A method and platform for extracting and analyzing time series data features

Abstract

The present invention relates to the field of Internet of Things technology, and specifically discloses a method and platform for extracting and analyzing time series data features, including: based on all class data feature parameters of time series data and all class device parameters of all Internet of Things devices, all assignable devices are obtained; based on all class data feature parameters of time series data and all class device parameters of all assignable devices, task requirement values ​​of time series data and device performance values ​​of all assignable devices are obtained; based on device performance values ​​and transmission distance values ​​of all assignable devices, edge allocation priority values ​​of all computing devices are obtained; based on edge allocation priority values ​​of all computing devices, time series data feature extraction and analysis results are obtained. The present invention dynamically adjusts the computing allocation of Internet of Things edge computing according to the characteristics of time series data and the performance of Internet of Things devices, and realizes the optimal configuration of resources in different scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a method and platform for extracting and analyzing time series data features. Background Art

[0002] At present, in IoT applications, sensors and smart devices continuously generate a large amount of time series data, which needs to be processed and analyzed in real time to provide valuable information. The feature extraction and analysis of these time series data often involve complex computing tasks and require efficient processing capabilities. However, the traditional processing method is to transmit all data to a remote data center or cloud for processing, which has lags and high power consumption. IoT edge computing allows data analysis and processing at the edge of IoT devices or networks. This technology distributes computing and data processing to other devices within the IoT, allowing devices to perform real-time data analysis and processing locally. Therefore, how to adjust the computing allocation of IoT edge computing to ensure the optimal configuration of resources in different scenarios has become the most critical issue at present.

[0003] However, the existing methods and platforms for extracting and analyzing time series data features only use some data characteristics carried by the time series data itself to accurately extract the information points given by the data itself and capture more dimensional data information, making the feature data smoother overall, which helps to improve the accuracy of subsequent modeling and prediction. However, they do not consider quantifying the priority of each computing device in the process of extracting and analyzing time series data features, nor do they consider how to dynamically adjust the computing allocation of IoT edge computing based on the characteristics of time series data and the performance of IoT devices to ensure the optimal configuration of resources in different scenarios. For example, the patent publication number is "CN110134839B" and the name is "Time Series Data Feature Processing Method, Device and Computer Readable Storage Medium". The method includes the following steps: clustering the time series data in combination with the K-MEANS clustering algorithm to form multiple cluster classes; using the time series feature words pre-set in the dictionary to search and compare in each cluster class, if the data consistent with the time series feature words in the dictionary is searched, the data in the corresponding cluster class is counted according to the time series feature words, and the data percentage corresponding to each time series feature word is counted; the time series data in each cluster class is clustered by mean shift, and the data in each cluster class is further clustered into cluster subclasses according to the data percentage. The above patent can accurately extract the information points assigned by the data itself based on some data characteristics carried by the time series data itself, and can also capture more dimensional data information, and make the feature data smoother as a whole, which is helpful to improve the accuracy of subsequent modeling prediction. However, this patent only uses some data characteristics carried by the time series data itself to accurately extract the information points given by the data itself and capture more dimensional data information, making the feature data smoother overall, which helps to improve the accuracy of subsequent modeling predictions. However, it does not consider quantifying the priority of each computing device in the process of time series data feature extraction and analysis, nor does it consider how to dynamically adjust the computing allocation of IoT edge computing based on the characteristics of time series data and the performance of IoT devices to ensure the optimal configuration of resources in different scenarios.

[0004] Therefore, the present invention proposes a method and platform for extracting and analyzing time series data features. Summary of the invention

[0005] The present invention provides a method and platform for extracting and analyzing time series data features, which are used to obtain all assignable devices according to all class data feature parameters of time series data and all class device parameters of all Internet of Things devices, and realize the screening of some Internet of Things devices that can perform computing task assignments in the process of time series data feature extraction and analysis from all Internet of Things devices, and then accurately obtain the task requirement value of the time series data according to all class data feature parameters of the time series data, realize the quantification of the difficulty of processing data when extracting and analyzing the features of the time series data, accurately obtain the device performance values ​​of all assignable devices according to all class device parameters of all assignable devices, and realize the ability level of each assignable device when extracting and analyzing the features of the time series data. Quantification is performed, and then the allocable device node graph is obtained according to the device performance values ​​and transmission distance values ​​of all allocable devices, which is convenient for the subsequent calculation of edge allocation priority values. According to the task requirement value of the time series data and the allocable device node graph, the edge allocation priority values ​​of all computing devices are accurately obtained, and the priority of each computing device in the time series data feature extraction and analysis process is quantified. Finally, the time series data feature extraction and analysis results are obtained according to the edge allocation priority values ​​of all computing devices. The computing allocation of the IoT edge computing is dynamically adjusted according to the characteristics of the time series data and the performance of the IoT devices, ensuring the optimal configuration of resources in different scenarios and improving the efficiency of time series data feature extraction and analysis.

[0006] The present invention provides a method for extracting and analyzing time series data features, comprising:

[0007] S1: Based on all class data feature parameters of time series data and all class device parameters of all IoT devices, all assignable devices are obtained;

[0008] S2: based on all class data characteristic parameters of the time series data and all class device parameters of all assignable devices, obtain the task requirement value of the time series data and the device performance value of all assignable devices;

[0009] S3: Based on the device performance values ​​and transmission distance values ​​of all the allocatable devices, an allocatable device node graph is obtained, and based on the task requirement value of the time series data and the allocatable device node graph, an edge allocation priority value of all computing devices is obtained;

[0010] S4: Based on the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained.

[0011] Preferably, the method for extracting and analyzing time series data features, S1: based on all class data feature parameters of the time series data and all class device parameters of all IoT devices, obtaining all assignable devices, including:

[0012] Obtain all class data feature parameters of time series data, where all class data feature parameters include data volume, data collection frequency, and data dimension;

[0013] Obtain all device parameters of all IoT devices, including device processor main frequency, device processor core number, device memory amount, and device memory speed;

[0014] Based on the device memory amount and all class data characteristic parameters of the time series data of all IoT devices, all assignable devices are obtained.

[0015] Preferably, the method for extracting and analyzing time series data features obtains all allocatable devices based on the device memory of all IoT devices and all class data feature parameters of the time series data, including:

[0016] The quotient of the value of the data volume of the time series data and the value of the data collection frequency of the time series data, and the product of the value of the data dimension of the time series data, are used as the screening data volume value;

[0017] If the value of the device memory amount of each IoT device is greater than the value of the filtered data amount, the corresponding IoT device is determined to be an allocatable device, and all the allocatable devices are obtained.

[0018] Preferably, the method for extracting and analyzing time series data features, S2: based on all class data feature parameters of the time series data and all class device parameters of all assignable devices, obtaining the task requirement value of the time series data and the device performance value of all assignable devices, comprises:

[0019] Obtaining a time series data feature extraction method, and obtaining a time series data feature extraction scoring value based on the time series data feature extraction method and a preset feature extraction method scoring table;

[0020] Obtaining a time series data feature analysis method, and obtaining a time series data feature analysis scoring value based on the time series data feature analysis method and a preset feature analysis method scoring table;

[0021] Based on the time series data feature extraction and scoring, time series data feature analysis and scoring, and all class data feature parameters of the time series data, the task requirement value of the time series data is obtained, which is:

[0022]

[0023] Among them, β is the task requirement value of time series data, α1 is the score value of time series data feature extraction, α2 is the score value of time series data feature analysis, e is a natural constant, A is the value of the data volume of time series data, B is the value of the data collection frequency of time series data, C is the data dimension of time series data, (α1, α2) maxis the maximum value of α1 and α2, ln is the natural logarithm, and the value of the natural constant e is 2.718;

[0024] Based on all class device parameters of all allocatable devices, a device performance value of each allocatable device is obtained.

[0025] Preferably, the method for extracting and analyzing time series data features, based on all class device parameters of all assignable devices, obtains the device performance value of each assignable device, including:

[0026]

[0027] Among them, γ is the device performance value of the currently calculated allocatable device, W is the device processor main frequency of the currently calculated allocatable device, W0 is the average of the device processor main frequencies of allocatable devices, E is the number of device processor cores of the currently calculated allocatable device, E0 is the average of the number of device processor cores of allocatable devices, R is the device memory amount of the currently calculated allocatable device, R0 is the average of the device memory amount of allocatable devices, H is the device memory speed of the currently calculated allocatable device, and H0 is the average of the device memory speed of allocatable devices.

[0028] Preferably, the method for extracting and analyzing time series data features, S3: based on the device performance values ​​and transmission distance values ​​of all assignable devices, obtaining an assignable device node graph, and based on the task requirement values ​​of the time series data and the assignable device node graph, obtaining edge assignment priority values ​​of all computing devices, including:

[0029] Acquire the transmission distance values ​​of all allocatable devices, and use the transmission distance values ​​of all the allocatable devices as the abscissa values, and the device performance values ​​of the corresponding allocatable devices as the ordinate values, obtain all the allocatable device nodes, and use the rectangular coordinate system containing all the allocatable device nodes as the allocatable device node graph;

[0030] Based on the task demand value of the time series data and the allocatable device node graph, the edge allocation priority values ​​of all computing devices are obtained.

[0031] Preferably, the method for extracting and analyzing time series data features obtains edge allocation priority values ​​of all computing devices based on the task requirement value of the time series data and the allocatable device node graph, including:

[0032] The analysis circle is obtained by taking the origin of the rectangular coordinate system of the allocable device node graph as the center and the task demand value of the time series data as the radius;

[0033] Determine whether the number of allocatable device nodes in the analysis circle is 0. If so, take the sum of the radius of the analysis circle and 1 as the radius of the new analysis circle, take the origin of the rectangular coordinate system of the allocatable device node graph as the center and the radius of the new analysis circle as the radius, obtain a new analysis circle, and continue to determine whether the number of allocatable device nodes in the new analysis circle is 0. If so, continue to increase the radius of the new analysis circle by a unit value until the number of allocatable device nodes in the latest analysis circle is not 0, then take all the allocatable device nodes in the latest analysis circle as computing device nodes, and take the radius of the latest analysis circle as the computing radius;

[0034] Otherwise, all allocatable device nodes within the analysis circle are regarded as computing device nodes, and the radius of the analysis circle is regarded as the computing radius;

[0035] Based on all computing device nodes, computing radius and task requirement values ​​of time series data, an edge allocation priority value of each computing device node is obtained, and the edge allocation priority value of each computing device node is used as the edge allocation priority value of the corresponding computing device.

[0036] Preferably, the method for extracting and analyzing time series data features obtains the edge allocation priority value of each computing device node based on all computing device nodes, computing radius and task requirement value of time series data, including:

[0037]

[0038] Among them, δ is the edge allocation priority value of the computing device node currently being calculated, β is the task requirement value of the time series data, x is the horizontal coordinate value of the computing device node currently being calculated, y is the vertical coordinate value of the computing device node currently being calculated, r is the distance between the computing device node currently being calculated and the origin of the rectangular coordinate system, r0 is the calculation radius, ln is the natural logarithm, and the value of the natural constant e is 2.718.

[0039] Preferably, the method for extracting and analyzing time series data features, S4: obtaining time series data feature extraction and analysis results based on edge allocation priority values ​​of all computing devices, including:

[0040] Sorting all computing devices in descending order of their edge allocation priority values ​​to obtain a computing device sorting result;

[0041] The computing tasks of the time series data feature extraction and analysis process are assigned to all computing devices, and the assignment order is consistent with the computing device sorting results.

[0042] The present invention provides a platform for extracting and analyzing time series data features, which is used to execute any one of the methods for extracting and analyzing time series data features in embodiments 1 to 9, including:

[0043] A screening module, used to obtain all assignable devices based on all class data feature parameters of the time series data and all class device parameters of all IoT devices;

[0044] A first calculation module is used to obtain the task requirement value of the time series data and the device performance value of all the assignable devices based on all the class data characteristic parameters of the time series data and all the class device parameters of all the assignable devices;

[0045] A second computing module is used to obtain an allocatable device node graph based on device performance values ​​and transmission distance values ​​of all the allocatable devices, and to obtain edge allocation priority values ​​of all computing devices based on the task requirement value of the time series data and the allocatable device node graph;

[0046] The allocation module is used to obtain time series data feature extraction and analysis results based on edge allocation priority values ​​of all computing devices.

[0047] The beneficial effects of the present invention compared with the prior art are as follows: according to all class data characteristic parameters of time series data and all class device parameters of all Internet of Things devices, all assignable devices are obtained, and some Internet of Things devices that can perform computing task assignments for time series data feature extraction and analysis processes are screened out from all Internet of Things devices, and then according to all class data characteristic parameters of time series data, the task requirement value of time series data is accurately obtained, and the difficulty of processing data when extracting and analyzing time series data is quantified; according to all class device parameters of all assignable devices, the device performance values ​​of all assignable devices are accurately obtained, and the ability of each assignable device when extracting and analyzing time series data is quantified. , and then according to the device performance values ​​and transmission distance values ​​of all assignable devices, the assignable device node graph is obtained, which is convenient for the subsequent calculation of edge allocation priority values, and according to the task requirement value of the time series data and the assignable device node graph, the edge allocation priority values ​​of all computing devices are accurately obtained, and the priority of each computing device in the time series data feature extraction and analysis process is quantified. Finally, according to the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained, and the computing allocation of the IoT edge computing is dynamically adjusted according to the characteristics of the time series data and the performance of the IoT devices, ensuring the optimal configuration of resources in different scenarios and improving the efficiency of time series data feature extraction and analysis.

[0048] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structure specifically pointed out in the written application document.

[0049] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0051] Figure 1 A flow chart of a method for extracting and analyzing time series data features in an embodiment of the present invention;

[0052] Figure 2 Schematic diagram of a platform for extracting and analyzing time series data features in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0054] Embodiment 1:

[0055] The present invention provides a method for extracting and analyzing time series data features. Figure 1 ,include:

[0056] S1: Based on all class data feature parameters of time series data and all class device parameters of all IoT devices, all assignable devices are obtained;

[0057] S2: based on all class data characteristic parameters of the time series data and all class device parameters of all assignable devices, obtain the task requirement value of the time series data and the device performance value of all assignable devices;

[0058] S3: Based on the device performance values ​​and transmission distance values ​​of all the allocatable devices, an allocatable device node graph is obtained, and based on the task requirement value of the time series data and the allocatable device node graph, an edge allocation priority value of all computing devices is obtained;

[0059] S4: Based on the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained.

[0060] In this embodiment, the time series data is time series data that undergoes feature extraction and analysis through a time series data feature extraction and analysis platform (data arranged in chronological order, and the time series data records the measurement values ​​of a variable or a group of variables at different time points, and can reflect how a phenomenon or indicator changes over time).

[0061] In this embodiment, the assignable devices are some IoT devices that are selected from all IoT devices based on all class data feature parameters of time series data and all class device parameters of all IoT devices and can be assigned computing tasks for time series data feature extraction and analysis processes.

[0062] In this embodiment, the task requirement value of the time series data is a numerical value obtained based on the characteristic parameters of all types of data of the time series data, which can characterize the difficulty of processing data when extracting and analyzing features of the time series data.

[0063] In this embodiment, the device performance value of the allocable device is a value obtained based on all types of device parameters of all allocable devices, and can characterize the capability of each allocable device when performing time series data feature extraction and analysis.

[0064] In this embodiment, the transmission distance value of the allocatable device is the actual distance between the device that collects the time series data in the Internet of Things and each allocatable device.

[0065] In this embodiment, the allocatable device node graph is a node graph obtained based on the device performance values ​​and transmission distance values ​​of all the allocatable devices and used to analyze and calculate the edge allocation priority values ​​of all computing devices.

[0066] In this embodiment, the edge allocation priority value of the computing device is a numerical value obtained based on the task requirement value of the time series data and the allocatable device node graph, which can characterize the priority of each computing device in assuming computing task allocation during the process of time series data feature extraction and analysis.

[0067] In this embodiment, the time series data feature extraction and analysis results are obtained after the computing tasks of the time series data feature extraction and analysis process are allocated to all computing devices based on the edge allocation priority values ​​of all computing devices.

[0068] The beneficial effects of the above technology are as follows: according to all the class data characteristic parameters of the time series data and all the class device parameters of all the IoT devices, all the assignable devices are obtained, and it is realized that some IoT devices that can be used for the computing task allocation of the time series data feature extraction and analysis process are screened out from all the IoT devices, and then according to all the class data characteristic parameters of the time series data, the task requirement value of the time series data is accurately obtained, and the difficulty of processing data when extracting and analyzing the time series data is quantified; according to all the class device parameters of all the assignable devices, the device performance values ​​of all the assignable devices are accurately obtained, and the ability of each assignable device when extracting and analyzing the time series data is quantified, and then based on According to the device performance values ​​and transmission distance values ​​of all assignable devices, the assignable device node graph is obtained to facilitate the subsequent calculation of edge allocation priority values. According to the task requirement value of the time series data and the assignable device node graph, the edge allocation priority values ​​of all computing devices are accurately obtained, and the priority of each computing device in the time series data feature extraction and analysis process is quantified. Finally, according to the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained, and the computing allocation of the IoT edge computing is dynamically adjusted according to the characteristics of the time series data and the performance of the IoT devices, ensuring the optimal configuration of resources in different scenarios and improving the efficiency of time series data feature extraction and analysis.

[0069] Embodiment 2:

[0070] Based on Example 1, the method for extracting and analyzing time series data features, S1: based on all class data feature parameters of time series data and all class device parameters of all IoT devices, all assignable devices are obtained, including:

[0071] Obtain all class data feature parameters of time series data, where all class data feature parameters include data volume, data collection frequency, and data dimension;

[0072] Obtain all device parameters of all IoT devices, including device processor main frequency, device processor core number, device memory amount, and device memory speed;

[0073] Based on the device memory amount and all class data characteristic parameters of the time series data of all IoT devices, all assignable devices are obtained.

[0074] In this embodiment, the data volume is the data volume of time series data, and the unit of the data volume is MB.

[0075] In this embodiment, the data collection frequency is the collection frequency of the time series data (the number of collections per unit time), and the unit of the collection frequency is Hertz (Hz).

[0076] In this embodiment, the data dimension quantity is the number of variable types (such as time variables, temperature variables, pressure variables, etc.) contained in the time series data.

[0077] In this embodiment, the device processor main frequency is the operating frequency of the processor of the Internet of Things device, and the unit of the device processor main frequency is Hertz (Hz).

[0078] In this embodiment, the number of device processor cores is the number of cores of the processor of the Internet of Things device.

[0079] In this embodiment, the device memory capacity is the memory capacity of the IoT device, and the unit of the device memory capacity is MB.

[0080] In this embodiment, the device memory speed is the reading and writing speed of data in the memory of the IoT device, and the unit of the device memory speed is MB / s.

[0081] The beneficial effects of the above technology are: clarifying the specific parameter items of all class data characteristic parameters of time series data and all class device parameters of IoT devices, and obtaining all assignable devices based on the device memory capacity of all IoT devices and all class data characteristic parameters of time series data, thereby realizing the screening of some IoT devices from all IoT devices that can perform computing task allocation for time series data feature extraction and analysis processes.

[0082] Embodiment 3:

[0083] On the basis of Example 2, the method for extracting and analyzing time series data features obtains all allocatable devices based on the device memory of all IoT devices and all class data feature parameters of time series data, including:

[0084] The quotient of the value of the data volume of the time series data and the value of the data collection frequency of the time series data, and the product of the value of the data dimension of the time series data, are used as the screening data volume value;

[0085] If the value of the device memory amount of each IoT device is greater than the value of the filtered data amount, the corresponding IoT device is determined to be an allocatable device, and all the allocatable devices are obtained.

[0086] The beneficial effects of the above technology are: according to all class data feature parameters of time series data, the filter data value is obtained, and then according to the filter data value and the device memory of the IoT device, all assignable devices are obtained, thereby realizing the screening of some IoT devices from all IoT devices that can perform computing task allocation for time series data feature extraction and analysis processes.

[0087] Embodiment 4:

[0088] Based on Example 1, the method for extracting and analyzing time series data features, S2: based on all class data feature parameters of the time series data and all class device parameters of all assignable devices, obtaining the task requirement value of the time series data and the device performance value of all assignable devices, including:

[0089] Obtaining a time series data feature extraction method, and obtaining a time series data feature extraction scoring value based on the time series data feature extraction method and a preset feature extraction method scoring table;

[0090] Obtaining a time series data feature analysis method, and obtaining a time series data feature analysis scoring value based on the time series data feature analysis method and a preset feature analysis method scoring table;

[0091] Based on the time series data feature extraction and scoring, time series data feature analysis and scoring, and all class data feature parameters of the time series data, the task requirement value of the time series data is obtained, which is:

[0092]

[0093] Among them, β is the task requirement value of time series data, α1 is the score value of time series data feature extraction, α2 is the score value of time series data feature analysis, e is a natural constant, A is the value of the data volume of time series data, B is the value of the data collection frequency of time series data, C is the data dimension of time series data, (α1, α2) max is the maximum value of α1 and α2, ln is the natural logarithm, and the value of the natural constant e is 2.718;

[0094] Based on all class device parameters of all allocatable devices, a device performance value of each allocatable device is obtained.

[0095] In this embodiment, the time series data feature extraction method is a method of extracting features from time series data, including feature extraction based on statistics, feature extraction based on frequency domain analysis, feature extraction based on wavelet transform, feature extraction based on machine learning, etc.

[0096] In this embodiment, the preset feature extraction method scoring table is a pre-set scoring table for obtaining time series data feature extraction scoring values, wherein the scoring table contains multiple time series data feature extraction methods, and each time series data feature extraction method corresponds to a time series data feature extraction scoring value separately.

[0097] In this embodiment, the scoring value for time series data feature extraction is a numerical value for scoring the complexity of the time series data feature extraction method.

[0098] In this embodiment, the time series data feature analysis method is a method of performing feature analysis on time series data, including time series modeling, similarity measurement, classification and cluster analysis, etc.

[0099] In this embodiment, the preset feature analysis method scoring table is a pre-set scoring table for obtaining time series data feature analysis scoring values, wherein the scoring table contains multiple time series data feature analysis methods, and each time series data feature analysis method corresponds to a time series data feature analysis scoring value.

[0100] In this embodiment, the time series data feature analysis scoring value is a numerical value that scores the complexity of the time series data feature analysis method.

[0101] The beneficial effects of the above technology are: based on all class data characteristic parameters of the time series data, the task requirement value of the time series data is accurately obtained, and the difficulty of processing data when extracting and analyzing features of the time series data is quantified; based on all class device parameters of all assignable devices, the device performance values ​​of all assignable devices are accurately obtained. This embodiment gives in detail a specific method for accurately obtaining the task requirement value of the time series data based on all class data characteristic parameters of the time series data.

[0102] Embodiment 5:

[0103] On the basis of Example 4, the method for extracting and analyzing time series data features obtains the device performance value of each allocatable device based on all device parameters of all allocatable devices, including:

[0104]

[0105] Among them, γ is the device performance value of the currently calculated allocatable device, W is the device processor main frequency of the currently calculated allocatable device, W0 is the average of the device processor main frequencies of allocatable devices, E is the number of device processor cores of the currently calculated allocatable device, E0 is the average of the number of device processor cores of allocatable devices, R is the device memory amount of the currently calculated allocatable device, R0 is the average of the device memory amount of allocatable devices, H is the device memory speed of the currently calculated allocatable device, and H0 is the average of the device memory speed of allocatable devices.

[0106] The beneficial effects of the above technology are: based on all types of device parameters of all allocatable devices, the device performance values ​​of all allocatable devices are accurately obtained, and the capabilities of each allocatable device during time series data feature extraction and analysis are quantified, which facilitates the subsequent determination of the allocatable device node graph.

[0107] Embodiment 6:

[0108] Based on Example 1, the method for extracting and analyzing time series data features, S3: based on the device performance values ​​and transmission distance values ​​of all assignable devices, obtaining an assignable device node graph, and based on the task requirement value of the time series data and the assignable device node graph, obtaining the edge assignment priority value of all computing devices, including:

[0109] Acquire the transmission distance values ​​of all allocatable devices, and use the transmission distance values ​​of all the allocatable devices as the abscissa values, and the device performance values ​​of the corresponding allocatable devices as the ordinate values, obtain all the allocatable device nodes, and use the rectangular coordinate system containing all the allocatable device nodes as the allocatable device node graph;

[0110] Based on the task demand value of the time series data and the allocatable device node graph, the edge allocation priority values ​​of all computing devices are obtained.

[0111] The beneficial effects of the above technology are: based on the device performance values ​​and transmission distance values ​​of all allocatable devices, all allocatable device nodes are obtained, and then based on allocatable device nodes, an allocatable device node graph is obtained, which is convenient for the subsequent calculation of edge allocation priority values, and the edge allocation priority values ​​of all computing devices are obtained based on the task requirement values ​​of the timing data and the allocatable device node graph.

[0112] Embodiment 7:

[0113] On the basis of Example 6, the method for extracting and analyzing time series data features obtains edge allocation priority values ​​of all computing devices based on the task requirement value of the time series data and the allocatable device node graph, including:

[0114] The analysis circle is obtained by taking the origin of the rectangular coordinate system of the allocable device node graph as the center and the task demand value of the time series data as the radius;

[0115] Determine whether the number of allocatable device nodes in the analysis circle is 0. If so, take the sum of the radius of the analysis circle and 1 as the radius of the new analysis circle, take the origin of the rectangular coordinate system of the allocatable device node graph as the center and the radius of the new analysis circle as the radius, obtain a new analysis circle, and continue to determine whether the number of allocatable device nodes in the new analysis circle is 0. If so, continue to increase the radius of the new analysis circle by the unit value (1) until the number of allocatable device nodes in the latest analysis circle is not 0, then take all the allocatable device nodes in the latest analysis circle as the computing device nodes, and take the radius of the latest analysis circle as the computing radius;

[0116] Otherwise, all allocatable device nodes within the analysis circle are regarded as computing device nodes, and the radius of the analysis circle is regarded as the computing radius;

[0117] Based on all computing device nodes, computing radius and task requirement values ​​of time series data, an edge allocation priority value of each computing device node is obtained, and the edge allocation priority value of each computing device node is used as the edge allocation priority value of the corresponding computing device.

[0118] The beneficial effects of the above technology are: according to the task requirement value of the allocatable device node graph and timing data, all computing device nodes and computing radius are obtained, and according to the task requirement value of all computing device nodes, computing radius and timing data, the edge allocation priority value of each computing device is obtained. This embodiment gives in detail a specific method for obtaining all computing device nodes and computing radius according to the task requirement value of the allocatable device node graph and timing data.

[0119] Embodiment 8:

[0120] On the basis of Example 7, the method for extracting and analyzing time series data features obtains the edge allocation priority value of each computing device node based on all computing device nodes, computing radius and task requirement value of time series data, including:

[0121]

[0122] Among them, δ is the edge allocation priority value of the computing device node currently being calculated, β is the task requirement value of the time series data, x is the horizontal coordinate value of the computing device node currently being calculated, y is the vertical coordinate value of the computing device node currently being calculated, r is the distance between the computing device node currently being calculated and the origin of the rectangular coordinate system, r0 is the calculation radius, ln is the natural logarithm, and the value of the natural constant e is 2.718.

[0123] The beneficial effects of the above technology are: based on the task requirement values ​​of all computing device nodes, computing radius and time series data, the edge allocation priority value of each computing device node is obtained, and the priority of each computing device in the allocation of computing tasks during the time series data feature extraction and analysis process is quantified. This embodiment gives in detail a specific method for obtaining the edge allocation priority value of each computing device node based on the task requirement values ​​of all computing device nodes, computing radius and time series data.

[0124] Embodiment 9:

[0125] Based on Example 1, the method for extracting and analyzing time series data features, S4: obtaining time series data feature extraction and analysis results based on edge allocation priority values ​​of all computing devices, including:

[0126] Sorting all computing devices in descending order of their edge allocation priority values ​​to obtain a computing device sorting result;

[0127] The computing tasks of the time series data feature extraction and analysis process are assigned to all computing devices, and the assignment order is consistent with the computing device sorting results.

[0128] The beneficial effects of the above technology are: according to the edge allocation priority value of all computing devices, the time series data feature extraction and analysis results are obtained, and the computing allocation of IoT edge computing is dynamically adjusted according to the characteristics of the time series data and the performance of the IoT devices, ensuring the optimal configuration of resources in different scenarios and improving the efficiency of time series data feature extraction and analysis.

[0129] Embodiment 10:

[0130] The present invention provides a platform for extracting and analyzing time series data features, which is used to perform any one of the time series data feature extraction and analysis methods in embodiments 1 to 9, referring to Figure 2 ,include:

[0131] A screening module, used to obtain all assignable devices based on all class data feature parameters of the time series data and all class device parameters of all IoT devices;

[0132] A first calculation module is used to obtain the task requirement value of the time series data and the device performance value of all the assignable devices based on all the class data characteristic parameters of the time series data and all the class device parameters of all the assignable devices;

[0133] A second computing module is used to obtain an allocatable device node graph based on device performance values ​​and transmission distance values ​​of all the allocatable devices, and to obtain edge allocation priority values ​​of all computing devices based on the task requirement value of the time series data and the allocatable device node graph;

[0134] The allocation module is used to obtain time series data feature extraction and analysis results based on edge allocation priority values ​​of all computing devices.

[0135] The beneficial effects of the above technology are as follows: according to all the class data characteristic parameters of the time series data and all the class device parameters of all the IoT devices, all the assignable devices are obtained, and it is realized that some IoT devices that can be used for the computing task allocation of the time series data feature extraction and analysis process are screened out from all the IoT devices, and then according to all the class data characteristic parameters of the time series data, the task requirement value of the time series data is accurately obtained, and the difficulty of processing data when extracting and analyzing the time series data is quantified; according to all the class device parameters of all the assignable devices, the device performance values ​​of all the assignable devices are accurately obtained, and the ability of each assignable device when extracting and analyzing the time series data is quantified, and then based on According to the device performance values ​​and transmission distance values ​​of all assignable devices, the assignable device node graph is obtained to facilitate the subsequent calculation of edge allocation priority values. According to the task requirement value of the time series data and the assignable device node graph, the edge allocation priority values ​​of all computing devices are accurately obtained, and the priority of each computing device in the time series data feature extraction and analysis process is quantified. Finally, according to the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained, and the computing allocation of the IoT edge computing is dynamically adjusted according to the characteristics of the time series data and the performance of the IoT devices, ensuring the optimal configuration of resources in different scenarios and improving the efficiency of time series data feature extraction and analysis.

[0136] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention, and the present invention is also intended to include these changes and modifications.

Claims

1. A method for extracting and analyzing time series data features, characterized in that: include: S1: Based on all class data feature parameters of time series data and all class device parameters of all IoT devices, all assignable devices are obtained; S2: based on all class data characteristic parameters of the time series data and all class device parameters of all assignable devices, obtain the task requirement value of the time series data and the device performance value of all assignable devices; S3: Based on the device performance values ​​and transmission distance values ​​of all the allocatable devices, an allocatable device node graph is obtained, including: Acquire the transmission distance values ​​of all allocatable devices, and use the transmission distance values ​​of all the allocatable devices as the abscissa values, and the device performance values ​​of the corresponding allocatable devices as the ordinate values, obtain all the allocatable device nodes, and use the rectangular coordinate system containing all the allocatable device nodes as the allocatable device node graph; Based on the task demand value of the time series data and the allocatable device node graph, the edge allocation priority values ​​of all computing devices are obtained, including: The analysis circle is obtained by taking the origin of the rectangular coordinate system of the allocable device node graph as the center and the task demand value of the time series data as the radius; Determine whether the number of allocatable device nodes in the analysis circle is 0. If so, take the sum of the radius of the analysis circle and 1 as the radius of the new analysis circle, take the origin of the rectangular coordinate system of the allocatable device node graph as the center and the radius of the new analysis circle as the radius, obtain a new analysis circle, and continue to determine whether the number of allocatable device nodes in the new analysis circle is 0. If so, continue to increase the radius of the new analysis circle by a unit value until the number of allocatable device nodes in the latest analysis circle is not 0, then take all the allocatable device nodes in the latest analysis circle as computing device nodes, and take the radius of the latest analysis circle as the computing radius; Otherwise, all allocatable device nodes within the analysis circle are regarded as computing device nodes, and the radius of the analysis circle is regarded as the computing radius; Based on the task requirement values ​​of all computing device nodes, computing radius and time series data, an edge allocation priority value of each computing device node is obtained, and the edge allocation priority value of each computing device node is used as the edge allocation priority value of the corresponding computing device; S4: Based on the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained.

2. The method for extracting and analyzing time series data features according to claim 1, characterized in that: S1: Based on all class data feature parameters of time series data and all class device parameters of all IoT devices, all assignable devices are obtained, including: Obtain all class data feature parameters of time series data, where all class data feature parameters include data volume, data collection frequency, and data dimension; Obtain all device parameters of all IoT devices, including device processor main frequency, device processor core number, device memory amount, and device memory speed; Based on the device memory amount and all class data characteristic parameters of the time series data of all IoT devices, all assignable devices are obtained.

3. The method for extracting and analyzing time series data features according to claim 2, characterized in that: Based on the device memory and all class data characteristic parameters of the time series data of all IoT devices, all allocatable devices are obtained, including: The quotient of the value of the data volume of the time series data and the value of the data collection frequency of the time series data, and the product of the value of the data dimension of the time series data, are used as the screening data volume value; If the value of the device memory amount of each IoT device is greater than the value of the filtered data amount, the corresponding IoT device is determined to be an allocatable device, and all the allocatable devices are obtained.

4. The method for extracting and analyzing time series data features according to claim 1, characterized in that: S2: Based on all class data characteristic parameters of the time series data and all class device parameters of all assignable devices, the task requirement value of the time series data and the device performance value of all assignable devices are obtained, including: Obtaining a time series data feature extraction method, and obtaining a time series data feature extraction scoring value based on the time series data feature extraction method and a preset feature extraction method scoring table; Obtaining a time series data feature analysis method, and obtaining a time series data feature analysis scoring value based on the time series data feature analysis method and a preset feature analysis method scoring table; Based on the time series data feature extraction and scoring, time series data feature analysis and scoring, and all class data feature parameters of the time series data, the task requirement value of the time series data is obtained, which is: Among them, β is the task requirement value of time series data, α1 is the score value of time series data feature extraction, α2 is the score value of time series data feature analysis, e is a natural constant, A is the value of the data volume of time series data, B is the value of the data collection frequency of time series data, C is the data dimension of time series data, (α1, α2) max is the maximum value of α1 and α2, ln is the natural logarithm, and the value of the natural constant e is 2.718; Based on all class device parameters of all allocatable devices, a device performance value of each allocatable device is obtained.

5. The method for extracting and analyzing time series data features according to claim 4, characterized in that: Based on all class device parameters of all allocatable devices, obtain device performance values ​​for each allocatable device, including: Among them, γ is the device performance value of the currently calculated allocatable device, W is the device processor main frequency of the currently calculated allocatable device, W0 is the average of the device processor main frequencies of allocatable devices, E is the number of device processor cores of the currently calculated allocatable device, E0 is the average of the number of device processor cores of allocatable devices, R is the device memory amount of the currently calculated allocatable device, R0 is the average of the device memory amount of allocatable devices, H is the device memory speed of the currently calculated allocatable device, and H0 is the average of the device memory speed of allocatable devices.

6. The method for extracting and analyzing time series data features according to claim 1, characterized in that: Based on the task requirement values ​​of all computing device nodes, computing radius and time series data, the edge allocation priority value of each computing device node is obtained, including: Among them, δ is the edge allocation priority value of the computing device node currently being calculated, β is the task requirement value of the time series data, x is the horizontal coordinate value of the computing device node currently being calculated, y is the vertical coordinate value of the computing device node currently being calculated, r is the distance between the computing device node currently being calculated and the origin of the rectangular coordinate system, r0 is the calculation radius, ln is the natural logarithm, and the value of the natural constant e is 2.

718.

7. The method for extracting and analyzing time series data features according to claim 1, characterized in that: S4: Based on the edge allocation priority values ​​of all computing devices, the time series data feature extraction and analysis results are obtained, including: Sorting all computing devices in descending order of their edge allocation priority values ​​to obtain a computing device sorting result; The computing tasks of the time series data feature extraction and analysis process are assigned to all computing devices, and the assignment order is consistent with the computing device sorting results.

8. A platform for extracting and analyzing time series data features, characterized in that: A method for extracting and analyzing time series data features according to any one of claims 1 to 7, comprising: A screening module, used to obtain all assignable devices based on all class data feature parameters of the time series data and all class device parameters of all IoT devices; A first calculation module is used to obtain the task requirement value of the time series data and the device performance value of all the assignable devices based on all the class data characteristic parameters of the time series data and all the class device parameters of all the assignable devices; A second computing module is used to obtain an allocatable device node graph based on device performance values ​​and transmission distance values ​​of all the allocatable devices, and to obtain edge allocation priority values ​​of all computing devices based on the task requirement value of the time series data and the allocatable device node graph; The allocation module is used to obtain time series data feature extraction and analysis results based on edge allocation priority values ​​of all computing devices.

Images