple task monitoring coefficients by a preset first monitoring step and a preset second monitoring step, respectively, to obtain the multiple first monitoring steps and the multiple second monitoring steps.

[0063] Furthermore, the construction module 15 is used to execute the following method:

[0064] Pre-construct an initial task monitor, wherein the initial task monitor includes a preset first monitoring branch, a preset second monitoring branch and a monitoring fusion fully connected layer, the preset first monitoring branch includes a preset first monitoring step, and the preset second monitoring branch includes a preset second monitoring step; respectively input the multiple first monitoring steps and the multiple second monitoring steps into the initial task monitor to obtain the multiple task monitors.

[0065] Furthermore, the construction module 15 is used to execute the following method:

[0066] Acquire multiple sample monitoring data sets, multiple sample first extraction data sets, multiple sample second extraction data sets and multiple sample monitoring result sets as training data sets; use the multiple sample monitoring data sets and multiple sample first extraction data sets in the training data set in combination with the preset first monitoring step to perform supervised training to obtain a preset first monitoring branch; use the multiple sample monitoring data sets and multiple sample second extraction data sets in the training data set in combination with the preset second monitoring step to perform supervised training to obtain a preset second monitoring branch; use the multiple sample first extraction data sets, multiple sample second extraction data sets and multiple sample monitoring result sets to perform supervised training to obtain a monitoring fusion fully connected layer; connect the preset first monitoring branch and the preset second monitoring branch in parallel, and connect the output layers of the preset first monitoring branch and the preset second monitoring branch to the input layer of the monitoring fusion fully connected layer to obtain the initial task monitor.

[0067] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0069] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A multidimensional data monitoring method for wearable devices, characterized in that: The method comprises: Interact with a patch layout array of a target wearable device, and obtain K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions; Acquire multiple recognition tasks of the target wearable device, and map and associate them with the K patch monitoring functions of the K deployed patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to one recognition task; Performing data center fluctuation interval endpoint screening on the K monitoring data sets to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients; Based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients, identification task monitoring step configuration is performed to obtain multiple first monitoring step lengths and multiple second monitoring step lengths; constructing a plurality of task monitors based on the plurality of first monitoring steps and the plurality of second monitoring steps; Performing long-term extraction and analysis on the monitoring data of the K deployed patches according to the multiple task monitors and the multiple task-patch set mapping relationships to obtain multiple monitoring results; Among them, the identification task monitoring step configuration is performed based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients to obtain multiple first monitoring step sizes and multiple second monitoring step sizes, including: Taking the multiple recognition tasks as indexes and combining the multiple task-patch set mapping relationships, the K monitoring data fluctuation coefficients are extracted to obtain multiple monitoring data fluctuation coefficient sets; Taking the K monitoring data fluctuation coefficients as indexes, and combining the plurality of task-patch set mapping relationships, mapping amount statistics are performed to obtain K mapping amounts; respectively taking the ratios of the K mapping amounts to the sum of the K mapping amounts as K mapping coefficients; Performing weighted calculation according to the K mapping coefficients and the plurality of monitoring data fluctuation coefficient sets to obtain a plurality of task monitoring coefficients corresponding to the plurality of identification tasks; The plurality of task monitoring coefficients are respectively multiplied by a preset first monitoring step length and a preset second monitoring step length to obtain the plurality of first monitoring step lengths and the plurality of second monitoring step lengths.

2. The multi-dimensional data monitoring method for wearable devices according to claim 1, characterized in that: The K monitoring data sets are subjected to data center fluctuation interval endpoint screening to determine the medians of the K monitoring data sets and the K monitoring data fluctuation coefficients, including: Randomly extracting a first monitoring data set from the K monitoring data sets, and performing mean calculation on the first monitoring data set to obtain a first monitoring comprehensive concentration value; Taking the first monitoring comprehensive concentration value as the midpoint of the first fluctuation interval, iterating respectively to the left and right ends of the midpoint of the first fluctuation interval according to the preset screening bandwidth to obtain the first fluctuation leftward interval and the first fluctuation rightward interval; Iterating the left end point of the first fluctuation leftward interval to the left according to the preset screening bandwidth to obtain a first iterative fluctuation leftward interval, and iterating the right end point of the first fluctuation rightward interval to the right according to the preset screening bandwidth to obtain a first iterative fluctuation rightward interval; Respectively determine whether the concentration density of the first iterative fluctuation leftward interval is greater than or equal to the concentration density of the first fluctuation leftward interval, and determine whether the concentration density of the first iterative fluctuation rightward interval is greater than or equal to the concentration density of the first fluctuation rightward interval, and if so, obtain the left-direction iteration authentication pass result and the right-direction iteration authentication pass result; Iterating leftwardly the left endpoint of the first iterative fluctuation leftward interval and iterating rightwardly the left endpoint of the first iterative fluctuation rightward interval according to the preset screening bandwidth until a preset iteration stop condition is met, thereby obtaining a fluctuation interval in the first data set; Performing mean calculation on the fluctuation interval in the first data set to obtain a median value of the first monitoring data set, and calculating the fluctuation variance of the fluctuation interval in the first data set to obtain a fluctuation coefficient of the first monitoring data; The K monitoring data sets are screened for endpoints of fluctuation intervals in the data set according to a preset screening bandwidth to obtain medians of the K monitoring data sets and fluctuation coefficients of the K monitoring data.

3. The multi-dimensional data monitoring method for wearable devices according to claim 2, characterized in that: The preset iteration stopping condition is that the concentration density difference between two adjacent iterations is less than or equal to the preset concentration density difference.

4. The multi-dimensional data monitoring method for wearable devices according to claim 2, characterized in that: The amount of monitoring data in the first iterative fluctuation leftward interval is counted, and the calculated result is divided by the interval length of the first iterative fluctuation leftward interval to obtain the concentration density of the first iterative fluctuation leftward interval.

5. The multi-dimensional data monitoring method for wearable devices according to claim 1, characterized in that: Building a plurality of task monitors based on the plurality of first monitoring steps and the plurality of second monitoring steps includes: Pre-constructing an initial task monitor, wherein the initial task monitor includes a preset first monitoring branch, a preset second monitoring branch and a monitoring fusion fully connected layer, the preset first monitoring branch includes a preset first monitoring step length, and the preset second monitoring branch includes a preset second monitoring step length; The multiple first monitoring step lengths and the multiple second monitoring step lengths are respectively input into the initial task monitor to obtain the multiple task monitors.

6. The multi-dimensional data monitoring method for wearable devices according to claim 5, characterized in that: include: Acquire multiple sample monitoring data sets, multiple sample first extraction data sets, multiple sample second extraction data sets, and multiple sample monitoring result sets as training data sets; Perform supervised training using multiple sample monitoring data sets and multiple sample first extraction data sets in the training data set in combination with the preset first monitoring step length to obtain a preset first monitoring branch; Perform supervised training using multiple sample monitoring data sets and multiple sample second extracted data sets in the training data set in combination with the preset second monitoring step length to obtain a preset second monitoring branch; Perform supervised training using a plurality of sample first extraction data sets, a plurality of sample second extraction data sets, and a plurality of sample monitoring result sets to obtain a monitoring fusion fully connected layer; The preset first monitoring branch and the preset second monitoring branch are connected in parallel, and the output layers of the preset first monitoring branch and the preset second monitoring branch are connected to the input layer of the monitoring fusion fully connected layer to obtain the initial task monitor.

7. A multidimensional data monitoring device for wearable devices, characterized in that: The device is used to implement the multi-dimensional data monitoring method for a wearable device according to any one of claims 1 to 6, comprising: A data acquisition module, the data acquisition module is used to interact with the patch layout array of the target wearable device, and acquire K monitoring data sets of K layout patches of the patch layout array within a preset monitoring window, wherein the K layout patches include K layout positions; A mapping module, the mapping module is used to obtain multiple recognition tasks of the target wearable device, and map and associate the K patch monitoring functions of the K deployed patches to obtain multiple task-patch set mapping relationships, wherein each task-patch set mapping relationship corresponds to one recognition task; A screening module, the screening module is used to screen the endpoints of the fluctuation interval of the K monitoring data sets, and determine the medians of the K monitoring data sets and the fluctuation coefficients of the K monitoring data; A configuration module, the configuration module is used to perform identification task monitoring step configuration based on the multiple task-patch set mapping relationships and the K monitoring data fluctuation coefficients to obtain multiple first monitoring step sizes and multiple second monitoring step sizes; A construction module, the construction module is used to construct a plurality of task monitors based on the plurality of first monitoring steps and the plurality of second monitoring steps; An analysis module is used to perform long-term extraction analysis on the monitoring data of the K deployed patches according to the mapping relationships between the multiple task monitors and the multiple task-patch sets to obtain multiple monitoring results.

Images