CN111064705B

CN111064705B - Data compression acquisition and transmission method suitable for advanced measurement system

Info

Publication number: CN111064705B
Application number: CN201911154356.XA
Authority: CN
Inventors: 袁博
Original assignee: State Grid Corp of China SGCC; Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Current assignee: State Grid Corp of China SGCC; Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Priority date: 2019-11-22
Filing date: 2019-11-22
Publication date: 2022-01-28
Anticipated expiration: 2039-11-22
Also published as: CN111064705A

Abstract

The invention provides a data compression acquisition and transmission method, which comprises the following steps: the intelligent ammeter acquires monitoring data of a first target data type and monitoring data of a second target data type from the acquired monitoring data; the intelligent electric meter performs compression sampling on the monitoring data of the first target data type through a compression sensing time model to obtain first compressed data, and transmits the first compressed data and the monitoring data of the second target data type to the data concentrator; the data concentrator performs compression sampling on the received monitoring data of the second target data type through a compressed sensing space model to obtain second compressed data, transmits the first compressed data and the second compressed data to a measurement data management center, and the measurement data management center performs data reconstruction on the first compressed data and the second compressed data. Compared with the prior art, the invention reduces the sampling frequency and the complexity of the compression process of the sampling end, and reduces the error of the data transmission result under the condition of packet loss or error code.

Description

Data compression acquisition and transmission method suitable for advanced measurement system

Technical Field

The invention relates to the technical field of smart power grids, in particular to a data compression acquisition and transmission method suitable for an advanced measurement system.

Background

Advanced Measurement Infrastructure (AMI) is a set of support processing systems used to measure, collect, store, analyze and utilize user electricity consumption information. As shown in fig. 1, the generalized advanced measurement system is composed of most of 5 parts, namely, a user end/user indoor network (HAN), an intelligent electric meter (acquisition terminal), a data concentrator (data transfer station), a communication network, and a measurement data management center, wherein the intelligent electric meter, the data concentrator, and the measurement data management system are sequentially connected by the data communication network, and data acquisition is realized through sampling, transmission, and the like. Specifically, after the intelligent electric meter acquires data from the HAN, the data are transmitted to the data concentrator through the communication network for simple integration, then the data are further transmitted to the measurement data management center through the communication network, and processing, analysis, display, storage, filing and the like are realized, wherein the measurement data management center has extremely strong data processing capacity and powerful hardware facility support. At present, in an advanced measurement system, the HAN is used as a whole to sample data of the HAN by a smart meter, and a management organization of the advanced measurement system usually cannot or cannot intervene in an internal structure of an indoor network of a user. Therefore, the AMI in a narrow sense includes only three core devices, i.e., a smart meter, a data concentrator, and a measurement data management center, and is connected through a communication network.

The method is used for collecting and transmitting data in a power grid, is a core function of a high-level measurement system, and is also a primary link for realizing other functions. In the future, the data of the advanced measurement system comprises three-phase voltage/current, power consumption, active/reactive power, time-of-use electricity price, electric energy quality parameters, alarm information and the like. Therefore, the collected data in the advanced measurement system is increased explosively, and the storage and transmission of mass data urgently require the compression of the collected data. The existing advanced measurement system usually adopts the traditional data compression methods such as discrete cosine transform, fourier transform, wavelet transform and the like to solve the problem of mass data transmission, however, the existing advanced measurement system has certain defects: (1) the sampling frequency of the sampling end is high, the complexity of the compression process is high, the data processing capacity of the intelligent ammeter and the data concentrator is weak, and the burden of the intelligent ammeter and the data concentrator can be increased seriously; (2) the data transmission result has great error under the condition of packet loss or error code, and the transmission effect of compressed data is poor.

Disclosure of Invention

In view of this, embodiments of the present invention provide a data compression acquisition and transmission method suitable for an advanced measurement system, so as to solve the problems in the prior art that when monitoring data acquisition and transmission are performed based on a conventional data compression method, a sampling frequency at an intelligent electric meter or a data concentrator is high, a compression complexity is high, and a data transmission effect is poor under a packet loss or error code condition.

The first aspect of the embodiments of the present invention provides a data compression acquisition and transmission method, which is suitable for an advanced measurement system including an intelligent electric meter, a data concentrator and a measurement data management center, and includes the steps of:

the intelligent ammeter acquires monitoring data of a first target data type and monitoring data of a second target data type from the acquired monitoring data;

the intelligent ammeter performs compression sampling on the monitoring data of the first target data type through a measurement matrix in the compressed sensing time model to obtain first compressed data, and transmits the first compressed data and the monitoring data of the second target data type to a data concentrator accessed by the intelligent ammeter;

the data concentrator performs compression sampling on the received monitoring data of the second target data type through a measurement matrix in the compressed sensing space model to obtain second compressed data, and transmits the first compressed data and the second compressed data to the measurement data management center;

and the measurement data management center performs data reconstruction on the first compressed data and the second compressed data through a reconstruction algorithm, a sparse basis and a measurement matrix which respectively correspond to the first compressed data and the second compressed data.

A second aspect of the embodiments of the present invention provides a power grid advanced measurement system, which includes an intelligent electric meter, a data concentrator, and a measurement data management center, where the intelligent electric meter, the data concentrator, and the measurement data management center compress and transmit monitoring data by using the data compression acquisition and transmission method described above.

Compared with the prior art, the data compression acquisition and transmission method has the advantages that the monitoring data of a first target data type are subjected to compression sampling at the intelligent electric meter, the monitoring data of a second target data type are subjected to compression sampling at the data concentrator, the compressed first compressed data and the compressed second compressed data are transmitted to the measurement data management center, the measurement data management center performs data reconstruction on the first compressed data and the compressed second data, the data compression acquisition and transmission method is used for acquiring and transmitting the monitoring data in the advanced measurement system of the power grid, the sampling frequency and the complexity of a compression process of a sampling end are reduced to a great extent, the compression complexity is transferred to the measurement data management center with strong data processing capacity, and meanwhile, the error of a data transmission result under the condition of packet loss or error code is reduced; in addition, according to different types of monitoring data, different types of monitoring data are selected to be compressed and sampled at the intelligent electric meter or the data concentrator, and the transmission effect of the monitoring data under each target data type is improved.

Drawings

FIG. 1 is a schematic diagram of an embodiment of a high level measurement system in the prior art;

FIG. 2 is a schematic diagram of an embodiment of a compressed sensing temporal model according to the present invention;

FIG. 3 is a schematic diagram of an embodiment of a compressed perceptual space model according to the present invention;

FIG. 4 is a schematic flow chart illustrating an implementation of a data compression collection and transmission method according to an embodiment of the present invention;

FIG. 5 is a reconstructed SNR for compressive sampling of monitored data using different types of measurement matrices;

FIG. 6 shows sparsity K of monitor data in Discrete Cosine Transform (DCT) basis, Discrete Fourier Transform (DFT) basis, and Discrete Wavelet Transform (DWT) basis;

FIG. 7 is a reconstructed SNR for reconstructing compressed monitor data using different reconstruction algorithms;

fig. 8 is a schematic flow chart illustrating a method for establishing a high-level grid measurement system according to an embodiment of the present invention;

FIG. 9 is a schematic diagram of a compression measurement control in the case of compression sampling of raw data by a measurement matrix;

fig. 10 shows a reconstructed signal-to-noise ratio after compressed acquisition and transmission of monitoring data of an initial target data type by the advanced power grid measurement system of the present invention;

fig. 11 shows reconstructed snrs of the compressive sampling method, the conventional DCT compression method, and the conventional DFT compression method according to the present invention at different transmission packet loss rates.

Detailed Description

In order to explain the technical means of the present invention, the following description will be given by way of specific examples.

Example 1

The embodiment provides a data compression acquisition and transmission method which is suitable for a power grid advanced measurement system. The advanced measurement system of the power grid comprises an intelligent electric meter, a data concentrator and a measurement data management center, wherein the intelligent electric meter is connected with the data concentrator and used for transmitting acquired monitoring data to the corresponding data concentrator, and the data concentrator is connected with the measurement data management center and used for transmitting the monitoring data received from the intelligent electric meter to the measurement data management center.

In a high-level measurement system of a power grid, an intelligent electric meter and a data concentrator are influenced most by mass data, and network congestion is most likely to occur. In order to relieve network congestion, the data compression acquisition and transmission method performs compression sampling at the intelligent electric meter or the data concentrator, namely, the intelligent electric meter and the data concentrator are used as cores for realizing compression sampling of monitoring data, and compressed data after compression sampling is transmitted to the measurement data management center, the measurement data management center is used for performing data reconstruction on the compressed data to obtain original data before compression sampling, and the data compression and transmission method in the advanced measurement system comprises three main processes of compression sampling, data transmission and data reconstruction.

Specifically, compressive sampling, data transmission and data reconstruction are three major processes of compressive sensing. Compressed sensing is a novel theory for accurately reconstructing an original signal with high probability by using a small number of random projection values under the condition that the signal satisfies sparsity. The basic idea of compressed sensing is: n-dimensional (i.e. Nx 1-dimensional) original signal x satisfies the sparse basis Ψ ∈ R in a certain Nx N-dimension^M×NWhen the lower sparse vector s is sparse, the matrix phi epsilon R can be measured through the dimension of M multiplied by N^M×NCarrying out linear dimensionality reduction observation (namely compression sampling) on the original signal x to obtain an M-dimensional (namely M multiplied by 1-dimensional) compression measurement signal y (M)<<N), then the compressed data is transmitted and the original signal x is reconstructed using a compressed sensing reconstruction algorithm. Press and pressThe mathematical model of down-sampling is:

y＝Φx＝ΦΨs＝Θs (1)

in the formula (1), Θ ═ Φ Ψ is referred to as a sensing matrix, the number K of non-zero elements of the sparse vector s is referred to as the sparsity of the original signal x under the sparsity basis Ψ, and M/N is the compression ratio of the compressed sampling. The measurement matrix, the sparse basis and the reconstruction algorithm are three elements of compressed sensing, meanwhile, the compression ratio M/N is also an important factor needing to be considered in practical application of the compressed sensing, and the core parameters of the specific measurement matrix are constructed according to the type of the measurement matrix. Therefore, the compressed sensing has four application elements in practical application, namely a measurement matrix type, a sparse basis, a reconstruction algorithm and a compression ratio, and a compressed sensing model can be constructed according to the measurement matrix (generated by the measurement matrix type and the compression ratio), the sparse basis and the reconstruction algorithm.

Specifically, in the advanced measurement system of the power grid, the smart meter has two parameters, namely, a raw data sensing (acquisition) time interval T in each transmission₀And the total sensing (acquisition) duration T of the original data, wherein T represents that the intelligent electric meter transmits the obtained monitoring data (the original data or the compressed data) to the data concentrator at intervals of T, and T₀Within the expression time T, the intelligent electric meter is arranged at intervals of time T₀And carrying out data perception once. Since the monitoring data at the smart meter is a time sequence, a compressed sensing model for performing compressed sampling at the smart meter is proposed to be a compressed sensing time model, and the compressed sensing time model is shown in fig. 2. As shown in fig. 2, in the compressed sensing time model, a measurement matrix (generated by the type of the measurement matrix and the compression ratio) is built in the smart meter, the measurement matrix, a reconstruction algorithm and a sparse basis are built in the measured data management center, the smart meter performs compressed sampling on monitoring data suitable for the compressed sensing time model according to the corresponding measurement matrix, transmits the compressed sampling data to the measured data management center through the data concentrator, and reconstructs the data according to the reconstruction algorithm. In the compressed sensing time model, the monitoring data sensed by the intelligent electric meter before each compressed sampling is recorded as an original time sequence, and the length of the original time sequence is recorded as the length N of the original time sequence_TIn which N is_T＝T/T₀。

Specifically, in the advanced power grid measurement system, each data concentrator has a plurality of directly connected smart meters as the smart meters managed by the data concentrator. The data concentrator is used for transferring and integrating monitoring data uploaded by the intelligent electric meters in a management range at a certain moment, and transmitting the monitoring data to the measurement data management center, wherein the number Ns of the intelligent electric meters managed by the data concentrator is a main parameter of the data concentrator. Since the monitored data (original data or compressed data) at the data concentrator is a spatial sequence, it is proposed to label a compressed sensing model for performing compressed sampling at the data concentrator as a compressed sensing spatial model, which is shown in fig. 3. As shown in fig. 3, in the compressed sensing space model, a measurement matrix (generated by the type of the measurement matrix and the compression ratio) is built in a data concentrator, the measurement matrix, a reconstruction algorithm and a sparse basis are built in a measured data management center, and the data concentrator performs compressed sampling on monitoring data suitable for the compressed sensing space model according to the corresponding measurement matrix, transmits the sampled data to the measured data management center, and performs reconstruction according to the corresponding reconstruction algorithm. In the compressed sensing space model, monitoring data obtained by a data concentrator before each compressed sampling is recorded as an original space sequence, and the length of the original space sequence can be expressed as the number Ns of smart meters in a management range of the data concentrator.

Based on this, in this embodiment, the compressed sensing model includes a compressed sensing time model and a compressed sensing space model, the compressed sensing time model is used for performing compressed sampling on the monitoring data at the smart meter, and the compressed sensing space model is used for performing compressed sampling on the monitoring data at the data concentrator.

As shown in fig. 4, the data compression collection and transmission method in this embodiment includes the following steps:

step S101, the intelligent ammeter acquires monitoring data of a first target data type and monitoring data of a second target data type from the acquired monitoring data.

In this embodiment, in order to implement a data compression acquisition and transmission method that can be used in a power grid advanced measurement system, monitoring data needs to be classified first. Specifically, the first target data type is an initial target data type which is suitable for the compressed sensing time model and is selected from a plurality of initial target data types obtained by preliminarily classifying monitoring data according to data characterization information, the type of the intelligent electric meter, the monitoring time period and data smoothness; the second target data type is an initial target data type which is suitable for the compressed sensing space model and is in a plurality of initial target data types obtained by preliminarily classifying the monitoring data according to data characterization information, the type of the intelligent electric meter, the electricity consumption monitoring time period and data smoothness.

Specifically, the step of classifying the monitoring data to obtain the first target data type and the second target data type includes the following steps S1011 to S1013:

and S1011, classifying the monitoring data obtained from the power grid according to the data characterization information, the type of the intelligent electric meter and the power consumption monitoring time period to obtain multiple types of monitoring data. In this embodiment, the data characterization information includes three classifications of power consumption, power, voltage, current and power quality, the smart meter type includes six classifications of city residential electric meters, county or rural residential electric meters, heavy industrial factory electric meters, light industrial factory electric meters, office buildings or service electric meters and microgrid electric meters, the power consumption monitoring period includes two classifications of power consumption peak period and power consumption valley period, and therefore, the monitoring data is preliminarily classified according to the data characterization information, the smart meter type and the power consumption monitoring period, and 36 classifications can be obtained.

Specifically, when monitoring data are classified according to data characterization information, the type of the intelligent electric meter and the electricity utilization monitoring time period, the process is as follows:

firstly, dividing monitoring data according to data representation information and types of the intelligent electric meters to obtain U multiplied by V classifications, wherein U is the number of the data representation information, V is the number of the types of the intelligent electric meters, and the types of the intelligent electric meters are divided according to installation positions. In this embodiment, the monitoring data is divided into 3 types according to the data characterization information, that is, U is 3, where the I type is power consumption and power, the II type is three-phase voltage and current, and the III type is power quality data, where the power quality data mainly includes information such as harmonic, voltage, and frequency offset; meanwhile, the monitoring data is divided into 6 types according to the type of the smart meter (namely, the mounting position of the smart meter), namely, V is 6, wherein the 1 st type is an urban residential meter, the 2 nd type is a county city or rural residential meter, the 3 rd type is a heavy industrial factory meter, the 4 th type is a light industrial factory meter, the 5 th type is an office building or a service industry meter, and the 6 th type is a micro-grid meter, wherein the micro-grid meter can contain smart meters such as a distributed renewable energy source and a novel load. Referring to table 1, table 1 is a monitoring data classification table when the monitoring data is divided into U × V classifications, as shown in table 1, a total of 18 classifications are obtained.

TABLE 1 monitoring data Classification Table

Then, the monitoring data under the U × V classifications are divided again according to the electricity consumption monitoring time interval to obtain U × V × W classifications, the monitoring data under the U × V × W classifications in a preset period are obtained, and whether the monitoring data under each classification meet preset sparsity conditions or not is sequentially judged, wherein W is the number of the electricity consumption monitoring time intervals. According to the relationship between the characteristics of the electricity consumption monitoring information and the time, the monitoring data in one day is generally divided into two electricity consumption monitoring periods, namely an electricity consumption peak period and an electricity consumption valley period, wherein W is 2. In some embodiments, the peak electricity consumption time is 7: 00-23: 00, and the valley electricity consumption time is 23: 00-7: 00 of the next day. Therefore, according to the data characterization information, the type of the smart meter and the electricity consumption monitoring period, the advanced measurement system obtains 36 types of monitoring data from the power grid.

Finally, qualitative statistical analysis is performed on whether the monitoring data under each classification is suitable for compressed sensing, and statistical data in the embodiment is obtained from an existing power grid measurement system. Specifically, the statistical data includes monitoring data obtained by V smart meters in a preset period and monitoring data obtained by at least one data concentrator in the preset period, where the sum of the types of the smart meters in the management range of all the data concentrators is V. That is, in an embodiment, one smart meter of 6 types may be selected, and then at least one data concentrator is selected, for example, 3 data concentrators are selected, where the 3 data concentrators are respectively marked as a data concentrator a, a data concentrator B, and a data concentrator C, where the smart meters in the management range of the data concentrator a include two types of urban residential meters, office buildings, or service electric meters, the smart meters in the management range of the data concentrator B include two types of county or rural residential meters and microgrid electric meters, the smart meters in the management range of the data concentrator C include two types of heavy industrial plant electric meters and light industrial plant electric meters, and finally, the monitoring data in the preset period at the 6 smart meters and the 3 data concentrators are used as statistical data. Meanwhile, since the monitoring data generally changes periodically on a daily basis, in order to accurately analyze the sparsity of the monitoring data, it is preferable that the preset period is 24 hours. The analysis of the statistical data shows that: (a) for the I-type power consumption and power, in time (namely in the same intelligent electric meter), the monitoring data collected by the intelligent electric meter changes obviously along with the peak valley of each day, the time sequence of the power consumption peak period shows small fluctuation, but the data at the power consumption valley stage is almost unchanged, but no matter the power consumption peak period or the power consumption valley period, the monitoring data does not change suddenly in a certain shorter period, the time domain data changes smoothly, for different types of intelligent electric meters, the absolute value difference of the data is larger, but no matter which type of intelligent electric meter, the relative change of the time sequence is very similar, and the smoothness of the instant domain data is irrelevant to the type of the intelligent electric meter; spatially (namely in the same data concentrator), data collected by the intelligent electric meters in the data concentrator at the same moment has certain fluctuation, wherein the data present smaller general fluctuation in the peak period of power utilization, and the data present smaller general fluctuation in the valley period of power utilization, but the variation among the data of the same type of intelligent electric meters is kept in a certain range, and the data present smaller general fluctuation only in the range, and for different monitoring moments, the absolute value difference of the data is larger, but no matter what monitoring moment, the spatial sequence smoothness presents similarity. (b) For voltage and current data of the type II, whether the data sequence in time or the data sequence in space, whether the peak time period or the valley time period of power utilization, the elements of the data sequence in the same voltage class range are almost unchanged, the data collected in the intelligent electric meter are smoothly changed, the data collected by the data concentrator also show smooth fluctuation in a certain small range, namely the data are almost irrelevant to the power utilization monitoring time period and the type of the intelligent electric meter, and the smoothness difference between the time sequence and the space sequence is small. (c) For the type III electric energy quality data, different electricity utilization monitoring time periods are different, in the electricity utilization peak time period, the space sequence and the time sequence both show obvious fluctuation, the fluctuation of the original space data sequence formed by the intelligent electric meters of the same type is slightly smaller than that of the original time data sequence, the fluctuation of the original space data sequence formed by the intelligent electric meters of different types is slightly larger than that of the original time data sequence, for the intelligent electric meters of different types, the absolute value difference of the data is larger, but the relative change curves of the data are very similar, and the instant domain data is irrelevant to the type of the intelligent electric meters; and in the electricity consumption valley period, the general fluctuation is only small no matter the time sequence or the space sequence, and the smoothness of the time sequence and the space sequence is similar and is not related to the type of the intelligent electric meter.

Comprehensive analysis shows that the space sequence and the time sequence of the class III data (power quality data) in the peak period of power utilization have large general fluctuation; the space sequence of the I type data (power consumption and power) in the peak period of the electricity utilization and the space sequence and the time sequence of the III type data (power quality data) in the valley period of the electricity utilization have small general fluctuation; the time sequence of the class I data (power consumption and power) in the peak period of power consumption and the space sequence of the class I data (power consumption and power) in the valley period of power consumption show relatively smooth fluctuation; the changes of the rest of the data sequences (spatial sequence and time sequence) have good smoothness. Therefore, the five types of data all show good smoothness except that a few types of monitoring data have general fluctuation, and are all suitable for compressed sensing, namely the compressed sensing has universality in an advanced measurement system.

Step S1012, performing classification and integration on the 36 classified monitoring data according to the smoothness of the monitoring data, and acquiring 5 types of monitoring data of the initial target data type.

According to the analysis result of the smoothness of the U × V × W-class monitoring data obtained by the compressed sensing applicability analysis in the step S1011, integration is performed according to the smoothness characteristics of various types of data, and the U × V × W classified monitoring data are further classified to obtain an initial target data type. The type of the intelligent electric meter has almost no influence on the data smoothness, so that the influence of the type of the intelligent electric meter can be ignored during the initial data classification, and the initial target data type is obtained by integrating according to the electricity utilization monitoring time period, the data representation information and the data smoothness. In this embodiment, 36 types of monitoring data are divided into 5 initial target data types, as shown in table 2. The initial target data types comprise I-A, I-B, II-AB, III-A and III-B, wherein I represents electricity consumption and power of the six types of smart meters, II represents voltage and current of the six types of smart meters, III represents power quality of the six types of smart meters, A represents monitoring data of the six types of smart meters in electricity consumption peak periods, and B represents monitoring data of the six types of smart meters in electricity consumption valley periods. The original time data sequence refers to monitoring data obtained by the intelligent electric meters within a period of time, and the original space data sequence refers to monitoring data obtained by the data concentrator at a certain moment and coming from all the intelligent electric meters within the management range of the data concentrator.

TABLE 2 initial target data types

Then, the sparsity of the monitoring data under the 5 initial target data types is verified and judged. The monitoring data under 5 types of initial target data types can be tested for multiple times by taking a discrete cosine transform basis as a sparse basis, and then whether K/N < a set value (such as 20%) is satisfied, where N represents the length of an original signal (the length of an original time data sequence or the length of an original space data sequence), such as N being 100, and K represents the sparsity of the monitoring data under the sparse basis. By the above determination, it can be seen that all of the initial target data types in table 2 satisfy: K/N < a set value (such as 20%), namely the monitoring data under the 5 types of initial target data have good sparsity and are suitable for a compressed sensing model. Therefore, according to judgment, the data acquired by all advanced measurement systems have good smoothness and meet the preset sparsity, and all types of monitoring data are suitable for the compressed sensing model.

Step S1013, according to the original data length and the original data smoothness information of the monitoring data of each initial target data type, determining the corresponding relation between the monitoring data of each initial target data type and the compressed sensing time model and the compressed sensing space model, and obtaining a first target data type and a second target data type.

The higher the sparsity of the compressed and sensed original data under the sparse basis is, the better the data reconstruction effect is, and the better the stability of the original data is, the longer the length is, and the higher the sparsity is. Therefore, for monitoring data under different initial target data types, two factors of 'good smoothness of data sequence of original signal' and 'large data length of original signal' should be considered comprehensively when selecting the compressed sensing model. The original data length of the monitoring data comprises an original time sequence length and an original space sequence length, the original time sequence length is a ratio of total data sensing (acquisition) time length to time interval of data sensing (acquisition) when the intelligent electric meter transmits the monitoring data of the initial target data type every time, the original space sequence length is the number of the intelligent electric meters connected with the data concentrator, and a plurality of space sequence lengths corresponding to all the data concentrators form a space sequence length interval. The original data smoothness includes original time sequence smoothness and original space sequence smoothness, wherein the original time sequence smoothness is the smoothness of the data sequence formed by the monitoring data of the original target data type at the intelligent electric meter, and the original space sequence smoothness is the smoothness of the data sequence formed by the monitoring data of the original target data type at the data concentrator.

Specifically, when the corresponding relationship between the monitoring data of each initial target data type and the compressed sensing time model and the compressed sensing space model is determined according to the original data length and the original data smoothness information of the monitoring data of each initial target data type:

first, the original time series smoothness, original spatial series length interval, and original time series length of all the initial target data types I-A, I-B, II-AB, III-A, III-B are analyzed.

Then, when the original time sequence smoothness and the original space sequence smoothness of the monitoring data of the initial target data type meet the similar conditions, determining the corresponding relation between the initial target data type and the compressed sensing time model and the compressed sensing space model according to the original space sequence length and the original time sequence length. Specifically, if the original time sequence length is greater than the maximum value of the space sequence length interval, the initial target data type corresponds to the compressed sensing time model; if the original time sequence length is smaller than the minimum value of the space sequence length interval, the initial target data type corresponds to the compressed sensing space model; if the original time sequence length is within the space sequence length interval, the compressed sensing reconstruction effect of the monitoring data of the initial target data type in different original time sequence lengths and original space sequence lengths is specifically analyzed, the data smoothness and the data length are comprehensively considered, the compressed transmission effect of the original time sequence and the original space sequence is compared by using a sampling statistical analysis method, and finally an applicable compressed sensing model is determined.

And finally, when the original time sequence length and the original space sequence length of the monitoring data of the original target data type meet the same condition (for example, the length difference is within a certain range), determining that the original target data type corresponds to the compressed sensing time model or the compressed sensing space model according to the original time sequence smoothness and the original space sequence smoothness. Specifically, if the original time sequence smoothness is superior to the original space sequence smoothness, it is determined that the initial target data type corresponds to the compressed sensing time model, and if the original space sequence smoothness is superior to the original time sequence smoothness, it is determined that the initial target data type corresponds to the compressed sensing space model.

In the specific embodiment, a compressed sensing temporal model and a compressed sensing spatial model related to two factors, namely, "the data sequence of the original signal is smooth" and "the data length of the original signal is large" are explained in detail first.

(a) The compressed sensing time model (as shown in fig. 2) is that monitoring data is compressed and sampled by using a linear dimensionality reduction mode at each smart meter, a compressed time data sequence obtained by the smart meter is a compressed measurement signal y in formula (1) (a compressed sampling mathematical model), and at this time, the length N of an N × 1-dimensional original time sequence x is the length N of the original time sequence_T. In the advanced measurement system of the power grid, a smart meter performs data centralized processing and transmission every 2 hours, namely the total sensing time T of original data transmitted by the smart meter every time is 120 min; and the original data perception time interval T of the intelligent electric meter₀For different monitoring data types, T is divided by T₀I.e. the original time series length N_T。

(b) The compressed sensing space model (as shown in fig. 3) is that a linear dimensionality reduction method is used for performing compressed sampling on acquired collection values of all the smart meters in a management range of each data concentrator at a certain moment. The length of the original space sequence at the data concentrator is the number N of the intelligent electric meters in the management range_SThe original spatial data sequence is the original signal x in equation (1) (compressive sampling mathematical model), and the length N of the N × 1 dimensional original signal x is the length N of the original spatial data sequence_S。

In both the compressed sensing time model and the compressed sensing space model, N represents the original data length. Specifically, in the compressed sensing time model, N is the original time sequence length N_T＝T/T₀(ii) a In the compressed sensing space model, N is the number N of the smart meters in the range of the data concentrator_S. Thus, N is mainly influenced by three parameters of the smart meter and the data concentrator, respectivelyOriginal data sensing total duration T and original data sensing time interval T₀Number N of intelligent electric meters in management range of data concentrator_S. In this embodiment, the settings of three main parameters of the smart meter and the data concentrator are shown in table 3.

TABLE 3T, T for each initial target data type in the advanced metrology system₀、N_S

In one embodiment, the smoothness information of the monitored data under different initial target data types is shown in table 2. From the original time-series smoothness and original space-series smoothness of the 5 types of initial target data types I-A, I-B, II-AB and III-A, III-B analyzed in Table 2, it can be seen that: the original time sequence and the original space sequence of II-AB and III-B have similar smoothness; the original time sequence smoothness of I-A, I-B is superior to the original spatial sequence smoothness; the original time sequence and the original space sequence of the III-A fluctuate within a certain range, the data smoothness is general, the smoothness of the original time sequence and the smoothness of the original space sequence of the two types of initial target data types are similar, and in the original space data sequence, the fluctuation of the original space data sequence formed by different types of intelligent electric meters is slightly larger than that of the original space data sequence formed by the same type of intelligent electric meters; the smoothness of the III-A is related to the connection relation of the intelligent electric meters, the volatility of the original space data sequence formed by the intelligent electric meters of the same type is slightly smaller than that of the original time data sequence, and the volatility of the original space data sequence formed by the intelligent electric meters of different types is similar to that of the original time data sequence.

Therefore, in this embodiment, the influence factor of "the data length of the original signal is large" needs to be analyzed. Original time series length N_TThe longer the sum of the original spatial sequence length NsThe larger the better. In this embodiment, based on the comprehensive consideration of the transmission network and other factors, the number Ns of the smart meters in the management range of the data concentrator is 40 to 100, i.e. the length interval of the original spatial sequence is [40,100]]. For the monitoring data under each initial target data type, if T₀The time sequence length is more than or equal to 3min, and the original time sequence length is inevitably smaller than the minimum value of the original space sequence length interval; if T₀Less than or equal to 1.2min, the length of the original time sequence is inevitably larger than the maximum value of the length interval of the original space sequence. Thus, when the smoothness of the monitored data is similar at the smart meter and the data concentrator, T₀Compression sampling is carried out on monitoring data in the initial target data type of more than or equal to 3min through a compressed sensing space model, and T₀The monitoring data under the initial target data type less than or equal to 1.2min is subjected to compression sampling through a compression sensing time model for 1.2min<T₀<The monitoring data under the initial target data type of 3min was determined by statistical analysis and taking into account data smoothness.

In this embodiment, tables 2 and 3 are analyzed comprehensively, and the compressed sensing model is selected as follows:

(a) and (5) selecting a compressed sensing time model from the monitoring data under the initial target data types II-AB and I-A.

(b) And (5) selecting a compressed sensing space model from the monitoring data under the initial target data type III-B.

(c) Monitoring data under the initial target data type I-B can adopt a compressed sensing time model or a compressed sensing space model. And determining the correspondence of the compressed sensing model by statistically analyzing the effect under different original data lengths. If the intelligent electric meters in the management range of the data concentrator are of the same type and the number of the intelligent electric meters is more than 60, preferentially selecting a compressed sensing space model; if the smart meters in the management range of the data concentrator are different types and the number of the smart meters is more than 70, the compressed sensing time model is preferably selected. It should be noted that, when the smart meters in the management range of the data concentrator are of the same type and the number of the smart meters is greater than 60, the compressed sensing spatial model is preferentially selected, and when the smart meters in the management range of the data concentrator are of different types and the number of the smart meters is greater than 70, the compressed sensing temporal model is preferentially selected, the compressed sensing spatial model is obtained by performing statistical analysis on the monitoring data in the initial target data type I-B, and this selection condition may also be adjusted to other conditions according to actual conditions.

(d) Monitoring data under the initial target data type III-A can adopt a compressed sensing time model or a compressed sensing space model. And determining the correspondence of the compressed sensing model by statistically analyzing the effect under different original data lengths. If the intelligent electric meters in the management range of the data concentrator are of the same type and the number of the intelligent electric meters is more than 80, preferentially selecting a compressed sensing space model; and if the smart meters in the data concentrator management range are of different types, selecting a compressed sensing time model (when the smart meters in the data concentrator management range are of different types, the smoothness of the original spatial data sequence is the same as that of the original temporal data sequence, but the length of the original temporal data sequence is necessarily larger than that of the original spatial sequence). It should be noted that, when the smart meters in the management range of the data concentrator are of the same type and the number of the smart meters is greater than 80, the compressed sensing space model is preferentially selected, which is obtained by performing statistical analysis on the monitoring data in the initial target data type III-a, and this selection condition may be adjusted to other conditions according to actual conditions.

Step S102, the smart electric meter compresses and samples the monitoring data of the first target data type through a measurement matrix in the compressed sensing time model to obtain first compressed data.

Specifically, a measurement matrix (generated by a measurement matrix type and a compression ratio) of a compressed sensing time model corresponding to a first target data type is built in the smart meter, and a first compression measurement control method is generated based on the measurement matrix, and the first compression measurement control method is used for controlling the smart meter to perform compression sampling on monitoring data of the first target data type to obtain first compression data. The measurement matrix type is determined according to the operation complexity and the data reconstruction effect of the monitoring data of each initial target data type under different measurement matrix types, and the measurement matrix type is a binary sparse measurement matrix in the embodiment; the compression ratio is determined according to the data reconstruction effect required by the monitoring data of each initial target data type, and the compression ratio is sequentially determined according to the lowest compression ratio, the required compression ratio and the actual compression ratio.

Specifically, step S102 includes the following steps S1021 to S1023:

and step S1021, reading the measurement matrix built in the intelligent ammeter in advance by the intelligent ammeter.

The key of reading the measurement matrix by the intelligent ammeter is to determine the specific measurement matrix type and the compression ratio, and to arrange the measurement matrix type and the compression ratio in the intelligent ammeter in advance. No matter the compressed sensing time model or the compressed sensing space model, a measurement matrix, a sparse basis and a reconstruction algorithm are all three elements of compressed sensing, wherein the compressed sampling process in the compressed sensing time model is realized through the measurement matrix built in the smart meter. Specifically, the measurement matrix is generated by the measurement matrix type and the compression ratio, and the measurement matrix type and the compression ratio are only related to the initial target data type and are not related to the compressed sensing model. Therefore, to determine the measurement matrix type and the compression ratio, the following steps are performed:

firstly, the operation complexity of a plurality of types of measurement matrixes and the data reconstruction effect when the monitoring data under the initial target data type are compressed and sampled are analyzed, and the measurement matrix type suitable for the monitoring data under all the initial target data types is selected according to the analysis result. The measurement matrix type is the key for realizing compression sampling, and different measurement modes not only influence the data reconstruction effect, but also influence the realization of a sampling end. The measurement matrix types commonly used at present comprise a plurality of random measurement matrix types such as a Gaussian measurement matrix, a partial Hadamard measurement matrix, a partial Fourier measurement matrix, a Toeplitz matrix and the like, wherein the Gaussian measurement matrix is widely applied to theory with good performance. Compared with the measurement mode, the binary sparse measurement matrix has the advantages of being small in storage capacity, low in sampling complexity, easy to update and the like. In order to select the measurement matrix type suitable for the monitoring data under each initial target data type, different types of measurement matrices are needed to be adopted to perform compression sampling on the monitoring data of each initial target data type, when a compression sampling test is performed, the sparse basis, the reconstruction algorithm and the compression ratio are all set to be the same, and only the measurement matrix types are different. Referring to fig. 5, fig. 5 shows reconstructed snrs when different types of measurement matrices are used to perform compression sampling on target data, where the reconstructed snrs are used to represent data reconstruction effects of the compression sampling, 50 pieces of monitoring data are respectively selected in 5 types of initial target data types during a test, and the reconstructed snrs corresponding to the compression ratios in each measurement matrix type in fig. 5 are obtained by averaging the reconstructed snrs obtained according to 350 pieces of data. In the test, discrete cosine transform (DFT) bases are adopted for sparse bases in the compressed sensing model, an Orthometric Matching Pursuit (OMP) algorithm is adopted for a reconstruction algorithm, and the types of measurement matrixes are a Gaussian random measurement matrix, a partial Hadamard measurement matrix, a Toeplitz measurement matrix, a Partial Discrete Cosine Transform (PDCT) measurement matrix, a partial Fourier transform (PFFT) measurement matrix and a binary sparse measurement matrix. As can be seen from fig. 5, the binary sparse measurement matrix and the gaussian random measurement matrix both have a better sampling effect, and are significantly better than other measurement matrix types. However, due to the extreme sparsity of the binary sparse measurement matrix, the hardware implementation difficulty and the measurement operation complexity of the binary sparse measurement matrix are obviously superior to those of the gaussian measurement matrix. Therefore, in this embodiment, for the monitoring data of each initial target data type, a binary sparse matrix is selected as the measurement matrix type.

And then, selecting the compression ratio applicable to the monitoring data under each initial target data type according to the data reconstruction effect. In compressed sampling sensing, the compression ratio directly influences the reconstruction or decompression precision of a signal, and the process of determining the compression ratio is as follows: firstly, acquiring a preset value of a reconstructed signal-to-noise ratio, a preset compression ratio interval and a preset compression ratio margin; secondly, analyzing the compression ratio of the monitoring data of each initial target data type when the reconstruction signal-to-noise ratio is a reconstruction signal-to-noise ratio preset value, and recording the compression ratio obtained by analysis as a lowest compression ratio; thirdly, if the lowest compression ratio is smaller than the lowest value of the preset compression ratio interval, updating the lowest compression ratio to the lowest value of the preset compression ratio interval, and if the lowest compression ratio is larger than the maximum value of the preset compression ratio interval, verifying the accuracy of the analysis process of the lowest compression ratio and not updating the lowest compression ratio when the analysis process of the lowest compression ratio is accurate; fourthly, calculating a required compression ratio corresponding to the lowest compression ratio according to a preset compression ratio margin; and fifthly, when the compression ratio of each initial target data type has a unification requirement, taking the required compression ratio with the largest value in the required compression ratios corresponding to each initial target data type as the actual compression ratio of each initial target data type, and when the compression ratio of each initial target data type does not have the unification requirement, taking the required compression ratio of each initial target data type as the actual compression ratio of each initial target data type, wherein the actual compression ratio is the compression ratio set in the compression sensing time model or the compression sensing space model corresponding to the initial target data type.

According to the related research of compressed sensing, the reconstruction signal-to-noise ratio is larger than 40dB, and the application effect of compressed sensing can be ensured. Also, when selecting the compression ratio, the analysis should be made in conjunction with the trend of change in the compression ratio. When the compressive sensing model is constructed, the compression ratio is reasonably selected according to the requirement of specific data reconstruction effect and the data reconstruction effect statistical analysis conclusion of the collected data in the region. Without specific requirements, it can be set according to the minimum requirement of 40dB and the compression ratio recommendation between 25% and 50%. In a preferred embodiment, the compression ratio is determined according to the data reconstruction effect required by the monitoring data of each initial target data type, and in an embodiment, the determination process of the compression ratio comprises the following steps:

(a) acquiring a reconstructed signal-to-noise ratio preset value of 40dB, a preset compression ratio interval [ 25%, 50% ] and a preset compression ratio margin of 10%;

(b) according to the requirements that the total number of the samples to be statistically analyzed is not less than 1050 and the total number of the samples to be statistically analyzed is not less than 150 for each initial target data type, the present embodiment performs statistical analysis on 1400 data samples, wherein the number of the samples for each initial target data type is 200. Analyzing the compression ratio of the monitoring data of each initial target data type when the reconstructed signal-to-noise ratio is a preset value of 40dB, and recording the compression ratio of each initial target data type (II-AB, I-A, I-B, III-A, III-B) obtained by analysis as the lowest compression ratio, as shown in Table 4;

(c) judging whether the lowest compression ratio is in a preset compression ratio interval [ 25%, 50% ], if the lowest compression ratio is smaller than the minimum value of the preset compression ratio interval by 25%, updating the lowest compression ratio to the minimum value of the interval, and if the lowest compression ratio is larger than the maximum value of the preset compression ratio interval by 50%, verifying the accuracy of the lowest compression ratio analysis process, and not updating the lowest compression ratio on the premise of ensuring the accuracy; judging that the lowest compression ratio of each initial target type data does not need to be updated;

(d) the required compression ratio corresponding to the lowest compression ratio, that is, the required compression ratio is calculated from the preset compression ratio margin 10%, that is, the required compression ratio is the lowest compression ratio x (1+ 10%). In the embodiment, a certain error is allowed in the lowest compression ratio, so that a preset margin is set, and the required compression ratio of each initial target data type is shown in table 4;

(e) when the compression ratio of each initial target data type has a unification requirement, taking the required compression ratio with the largest value in the required compression ratios corresponding to each initial target data type as the actual compression ratio of each initial target data type, and when the compression ratio of each initial target data type does not have the unification requirement, taking the required compression ratio of each initial target data type as the actual compression ratio of each initial target data type; in this embodiment, for convenience of construction of the compressed sensing model, solution equations of the compression measurement device and the reconstruction algorithm are unified, and for different initial target data types (5 types in total, I-A, I-B, II-AB and III-A, III-B), a unified compression ratio is selected, that is, the compression ratios of the monitored data under the various initial target data types are all uniformly set according to the maximum required compression ratio, and the actual compression ratio is uniformly set to 36%.

TABLE 4 correspondence table of each initial target data type and compression ratio

Step S1022, on the basis of the correspondence between the initial target data type and the compressed sensing model, according to the built-in measurement matrix (generated by the binary sparse matrix type and the compression ratio), a first compression measurement control method for performing compression sampling on the monitoring data of the first target type is generated at the smart meter.

Specifically, the first compression measurement control method controls the smart meter to perform compression sampling on the monitoring data in the first target data type. In the compression sampling, the control signal in the first compression measurement control method is expressed as P_j(i) Representing the M x N dimensional measurement matrix phi and the control signal P_j(i) The relationship of (a) to (b) is as follows:

wherein when the control signal P_j(i) When the value is 1, controlling a switch j to be opened and reading the ith element value x of the N multiplied by 1 dimensional original time sequence x sensed by the intelligent electric meter at a certain time_iOtherwise, the switch is closed and the read value is zero. In an embodiment, i-1, 2,3 … N denotes the index of an element in the original temporal sequence, where N-N in the compressed perceptual temporal model_T＝T/T₀(ii) a j is 1,2,3 … M, and represents the control switch label, M control switches, M/N is the actual compression ratio is 36%, each time M switches simultaneously to the ith element value x of the original time sequence x_iAnd (5) controlling.

In step S1023, the smart meter performs compression sampling on the monitoring data of the first target data type according to the generated first compression measurement control method to obtain first compression data.

And S103, the intelligent electric meter transmits the first compressed data and the monitoring data of the second target data type to a data concentrator accessed by the intelligent electric meter.

In this embodiment, in order to adapt to compression sampling of monitoring data at the data concentrator, a corresponding relationship between the data concentrator and the smart meter needs to be set, that is, a connection relationship between the smart meter and the data concentrator in the advanced power grid measurement system needs to be set. Wherein, smart electric meter and data concentrator's relation of connection satisfies: a data concentrator is connected with a plurality of intelligent electric meters, one intelligent electric meter is connected with one data concentrator, and intelligent electric meters of the same type are preferentially connected with the same data concentrator. Specifically, through the sparsity analysis of the monitoring data of the initial target data type, the spatial data sequence of the intelligent electric meters of the same type presents relatively gentle volatility, and therefore it is very important to reasonably set the corresponding relation between the data concentrator and the intelligent electric meters. In the advanced measurement system of the power grid, based on the consideration of comprehensive factors such as a transmission network, the number Ns of the smart meters connected to the data concentrator is generally 40 to 100, and on this basis, the corresponding relationship between the data concentrator and the smart meters in this embodiment is set according to the following principle: the number of the intelligent electric meters in the management range of the data concentrator is controlled to be 40-100, and the intelligent electric meters of the same type are connected with the same data concentrator as much as possible.

Specifically, in this embodiment, the setting of the corresponding relationship between the data concentrator and the smart electric meter, and the sequentially accessing the smart electric meter to the data concentrator includes the following steps:

(a) acquiring a preset range interval [ X1, X2] of the number of the smart meters connected to the data concentrator, wherein in the embodiment, the preset range interval is [40,100 ]; the number Ns of the intelligent electric meters connected with the data concentrator is equal to or greater than 0 and equal to or less than 100, the intelligent electric meters of the same type are preferentially connected with the same data concentrator, and the intelligent electric meters connected with the data concentrator are not connected with other data concentrators;

(b) if the number of certain intelligent electric meters in the 6 types of intelligent electric meters is within a preset range interval, namely the number of the intelligent electric meters is within a range of [40,100], accessing the intelligent electric meters of the type into the same data concentrator;

(c) if the number of certain intelligent electric meters in the 6 types of intelligent electric meters is larger than the maximum value X2 of the preset range interval [ X1, X2], namely the number of the intelligent electric meters is larger than 100, selecting X2 (namely 100) intelligent electric meters from the intelligent electric meters, accessing the selected intelligent electric meters into the same data concentrator, and enabling the data concentrator not to be connected with other intelligent electric meters any more; then, if the number of the type of smart meters which are not allocated with the data concentrator is in the range of [ X1, X2] (namely [40,100]), the type of smart meters which are not allocated with the data concentrator are independently connected to one data concentrator, and if the number of the type of smart meters which are not allocated with the data concentrator is less than X1 (namely 40), the smart meters are combined with other types of smart meters and are jointly placed in one data concentrator; for example, in this embodiment, 256 smart meters of the city resident type are placed in the management range of the same data concentrator, that is, the first 100 meters are placed in one data concentrator, the 101 th to 200 th meters are placed in another data concentrator, and no other smart meters are connected to the two data concentrators, while the remaining 201 th to 256 th meters are placed in another data concentrator, and no other smart meters are connected to the data concentrator.

(d) If the number of certain types of smart meters in the 6 types of smart meters is smaller than the minimum value X1 in the preset range interval, and other types of smart meters which are not accessed to the data concentrator and have the number smaller than X1 exist, combining the different types of smart meters which are not accessed to the data concentrator and have the number smaller than X1, wherein the total number of the combined smart meters is in [ X1, X2 ]; for example, in this embodiment, 32 intelligent electric meters of county or rural residential type, and 125 intelligent electric meters of office building or service type, the first 100 of the office building or service electric meters are placed in the management range of the same data concentrator, and the remaining 25 office building or service electric meters, 32 county or rural residential electric meters are all smaller than 40, so that the 25 office building or service electric meters and the 32 county or rural residential electric meters are connected to another data concentrator;

(e) if the number of certain intelligent electric meters in the 6 types of intelligent electric meters is smaller than the minimum value X1 of the preset range interval, and no other types of intelligent electric meters which are not accessed to the data concentrator and the number of which is smaller than X1 exist, accessing the type of intelligent electric meters to the data concentrator connected with the intelligent electric meters, and ensuring that the number of the intelligent electric meters connected with the data concentrator is in the range of [ X1, X2 ]; for example, in this embodiment, only 15 remaining microgrid-type smart meters are not connected to the data concentrator, and 57 smart meters (25 office buildings or service electric meters and 32 city or rural residential meters) are connected to one data concentrator, and then the 15 microgrid meters are also connected to the data concentrator connected to the 57 smart meters.

Because the spatial data sequence of the intelligent electric meters of the same type presents relatively gentle volatility, the monitoring data at the data concentrator can present relatively gentle volatility by configuring the connection relation between the intelligent electric meters and the data concentrator in the steps (a) to (e).

And step S104, the data concentrator performs compression sampling on the received monitoring data of the second target data type through a measurement matrix in the compressed sensing space model to obtain second compressed data.

Specifically, the data concentrator is internally provided with a measurement matrix of a compressed sensing space model corresponding to a second target data type and a second compression measurement control method generated based on the measurement matrix, and the second compression measurement control method is used for controlling the data concentrator to perform compression sampling on monitoring data of the second target data type to obtain second compression data. The measurement matrix type is determined according to the operation complexity and the data reconstruction effect of the monitoring data of each initial target data type under different measurement matrix types, and in the embodiment, the measurement matrix type is a binary sparse measurement matrix; the compression ratio is determined according to the data reconstruction effect required by the monitoring data of each initial target data type, and the compression ratio is sequentially determined according to the lowest compression ratio, the required compression ratio and the actual compression ratio.

Specifically, step S104 includes the following steps S1041 to S1043:

in step S1041, the data concentrator reads the measurement matrix type and compression ratio built in it in advance.

In order to ensure the uniformity, in this embodiment, the type and the compression ratio of the measurement matrix built in the data concentrator are the same as those of the measurement matrix built in the smart meter, the type of the measurement matrix is a binary sparse measurement matrix, and the compression ratio is 36%. The detailed process may refer to step S1021.

Step S1042, on the basis of the correspondence between the initial target data type and the compressed sensing model, a second compressed measurement control method for performing compressed sampling on the monitoring data of the second target data type is generated at the data concentrator according to the built-in binary sparse measurement matrix and the compression ratio.

Specifically, the second compression measurement control method is to control the data concentrator to perform compression sampling on the monitoring data in the second target data type. In the compression sampling, the control signal in the second compression measurement control method still uses P_j(i) Representing the M x N dimensional measurement matrix phi and the control signal P_j(i) Still as shown in equation (2), the control method may refer to step S1022.

Step S1043, the data concentrator performs compression sampling on the monitored data of the second target data type according to the generated second compression measurement control method to obtain second compressed data.

And S105, the data concentrator transmits the first compressed data and the second compressed data to the measurement data management center, and the measurement data management center performs data reconstruction on the first compressed data according to the reconstruction algorithm, the sparse basis and the measurement matrix corresponding to the first compressed data and performs data reconstruction on the second compressed data according to the reconstruction algorithm, the sparse basis and the measurement matrix corresponding to the second compressed data.

The sparse basis is determined according to the sparsity of the monitoring data of each initial target data type under different sparse basis, and in the embodiment, the sparse basis is a discrete cosine transform basis; the reconstruction algorithm is determined according to the data reconstruction effect and the operation speed of the reconstruction algorithm when the data reconstruction is carried out on the monitoring data of each initial target data type, and in the embodiment, the reconstruction algorithm is a gradient tracking algorithm or a spectral projection gradient algorithm; the type and compression ratio of the measuring matrix are the same as those of the measuring matrix built in the intelligent ammeter and the data concentrator.

Specifically, step S105 includes the following steps S1051 to S1055:

step S1051, the data concentrator transmits the first compressed data and the second compressed data to the metrology data management center.

Step S1052, the measurement data management center reads the reconstruction algorithm, sparse basis and measurement matrix built in the measurement data management center in advance.

The key of reading the reconstruction algorithm, the sparse basis and the measurement matrix by the measurement data management center is to predetermine the specific reconstruction algorithm, the sparse basis and the measurement matrix, and to arrange the reconstruction algorithm, the sparse basis and the measurement matrix in the measurement data management center in advance. In this embodiment, according to the requirement of consistency, the type and the compression ratio of the measurement matrix are the same as those of the measurement matrix built in the intelligent ammeter and the data concentrator, that is, the type of the measurement matrix is a binary sparse measurement matrix, and the compression ratio is 36%. No matter a compressed sensing time model or a compressed sensing space model, a measurement matrix, a sparse basis and a reconstruction algorithm are all three elements of compressed sensing, wherein the reconstruction process is mainly realized through the reconstruction algorithm. Whether the reconstruction algorithm or the sparse basis is related to the initial target data type only and is not related to the compressed sensing model. Therefore, to determine the sparse basis and the reconstruction algorithm, the following operations are performed:

firstly, the sparsity of monitoring data of an initial target data type under a plurality of sparse bases is analyzed, and an applicable sparse base of the monitoring data of the initial target data type is selected according to the sparsity.

The sparsity of the same signal under different sparsity bases is greatly different. At present, the construction of sparse basis is mainly based on fixed orthogonal transformation basis, which is mainly due to the simple realization of the fixed orthogonal transformation basis and the strong applicability to most signals. In addition, the common sparse basis also has multi-scale set analysis, but the realization difficulty of the sparse basis is higher. The fixed orthogonal transform basis is a sparse basis commonly used in compressed sensing and includes a Discrete Cosine Transform (DCT) basis, a Discrete Fourier Transform (DFT) basis, a Discrete Wavelet Transform (DWT) basis, and the like. The present embodiment uses a fixed orthogonal transformation basis as the sparse basis, in consideration of both the implementation effect and the ease of implementation. In the fixed orthogonal transformation base, DCT base, DFT base and DWT base can be used for sinusoidal signals, but the DCT base, DFT base and DWT base are simpler to realize compared with the DWT base. The three fixed orthogonal transformation bases are adapted to the initial target data type II-AB. Meanwhile, the DCT-based is significantly better than the DFT-based in effect for other signals without sinusoidal regularity, and is widely used in data compression in many fields due to its high-quality practicability, for example, in the international video protocol, DCT compression is used as a general compression method, which is also because DCT compression has a better compression effect on smooth natural images. And the monitoring data under the initial target data type I-A, I-B, III-A, III-B has the characteristic of smooth change and is similar to the change rule of natural images, so the DCT base is also suitable for the monitoring data under the target data types I-A, I-B, II-AB and III-A, III-B. In order to select sparse bases suitable for monitoring data under an initial target data type, different sparse bases are needed to be adopted for carrying out compression sampling on the monitoring data, when a compression sampling test is carried out, the type of a measurement matrix, a reconstruction algorithm and a compression ratio are set to be the same, and only the sparse bases are set to be different. Referring to fig. 6, fig. 6 shows sparsity K of the monitoring data of the initial target data type under the DCT basis, DFT basis, and DWT basis, respectively, when performing an experiment, 50 monitoring data of 5 types of initial target data types are selected, and the original time sequence length or the original space sequence length of each monitoring data is the same, and the sparsity K obtained in fig. 6 is an average value of 350 sparsity obtained by calculating the monitoring data of 350 initial target data types. As can be seen from fig. 6, the sparsity of the monitored data of each initial target data type under the DCT basis and the DWT basis is better than that of the DFT basis, but the DWT implementation is complex. Therefore, in the present embodiment, for the monitored data in each initial target data type, a Discrete Cosine Transform (DCT) basis is preferentially selected as the sparse basis.

And then, analyzing the data reconstruction effect of the multiple reconstruction algorithms when the monitoring data of the initial target data type are subjected to compression sampling, and selecting the reconstruction algorithm suitable for the monitoring data of the initial target data type according to the data reconstruction effect and the operation speed of the reconstruction algorithm.

Commonly used reconstruction algorithm for compressed sensing is mainly based on l₀Greedy algorithm of norm based on l₁Norm convex optimization algorithm and other combined recovery algorithms represented by iterative threshold algorithm, wherein the reconstruction effect of the convex optimization algorithm is generally better than that of the greedy algorithm. In the convex optimization algorithm, the spectral projection gradient algorithm (SPG) adopting the new step search strategy has higher signal reconstruction precision and good performance in each convex optimization algorithm. The greedy algorithm gradually approaches an original signal by adopting a circular iterative search mode, the convergence speed of the algorithm is improved by sacrificing the data reconstruction effect, the greedy algorithm mainly comprises algorithms such as Orthogonal Matching Pursuit (OMP), compressive sampling matching pursuit (CoSAMP), Subspace Pursuit (SP) and the like, and the algorithms belong to the matching pursuit class. With the intensive research of the greedy algorithm, a gradient tracking (GP) algorithm, which takes direction tracking as a basic idea, shows a better data reconstruction effect and a faster operation speed, but the data reconstruction effect of the GP algorithm is worse than that of the SPG algorithm. Meanwhile, the iterative threshold algorithm is taken as a representative of other algorithms, and has important bits in a reconstruction algorithm of compressed sampling, the iterative threshold algorithm introduces a threshold function, most typically an Iterative Hard Threshold (IHT) algorithm and a Fast Iterative Shrinkage Threshold (FISTA) algorithm, but the data reconstruction effect of the iterative threshold algorithm is poor. For the monitoring data under each initial target data type, the data reconstruction effect of compressed sensing is related to the data type. The monitoring data under the initial target data type II-AB has good smoothness and regularity, the monitoring data under the initial target data type I-B has good smoothness, and the monitoring data under the two data types have good reconstruction effect under the same condition, namely the data reconstruction precision is high. And the monitoring data under the initial target data types I-A, III-A and III-B have slightly poor smoothness and do not have sinusoidal regularity, and the reconstruction effect is poor under the same condition, namely the data areThe reconstruction accuracy is low. Based on the above analysis, in this embodiment, the data type with the poor reconstruction effect selects the SPG algorithm with the good reconstruction effect under the same condition, and the data type with the good reconstruction effect selects the GP algorithm with the slightly poor reconstruction effect but the fast speed, and for the monitoring data in each initial target data type, the reconstruction algorithm of this embodiment is selected as follows: a) for the monitoring data under the initial target data types II-AB and I-B, the GP algorithm is preferably selected as the reconstruction algorithm; b) the reconstruction algorithm preferably selects the SPG algorithm for data used under the initial target data types I-A, III-A and III-B. Referring to fig. 7, fig. 7 shows reconstructed snr using different reconstruction algorithms. During the test, 50 monitoring data under 5 types of initial target data are respectively selected, the original time sequence length or the original space sequence length of each monitoring data is the same, and the reconstructed signal-to-noise ratio in fig. 7 is an average value of 350 reconstructed signal-to-noise ratios obtained by compressing and reconstructing 350 monitoring data. In the test, DFT bases are adopted for sparse bases, and the types of measurement matrixes are binary sparse measurement matrixes. As can be seen from fig. 7, the reconstruction effect of the SPG algorithm and the GP algorithm is significantly better than that of other greedy algorithms and iterative threshold algorithms, and the GP algorithm is better than the GPSR algorithm but slightly worse than the SPG algorithm.

Step S1053, the measurement data management center solves the equation y according to the built-in reconstruction algorithm (GP algorithm or SPG algorithm)⁽¹⁾＝ΦΨs⁽¹⁾Reconstructing a sparse vector s of the monitored data of the first target data type⁽¹⁾Wherein, y⁽¹⁾Φ, Ψ respectively represent the first compressed data (i.e., the compressed data after the compressed sampling of the monitored data of the first target data type), the measurement matrix, and the sparse basis.

Step S1054, the measurement data management center solves the equation y according to the built-in reconstruction algorithm (GP algorithm or SPG algorithm)⁽²⁾＝ΦΨs⁽²⁾Reconstructing a sparse vector s of the monitored data of the second target data type⁽²⁾Wherein, y⁽²⁾Φ, Ψ respectively represent the second compressed data (i.e., the compressed data after the compressed sampling of the monitored data of the second target data type), the measurement matrix, and the sparse basis.

Step S1055, the measurement data management center obtains the finally reconstructed monitoring data under the first target data type and the second target data type through x ═ Φ S, and stores and archives the monitoring data of each initial target type in the measurement data management center; wherein s is for s in the first compressed data reconstruction process⁽¹⁾And s in the second compressed data reconstruction process⁽²⁾Are all applicable.

Compared with the prior art, the data compression acquisition and transmission method of the embodiment comprises the steps of after the monitoring data are obtained from a power grid by the intelligent electric meter, carrying out compression sampling on the monitoring data of a first target data type at the intelligent electric meter through a compression sensing time model, carrying out compression sampling on the monitoring data of a second target data type at a data concentrator, transmitting the compressed first compression data and the compressed second compression data to a measurement data management center, respectively carrying out data reconstruction on the first compression data and the compressed second compression data by the measurement data management center according to a corresponding compression sensing reconstruction algorithm, carrying out compression transmission on the monitoring data under an initial target data type through a compression sampling technology, greatly reducing the coding complexity of a sampling end, and transferring the coding complexity to the measurement data management center with strong data processing capability, meanwhile, the error of the data transmission result under the condition of packet loss or error code is reduced; in addition, according to different initial target data types, a compressed sensing time model and a compressed sensing space model are established, and different types of monitoring data are selected to be compressed and sampled at the intelligent electric meter or the data concentrator, so that the transmission effect of monitoring electric data under each target data type is improved.

Example 2:

the embodiment provides a power grid advanced measurement system based on a data compression acquisition and transmission method, which comprises an intelligent electric meter, a data concentrator and a measurement data management center, wherein the data compression acquisition and transmission method shown in the embodiment 1 is adopted when the intelligent electric meter, the data concentrator and the measurement data management center are used for acquiring and transmitting monitoring data.

The following describes, with reference to specific examples, a procedure when the advanced power grid measurement system in embodiment 2 is built. Example (c): a power grid advanced measurement system is planned to be built in a certain area, and 382 intelligent electric meters, a plurality of data concentrators and 1 measurement data management center are arranged in the area. Referring to fig. 8, when establishing the advanced grid measurement system in this area, the method includes the following steps:

step S201, obtaining the type of the intelligent electric meter and the initial target data type of the monitoring data in the advanced power grid measurement system, and configuring the total sensing (acquisition) time length T of the original data and the sensing (acquisition) time interval T of the original data of the intelligent electric meter₀。

In the advanced measurement system of the power grid, the data representation information, the power utilization monitoring time interval and the type of the intelligent electric meter are taken as the basis, the data smoothness of the monitoring data is taken as the classification basis, and the monitoring data is divided into 5 types of initial target data types. In this embodiment, the collected data in the advanced power grid measurement system has all 5 types of initial target data, I-A, I-B, II-AB and III-A, III-B, respectively. The monitoring data of the six types of intelligent electric meters in the electricity consumption peak period are represented by A, and the monitoring data of the six types of intelligent electric meters in the electricity consumption valley period are represented by B. Reference is made in detail to step S101 in example 1.

And according to the information collection characteristics of the intelligent electric meters, dividing the intelligent electric meters into 6 types. In this embodiment, the advanced measurement system has all 6 types of smart meters. The number of various types of smart meters of 382 smart meters in the region and the type of initial target data collected by each smart meter are shown in table 5.

TABLE 5 number of intelligent electric meters classified and data type of data collected

Except the type and the number of the intelligent electric meters and the initial target of acquisitionBesides the data type, the parameters of the intelligent ammeter further comprise the total sensing (acquisition) time length T of the original data and the sensing (acquisition) time interval T of the original data during each data transmission of the intelligent ammeter₀T and T₀The meaning of (a) is the same as in example 1. The monitoring data of different initial target data types are analyzed, and the T suitable for the monitoring data of different initial target data types can be determined₀And T. In this example, T₀The setting results of T and T are shown in Table 6, and according to the setting results of Table 6, the original time series length N of the monitoring data of different initial target data types can be calculated_T＝T/T₀。

TABLE 6 Smart electric meter T under different initial target data types₀And setting of T

Step S202, setting the connection relation between the data concentrator and the intelligent electric meter in the advanced power grid measurement system, and determining the configuration scheme of the data concentrator according to the intelligent electric meter. The data concentrator configuration scheme comprises the number of data concentrators, the number of intelligent electric meters collected by each data concentrator and the number of specific intelligent electric meters (namely specific intelligent electric meters). In the configuration, one data concentrator is connected to a plurality of smart meters, and one smart meter can only access one data concentrator, and for the method for determining the connection relationship between the data concentrator and the smart meters in the power grid advanced measurement system, reference may also be made to step S103 in embodiment 1.

Specifically, step S202 includes the following steps S2021 to S2024:

step S2021, analyzing the number of the intelligent electric meters of different types in the advanced measurement system, numbering the intelligent electric meters, and obtaining a mapping table of the intelligent electric meter types and the intelligent electric meter numbers. In this embodiment, after analyzing and numbering 382 smart meters, a mapping table of smart meter types and smart meter numbers is obtained, as shown in table 7: in 382 intelligent electric meters, the urban resident electric meter is an intelligent electric meter with the number of 1-156, the county city or rural resident electric meter is an intelligent electric meter with the number of 157-188, the heavy industry factory electric meter is an intelligent electric meter with the number of 189-225, the light industry factory electric meter is an intelligent electric meter with the number of 226-277, the office building or service industry electric meter is an intelligent electric meter with the number of 278-372, and the micro grid electric meter is an intelligent electric meter with the number of 373-382.

TABLE 7 numbering of all types of Smart meters

Step S2022, all the intelligent electric meters are used as elements of the set Q, the intelligent electric meter types of which the number of the intelligent electric meters in the set is within the preset range of the number of the intelligent electric meters connected with the data concentrator are obtained, and the connection relation between the intelligent electric meters of the types and the data concentrator is set.

Specifically, in this embodiment, 382 smart meters in the advanced power grid measurement system are all placed in a set Q, where the set Q represents a smart meter that is not set, and at an initial time, the set Q has 6 types of smart meters in common, and the smart meters numbered 1 to 382 are all elements of the set Q, and the initial element number of Q is 382. The embodiment continues to be described by taking the preset range interval of the number of the smart meters connected to the data concentrator as [40,100 ]. When the configuration of the data concentrator is realized by setting the connection of the intelligent electric meters, firstly, the type of the intelligent electric meters in the range of [40,100] of the number of the intelligent electric meters in the set Q is searched, and the intelligent electric meters of the type are set. According to the number of the intelligent electric meters of each type in table 7, the number of the intelligent electric meters in the light industry factory is 52, the number of the intelligent electric meters in the office building or the service industry is 95, the number of the intelligent electric meters of each type is not within the preset range interval [40,100], and at the moment, the intelligent electric meters of the 2 types are set, and the specific setting process is as follows: if the number of the intelligent electric meters of the same type in the advanced measurement system is within a preset range interval [40,100], correspondingly connecting the intelligent electric meters of the type with a data concentrator, numbering the data concentrator, confirming the condition of the connected intelligent electric meters, and deleting the intelligent electric meters connected with the data concentrator from the set Q. Through the process, the power grid advanced measurement system in the embodiment establishes 2 data concentrators, the data concentrator A is connected with all light industrial factory electric meters, namely, the intelligent electric meters with the serial numbers of 226-277 are placed in the management range of the data concentrator A; the data concentrator B is connected with all office buildings or service electric meters, namely, the intelligent electric meters with the numbers of 278-372 are placed in the range of the data concentrator B; at this time, the elements in the set Q are the smart meters with the numbers of 1-225 and the smart meters with the numbers of 373-382.

Step S2023, acquiring the type of the intelligent electric meters of which the number of the intelligent electric meters in the set Q is larger than the maximum value of the preset range interval of the number of the intelligent electric meters connected with the data concentrator, and setting the connection relation between the intelligent electric meters of the type and the data concentrator.

Specifically, for the type of the smart meters with the number greater than 100 in the set Q, the smart meters of the type are set. According to the number of the various types of smart meters in table 7, the number of the "urban resident meters" is 156, the number of the smart meters is greater than the maximum value of the preset range interval of 100, at this time, the smart meters are set, and the specific setting process is as follows: placing the first 100 intelligent electric meters in the management range of the same data concentrator, numbering the data concentrator without connecting other intelligent electric meters, confirming the connected intelligent electric meter conditions, and deleting the connected intelligent electric meters from the set Q; at this time, if the number of the type of smart meters in the set Q still exceeds 100, continuously placing the remaining first 100 smart meters in the management range of a new data concentrator, numbering the data concentrator, confirming the connected smart meters, and deleting the connected smart meters from the set Q until the number of the smart meters not yet set in the type of smart meters is less than 100; then, if the number of the smart meters of the type in the set Q is greater than or equal to 40, the smart meters are separately placed in a new data concentrator, the data concentrator is numbered, the connected smart meters are confirmed, and the connected smart meters are deleted from the set Q; if the number of the smart meters of the type in the set Q is less than 40, execute step S2024. Through the above process, in this embodiment, 2 data concentrators are newly established in the power grid measurement system and are marked as a data concentrator C and a data concentrator D, wherein the data concentrator C is connected with the first 100 urban residential electric meters with smaller numbers, that is, the intelligent electric meters with numbers of 1-100 are placed in the management range of the data concentrator C; the data concentrator D is connected with the rest 56 urban resident electric meters, namely the intelligent electric meters with the numbers of 101-156 are placed in the management range of the data concentrator D; at this time, the elements in the set Q are the smart meters with numbers 157 to 225 and the smart meters with numbers 373 to 382.

Step S2024, acquiring the type of the smart electric meters of which the number of the smart electric meters in the set Q is smaller than the minimum value of the preset range interval of the number of the smart electric meters connected with the data concentrator, and setting the connection relation between the smart electric meters of the type and the data concentrator.

Specifically, for the type in which the number of the smart meters in the set Q is less than 40, the smart meter of the type is set. According to the number of the intelligent electric meters of each type in table 7, the number of the remaining three types of intelligent electric meters in the set Q is less than the minimum value of 40 blocks in the preset range interval, which are respectively: "county city or rural residential electric meter" with the serial number of 157-188, "heavy industry factory electric meter" with the serial number of 189-255, "microgrid electric meter" with the serial number of 373-382. At this moment, the three types of intelligent electric meters are set, and the specific setting process is as follows: sequentially carrying out combination analysis on all types of intelligent electric meters in the set Q, and ensuring that the total number of the intelligent electric meters in all the combinations meets the condition that Ns is more than or equal to 40 and less than or equal to 100 according to each two-type combination or multi-type combination; if the smart meters which cannot meet the combination condition exist, the smart meters are placed in other data concentrators which are not up to 100 in a centralized mode. Through the process, 1 data concentrator is newly established in the advanced power grid measurement system and is recorded as a data concentrator E, the data concentrator E is connected with the remaining three types of intelligent electric meters, namely 32 'county or rural residential electric meters', 37 'heavy industrial plant electric meters' and 10 'microgrid electric meters', and 79 intelligent electric meters with numbers of 157-188, 189-255 and 373-382 are placed in the management range of the data concentrator E in total.

Through the steps from the step S2021 to the step S2024, 5 data concentrators are configured for 382 smart meters, at this time, the set Q becomes an empty set, all the smart meters complete connection setting with the data concentrators, and the data concentrators in the power grid advanced measurement system also complete configuration. The specific configuration is shown in table 8.

TABLE 8 data concentrator configuration scenarios

Step S203, establishing a compressed sensing time model and a compressed sensing space model, analyzing and determining a compressed sensing model applicable to each initial target data type, analyzing and determining a measurement matrix type, a sparse basis, a reconstruction algorithm and a compression ratio applicable to the monitoring data of each initial target data type under the compressed sensing model, wherein the initial target data type applicable to the compressed sensing time model is recorded as a first target data type, and the initial target data type applicable to the compressed sensing space model is recorded as a second target data type.

Specifically, step S203 includes the following steps S2031 to S2032:

step S2031, establishing a compressed sensing time model and a compressed sensing space model corresponding to each link (smart meter, data concentrator, measurement data management center) of the advanced power grid measurement system, which can refer to fig. 2 and 3.

Specifically, in this embodiment, the measurement matrix types of the compressed sensing model applicable to the monitoring data in each initial target data type are binary sparse measurement matrices; the sparse bases are DCT bases; and (3) monitoring data under the initial target data types II-AB and I-B, selecting a GP algorithm by a reconstruction algorithm, monitoring data under the initial target data types I-A, III-A, III-B, and selecting an SPG algorithm by the reconstruction algorithm.

For the compression ratio, the data reconstruction effect required to be met by the constructed power grid advanced measurement system needs to be set. Specifically, in this embodiment, the determination process of the compression ratio includes the following steps:

(1) according to the requirements that the total number of the samples to be statistically analyzed is not less than 1050 and the total number of the samples to be statistically analyzed is not less than 150 for each initial target data type, the present embodiment performs statistical analysis on 1400 data samples, wherein the number of the samples for each initial target data type is 200. First, a preset range of the compression ratio is obtained, and in this embodiment, the preset range of the compression ratio is described as [ 25%, 50% ], and the preset range is used for verifying the lowest compression ratio. Because there is no uniform requirement for the data reconstruction effect in the advanced power grid measurement system, the present embodiment analyzes the compression ratios of all samples with the data reconstruction effect of 50dB as an example, and obtains the lowest compression ratio corresponding to each initial target data type, as shown in table 9, where the data reconstruction effect of the monitoring data in each initial target data type is not less than 50dB when the lowest compression ratio is adopted for compression transmission. As can be seen from table 9, there is no case where the lowest compression ratio is less than 25% or greater than 50%, i.e., both are within the preset range [ 25%, 50% ] of the compression ratio, so that the lowest compression ratio may not need to be adjusted or analyzed for accuracy;

(2) determining the required compression ratio corresponding to each initial target data type according to the lowest compression ratio of each initial target data type and reserving a 10% margin, namely, the required compression ratio is the lowest compression ratio x (1+ 10%), as shown in table 9;

(3) when the compression sensing models under the initial target data types need to use a uniform compression ratio, the compression ratios need to be combined, and the compression ratios are unified according to the largest required compression ratio in the required compression ratios during combination, namely the largest required compression ratio is used as the actual compression ratio of all the initial target data types; when the compression sensing model under each initial target data type is not required to use a uniform compression ratio, each initial target data type is respectively set according to the respective required compression ratio, namely the actual compression ratio of each initial target data type is the respective required compression ratio. In this embodiment, the monitoring data of each initial target data type is required to adopt a uniform compression ratio, so the actual compression ratio of each initial target data type is the maximum required compression ratio of 5 required compression ratios, that is, 39%, as shown in table 9.

TABLE 9 initial target data type to compression ratio correspondence table

And step S2032, analyzing and determining the applicable compressed sensing model of each initial target data type.

In the embodiment, a compressed sensing time model is selected for monitoring data under the initial target data types II-AB and I-A; monitoring data under an initial target data type I-B, III-B, and selecting a compressed sensing space model; monitoring data under the initial target data type III-A can adopt a compressed sensing time model or a compressed sensing space model, and if the intelligent electric meters in the management range of the data concentrator are of the same type and the number of the intelligent electric meters is more than 80, the compressed sensing space model is preferentially selected for the monitoring data under the initial target data type III-A.

And S204, establishing a construction scheme of the advanced power grid measurement system according to the corresponding relation between each initial target data type and the compressed sensing model, and the measurement matrix type, the sparse basis, the reconstruction algorithm and the compression ratio which are applicable to the monitoring data of each initial target data type.

The construction scheme of the advanced grid measurement system of this embodiment is shown in table 10.

Table 10 advanced measurement system construction scheme for power grid

And S205, according to the connection relation between the intelligent electric meter and the data concentrator, realizing data communication among the intelligent electric meter, the data concentrator and the measurement data management center through a communication network, and establishing a hardware system of a high-level measurement system.

Step S206, generating a measurement matrix, a sparse basis and a reconstruction algorithm required by the advanced measurement system according to the compression ratio, the measurement matrix type, the sparse basis and the reconstruction algorithm applicable to the advanced measurement system construction scheme, arranging the measurement matrix in an intelligent ammeter (applicable to a compressed sensing time model) or a data concentrator (applicable to a compressed sensing space model) according to the compressed sensing model applicable to each initial target data type, and arranging the measurement matrix, the compression ratio, the reconstruction algorithm and the sparse basis in a measurement data management center.

Specifically, step S206 includes the following steps S2061 to S2064:

step S2061, the compression ratios required by the initial target data types obtained by analysis are built in all the intelligent electric meters and the data concentrators and are built in the measurement data management center at the same time. In this embodiment, the compression ratio of all the smart meters and the data concentrators is set to 39%. Step S2062, generating a sparse basis and placing the sparse basis in the measured data management center, and setting a trigger condition for calling the sparse basis to start a reconstruction process of the first compressed data or the second compressed data for the measured data management center. In this embodiment, sparse bases of monitoring data of all initial target data types are generated according to a discrete cosine transform mode. Step S2063, generating a reconstruction algorithm and arranging the reconstruction algorithm in the measured data management center, setting a triggering condition for calling the reconstruction algorithm as that the measured data management center starts a reconstruction process of the first compressed data or the second compressed data, and setting a judgment condition for calling which reconstruction algorithm is the initial target data type. In this embodiment, the reconstruction algorithms of the monitoring data of all the initial target data types are only the SPG algorithm and the GP algorithm, and when called, the corresponding reconstruction algorithm should be selected according to each initial target data type, specifically referring to the construction scheme of the power grid advanced measurement system in table 10. And S2064, generating a measurement matrix required in the advanced power grid measurement system according to the type of the measurement matrix and the determined compression ratio, and arranging the measurement matrix in all the intelligent electric meters and the data concentrator and in the measurement data management center at the same time. Setting a triggering condition for calling a measurement matrix as starting a compression sampling process in the intelligent ammeter and the data concentrator; and setting a triggering condition for calling the measurement matrix as a starting reconstruction process in the measurement data management center. In this embodiment, the measurement matrix types of all the smart meters, the data concentrators and the measurement data management center are binary sparse measurement matrices, and the compression ratios are 39%. Because the measurement matrix is generated in a binary random mode, the measurement matrix needs to be solidified and stored after being generated, so that the uniqueness of the measurement matrix used for the same initial target data type in the power grid advanced measurement system is guaranteed. Specifically, in this embodiment, the generation and internal process of the measurement matrix includes the following steps: (1) determining the original data length N of each initial target data type according to the compressed sensing model of each initial target data type, namely the column number of a measurement matrix; (2) according to the compression ratio, determining the number M of rows of the measurement matrix as a value of Nx compression ratio and then rounding; (3) initializing all elements of an M multiplied by N measurement matrix to be 0, randomly selecting alpha M positions from each row of the matrix, setting the elements of the positions to be 1, and then, the matrix is a binary sparse measurement matrix; wherein α should be set according to actual conditions, and is set to 0.1 in the embodiment; (4) and solidifying the measurement matrix, wherein each element in the matrix does not change randomly any more to obtain a final measurement matrix, and respectively arranging the final measurement matrix in the intelligent electric meter, the data concentrator and the measurement data management center.

Step S207, according to the measurement matrix built in the smart meter or the data concentrator, a first compression measurement control method for the monitoring data of the first target data type is generated at the smart meter, and a second compression measurement control method for the monitoring data of the second target data type is generated at the data concentrator.

In this embodiment, please refer to fig. 9 for the control principle of the first compression measurement method and the second compression measurement method. Fig. 9 is a schematic diagram of a compression measurement control principle when N-dimensional raw data x at the smart meter or the data concentrator is compression-sampled by the measurement matrix. As can be seen from FIG. 9, the ith element x in the N-dimensional original data is sequentially aligned_iProcessing (i ═ 1,2.. N) to obtain a compressed M-dimensional compressed signal y ═ y ·₁,y₂...y_M]^T. In the case of compressed sampling, a compression measurement method and a control signal P for a second compression measurement method_j(i) Can be read from the measuring matrix, M × N dimension measuring matrix phi and control signal P_j(i) The relationship of (a) to (b) is as follows:

wherein when the control signal P_j(i) When 1, the switch is turned on to read x_iTo x_iThe compression measurement is performed, otherwise the switch is closed and the read value is zero. Since the measurement matrix Φ is a binary sparse measurement matrix, and its internal elements are only 0 and 1, in this embodiment, it is proposed to implement the control of the compression measurement by using a switch control method.

The building of the power grid advanced measurement system is completed through the steps S201 to S207.

In order to verify the effect of the power grid advanced measurement system established by the invention, the invention obtains the reconstructed signal-to-noise ratio of the monitoring data under each initial target data type by carrying out compression sampling and transmission on the monitoring data under different initial target data types. Referring to fig. 10, fig. 10 is a reconstructed snr after the advanced power grid measurement system of the present invention performs compression acquisition and transmission on the monitoring data of the initial target data type. In fig. 10, for an initial target data type using a compressed sensing time model, a calculation method of a reconstructed signal-to-noise ratio is as follows: performing compression sampling and reconstruction analysis on the original data of all the intelligent electric meters, and solving an average value of all the obtained reconstruction signal-to-noise ratios, wherein the average value is used as the reconstruction signal-to-noise ratio of the monitoring data under the initial target data type; for an initial target data type adopting a compressed sensing space model, the reconstruction signal-to-noise ratio is as follows: performing compression sampling and reconstruction analysis on the original data of all the data concentrators, and solving an average value of all the obtained reconstruction signal-to-noise ratios, wherein the average value is used as the reconstruction signal-to-noise ratio of the monitoring data under the initial target data type; for an initial target data type which adopts a compressed sensing time model and a compressed sensing space model, namely a target data type III-A, the reconstruction signal-to-noise ratio is as follows: performing compression sampling and reconstruction analysis on raw data of data concentrators (namely, the data concentrators B and C) applicable to the compressed sensing space model and raw signals of smart meters (namely, the smart meters in the management range of the data concentrator A, D, E) applicable to the compressed sensing time model, wherein each data concentrator takes 400 raw signals, each smart meter takes 5 raw signals, and an average value is obtained for all obtained reconstruction signal-to-noise ratios, and the average value is used as the reconstruction signal-to-noise ratio of the monitoring electrical data under the initial target data type. As can be seen from fig. 7, the power grid measurement system of the present invention has a higher reconstructed signal-to-noise ratio for the monitoring data in each initial target data type, and the reconstructed signal-to-noise ratio reaches above 40 dB.

In addition, in order to verify the packet loss resistance of the power grid advanced measurement system established by the invention, the signal-to-noise ratio is reconstructed after data compression and transmission are carried out by the compression sampling mode, the traditional DCT compression mode and the traditional DFT compression mode when different packet loss rates are analyzed. Referring to fig. 11, fig. 11 shows reconstructed snrs after data compression and transmission in the compression sampling method, the conventional DCT compression method, and the conventional DFT compression method with different packet loss rates. In fig. 11, the calculation method of the reconstructed signal-to-noise ratio at different packet loss rates is as follows: and under the packet loss rate, performing compression measurement and reconstruction analysis on the original monitoring data of all the initial target data types, and taking the average value of all the obtained reconstruction signal-to-noise ratios as the reconstruction signal-to-noise ratio under the packet loss rate. Specifically, for an initial target data type adopting a compressed sensing time model, selecting original time data sequences of all intelligent electric meters as original monitoring data, and selecting 5 pieces of original monitoring data for each intelligent electric meter; for an initial target data type adopting a compressed sensing space model, selecting original space data sequences of all data concentrators as original monitoring data, and selecting 400 original monitoring data from each data concentrator; for an initial target data type which adopts a compressed sensing space model and a compressed sensing time model, selecting a data concentrator which is suitable for the compressed sensing space model under the initial target data type and an original space data sequence or an original time data sequence of an intelligent electric meter which is suitable for the compressed sensing time model as original monitoring data, wherein each data concentrator selects 400 original signals, and each intelligent electric meter selects 5 original signals. And for the packet loss rate setting, the setting is carried out in a mode of randomly deleting the data value according to the packet loss rate in the compressed measurement value obtained after each original monitoring data is subjected to compression sampling. As can be seen from fig. 11, with the increase of the packet loss rate, although the data reconstruction effect of the compression sampling decreases, the data reconstruction effect shows a trend of smooth decrease, and the influence on the data reconstruction effect is not great; the reconstruction effect of the conventional DCT compression mode and the conventional DFT compression mode is reduced approximately linearly, and the reconstruction signal-to-noise ratio is reduced to a serious step along with the increase of the packet loss rate, so that the reconstructed data has no availability. Therefore, the advanced power grid measurement system has obvious packet loss resistance.

Example 3

The present embodiment provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the computer program implements the steps of the data compression acquisition and transmission method shown in embodiment 1.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other ways. For example, the above-described embodiments of the apparatus/terminal device are merely illustrative, and for example, a module or a unit may be divided into only one logical function, and may be implemented in other ways, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted or not executed. The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method according to the embodiments of the present invention may also be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of the embodiments of the method. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain other components which may be suitably increased or decreased as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, in accordance with legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunications signals. The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.

Claims

1. A data compression acquisition and transmission method is suitable for an advanced measurement system comprising an intelligent ammeter, a data concentrator and a measurement data management center, and is characterized by comprising the following steps: the intelligent ammeter acquires monitoring data of a first target data type and monitoring data of a second target data type from the acquired monitoring data; the intelligent ammeter performs compression sampling on the monitoring data of the first target data type through a measurement matrix in the compressed sensing time model to obtain first compressed data, and transmits the first compressed data and the monitoring data of the second target data type to a data concentrator accessed by the intelligent ammeter; the data concentrator performs compression sampling on the received monitoring data of the second target data type through a measurement matrix in the compressed sensing space model to obtain second compressed data, transmits the first compressed data and the second compressed data to the measured data management center, and performs data reconstruction on the first compressed data and the second compressed data through a reconstruction algorithm, a sparse basis and the measurement matrix which respectively correspond to the first compressed data and the second compressed data by the measured data management center;

the first target data type is an initial target data type which is suitable for a compressed sensing time model and is in a plurality of initial target data types obtained by preliminarily classifying the monitoring data according to data characterization information, the type of the intelligent electric meter, the electricity consumption monitoring time period and data smoothness, and the second target data type is an initial target data type which is suitable for a compressed sensing space model and is in the plurality of initial target data types.

2. The data compression collection and transmission method according to claim 1, wherein the data representation information includes three categories of electricity consumption and power, voltage and current, and power quality, the types of smart meters include six categories of city residential meters, county or rural residential meters, heavy industrial factory meters, light industrial factory meters, office buildings or service industry meters, and micro-grid meters, the electricity consumption monitoring period includes two categories of peak electricity consumption period and valley electricity consumption period, the initial target data types include I-A, I-B, II-AB, III-a, and III-B, wherein I represents electricity consumption and power of six types of smart meters, II represents voltage and current of six types of smart meters, III represents power quality of six types of smart meters, and a represents monitoring data of six types of smart meters during peak electricity consumption period, and B represents monitoring data of the six types of intelligent electric meters in the electricity consumption valley period.

3. The method according to claim 1, wherein the correspondence relationship between the initial target data type and the compressed sensing time model and the compressed sensing space model is determined according to a raw data length and a raw data smoothness of the monitoring data of the initial target data type, the raw data length includes a raw time sequence length and a raw space sequence length, the raw data smoothness includes a raw time sequence smoothness and a raw space sequence smoothness, the raw time sequence length is a ratio of a total data acquisition time length to an acquisition time interval of each transmission of the monitoring data of the initial target data type by the smart meter, the raw space sequence length is a number of the smart meters connected to the data concentrator, and the raw space sequence lengths corresponding to all the data concentrators form a space sequence length interval, the original time sequence smoothness is the smoothness of a data sequence formed by the monitoring data of the initial target data type at the smart meter, and the original space sequence smoothness is the smoothness of the data sequence formed by the monitoring data of the initial target data type at the data concentrator.

4. The data compression acquisition and transmission method according to claim 3, wherein the step of determining the corresponding relationship between the monitoring data of the initial target data type and the compressed sensing time model and the compressed sensing space model according to the original data length and original data smoothness of the monitoring data of the initial target data type comprises:

when the original time sequence smoothness and the original space sequence smoothness of the monitoring data of the initial target data type meet similar conditions, if the original time sequence length is larger than the maximum value of the space sequence length interval, the initial target data type corresponds to the compressed sensing time model, and if the original time sequence length is smaller than the minimum value of the space sequence length interval, the initial target data type corresponds to the compressed sensing space model;

and when the original time sequence length and the original space sequence length of the monitoring data of the initial target data type meet the same condition, determining that the initial target data type corresponds to the compressed sensing time model or the compressed sensing space model according to the original time sequence smoothness and the original space sequence smoothness.

5. The data compression acquisition and transmission method according to claim 1, wherein the connection relationship between the smart meter and the data concentrator satisfies the following conditions: a data concentrator connects a plurality of smart electric meters, a smart electric meter inserts a data concentrator, the quantity of the smart electric meter that the data concentrator is connected is located and predetermines within a range interval and the smart electric meter of the same type is preferential to be connected with same data concentrator.

6. The data compression acquisition and transmission method according to claim 5, wherein the smart meter is connected with the data concentrator by the following method:

if the number of the intelligent electric meters which are not accessed to the data concentrator is within a preset range interval, accessing the intelligent electric meters of the type to one data concentrator;

if the number of the intelligent electric meters which are not accessed to the data concentrator is larger than the maximum value of a preset range interval, selecting the intelligent electric meters with the maximum value number from the intelligent electric meters of the type, accessing the selected intelligent electric meters to one data concentrator, and enabling the data concentrator not to be connected with other intelligent electric meters any more;

if the number of the intelligent electric meters which are not accessed to the data concentrator is smaller than the minimum value of the preset range interval, and other intelligent electric meters which are not accessed to the data concentrator and the number of which is smaller than the minimum value of the preset range interval exist, accessing the intelligent electric meters of the type and other intelligent electric meters which are not accessed to the data concentrator and the number of which is smaller than the minimum value of the preset range interval into one data concentrator, wherein the number of the intelligent electric meters connected to the data concentrator meets the preset range interval;

if the number of the intelligent electric meters which are not accessed to the data concentrator is smaller than the minimum value of the preset range interval, and other intelligent electric meters which are not accessed to the data concentrator and the number of which is smaller than the minimum value of the preset range interval do not exist, the intelligent electric meters of the type are accessed to the data concentrator which is connected with the intelligent electric meters, and the number of the intelligent electric meters connected with the data concentrator meets the preset range interval.

7. The data compression collection and transmission method according to claim 1, wherein the compressed sensing temporal model and the compressed sensing spatial model each include a measurement matrix, a sparse basis and a reconstruction algorithm, the measurement matrix is obtained according to the type of the measurement matrix and the compression ratio, the type of the measurement matrix is determined according to the operation complexity and the data reconstruction effect of the monitoring data of each initial target data type under different types of the measurement matrix, the sparse basis is determined according to the sparsity of the monitoring data of each initial target data type under different sparse basis, the reconstruction algorithm is determined according to the data reconstruction effect and the operation speed of the reconstruction algorithm when the data reconstruction is carried out on the monitoring data of each initial target data type, the compression ratio is determined according to the data reconstruction effect required by the monitoring data of each initial target data type.

8. The data compression acquisition and transmission method according to claim 7, wherein the step of determining the compression ratio based on the data reconstruction effect required for the monitored data of each initial target data type comprises:

acquiring a preset value of a reconstructed signal-to-noise ratio, a preset compression ratio interval and a preset compression ratio margin;

analyzing the compression ratio of the monitoring data of each initial target data type when the reconstruction signal-to-noise ratio is the preset value of the reconstruction signal-to-noise ratio, and recording the compression ratio obtained by analysis as the lowest compression ratio, wherein the reconstruction signal-to-noise ratio is used for representing the data reconstruction effect;

if the lowest compression ratio is smaller than the lowest value of the preset compression ratio interval, updating the lowest compression ratio to the lowest value of the preset compression ratio interval, and if the lowest compression ratio is larger than the maximum value of the preset compression ratio interval, verifying the accuracy of the analysis process of the lowest compression ratio and not updating the lowest compression ratio when the analysis process of the lowest compression ratio is accurate;

calculating a required compression ratio corresponding to the lowest compression ratio according to the preset compression ratio margin;

when the compression ratio of each initial target data type has a unification requirement, the required compression ratio with the largest value in the required compression ratios corresponding to each initial target data type is used as the actual compression ratio of each initial target data type, and when the compression ratio of each initial target data type does not have the unification requirement, the required compression ratio of each initial target data type is used as the actual compression ratio of each initial target data type, and the actual compression ratio is the compression ratio set in the compression sensing time model or the compression sensing space model corresponding to the initial target data type.

9. The data compression collection and transmission method according to claim 7, wherein the measurement matrix type is a binary sparse measurement matrix, the sparse basis is a discrete cosine transform basis, and the reconstruction algorithm is a gradient tracking algorithm or a spectral projection gradient algorithm.

10. An advanced power grid measurement system, which comprises an intelligent electric meter, a data concentrator and a measurement data management center, wherein the intelligent electric meter, the data concentrator and the measurement data management center compress and transmit monitoring data by the data compression acquisition and transmission method according to any one of claims 1 to 9.