Data acquisition method for Internet of Things data

Abstract

The invention relates to the technical field of data processing, in particular to an Internet of Things data-oriented data acquisition method, which comprises the following steps of: acquiring each Internet of Things data sequence, segmenting the Internet of Things data sequence, and acquiring the core fluctuation degree of each segment of each Internet of Things data sequence; on the basis of the core fluctuation degree, obtaining the number of fitting times of each segment of each Internet of Things data sequence; obtaining the continuity of adjacent segments in each Internet of Things data sequence; according to the continuity, obtaining the final fitting weight of each data in each segment of each Internet of Things data sequence, according to the final fitting weight and the fitting times, performing polynomial fitting on the data in each segment of each Internet of Things data sequence to obtain a polynomial of each segment, and based on the polynomial, obtaining the data in each segment of each Internet of Things data sequence. And polynomial filtering is carried out on each segment of each Internet of Things data sequence. According to the method, the accuracy of data filtering is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing. More specifically, the present invention relates to a data acquisition method for Internet of Things (IoT) data. Background Art

[0002] In traditional industrial scenarios, enterprises generally face problems such as data islands (e.g., the separation of production data and management data) and the lack of intelligent top-level design. For example, although industrial equipment can collect operation data through sensors, due to poor protocol compatibility and insufficient data processing capabilities, the quality of the collected data is difficult to guarantee, which hinders subsequent data processing. Therefore, it is necessary to filter the operation data of industrial equipment.

[0003] In the process of filtering and cleaning IoT data sequences using polynomial smoothing filtering, a globally fixed polynomial order is usually set for local polynomial fitting of the IoT data sequences. Due to the different states of IoT data, that is, there are data segments with large fluctuations and data segments with small fluctuations in the IoT data sequence. Therefore, setting a globally fixed polynomial order for local polynomial fitting in polynomial smoothing filtering will result in inappropriate fitting effects for data segments with large fluctuations, while data segments with small fluctuations are prone to overfitting, resulting in poor overall fitting and filtering effects of the IoT data sequence. Summary of the Invention

[0004] To solve the problem that inappropriate fitting effects may occur for data segments with large fluctuations or overfitting may easily occur for data segments with small fluctuations when setting a globally fixed polynomial order for local polynomial fitting in polynomial smoothing filtering, the present invention proposes a data acquisition method for IoT data, which includes the following steps: Collect each IoT data sequence; segment each IoT data sequence; obtain the core fluctuation degree of each segment of each IoT data sequence; based on the core fluctuation degree, obtain the fitting times of each segment of each IoT data sequence. Obtain the adjacent data segments of each segment in adjacent segments of each IoT data sequence; obtain the average weighted slope of the adjacent data segments of each segment in adjacent segments of each IoT data sequence; according to the difference in the average weighted slopes of the adjacent data segments of two segments in adjacent segments, obtain the continuity of adjacent segments in each IoT data sequence. According to the continuity, obtain the fitting weight of each data in each segment of each Internet of Things data sequence; according to the fitting weight, obtain the final fitting weight of each data in each segment of each Internet of Things data sequence; according to the final fitting weight and the number of fitting times, perform weighted least squares fitting on the data in each segment of each Internet of Things data sequence to obtain the polynomial of each segment of each Internet of Things data sequence; based on the polynomial, perform polynomial filtering on each segment of each Internet of Things data sequence to obtain the filtered Internet of Things data sequence for each item.

[0005] The innovation of the present invention lies in dividing each Internet of Things data sequence into several segments, which is convenient for subsequent separate fitting of the data in each segment, and can take into account the fluctuation of the data in each segment; by analyzing the fluctuation of the data in each segment of each Internet of Things data sequence, the core fluctuation degree of each segment of each Internet of Things data sequence is obtained; based on the core fluctuation degree, the number of fitting times of each segment of each Internet of Things data sequence is adaptively obtained, so that when subsequent polynomial fitting is performed on each segment, the fluctuation of the data in each segment can be taken into account; then when the present invention performs fitting on each segment, it analyzes the continuity between adjacent segments to obtain the final fitting weight of each data in each segment of each Internet of Things data sequence, so that the continuity can be maintained after fitting between segments. Finally, according to the final fitting weight and the number of fitting times, weighted least squares fitting is performed on the data in each segment to obtain the polynomial of each segment of each Internet of Things data sequence for polynomial filtering, improving the filtering effect.

[0006] Preferably, the segmentation of each Internet of Things data sequence includes: Preset the minimum length L of the data segment. The data segments formed by the L data before and after each data in the j-th Internet of Things data sequence are used as the adjacent data segments of each data in the j-th Internet of Things data sequence. According to the adjacent data segments of each data in the j-th Internet of Things data sequence, obtain the coefficient of variation of each data in the j-th Internet of Things data sequence. Preset the difference threshold T. If the absolute value of the difference between the coefficients of variation of any two adjacent data in the j-th Internet of Things data sequence is greater than the difference threshold T, one of the data is recorded as a segmentation point, and multiple segmentation points in the j-th Internet of Things data sequence are obtained. According to the multiple segmentation points in the j-th Internet of Things data sequence, the j-th Internet of Things data sequence is divided into several segments.

[0007] It is convenient for subsequent analysis of the fluctuation of the data in each segment to obtain the number of fitting times of each segment.

[0008] Preferably, the obtaining of the core fluctuation degree of each segment of each Internet of Things data sequence includes: ; In the formula, represents the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence; represents the number of data in the k-th segment of the j-th Internet of Things data sequence; represents the index of the m-th data in the k-th segment of the j-th Internet of Things data sequence; represents the average value of all data in the k-th segment of the j-th Internet of Things data sequence; represents the value of the m-th data in the k-th segment of the j-th Internet of Things data sequence; exp() represents the exponential function with the natural constant as the base; || represents the absolute value symbol.

[0009] It is convenient to adaptively obtain the fitting times of each segment of each Internet of Things data sequence according to the core fluctuation degree of each segment.

[0010] Preferably, the obtaining of the fitting times of each segment of each Internet of Things data sequence includes: ; In the formula, represents the fitting times of the k-th segment of the j-th Internet of Things data sequence; represents the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence; represents the minimum value among the core fluctuation degrees of all segments of the j-th Internet of Things data sequence; represents the logarithmic function with 2 as the base.

[0011] Preferably, the obtaining of the adjacent data segments of each segment in the adjacent segments of each Internet of Things data sequence includes: For the j-th Internet of Things data sequence, the k-th segment and the (k + 1)-th segment are adjacent segments. Denote the number of all data in the k-th segment as a, denote the downward rounding value of the product between 0.5 and a as b, and denote the data segment formed by the last data in the k-th segment and the previous b data as the adjacent data segment of the k-th segment; Denote the number of all data in the (k + 1)-th segment as c, and denote the downward rounding value of the product between 0.5 and c as d; Denote the data segment formed by the first data in the (k + 1)-th segment and the d-th data as the adjacent data segment of the (k + 1)-th segment.

[0012] Preferably, the obtaining of the average weighted slope of the adjacent data segments of each segment in the adjacent segments of each Internet of Things data sequence includes: ; In the formula, represents the average weighted slope of the adjacent data segment of the k-th segment; represents the index difference of the x-th data in the adjacent data segment of the k-th segment; represents the number of all data in the adjacent data segment of the k-th segment; represents the value of the x-th data in the adjacent data segment of the k-th segment; represents the value of the (x - 1)-th data in the adjacent data segment of the k-th segment; exp() represents the exponential function with the natural constant as the base; According to the method for obtaining the average weighted slope of the adjacent data segment of the k-th segment, obtain the average weighted slope of the adjacent data segment of the (k + 1)-th segment.

[0013] It is convenient to subsequently obtain the continuity of adjacent segments in each Internet of Things data sequence based on the difference in the average weighted slopes of the adjacent data segments of adjacent segments.

[0014] Preferably, the obtaining of the continuity of adjacent segments in each Internet of Things data sequence includes: ; In the formula, represents the continuity of the -th segment and the -th segment of the j-th Internet of Things data sequence; represents the average value of all data in the adjacent data segment of the -th segment; represents the average value of all data in the adjacent data segment of the -th segment; represents the average weighted slope of the adjacent data segment of the -th segment; represents the average weighted slope of the adjacent data segment of the -th segment; | | represents taking the absolute value; exp() represents the exponential function with the natural constant as the base.

[0015] It is convenient to subsequently adaptively obtain the fitting weight of each data in each segment of each Internet of Things data sequence according to the continuity of adjacent segments.

[0016] Preferably, the obtaining of the fitting weight of each data in each segment of each Internet of Things data sequence includes: ; In the formula, represents the fitting weight of the m-th data in the k-th segment of the j-th Internet of Things data sequence; represents the number of data in the k-th segment of the j-th Internet of Things data sequence; represents the index of the m-th data in the k-th segment of the j-th Internet of Things data sequence; represents the standard deviation of the data in the k-th segment of the j-th Internet of Things data sequence; represents the continuity of the k-th segment and the (k + 1)-th segment of the j-th Internet of Things data sequence; Represents the continuity between the k-th segment and the (k - 1)-th segment of the j-th Internet of Things data sequence.

[0017] Preferably, obtaining the final fitting weight of each data in each segment of each Internet of Things data sequence includes: ; In the formula, Represents the final fitting weight of the m-th data in the k-th segment of the j-th Internet of Things data sequence; Represents the fitting weight of the m-th data in the k-th segment of the j-th Internet of Things data sequence; Represents the number of data in the k-th segment of the j-th Internet of Things data sequence.

[0018] Preferably, collecting each Internet of Things data sequence includes: Collecting each Internet of Things data in the past month from the database. Each Internet of Things data contains several items of data, namely the current data, voltage data, and vibration data of the device at each sampling time in the Internet of Things; arranging each item of data of all the collected Internet of Things data in the order of sampling time, and obtaining each Internet of Things data sequence.

[0019] The present invention has the following beneficial effects: The purpose of the present invention is to divide each Internet of Things data sequence into several segments, which is convenient for subsequent separate fitting of the data in each segment, and can take into account the fluctuation situation of the data in each segment; by analyzing the fluctuation situation of the data in each segment of each Internet of Things data sequence, the core fluctuation degree of each segment of each Internet of Things data sequence is obtained; based on the core fluctuation degree, the fitting times of each segment of each Internet of Things data sequence are adaptively obtained, so that when performing polynomial fitting on each segment later, the fluctuation situation of the data in each segment can be taken into account; then when the present invention fits each segment, it analyzes the continuity between adjacent segments to obtain the final fitting weight of each data in each segment of each Internet of Things data sequence, so that the continuity can be maintained after fitting between segments. Finally, according to the final fitting weight and the fitting times, weighted least squares fitting is performed on the data in each segment to obtain the polynomial of each segment of each Internet of Things data sequence for polynomial filtering, improving the filtering effect. Description of the Drawings

[0020] By reading the following detailed description with reference to the drawings, the above and other purposes, features, and advantages of the exemplary embodiments of the present invention will become easy to understand. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 Is the step flow chart of a data acquisition method for Internet of Things data according to an embodiment of the present invention; Figure 2 It is a schematic diagram of an adjacent data segment of adjacent segments. Specific implementation manners

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] The specific implementation manners of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Please refer to Figure 1 , which shows a flowchart of steps of a data acquisition method for Internet of Things data provided by an embodiment of the present invention. The method includes the following steps: S001. Collect each Internet of Things data sequence.

[0024] In an embodiment of the present invention, each Internet of Things data in the past month is collected from a database. Each Internet of Things data includes several items of data, that is, each Internet of Things data includes the current data, voltage data, and vibration data of a device at each sampling time in the Internet of Things; Each item of data of all the collected Internet of Things data is arranged in the order of sampling time, so as to construct each Internet of Things data sequence.

[0025] S002. Segment each Internet of Things data sequence, and obtain the core fluctuation degree of each segment of each Internet of Things data sequence; based on the core fluctuation degree, obtain the fitting times of each segment of each Internet of Things data sequence.

[0026] It should be noted that in the process of filtering and cleaning the Internet of Things data sequence by polynomial smoothing filtering, due to the different states of the Internet of Things data, if a globally fixed polynomial order is set for polynomial smoothing filtering for local polynomial fitting, data segments with a large fluctuation degree cannot achieve a suitable fitting effect, while data segments with a small fluctuation degree are prone to overfitting, resulting in a poor overall fitting and filtering effect of the Internet of Things data sequence; therefore, the present invention first segments each Internet of Things data sequence, analyzes the fluctuation of the data in each segment of each Internet of Things data sequence, obtains the fitting times of each segment of each Internet of Things data sequence, considers the fluctuation of the data in the segment, then obtains the fitting weight of each data in each segment according to the continuity between segments, and finally uses the least squares method to perform polynomial fitting on the data in each segment according to the fitting times of each segment and the fitting weight of each data in each segment.

[0027] It should be further noted that, first, the adjacent data segments of each data in each Internet of Things data sequence are obtained. According to the adjacent data segments, the coefficient of variation of each data in each Internet of Things data sequence is obtained. If the difference in the coefficient of variation of adjacent data in the Internet of Things data sequence is larger, it indicates that there is a large change in the fluctuation of the surrounding data of the adjacent data, that is, the characteristics of the data have changed significantly. At this time, the adjacent data should be divided more. Therefore, each Internet of Things data sequence is segmented according to this feature.

[0028] In the embodiment of the present invention, the preset minimum length of the data segment L = 50. In other embodiments, the implementer can preset the value of the minimum length L of the data segment according to the specific implementation situation; The process of segmenting each Internet of Things data sequence is as follows: The data segments formed by the L data before and after each data in the j-th Internet of Things data sequence are used as the adjacent data segments of each data in the j-th Internet of Things data sequence. According to the adjacent data segments of each data in the j-th Internet of Things data sequence, the coefficient of variation of each data in the j-th Internet of Things data sequence is obtained; It should be noted that the method for obtaining the coefficient of variation is a well-known technology. In the embodiment of the present invention, it will not be elaborated too much.

[0029] The preset difference threshold T = 0.3. If the absolute value of the difference between the coefficients of variation of any two adjacent data in the j-th Internet of Things data sequence is greater than the difference threshold T, one of the data is recorded as a segmentation point, and multiple segmentation points in the j-th Internet of Things data sequence are obtained. According to the multiple segmentation points in the j-th Internet of Things data sequence, the j-th Internet of Things data sequence is divided into several segments; Similarly, each Internet of Things data sequence is divided into several segments.

[0030] It should be noted that if the data in any segment of any Internet of Things data sequence fluctuates more, at this time, the larger the number of fitting times required for this segment to satisfy the accurate fitting of its fluctuating part. Therefore, the present invention analyzes the data fluctuation in this segment of the Internet of Things data sequence, obtains the core fluctuation degree of this segment of the Internet of Things data sequence, and then adaptively obtains the number of fitting times of this segment of the Internet of Things data sequence; If the difference between the value of each data in the segment and the mean value of all data is larger, it indicates that the core fluctuation degree of the segment is larger. Also, since the edges between segments need to maintain continuity during the process of fitting the segmented data, the determination of the number of fitting times should be mainly based on the fluctuation of the data near the middle in the segment. Because the difference between the value of each data and the mean value of all data is corrected according to the degree of each data approaching the middle. If the degree of a data approaching the middle is larger, the difference between the value of this data and the mean value of all data is more concerned.

[0031] In the embodiment of the present invention, obtain the core fluctuation degree of each segment of each Internet of Things data sequence: ; In the formula, represents the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence; represents the number of data in the k-th segment of the j-th Internet of Things data sequence; represents the index of the m-th data in the k-th segment of the j-th Internet of Things data sequence; represents the mean value of all data in the k-th segment of the j-th Internet of Things data sequence; represents the value of the m-th data in the k-th segment of the j-th Internet of Things data sequence; exp() represents the exponential function with the natural constant as the base; || represents the absolute value symbol; The larger the value of, the greater the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence; The larger the value of, the greater the degree that the m-th data in the k-th segment of the j-th Internet of Things data sequence is close to the middle. At this time, more attention should be paid to the difference between the m-th data and the mean value of all data in the k-th segment.

[0032] Obtain the fitting times of each segment of each Internet of Things data sequence: ; In the formula, represents the fitting times of the k-th segment of the j-th Internet of Things data sequence; represents the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence; represents the minimum value among the core fluctuation degrees of all segments of the j-th Internet of Things data sequence; since the greater the core fluctuation degree of the k-th segment of the j-th Internet of Things data sequence, the greater the fitting times required for the k-th segment of the j-th Internet of Things data sequence to satisfy the accurate fitting of its fluctuating part.

[0033] S003. Obtain the average weighted slope of each adjacent data segment of each segment in each Internet of Things data sequence. According to the difference between the average weighted slopes of the adjacent data segments of two segments in the adjacent segments and the difference in the data values in the adjacent data segments, obtain the continuity of each adjacent segment in each Internet of Things data sequence; according to the continuity, obtain the final fitting weight of each data in each segment of each Internet of Things data sequence.

[0034] It should be noted that when fitting each segment in each item of IoT data sequence, the adjacent data segments between adjacent segments should be kept continuous as much as possible. If the continuity of the adjacent data segments between adjacent segments is poor, it indicates that the change trends of the neighborhood data segments between adjacent segments are inconsistent. Then, when fitting the segments, it is necessary to set lower fitting weights for the data at the adjacent data segments of the segments (i.e., the edge data), so that the edge data of the segments with low continuity is allowed to have higher errors, and the continuity between the segments is ensured as much as possible after fitting. Therefore, the present invention needs to first obtain the adjacent data segments of two adjacent segments, then obtain the continuity of the two adjacent segments according to the difference between the adjacent data segments of the two adjacent segments, and finally set appropriate fitting weights for the data in the segments according to the continuity of the two adjacent segments; First, it is necessary to obtain the adjacent data segments of two adjacent segments. The schematic diagram of the adjacent data segments is as Figure 2 shown. If data segment 1 is adjacent to data segment 2, then the adjacent data segment of data segment 1 and the adjacent data segment in data segment 2 are Figure 2 the data in the green box in.

[0035] In the embodiment of the present invention, for the k-th segment and the (k + 1)-th segment of the j-th item of IoT data sequence, the number of all data in the k-th segment is denoted as a, the downward integer value of the product of 0.5 and a is denoted as b, and the data segment formed by the last data in the k-th segment and the previous b data is denoted as the adjacent data segment of the k-th segment; the number of all data in the (k + 1)-th segment is denoted as c, and the downward integer value of the product of 0.5 and c is denoted as d; the data segment formed by the first data in the (k + 1)-th segment and the d-th data is denoted as the adjacent data segment of the (k + 1)-th segment; It should be noted that when analyzing the continuity of adjacent segments, first, it is necessary to analyze the change trend of the data values in the adjacent data segments of each segment in the adjacent segments, obtain the average weighted slope of the adjacent data segments of each segment in the adjacent segments, and use the weighted method in the calculation. If the data in the neighborhood data segment of one segment in the adjacent segments is closer to another segment, a greater weight is assigned to this data.

[0036] In the embodiment of the present invention, for any data in the adjacent data segment of the k-th segment of the j-th item of IoT data sequence, the absolute value of the difference between the index value of the data and the total number of data in the k-th segment is denoted as the index difference of the data; for any data in the adjacent data segment of the (k + 1)-th segment, the absolute value of the difference between the index value of the data and 1 is denoted as the index difference of the data; it should be noted that the k-th segment and the (k + 1)-th segment of the j-th item of IoT data sequence are adjacent segments; Obtain the average weighted slope of the adjacent data segment of the k-th segment: ; In the formula, represents the average weighted slope of the adjacent data segment of the k-th segment; represents the index difference of the x-th data in the adjacent data segment of the k-th segment; represents the number of all data in the adjacent data segment of the k-th segment; represents the value of the x-th data in the adjacent data segment of the k-th segment; represents the value of the (x - 1)-th data in the adjacent data segment of the k-th segment; exp() represents the exponential function with the natural constant as the base; The value of represents the difference between the x-th data and the (x - 1)-th data in the adjacent data segment of the k-th segment, representing the slope between adjacent data in the adjacent data segment; represents the inverse proportional value of the index difference of the x-th data in the adjacent data segment of the k-th segment. The larger its value, the closer the x-th data in the adjacent data segment of the k-th segment is to the (k + 1)-th segment. At this time, more attention should be paid to the difference between the x-th data and the (x - 1)-th data in the adjacent data segment of the k-th segment. Therefore, after weighted summation of the slopes between adjacent data according to the index difference, the average weighted slope of the neighborhood data segment is obtained.

[0037] Similarly, the average weighted slope of the adjacent data segment of the (k + 1)-th segment is obtained by the above method.

[0038] It should be noted that when the difference in the average weighted slope between the adjacent data segments of adjacent segments is smaller, it indicates that the continuity of adjacent segments is greater. And when the difference in the average value of all data in the adjacent data segments of adjacent segments is smaller, it indicates that the continuity of adjacent segments is greater. Therefore, in the present invention, the continuity of adjacent segments is obtained according to the difference in the average value of all data in the adjacent data segments of adjacent segments and the difference in the average weighted slope between the adjacent data segments of adjacent segments.

[0039] In the embodiment of the present invention, the continuity between the k-th segment and the (k + 1)-th segment of the j-th item of Internet of Things data sequence is obtained: ; In the formula, represents the continuity between the -th segment and the -th segment of the j-th item of Internet of Things data sequence; represents the average value of all data in the adjacent data segment of the -th segment; represents the average value of all data in the adjacent data segment of the -th segment; represents the average weighted slope of the adjacent data segment of the -th segment; represents the The average weighted slope of the adjacent data segments of each segment; Indicates taking the absolute value; exp() represents the exponential function with the natural constant as the base; The closer the value of is to 1, the smaller the average weighted slope between the adjacent data segments of adjacent segments, and the greater the continuity of adjacent segments at this time.

[0040] It should be noted that when fitting each segment in each item of the Internet of Things data sequence, the fitting weight of each data is obtained according to the position of each data in the segment. Generally, the fitting weight of the middle data of the segment is larger, and the fitting weight of the edge data of the segment is smaller; however, in order to maintain the continuity between segments after fitting, it is necessary to combine the continuity of adjacent segments to obtain the fitting weight of the data in each segment, so that the fitting weight of the middle data of the segment with poor continuity is even larger and the fitting weight of the edge data is even smaller.

[0041] If the continuity of any segment of any item of the Internet of Things data sequence with its two adjacent segments is poor, it is necessary to set a larger fitting weight for the middle data of the segment and a lower fitting weight for the edge data of the segment, so that the edge data of the segment with low continuity is allowed to have a higher error, so as to ensure the continuity between segments as much as possible after fitting; if the continuity of the segment with its two adjacent segments is good, then fitting is performed with a more average fitting weight.

[0042] In the embodiment of the present invention, the fitting weight of the m-th data in the k-th segment of the j-th item of the Internet of Things data sequence is obtained: ; In the formula, represents the fitting weight of the m-th data in the k-th segment of the j-th item of the Internet of Things data sequence; represents the number of data in the k-th segment of the j-th item of the Internet of Things data sequence; represents the index of the m-th data in the k-th segment of the j-th item of the Internet of Things data sequence; represents the standard deviation of the data in the k-th segment of the j-th item of the Internet of Things data sequence; represents the continuity between the k-th segment and the (k + 1)-th segment of the j-th item of the Internet of Things data sequence; represents the continuity between the k-th segment and the (k - 1)-th segment of the j-th item of the Internet of Things data sequence; represents the position situation of the m-th data in the k-th segment of the j-th item of the Internet of Things data sequence. The larger its value, the more the position of the m-th data is at the edge of the k-th segment, and the smaller its value, the more the position of the m-th data is in the middle of the k-th segment; It is known that The smaller the value of, the smaller the continuity between the k-th segment of the j-th item of the Internet of Things data sequence and its two adjacent segments. At this time, Reduce it so that has a larger value. At this time, it is necessary to set a larger fitting weight for the intermediate data of the k-th segment and a lower fitting weight for the edge data of the k-th segment. Therefore, if the position of the m-th data in the k-th segment is at the edge of the k-th segment, then according to Adjust by a larger amount so that has a larger value, then has a smaller value; if the position of the m-th data in the k-th segment is in the middle of the k-th segment, according to Adjust by a smaller amount so that changes little, then has a larger value; It is known that the larger the value, the greater the continuity between the k-th segment of the j-th item of IoT data sequence and its two adjacent segments. At this time, adjust to make have a smaller value. Therefore, if the position of the m-th data in the k-th segment is at the edge of the k-th segment, then according to Adjust by a smaller amount so that has a smaller value; if the position of the m-th data in the k-th segment is in the middle of the k-th segment, according to Adjust by a smaller amount so that has a smaller value. Therefore, for the data in the segment with greater continuity, the fitting weights of the data in the segment can be adjusted more towards being average according to the continuity.

[0043] Obtain the final fitting weight of the m-th data in the k-th segment of the j-th item of IoT data sequence: ; In the formula, represents the final fitting weight of the m-th data in the k-th segment of the j-th item of IoT data sequence; represents the fitting weight of the m-th data in the k-th segment of the j-th item of IoT data sequence; represents the number of data in the k-th segment of the j-th item of IoT data sequence; is used to normalize the value.

[0044] S004. According to the final fitting weight of each data in each segment of each Internet of Things data sequence and the fitting times of each segment of each Internet of Things data sequence, perform polynomial fitting on the data in each segment of each Internet of Things data sequence to obtain the polynomial of each segment of each Internet of Things data sequence, and based on the polynomial, perform polynomial filtering on each segment of each Internet of Things data sequence.

[0045] It should be noted that in the present invention, the fitting times of each segment are obtained through the fluctuation of the data in each segment of each Internet of Things data sequence, and the final fitting weight of each data in each segment is confirmed by analyzing the continuity between adjacent segments. Therefore, then, according to the final fitting weight of each data in each segment of each Internet of Things data sequence and the fitting times of each segment of each Internet of Things data sequence, perform fitting on the data in each segment of each Internet of Things data sequence.

[0046] In the embodiment of the present invention, according to the final fitting weight of each data in each segment of each Internet of Things data sequence and the fitting times of each segment of each Internet of Things data sequence, perform weighted least squares fitting on each segment of each Internet of Things data sequence to obtain the polynomial of each segment of each Internet of Things data sequence; According to the polynomial of each segment of each Internet of Things data sequence, perform polynomial filtering on each segment of each Internet of Things data sequence to obtain each Internet of Things data sequence after filtering.

[0047] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data collection method for Internet of Things data, characterized in that: include: Collect each IoT data sequence; Segment each IoT data sequence; Get the core volatility of each segment of each IoT data series; Based on the core volatility, obtaining the number of fittings for each segment of each IoT data series; Obtain the adjacent data segments of each segment in the adjacent segments of each IoT data sequence; obtain the average weighted slope of the adjacent data segments of each segment in the adjacent segments of each IoT data sequence; obtain the continuity of the adjacent segments in each IoT data sequence according to the difference of the average weighted slopes of the adjacent data segments of two segments in the adjacent segments; According to the continuity, a fitting weight of each data in each segment of each IoT data sequence is obtained; according to the fitting weight, a final fitting weight of each data in each segment of each IoT data sequence is obtained; According to the final fitting weight and the number of fittings, weighted least squares fitting is performed on the data in each segment of each IoT data sequence to obtain a polynomial for each segment of each IoT data sequence; Based on the polynomial, each segment of each IoT data sequence is subjected to polynomial filtering to obtain each IoT data sequence after filtering.

2. The data collection method for Internet of Things data according to claim 1 is characterized in that: The segmenting of each IoT data sequence includes: The minimum length of the data segment is preset to be L, and the data segment composed of L data before and after each data in the j-th IoT data sequence is used as the adjacent data segment of each data in the j-th IoT data sequence, and the coefficient of variation of each data in the j-th IoT data sequence is obtained according to the adjacent data segment of each data in the j-th IoT data sequence; A difference threshold T is preset. If the absolute value of the difference between the coefficients of variation of any two adjacent data in the j-th IoT data sequence is greater than the difference threshold T, one of the data is recorded as a segmentation point to obtain multiple segmentation points in the j-th IoT data sequence. According to the multiple segmentation points in the j-th IoT data sequence, the j-th IoT data sequence is divided into several segments.

3. The data collection method for Internet of Things data according to claim 1 is characterized in that: The method of obtaining the core volatility of each segment of each IoT data sequence includes: ； In the formula, represents the core volatility of the kth segment of the jth IoT data sequence; represents the number of data in the kth segment of the jth IoT data sequence; Represents the index of the mth data in the kth segment of the jth IoT data sequence; represents the mean of all data in the kth segment of the jth IoT data sequence; Represents the value of the mth data in the kth segment of the jth IoT data sequence; exp() represents an exponential function with a natural constant as the base; || represents the absolute value symbol.

4. The data collection method for Internet of Things data according to claim 1 is characterized in that: The obtaining of the fitting times of each segment of each IoT data sequence includes: ； In the formula, represents the number of fittings of the kth segment of the jth IoT data sequence; represents the core volatility of the kth segment of the jth IoT data sequence; Represents the minimum value of the core volatility of all segments of the j-th IoT data sequence; Represents the base 2 logarithm function.

5. The data collection method for Internet of Things data according to claim 1 is characterized in that: The step of obtaining the adjacent data segments of each segment in the adjacent segments of each IoT data sequence includes: The kth segment and the k+1th segment of the jth IoT data sequence are adjacent segments. The number of all data in the kth segment is denoted as a, the floor value of the product of 0.5 and a is denoted as b, and the data segment consisting of the last data in the kth segment and the previous b data is denoted as the adjacent data segment of the kth segment; the number of all data in the k+1th segment is denoted as c, and the floor value of the product of 0.5 and c is denoted as d; the data segment consisting of the first data in the k+1th segment and the dth data is denoted as the adjacent data segment of the k+1th segment.

6. The data collection method for Internet of Things data according to claim 1 is characterized in that: The step of obtaining the average weighted slope of the adjacent data segments of each segment in the adjacent segments of each IoT data sequence includes: ； In the formula, represents the average weighted slope of the adjacent data segments of the kth segment; Indicates the index difference of the xth data in the adjacent data segment of the kth segment; Represents the number of all data in the adjacent data segments of the kth segment; Indicates the value of the xth data in the adjacent data segment of the kth segment; It represents the value of the x-1th data in the adjacent data segment of the kth segment; exp() represents an exponential function with a natural constant as the base; according to the method for obtaining the average weighted slope of the adjacent data segments of the kth segment, the average weighted slope of the adjacent data segments of the k+1th segment is obtained.

7. The data collection method for Internet of Things data according to claim 1 is characterized in that: The obtaining of continuity of adjacent segments in each IoT data sequence includes: ； In the formula, represents the jth IoT data sequence Section and Continuity of the segments; Indicates The average value of all data in the adjacent data segments of a segment; Indicates The average value of all data in the adjacent data segments of a segment; Indicates The average weighted slope of the adjacent data segments of the segment; Indicates The average weighted slope of the adjacent data segments of the segment; It means taking the absolute value; exp() represents the exponential function with a natural constant as the base.

8. The data collection method for Internet of Things data according to claim 1 is characterized in that: The step of obtaining the fitting weight of each data in each segment of each IoT data sequence includes: ； In the formula, represents the fitting weight of the mth data in the kth segment of the jth IoT data sequence; represents the number of data in the kth segment of the jth IoT data sequence; Represents the index of the mth data in the kth segment of the jth IoT data sequence; Represents the standard deviation of the data in the kth segment of the jth IoT data sequence; Represents the continuity between the kth segment and the k+1th segment of the jth IoT data sequence; Represents the continuity between the kth segment and the k-1th segment of the jth IoT data sequence.

9. The data collection method for Internet of Things data according to claim 1, characterized in that: The method of obtaining the final fitting weight of each data in each segment of each IoT data sequence includes: ； In the formula, represents the final fitting weight of the mth data in the kth segment of the jth IoT data sequence; represents the fitting weight of the mth data in the kth segment of the jth IoT data sequence; Represents the number of data in the kth segment of the jth IoT data sequence.

10. The data collection method for Internet of Things data according to claim 1, characterized in that: The collection of each IoT data sequence includes: Each IoT data of the past month is collected from the database. Each IoT data contains several items of data, namely the current data, voltage data and vibration data of the equipment at each sampling time in the IoT. Each item of all collected IoT data is arranged in sequence according to the sampling time to obtain each IoT data sequence.

Images