An intelligent electric meter data optimization storage method

Abstract

The present invention relates to the field of data processing, and particularly to an intelligent electric meter data optimization storage method. It performs normalization, smoothing, and outlier processing on multi-dimensional time series data such as electric energy, current, voltage, and power. Subsequently, it uses the autocorrelation function to perform periodic analysis on time series such as power, voltage, and current to determine the periodic intensity of the data. Finally, it compresses the data through pulse code modulation (PCM), differential pulse code modulation (DPCM), and Huffman coding, thereby effectively improving the data compression rate and reducing redundancy, ensuring the optimization of data transmission efficiency and storage space. The present invention compresses the electric meter data through periodic analysis and Huffman coding, thereby improving the storage efficiency and transmission efficiency.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and particularly to an intelligent electric meter data optimization storage method. Background Art

[0002] An intelligent electric meter is an electronic device that can collect, transmit, and record electrical energy usage data in real time. Its main functions include electrical energy metering, data storage, and remote transmission. The intelligent electric meter can upload power data to the cloud or server in real time through wireless communication technology, facilitating centralized management and analysis by power companies. With the development of smart grids, the popularity of intelligent electric meters has been continuously increasing, and the amount of real-time data collected by electric meters has also increased sharply. Due to the large amount of data generated per second, this poses higher requirements for data storage and transmission in intelligent electric meters.

[0003] First, the amount of data collected by the electric meter is huge. Especially at high sampling frequencies, the electric meter needs to store a large amount of data, which poses high requirements for storage space and computing power. Second, since the electric meter data is frequently uploaded to the server, how to reduce the amount of data uploaded and reduce the consumption of network bandwidth has also become an urgent problem to be solved. In addition, when the network transmission is interrupted, the electric meter data will be saved locally, and at this time, the data storage pressure will also increase continuously. In order to reduce the storage and transmission pressure and improve data processing efficiency, the intelligent electric meter needs to adopt an efficient compression algorithm to reduce redundant data and optimize data storage.

[0004] Currently, Huffman coding, as a classic lossless data compression technology, is applied to various data compression scenarios. Although Huffman coding can effectively compress data, in the application scenario of intelligent electric meter data, the effect of Huffman coding may be limited. For example, the electricity count may not change much within a sampling period, but the value of each data point is not exactly the same. In addition, the variation range of other parameters (such as current, voltage) in the electric meter data is large, which reduces the efficiency of Huffman coding. Summary of the Invention

[0005] In view of the above problem of reduced efficiency of Huffman coding, the present invention proposes an intelligent meter data optimization storage method, including: obtaining multiple operating parameters of the intelligent meter, and the data at multiple consecutive acquisition times of each operating parameter constitute the corresponding parameter time series; calculating the period of each parameter time series, and denoting any data in the parameter time series as the target data; the acquisition time corresponding to the target data is the target time, and the acquisition time corresponding to every other period of the target time is denoted as the comparison time; the data with a set number before and after the target time constitute the neighborhood sequence; obtaining the neighborhood sequences of each comparison time and the target time to obtain multiple neighborhood sequences, and obtaining the periodic intensity of the target data based on the similarity between the multiple neighborhood sequences; obtaining the periodic intensity of all data in any parameter time series, and in response to the periodic intensity being greater than or equal to the set threshold, denoting the corresponding data as strong periodic data; in response to the periodic intensity being less than the set threshold, denoting the corresponding data as weak periodic data; using pulse code modulation for all strong periodic data and using differential pulse code modulation for all weak periodic data; using Huffman coding to compress all the modulated data to reduce the storage space of the intelligent meter data.

[0006] The present invention dynamically detects and quantifies the periodic characteristics of each parameter time series of the intelligent meter, realizes the accurate comparison of the similarity between the target data and its periodic neighborhood, thereby effectively distinguishing strong periodic and weak periodic data, and respectively using pulse code modulation and differential pulse code modulation for processing, and then combining Huffman coding compression, significantly reducing the data storage space and the upload bandwidth requirement, while ensuring the data accuracy and integrity. Compared with the existing method that directly uses Huffman coding, the present invention greatly improves the compression efficiency and data restoration rate.

[0007] Further, calculating the period of each parameter time series further includes: obtaining the autocorrelation function corresponding to different time lag values of each parameter time series; denoting the time lag value corresponding to the maximum value in all the autocorrelation functions as the period.

[0008] By obtaining the autocorrelation function corresponding to different time lag values of each parameter time series and selecting the time lag value corresponding to the maximum value in the autocorrelation function as the period, the present invention accurately reflects the inherent periodicity of the data. Compared with the traditional method that relies on experience to set the period, the present invention provides an objective and automatic period identification means, ensuring the accuracy and stability of the period calculation.

[0009] Further, the specific calculation method of the periodic intensity is as follows:

[0010] ;

[0011] where represents the periodic intensity of the power corresponding to the time stamp ; Represents the maximum value of the cycle sequence number in the power time series; Represents the timestamp The neighborhood sequence at the reference time after the Represents the timestamp The neighborhood sequence at the reference time after the Represents the calculation function of the Dtw distance; Represents the tuning parameter factor.

[0012] The present invention adopts a periodic intensity calculation formula based on the dynamic time warping (DTW) distance to quantify the similarity between neighborhood sequences in different cycles, effectively capturing cycle repeatability and avoiding reducing the compression efficiency due to inconsistent data continuity. Compared with traditional methods, this formula has higher accuracy and robustness in data repeatability recognition.

[0013] Further, obtaining multiple operating parameters of the smart meter further includes: collecting the electric energy reading, power, voltage, and current parameters of the smart meter; setting the frequency range of data collection by the smart meter to [1, 15].

[0014] Further, it also includes temporarily storing the parameter time series in the meter memory and regularly uploading them to the cloud platform through the RS485 interface.

[0015] Further, it also includes not calculating the cycle of the electric energy reading time series, and performing data compression processing on the electric energy reading time series using pulse code modulation and then using Huffman coding.

[0016] Aiming at the problem that the electric energy reading time series lacks periodicity due to the accumulation characteristic, the present invention does not perform periodic calculation on it, but directly uses pulse code modulation (PCM) processing to achieve data compression and retain key information. Compared with the traditional method of uniformly processing all data, this scheme better adapts to the characteristics of the electric energy reading, improving the compression efficiency and data recovery accuracy.

[0017] Further, it also includes performing preprocessing operations on the parameter time series, specifically: normalizing the parameter time series; performing noise reduction processing on the normalized parameter time series.

[0018] Further, the maximum-minimum normalization algorithm is used to normalize the parameter time series.

[0019] Further, the sliding average method is used to smooth the normalized parameter time series to remove noise.

[0020] The sliding average method is used to smooth the time series of each parameter after normalization, effectively eliminating noise and short-term fluctuations, ensuring that the data more accurately reflects the long-term trend. Compared with the traditional method without smoothing, the present invention improves the data stability and the quality of the data before compression, thus optimizing the overall compression effect.

[0021] Further, the Otsu method is used to obtain the set threshold.

[0022] The technical effects of the present invention are as follows:

[0023] After preprocessing the multi-dimensional data of the smart meter, the present invention uses the autocorrelation function and the neighborhood sequence similarity to quantify the periodic intensity, effectively distinguishing strong periodic and weak periodic data. Then, pulse code modulation and differential pulse code modulation are respectively adopted, and finally, Huffman coding is combined to achieve data compression. Compared with the existing method that only relies on Huffman coding to directly process the original data, the present invention makes full use of the inherent periodic characteristics of the data, greatly reducing the storage space and the upload bandwidth requirements. At the same time, it ensures the data accuracy and the restoration rate, and has automation, objectivity and robustness, significantly improving the overall performance of the smart meter data management. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of illustration and not limitation, and the same or corresponding reference numerals indicate the same or corresponding parts, wherein:

[0025] Figure 1 is a flowchart schematically showing a method for optimizing the storage of smart meter data in an embodiment of the present invention;

[0026] Figure 2 is a schematic diagram showing the power consumption reading and power time series with a period of 1 day in an embodiment of the present invention;

[0027] Figure 3 is a schematic diagram showing the power consumption reading and power time series with a period of 5 days in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0029] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings.

[0030] Embodiment of the intelligent electricity meter data optimization storage method:

[0031] As Figure 1 shown, the intelligent electricity meter data optimization storage method of the present invention includes:

[0032] S1. Obtain relevant data of the intelligent electricity meter and perform preprocessing.

[0033] During the operation of the intelligent electricity meter, a large number of high-frequency data records will be generated. These data records not only need to be stored, but also need to be uploaded to the cloud or other devices. Therefore, in this embodiment, Huffman coding is used to improve data storage efficiency, reduce storage space occupation, reduce redundant data, and improve upload efficiency.

[0034] In the actual application of the intelligent electricity meter, multiple data items are collected in real time, usually including the following types of information: electricity reading (unit: kilowatt-hour, kWh), power (unit: watt, W), voltage (unit: volt, V), and current (unit: ampere, A). In this embodiment, the data collection frequency of the intelligent electricity meter can be set to 1 Hz, and each collected data record can be represented as a vector:

[0035] ;

[0036] Among them, represents the timestamp of the current record; represents the data vector at the collection moment ; represents the electricity reading at the collection moment ; represents the power at the collection moment ; represents the voltage at the collection moment ; represents the current at the collection moment . In addition to 1 Hz for the above data collection frequency, the implementer can also adjust it to 5 Hz, 10 Hz, 15 Hz, etc. according to the application scenario of the intelligent electricity meter.

[0037] The collected data will be temporarily stored in the electricity meter memory and will be regularly uploaded to the cloud platform through the RS485 interface later. However, since it cannot be guaranteed that the network and communication of the intelligent electricity meter are always connected, in the case where the intelligent electricity meter cannot access the external network, the power data is more dependent on the local storage capacity of the electricity meter, and thus the necessity of improving data storage efficiency is more prominent.

[0038] The above use Indicates the moment All the recorded data, which includes multi-dimensional data such as electric energy, voltage, current, power, etc. Then the data records of the electricity meter will form a continuous data sequence, having:

[0039] ;

[0040] wherein Indicates a data set composed of multi-dimensional data at a total of acquisition moments; Indicates the time stamp sequence; Then indicates the time stamp corresponding multi-dimensional data. After obtaining the original data, it is necessary to perform certain preprocessing on the data, aiming to eliminate redundant information in the data, reduce the influence of outliers, standardize data with different dimensions, and provide a consistent data format for subsequent algorithms.

[0041] First, all the data is normalized. Here, the maximum-minimum normalization can be used for processing the time series of each parameter. It should be noted that the above-mentioned time series of each parameter refers to the time series composed of a total of acquisition moments. An exemplary illustration: The time series of the voltage parameter at a total of acquisition moments can be recorded as ; Then, the moving average method is used to perform a smoothing operation on the time series of each parameter to remove noise. The above-mentioned preprocessing-related algorithms belong to well-known technologies and will not be elaborated here. Denote the data set composed of the preprocessed multi-dimensional data as , having:

[0042] ;

[0043] wherein Indicates the data set composed of the preprocessed multi-dimensional data; Then indicates the data vector corresponding to the time stamp , and having:

[0044] ;

[0045] wherein Indicates the data vector corresponding to the time stamp ; Indicates the preprocessed electric energy reading corresponding to the time stamp ; Indicates the preprocessed power corresponding to the time stamp ; Indicates the preprocessed voltage corresponding to the time stamp ; Indicates the preprocessed current corresponding to the time stamp .

[0046] S2. Periodically analyze the power time series, voltage time series, and current time series based on the autocorrelation function; obtain the neighborhood sequence of the data, calculate the periodicity strength of the data based on the similarity between the neighborhood sequences of different periods; obtain the intensity threshold based on the maximum inter-class variance method to obtain strong periodic data and weak periodic data; complete the intelligent meter data compression based on pulse code modulation, differential pulse code modulation, and Huffman coding.

[0047] In step S1, processed multi-dimensional time series data including electric energy, current, voltage, and power is obtained. However, although these data have been optimized in format and quality, they still occupy a large storage space and require a high bandwidth for uploading. Therefore, to solve this problem, the Huffman coding algorithm can be used as the main compression method in this embodiment.

[0048] S2.1 Periodically analyze the power time series, voltage time series, and current time series based on the autocorrelation function.

[0049] Huffman Coding is an optimal prefix coding algorithm based on the frequencies of data symbols. It assigns codes to the symbols in the input data by constructing an optimal binary tree (Huffman tree). The higher the frequency of a symbol, the shorter the assigned code; the lower the frequency of a symbol, the longer the assigned code. In this way, Huffman coding can effectively reduce the storage space of data.

[0050] In step S1, a data set composed of preprocessed multi-dimensional data is obtained , and when expanded, it can be expressed as:

[0051] ;

[0052] Furthermore, each row of parameters can be expressed as a parameter time series, as follows:

[0053] ;

[0054] where represents the preprocessed electric energy reading time series; represents the preprocessed power time series; represents the preprocessed voltage time series; represents the preprocessed current time series. In this embodiment, if the Huffman coding is directly used to compress the data set , the possible problems are:

[0055] Due to the working principle of the intelligent meter, the electric energy reading is cumulative, that is, the electric energy reading time series If there are a large number of non-repeating electricity readings in [the data], then when directly constructing a Huffman tree based on the frequencies of different electricity readings, a large amount of raw data will be generated. An exemplary illustration is as follows: Suppose there are a total of 10 electricity readings, which are 100, 101, 102, …, 109 respectively. At this time, the scenario for obtaining electricity readings may be the peak electricity consumption period of users, where there are no identical electricity readings. And since the occurrence frequency of each electricity reading is 1, ultimately 10 different Huffman codes will be generated when passing through the Huffman tree, and at this time, the data cannot be effectively compressed. In addition, apart from electricity readings, there will also be a large amount of raw data in other current and voltage data collected by smart meters, which will lead to an increase in the length of Huffman codes and a reduction in compression efficiency.

[0056] Based on the above analysis content, first observe Figure 2 the schematic diagram of electricity reading and power time series with a period of 1 day as shown, which represents the household electricity consumption situation in a general scenario. For the electricity reading time series , it shows an increasing phenomenon in the morning and evening. And for the power time series , it also changes in the morning and shows a sharp increase and a sharp drop in the evening. Further observe Figure 3 the schematic diagram of electricity reading and power time series with a period of 5 days as shown. Although the power time series changes differently every day, with sudden increases and sudden drops, but looking at the overall power time series change curves of each day, due to the certain regularity of the user's own work and rest, the overall power time series change situations of each day are similar. That is to say, in the long run, the change of power time series is periodic. Similarly, for the voltage time series and the current time series also have similar properties, which will not be elaborated here.

[0057] In this embodiment, the power time series , the voltage time series and the current time series can be analyzed periodically first. For the data within a cycle, although there is a certain similarity in the data within each cycle, that is, there will be a large amount of repeated data between the data in one cycle and the data in another cycle. However, considering that the parameter time series such as the power time series change greatly within one cycle, if Huffman coding is directly applied, a large amount of raw data will still be generated. Therefore, in this embodiment, for the data within a cycle, pulse code modulation (PCM) can be used for processing first. Similarly, since the electricity reading time series shows an overall increasing trend, pulse code modulation (PCM) can also be used for processing first.

[0058] Specifically, taking the power time series For example, first, the autocorrelation function is used to measure its similarity with itself at different time lags, which can be expressed as:

[0059] ;

[0060] where represents the autocorrelation function of the power time series at a time lag of , which measures the similarity of the power data at different lag time intervals; represents the total number of data in the power time series ; represents the power value at the time stamp ; represents the mean value of the data in the power time series ; represents the time lag, that is, the time interval from one time point to the next.

[0061] After obtaining the autocorrelation function of the power time series , the time lag value is further dynamically adjusted to find the significant periodicity of the power time series. In this embodiment, based on the user's electricity consumption habits and the data acquisition frequency of the smart meter, the selected values of the time lag value can be 21600 (6h), 43200 (12h), and 86400 (24h). In addition to the above selected values, the implementer can also set the time lag value by himself according to the actual situation, such as 36000 (10h), 64800 (18h), and 108000 (30h), etc.

[0062] Calculate the autocorrelation functions corresponding to each time lag value. Taking the above example, we can get , and . It should be noted that: here is an example for illustration, and only the autocorrelation functions corresponding to 3 time lag values are obtained. In actual applications, more time lag values may be selected. After obtaining all the autocorrelation functions, the time lag value corresponding to the maximum autocorrelation function is used as the period of the power time series and is denoted as .

[0063] S2.2. Obtain the neighborhood sequence of the data, calculate the periodicity intensity of the data based on the similarity between the neighborhood sequences of different periods; obtain the intensity threshold based on the maximum inter-class variance method, and obtain strong periodic data and weak periodic data.

[0064] Next, to quantify the periodicity intensity of each data in the time series data, in this embodiment, based on the data and a period And calculate the neighborhood similarity of the corresponding data after multiple cycles. First, for the power time series the timestamps of the power values , obtain the first 5 power values adjacent to it and the last 5 power values, and get a power data neighborhood sequence containing a total of 11 timestamps, denoted as ; then obtain the corresponding data for several subsequent cycles of the timestamp , which can be respectively denoted as: , where represents the maximum value of the cycle order number included in the power time series , and each of the above acquisition times can also be denoted as a reference time; then obtain the neighborhood sequences of the above data, and the acquisition method is the same as that of the power value , and the specific process will not be elaborated. So far, multiple neighborhood sequences can be obtained, which can be respectively denoted as: , and further calculate the periodicity intensity of the power value corresponding to the timestamp , specifically:

[0065] ;

[0066] where represents the periodicity intensity of the power corresponding to the timestamp ; represents the maximum value of the cycle order number in the power time series ; represents the timestamp the th cycle after the power neighborhood sequence; represents the timestamp the th cycle after the power neighborhood sequence; represents the calculation function of the Dtw distance; represents a tuning parameter factor to avoid the denominator being zero. In this embodiment, an empirical value of 5E-6 can be taken.

[0067] When is smaller, it indicates that the similarity between the neighborhood sequences corresponding to the same position in different cycles is greater, then the power corresponding to the timestamp is more likely to be repeated data within the cycle, and its corresponding periodicity intensity is greater; when is larger, it indicates that the similarity between the neighborhood sequences corresponding to the same position in different cycles is smaller, then the timestamp The corresponding power The less likely the data within the period is, the smaller the corresponding periodic intensity will be

[0068] Obtain the power time series according to the above method for all the data in it, and use the Otsu method to obtain the intensity threshold. Denote the power corresponding to the periodic intensity greater than or equal to the intensity threshold as strong periodic power, and the power corresponding to the periodic intensity less than the intensity threshold as weak periodic power. Observe Figure 3 the power time series change line shown. There is a certain periodicity in the power change in the early and late periods. The power in these periods belongs to strong periodic power. At the same time, as Figure 3 shown, its data is continuous. If on a certain day, an unusual event such as a leave occurs to the user, the generated power data will no longer conform to Figure 3 the periodicity shown. At this time, the power data is weak periodic power. At the same time, when the smart meter collects power, certain errors may occur. At this time, the power data also does not belong to periodic data, so it can be classified as weak periodic power

[0069] To sum up, the weak periodic power is mainly the data generated by some unusual events. That is to say, the proportion of weak periodic power is relatively small and there may be mutations. If pulse code modulation (PCM) is used for processing, not only will more encoded data be generated, but also the encoding efficiency will be reduced. Therefore, in order to save these weak periodic powers, differential pulse code modulation (DPCM) can be used for processing in this embodiment

[0070] So far, the strong periodic power and weak periodic power in the power time series have been obtained. For the voltage time series and the current time series obtain the strong periodic voltage, weak periodic voltage, strong periodic current and weak periodic current in the same way as above

[0071] Similarly, taking the power time series as an example, for the strong periodic power in the power time series , use pulse code modulation (PCM) for processing, and for the weak periodic power in the power time series , use differential pulse code modulation (DPCM) for processing. Then process all the encoded data after modulation using Huffman coding to complete the data compression of the power time series . Correspondingly, for the voltage time series and the current time series complete the data compression in the same way as the power time series . For the electric energy reading time series , it also first uses pulse code modulation (PCM) for processing and then uses Huffman coding to complete data compression.

[0072] S2.3. Complete the data compression of the smart meter based on pulse code modulation, differential pulse code modulation, and Huffman coding.

[0073] Finally, an example is used to illustrate the data storage process of the smart meter:

[0074] Suppose there are 15 consecutive powers: 500(1), 505(2), 510(3), 510(4), 515(5), 510(6), 799(7), 800(8), 520(9), 525(10), 530(11), 535(12), 540(13), 545(14), 540(15); it should be noted that for the convenience of the example, the power values selected here are the actual power values, rather than the power data after the preprocessing in step S1.

[0075] After calculating the periodic intensity of each power and comparing it with the intensity threshold, the strongly periodic powers are: 500(1), 505(2), 510(3), 510(4), 515(5), 510(6), 505(9), 505(10), 530(11), 535(12), 540(13), 545(14), 540(15); and the weakly periodic powers are: 799(7), 800(8). First, use pulse code modulation (PCM) to process all consecutive strongly periodic powers, and the coding result is: 0(1), +++++(2), +++++(3), 0(4), +++++(5), ----- (6), 505(9), 0(10), +++++(11), +++++(12), +++++(13), +++++(14), ----- (15); then use differential pulse code modulation (DPCM) to process all consecutive weakly periodic powers, and the coding result is: +289(7), +1(8). It should be noted that: the 9th power 505 above does not use pulse code modulation (PCM) because the strongly periodic power and the weakly periodic power may not be consecutive. At this time, if the strongly periodic power connected to the weakly periodic power is directly subjected to pulse code modulation (PCM), a large number of symbols will be generated, thus affecting the subsequent coding efficiency. That is to say, the first strongly periodic power following the weakly periodic power retains the original data and does not undergo pulse code modulation (PCM).

[0076] The encoded and modulated result obtained so far is as follows: 0(1), +++++(2), +++++(3), 0(4), +++++(5), ----- (6), +289(7), +1(8), 505(9), 0(10), +++++(11), +++++(12), +++++(13), +++++(14), ----- (15). That is to say, the initial data only includes a total of 6 data as shown below: 0, +++++, -----, +289, +1, 505. The original data without encoding and modulation includes the following data: 500, 505, 510, 515, 530, 535, 540, 545, 799, 800, that is, a total of 10 data. If the Huffman algorithm is directly used to encode the above original data, the encoding length will be significantly longer than the data after encoding and modulation. At the same time, due to the small amount of data given in the above example, when a large amount of data is generated by the smart meter, directly using the Huffman algorithm for the original data without encoding and modulation will result in a longer encoding length. The encoding process of the above Huffman algorithm and the subsequent decoding process of the Huffman algorithm are well-known technologies, and the specific implementation methods will not be elaborated here.

Claims

1. An intelligent electric meter data optimization storage method, characterized in that, The method includes: Obtaining multiple operating parameters of the smart meter, and the data at multiple consecutive acquisition moments of each operating parameter constitute the corresponding parameter time series; Obtaining the autocorrelation functions corresponding to different time lag values of each parameter time series; recording the time lag value corresponding to the maximum value among all the autocorrelation functions as the period; designating any data in the parameter time series as the target data; the acquisition moment corresponding to the target data is the target moment, and recording the acquisition moment corresponding to each period after the target moment as the comparison moment; a set number of data before and after the target moment constitute the neighborhood sequence; Obtaining multiple neighborhood sequences from the neighborhood sequences of each comparison moment and the target moment, and obtaining the periodic intensity of the target data based on the similarity between the multiple neighborhood sequences; obtaining the periodic intensity of all the data in any parameter time series, and in response to the periodic intensity being greater than or equal to the set threshold, designating the corresponding data as strong periodic data; in response to the periodic intensity being less than the set threshold, designating the corresponding data as weak periodic data; Performing pulse code modulation on all the strong periodic data and differential pulse code modulation on all the weak periodic data; compressing all the modulated data using Huffman coding to reduce the storage space of the smart meter data.

2. The intelligent electric meter data optimization storage method according to claim 1, wherein The specific calculation method of the periodic intensity is: ； Among them represents the timestamp corresponding periodic intensity of power; represents the maximum value of the cycle order number in the power time series; represents the timestamp the neighborhood sequence at the reference time after the represents the timestamp the neighborhood sequence at the reference time after the represents the calculation function of the Dtw distance; represents the tuning parameter factor.

3. The intelligent electric meter data optimization storage method according to claim 1, characterized in that, Obtaining multiple operating parameters of the smart meter, including: Collecting the power reading, power, voltage, and current parameters of the smart meter; Setting the frequency range for the smart meter to collect data as [1, 15].

4. The intelligent electric meter data optimization storage method according to claim 3, wherein It also includes temporarily storing each parameter time series in the meter memory and regularly uploading it to the cloud platform through the RS485 interface.

5. The intelligent electric meter data optimization storage method according to claim 3, characterized in that It also includes not calculating the period of the power reading time series, performing pulse code modulation on the power reading time series, and then performing data compression processing using Huffman coding.

6. The intelligent electric meter data optimization storage method according to claim 3, characterized in that It also includes performing preprocessing operations on each parameter time series, specifically: Performing normalization processing on each parameter time series; Performing noise reduction processing on each parameter time series after the normalization processing.

7. A method for optimizing the storage of smart meter data according to claim 6, characterized in that, Using the maximum-minimum normalization algorithm to perform normalization processing on each parameter time series.

8. The intelligent electric meter data optimization storage method according to claim 6, wherein Using the moving average method to perform smoothing processing on each parameter time series after the normalization processing to remove noise.

9. A method for optimizing the storage of smart meter data according to claim 1, characterized in that, Using the maximum inter-class variance method to obtain the set threshold.

Images