Intelligent meter information extraction and intelligent analysis method and system

Abstract

The application discloses an intelligent electric meter information extraction and intelligent analysis method and system, relates to the technical field of intelligent electric meter detection, and comprises the following steps: extracting external information of an electric meter from an electric meter image; acquiring an electric meter pulse, a current and a voltage signal; performing standardization processing on the pulse signal, and performing accuracy evaluation and confidence verification according to the standardized pulse constant; constructing a feature matrix according to the current signal and the voltage signal of the electric meter to be detected and historical fault data, performing health scoring on the electric meter to be detected according to the feature matrix, and predicting the remaining life of the electric meter to be detected according to the health score of the electric meter to be detected at continuous time points; and completing monitoring analysis on the electric meter to be detected according to the extracted external information of the electric meter, the accuracy evaluation and the confidence verification result, and the health score and the remaining life. The electric meter is intelligently analyzed from the dimensions of the accuracy of the electric meter, the health score and the remaining life, so that the supervision ability for the electric meter is improved, and the accuracy of the electric meter data is further ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent electric meter detection, in particular to an intelligent electric meter information extraction and intelligent analysis method and system. BACKGROUND

[0002] An electric energy meter is a metering device for measuring and recording electric power consumption, mainly used for electric energy metering in residential, industrial and commercial electric power consumption scenarios. Its core function is to accurately record the active electric energy consumed by users to provide a basis for electric power settlement. Modern intelligent electric meters also have functions such as remote data transmission and load monitoring.

[0003] Power companies usually need to manage a number of electric meters, but the current technology still has deficiencies: since the data of the electric meter is an important source of power grid load analysis and fault diagnosis, the accuracy of the electric meter is closely related to the reliability of the data and subsequent power grid optimization scheduling, and as time goes by, the internal components of the electric meter will gradually age, thereby affecting the accuracy of the electric meter. Therefore, in order to more effectively manage the electric meter, a method is needed to extract and analyze the data of the electric meter. SUMMARY

[0004] To solve the above problems, the present application provides an intelligent electric meter information extraction and intelligent analysis method and system, which intelligently analyzes the electric meter from the aspects of electric meter accuracy, health score, and remaining life, thereby not only improving the supervision ability of the electric meter, but also being beneficial to ensuring the accuracy of the electric meter data.

[0005] To achieve the above purpose, the present application adopts the following technical solutions:

[0006] In a first aspect, the present application provides an intelligent electric meter information extraction and intelligent analysis method, comprising:

[0007] extracting electric meter external information from the obtained electric meter image to be tested;

[0008] obtaining electric meter internal information of the electric meter to be tested in use, the electric meter internal information including pulse signals, current signals, and voltage signals;

[0009] standardizing the pulse signals, and performing accuracy evaluation and confidence verification on the electric meter to be tested according to the obtained standardized pulse constant;

[0010] constructing a feature matrix according to the current signals and the voltage signals of the electric meter to be tested and historical fault data, performing health scoring on the electric meter to be tested according to the feature matrix, and predicting the remaining life according to the health score of the electric meter to be tested at consecutive time points;

[0011] According to the extracted external information of the electric meter, the results of accuracy evaluation and confidence verification, and the health score and remaining life, the monitoring analysis of the electric meter to be tested is completed.

[0012] As an alternative embodiment, the process of extracting external information of the electric meter includes:

[0013] The local feature map of the electric meter image is extracted by the convolutional neural network, and the local feature map is expanded into a sequence feature according to the spatial position. Then, the sequence feature is encoded by the bidirectional LSTM;

[0014] According to the local feature map and the encoded sequence feature map, the spatial attention weight of the position in the local feature map is calculated; and the channel attention weight of the local feature map is obtained by processing the local feature map through global average pooling and a fully connected layer.

[0015] The local feature map is enhanced according to the spatial attention weight and the channel attention weight, and the enhanced feature map is obtained. Then, the external information of the electric meter is extracted after the enhanced feature map is optically compensated.

[0016] As an alternative embodiment, the spatial attention weight is:

[0017] ; ;

[0018] The channel attention weight is: ;

[0019] wherein, represents the feature vector of the position in the local feature map ; the hidden state of the previous time step; , represents the weight matrix; v represents the weight vector; represents the unnormalized attention score of the position in the local feature map ; represents the unnormalized attention score of any position in the local feature map ; represents the normalized spatial attention weight at the position ; represents the global average pooling of the local feature map , and outputs a C-dimensional vector; represents the weight of the fully connected layer; represents the channel attention weight of the cth channel; is an activation function.

[0020] As an alternative embodiment, the accuracy evaluation is: ;

[0021] The confidence verification is: ;

[0022] wherein, is the pulse constant of the standard meter, is the pulse accuracy of the meter to be tested; is the pulse constant of the meter to be tested; represents the t-distribution critical value, and n is the historical data amount of the pulse constant of the meter to be tested; is the confidence of the displayed data of the meter to be tested; is the historical data mean value of the pulse constant of the meter to be tested, is the historical data standard deviation of the pulse constant of the meter to be tested.

[0023] As an alternative embodiment, the health score is:

[0024] ;

[0025] wherein, H is the comprehensive health score; represents the dynamic weight; represents the normalized value of the i-th feature ; represents the historical failure times of the meter to be tested; is the decay coefficient; and m is the total number of features.

[0026] As an alternative embodiment, the remaining life prediction is:

[0027] ;

[0028] wherein, represents the remaining life prediction value of the meter to be tested at time t; is the model fusion coefficient; is the shape parameter; is the characteristic life; represents the health score sequence of n time periods.

[0029] In a second aspect, the present application provides an intelligent meter information extraction and intelligent analysis system, comprising:

[0030] An image acquisition module configured to extract meter external information from the acquired meter image to be tested;

[0031] An information acquisition module configured to acquire meter internal information when the meter to be tested is in use, wherein the meter internal information includes pulse signals, current signals and voltage signals;

[0032] The accuracy evaluation module is configured to standardize the pulse signal, and evaluate the accuracy and verify the confidence of the meter to be tested according to the obtained standardized pulse constant.

[0033] The remaining life evaluation module is configured to construct a feature matrix according to the current signal and the voltage signal of the meter to be tested and historical failure data, evaluate the health score of the meter to be tested according to the feature matrix, and predict the remaining life of the meter to be tested according to the health score of the meter to be tested at continuous time points.

[0034] The supervision module is configured to complete the monitoring analysis of the meter to be tested according to the extracted external information of the meter, the results of the accuracy evaluation and the confidence verification, and the health score and the remaining life.

[0035] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, and computer instructions stored in the memory and running on the processor, when the computer instructions are run by the processor, the method of the first aspect is completed.

[0036] In a fourth aspect, the present application provides a computer readable storage medium for storing computer instructions, when the computer instructions are executed by the processor, the method of the first aspect is completed.

[0037] In a fifth aspect, the present application provides a computer program product comprising a computer program, when the computer program is executed by the processor, the method of the first aspect is completed.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] The present application provides an intelligent meter information extraction and intelligent analysis method and system, which can extract and upload the information of the meter to the supervision platform, analyze the extracted information, evaluate the accuracy and aging of the meter, and facilitate the staff to replace and maintain the meter in time, which not only improves the supervision ability of the meter, but also is beneficial to ensure the accuracy of the meter data.

[0040] The present application collects meter image information and extracts relevant information of the meter itself, extracts and records the nameplate information of the meter, and can also collect and analyze the current, voltage and pulse signal information inside the meter to evaluate the accuracy, health and remaining life of the meter, so that the staff can timely understand the specific situation of each meter, and the meter can be more effectively managed.

[0041] The present application processes the collected electric meter image, and the characters in the image can be inclined due to the shooting angle problem, the present application can identify and correct the inclination of the region with characters, and can also inhibit the problem that the characters are difficult to identify caused by reflection, thereby improving the accuracy of the present application in extracting the electric meter image information, and the extracted information can be used as the identity information of the electric meter, facilitating subsequent query of the electric meter condition.

[0042] The present application analyzes the collected current, voltage and pulse signals to evaluate the accuracy of the electric meter, and as the use time increases, the electric meter itself will age, the present application can evaluate the aging condition of the electric meter by calculating the health degree and remaining life of the electric meter, which helps the staff to determine whether to maintain or replace the electric meter, increases the supervision degree while obtaining information in time, and improves the practicability.

[0043] The advantages of the additional aspects of the present application will be partially given in the following description, partially become obvious from the following description, or be known through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description, obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of the provided drawings.

[0045] Figure 1 The intelligent electric meter information extraction and intelligent analysis method flow chart is provided for the embodiment 1 of the present application. DETAILED DESCRIPTION

[0046] The present application will be further described below in combination with the drawings and embodiments.

[0047] It should be pointed out that the following detailed description is exemplary and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0048] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0049] Embodiments in the application and features in the embodiments can be combined with each other in the case of no conflict.

[0050] Embodiment 1

[0051] The embodiment provides an intelligent meter information extraction and intelligent analysis method, as shown in the figure, which comprises the following steps. Figure 1

[0052] extracting meter external information from the acquired meter image;

[0053] acquiring meter internal information of the meter under use, the meter internal information comprising pulse signals, current signals and voltage signals;

[0054] performing standardization processing on the pulse signals, and performing accuracy evaluation and confidence verification on the meter under test according to the obtained standardized pulse constant;

[0055] constructing a feature matrix according to the current signals and the voltage signals of the meter under test and historical fault data, performing health scoring on the meter under test according to the feature matrix, and predicting the remaining life of the meter under test according to the health score of the meter under test at continuous time points;

[0056] completing monitoring analysis on the meter under test according to the extracted meter external information, the accuracy evaluation and the confidence verification result, and the health score and the remaining life.

[0057] The embodiment method will be described in detail below.

[0058] S1, acquiring meter images and meter internal information in the meter use process, and extracting meter external information from the meter images.

[0059] Specifically, it comprises the following steps.

[0060] S11, extracting local features in the meter image.

[0061] Further, specifically, it comprises the following steps.

[0062] ​S111, a CNN-LSTM hybrid architecture combining convolutional neural network (CNN) and long short-term memory network (LSTM) is adopted, and local features of the electric meter image are extracted by the convolutional neural network:

[0063] ;

[0064] wherein, represents the input electric meter image, the dimension is height H width W and the number of channels C. represents the local feature map output by the CNN, and the number of channels in this embodiment is 3.

[0065] The above method can extract local texture, edge and other low-level visual features of the electric meter image, which is convenient for subsequent image recognition work.

[0066] S112, after the local feature map output by the CNN is unfolded into a sequence feature according to the spatial position, the sequence feature is encoded by a bidirectional LSTM:

[0067] ;

[0068] wherein, represents the hidden state output at time step , which encodes the context information, and T represents the product of the length and width.

[0069] The above method is used to model the time sequence relationship between characters, and solves the problem that the characters are dislocated in space due to the inclination of the electric meter image caused by the shooting angle during shooting.

[0070] S12, dynamically focusing on the area related to characters in the electric meter image, for correcting the inclined character information, and suppressing the interference caused by the reflection during shooting on the character information extraction of the electric meter image.

[0071] Further, specifically:

[0072] S121, a spatial attention mechanism is established to solve the character inclination during shooting:

[0073] ;

[0074] ;

[0075] wherein, represents the local feature map position feature vector of the previous time step; hidden state of the previous time step; , denotes a weight matrix; v denotes a weight vector; denotes the unnormalized attention score of position in the local feature map ; denotes the unnormalized attention score of any position in the local feature map ; denotes the normalized spatial attention weight at position , reflecting the importance of the position feature; the weight value does not represent the degree of inclination of the character, but reflects the importance of a certain position in the ammeter image, for example to the key area detection, the greater the value, indicating that the area contains more relevant features, such as more text characters, etc., the smaller the value, indicating that the text character in the area is less.

[0076] The high-weight area usually contains features that can be used to detect the inclination, such as the edge of the ammeter, the corner point, etc., which need to be extracted from these areas in the subsequent through geometric analysis to extract the inclination angle and correct it, which can be achieved by Hough transform, etc. The above content in the embodiment is used to determine the inclination of the text character, and the inclined text character can be corrected through the prior art in the subsequent, which will not be described in detail here.

[0077] S122, a channel attention mechanism is established to suppress invalid high-frequency features caused by reflection and enhance useful channels:

[0078] ;

[0079] wherein, denotes global average pooling on the local feature map , outputting a C-dimensional vector; denotes the weight of the fully connected layer; denotes the channel attention weight of the cth channel; is an activation function.

[0080] S123, a dynamic feature enhancement formula is established to fuse spatial attention and channel attention for enhancing features in the ammeter image:

[0081] ;

[0082] wherein, denotes channel-wise multiplication; denotes the enhanced feature map, which has the same dimension as The same.

[0083] S13, compensate for overexposure of the reflective area, perform optical compensation on the feature-enhanced electric meter image to obtain a corrected electric meter image.

[0084] Further, specifically:

[0085] S131, suppress high-frequency details in the feature-enhanced electric meter image through a low-pass filter, and take the suppressed image as an ambient light intensity map; the ambient light intensity map reflects the global light distribution of the electric meter image and belongs to low-frequency information. After suppressing high-frequency details through a low-pass filter (such as a Gaussian filter), the remaining low-frequency components are the ambient light estimate.

[0086] S132, extract high-frequency components in the enhanced feature map through a high-pass filter method, denoted as .

[0087] S133, establish an optical property compensation formula:

[0088] ;

[0089] wherein, indicates an attenuation coefficient; indicates the electric meter image after optical compensation; and R is the ambient light intensity map.

[0090] In the above manner, the overexposure phenomenon of the reflective area in the electric meter image can be compensated for, and the original texture of the electric meter image can be restored, thereby helping to make subsequent image recognition more accurate.

[0091] S14, extract electric meter external information in the electric meter image after optical compensation.

[0092] Specifically, the electric meter external information mainly includes one or more of the accuracy level (such as 0.5S, 1.0 level, etc.), pulse constant (imp / kWh), production batch code, manufacturing unit information, calibration date and validity period, measurement license mark, and electric meter number on the electric meter. These information are used to upload to the electric meter supervision system, so as to facilitate the staff to understand various aspects of the electric meter.

[0093] Among them, the accuracy level will usually be marked on the panel, indicating the measurement accuracy level of the electric meter. For example, 0.5S, 1.0 level, etc.; but as the electric meter is used for a long time, the internal components of the electric meter will age, and the accuracy may change, so the accuracy of the electric meter needs to be evaluated in subsequent work.

[0094] The pulse constant (imp / kWh) may not be directly displayed on the panel, but some electric meters will be marked on the technical parameter area or nameplate.

[0095] Production batch code usually appears in the form of bar code or numbers on the meter housing or nameplate for tracing production information.

[0096] Manufacturing unit information includes manufacturer name, trademark and model, usually located in a prominent position on the panel.

[0097] The calibration date and validity period are usually printed on the calibration label or nameplate, and some meters will indicate the next calibration time.

[0098] The metering license mark is a mandatory identification indicating that the meter meets the relevant metrological standards, usually displayed with the license number.

[0099] S15, the internal information of the meter during use includes the pulse signal, voltage signal and current signal of the meter.

[0100] In this embodiment, a handheld camera can be used to take pictures of the meter, or a scanner can be used to obtain the meter image according to actual needs; a photoelectric sensor can be used to obtain the pulse signal generated by the rotation of the meter dial, and the voltage signal and current signal can be obtained by using a current transformer and a voltage sampling circuit. The obtained information is uploaded to the meter supervision system through the cloud server. The transmission method of the above information is known to those skilled in the art, and will not be described in detail here.

[0101] S2, standardizing the pulse signal in the collected internal information of the meter.

[0102] Specifically includes:

[0103] S21, calculating the meter pulse constant according to the obtained pulse signal :

[0104] ;

[0105] Where N represents the number of pulses of the pulse signal obtained in one cycle; E represents the actual power consumption in one cycle, with the unit of kWh; In this embodiment, 24 hours of data are continuously collected every 15 minutes; In this embodiment, a photoelectric sensor or a Hall element can be used to capture the pulse signal generated by the rotation of the meter dial, and the number of pulses is recorded. When installing, the sensor installation position needs to be ensured to be accurate to avoid signal distortion caused by mechanical vibration or electromagnetic interference.

[0106] S22, establishing a standardization formula:

[0107] ;

[0108] Where, is the standardized pulse constant; represents the historical data mean of the pulse constant, with the unit of imp / kWh; standard deviation of historical data of pulse constant; In order to prevent zero constant, 1x10 -5 The above method is used to eliminate the dimensional difference of different electric meters, so that the data conforms to N(0, 1) distribution.

[0109] S3, multi-dimensional evaluation of the electric meter, including accuracy evaluation and confidence verification;

[0110] Specifically, it includes:

[0111] S31, establishing an accuracy evaluation model:

[0112] ;

[0113] wherein, represents the pulse constant of the standard electric meter, represents the pulse accuracy of the electric meter;

[0114] Wherein, the standard electric meter can use a high-precision standard electric energy meter (such as 0.2 level or above) as a reference, synchronously collect electric energy data under the same load condition as the electric meter to be tested, and record the electric energy value (Wh) and pulse number of the standard electric meter as a reference for error calculation;

[0115] By comparing the pulse constant of the electric meter to be tested with that of the standard electric meter, the relative error is calculated, and finally the pulse accuracy of the electric meter is calculated. If the error exceeds the allowed range (such as 0.5S level table allows ±0.5%), it needs to be recalibrated or checked. For the case of large error, the pulse constant estimation value needs to be adjusted and the test is repeated until the error is within a reasonable range.

[0116] Before verification, the voltage, current parameters and pulse constant of the electric meter to be tested need to be correctly set, the verification instrument needs to be preset with the same number of turns as the tested meter, and the load needs to be kept stable during testing to avoid voltage or current fluctuations affecting the accuracy of pulse counting.

[0117] S32, establishing a confidence verification formula:

[0118] ;

[0119] wherein, represents the t-distribution critical value, n represents the amount of historical data of the pulse constant of the electric meter to be tested, because the electric meter needs to be tested regularly, if it has been measured for 3 times before, n=3; each period corresponds to a set of historical data, represents the confidence of the data displayed by the electric meter to be tested, the higher the confidence, the more accurate the data displayed by the electric meter.

[0120] In probability theory and statistics, the t-distribution (t-distribution) is used to estimate the mean of a population that is normally distributed and the variance is unknown according to a small sample. If the population variance is known or when the sample size is large enough, the normal distribution should be used to estimate the population mean, which can be selected according to the actual situation, that is, the actual number of samples, to use t-distribution or normal distribution.

[0121] S4, evaluate the comprehensive health degree of the electric meter by the remaining life, and predict the remaining life of the electric meter.

[0122] Specifically includes:

[0123] S41, combine the current fluctuation of the electric meter, the voltage fluctuation and the historical failure record to construct a feature matrix.

[0124] S42, extract the time domain features and frequency domain features of the feature matrix by the sliding window statistical method.

[0125] S43, establish a health score calculation formula:

[0126] ;

[0127] Wherein, H is the comprehensive health score, which is 0-100 in this embodiment; Indicates the dynamic weight of the i-th feature, which is determined by the entropy weight method, which will not be described here; Indicates the normalized value of the i-th feature ; Indicates the historical failure times of the electric meter to be tested; Is the decay coefficient.

[0128] Wherein, the feature matrix is constructed as follows:

[0129] (1) Select current fluctuation value ΔI, voltage fluctuation value ΔV and historical failure flag F, and historical failure flag F represents the number of occurrences of each type of failure, that is, the historical failure record.

[0130] Because the current and voltage data of the electric meter are recorded, when selecting data, the following method can be used: divide the historical time into multiple time periods, query the current fluctuation value ΔI, voltage fluctuation value ΔV and historical failure flag F of the electric meter in each time period.

[0131] (2) Then, construct a time sequence feature matrix, each row of the time sequence feature matrix represents a corresponding time period, and each row of the time sequence feature matrix is referred to as: .

[0132] Moreover, in the health score calculation formula There are multiple items, and each item needs to be added. In order to facilitate description, i is used to represent the characteristics of the item.

[0133] Therefore, in the present embodiment, the health score calculation formula in the present embodiment is There are two items, one is a frequency domain feature, and the other is a time domain feature. When calculating, the two items need to be added;

[0134] For example, when i=1, it represents a frequency domain feature, then The item contains data related to the frequency domain feature; when i=2, it represents a time domain feature, then The item contains data related to the time domain feature.

[0135] In the above manner, it is helpful to quantify the current state of the electric meter, and the above manner combines the collected real-time data and the once-collected historical records, avoiding single-index misjudgment.

[0136] S44, establish a remaining life prediction model:

[0137] ;

[0138] Wherein, represents the remaining life prediction value of the electric meter at time t, and the unit is hour; is a model fusion coefficient; is a shape parameter; is a characteristic life; represents the health state sequence of the recent n time periods.

[0139] Wherein, and The values of and can be learned from the failure distribution of similar electric meters, and the LSTM training data comes from a cloud aging experiment database, represents the continuous health score from the historical time t-n to the current time t, which is used to analyze the health trend and abnormal situation of the electric meter.

[0140] The above method is used to predict the combined effect of sudden failure (Weibull) and gradual attenuation (LSTM). Compared with a single prediction model, the prediction error of the present method is lower.

[0141] The present scheme can collect image information of the electric meter, process the image, and then extract the corresponding information of the electric meter through image recognition and upload it to the electric meter supervision system, which helps the staff to understand the information of the electric meter, and facilitates the query of the electric meter according to the code of the electric meter.

[0142] For the collected current, voltage, pulse quantity and other aspects, the accuracy of the ammeter, the health score and the remaining life of the ammeter are obtained by calculating through the corresponding module. Since the internal devices of the ammeter will age with use, it is necessary to evaluate through the above-mentioned manner, which helps the staff to understand the condition of the ammeter in time and determine whether to replace or repair it. Moreover, this part of information will also be uploaded to the ammeter supervision system, so that the situation of the ammeter in various aspects can be understood.

[0143] Preferably, the maintenance condition and disassembly condition of the ammeter can also be uploaded to the ammeter supervision system through the above-mentioned manner to realize whole-process tracking of the whole life cycle of the ammeter. Through the above-mentioned manner, complete traceability records can be established for the ammeter from factory detection to operation and maintenance, so that the staff can understand the condition of the ammeter in time, thereby implementing more effective supervision of the ammeter.

[0144] It should be noted that all data acquisition is based on compliance with laws and regulations and user consent, and legal application of data is performed.

[0145] Embodiment 2

[0146] The embodiment provides an intelligent ammeter information extraction and intelligent analysis system, which comprises:

[0147] An image acquisition module configured to extract ammeter external information from the acquired ammeter image to be tested;

[0148] An information acquisition module configured to acquire ammeter internal information when the ammeter to be tested is used, wherein the ammeter internal information comprises a pulse signal, a current signal and a voltage signal;

[0149] An accuracy evaluation module configured to perform standardization processing on the pulse signal, and perform accuracy evaluation and confidence verification on the ammeter to be tested according to the obtained standardized pulse constant;

[0150] A remaining life evaluation module configured to construct a feature matrix according to the current signal and the voltage signal of the ammeter to be tested and historical failure data, perform health scoring on the ammeter to be tested according to the feature matrix, and predict the remaining life according to the health score of the ammeter to be tested at continuous time;

[0151] A supervision module configured to complete monitoring analysis on the ammeter to be tested according to the extracted ammeter external information, the results of accuracy evaluation and confidence verification, and the health score and the remaining life.

[0152] It should be noted that the above modules correspond to the steps described in Embodiment 1, and the above modules have the same examples and application scenarios as the corresponding steps, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules can be executed in a computer system such as a set of computer executable instructions as part of a system.

[0153] In more embodiments, there are also provided:

[0154] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, when executed by the processor, complete the method described in Embodiment 1. For brevity, it will not be described here.

[0155] It should be understood that in the embodiments, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, ready-to-program gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0156] The memory can include read-only memory and random access memory, and provide instructions and data to the processor, and a part of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0157] A computer readable storage medium for storing computer instructions, when executed by a processor, completes the method described in Embodiment 1.

[0158] The method in Embodiment 1 can be directly embodied as a hardware processor to complete, or a combination of hardware and software modules in the processor to complete. The software module can be located in a storage medium mature in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, register, etc. The storage medium is located in the memory, and the processor reads the information in the memory to complete the steps of the above method in combination with the hardware. To avoid repetition, it will not be described in detail here.

[0159] A computer program product includes a computer program, which, when executed by a processor, implements the method described in Embodiment 1.

[0160] The present application also provides at least one computer program product tangibly stored on a non-transitory computer readable storage medium. The computer program product includes computer executable instructions, for example, instructions embodied in program modules, executed by devices at the target real or virtual processor to perform the processes / methods as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The functionality of the program modules can be combined or split between program modules as desired in various embodiments. Machine executable instructions for program modules can be executed within a local or distributed device. In a distributed device, program modules can be located in local and remote memory storage devices.

[0161] Computer program code for carrying out operations of the present application can be written in one or more programming languages. These computer program codes can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program codes, when executed by the computer or other programmable data processing apparatus, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program codes can be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0162] In the context of the present application, computer program code or related data can be carried by any suitable carrier, to enable the device, apparatus or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, and the like. Examples of signals can include electrical, optical, radio, sound or other forms of propagated signals, such as carrier waves, infrared signals, and the like.

[0163] Those skilled in the art can understand that the units and algorithm steps of the examples described in conjunction with the embodiments can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0164] The above describes the specific embodiments of the present application in conjunction with the accompanying drawings, but is not a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without inventive labor are still within the scope of protection of the present application.

Claims

1. A method for intelligent analysis and extraction of information from smart meters, characterized in that, include: Extract external information of the meter from the acquired image of the meter under test; Obtain the internal information of the meter under test when it is in use. The internal information of the meter includes pulse signals, current signals and voltage signals. The pulse signal is standardized, and the accuracy and confidence of the meter under test are evaluated based on the obtained standardized pulse constant. A feature matrix is ​​constructed based on the current and voltage signals of the meter under test and historical fault data. A health score is calculated for the meter under test based on the feature matrix. The remaining lifespan is predicted based on the health score of the meter under test at continuous time intervals. Based on the extracted external information of the electricity meter, the results of accuracy assessment and confidence verification, as well as the health score and remaining lifespan, the monitoring and analysis of the electricity meter under test is completed. Health score: ； Where H represents the overall health score; Indicates dynamic weights; Represents the i-th feature The normalized value; This indicates the number of historical faults of the meter under test; is the attenuation coefficient; m is the total number of features; The remaining lifespan is predicted to be: ； in, This represents the predicted remaining lifespan of the meter under test at time t. These are the model fusion coefficients; For shape parameters; Characteristic lifetime; This represents a sequence of health scores over n time periods.

2. The intelligent analysis method for extracting information from smart meters as described in claim 1, characterized in that, The process of extracting external information from the electricity meter includes: Local feature maps of the electricity meter image are extracted by a convolutional neural network. The local feature maps are then expanded into sequence features according to their spatial location and encoded using a bidirectional LSTM. Calculate the position in the local feature map based on the local feature map and the encoded sequence feature map. Spatial attention weights; channel attention weights of local feature maps are obtained by processing local feature maps through global average pooling and fully connected layers; The local feature map is enhanced by spatial attention weight and channel attention weight to obtain an enhanced feature map. After optical compensation, the external information of the meter is extracted from the enhanced feature map.

3. The intelligent analysis method for extracting information from smart meters as described in claim 2, characterized in that, Spatial attention weights are: ； ； The channel attention weights are: ; in, Representing local feature maps Middle position eigenvectors; The hidden state of the previous time step; , v represents the weight matrix; v represents the weight vector; Representing local feature maps Middle position Unnormalized attention score; Representing local feature maps any position Unnormalized attention score; Indicates the location Normalized spatial attention weights; Represents the local feature map Perform global average pooling to output a C-dimensional vector; Indicates the weights of the fully connected layer; This represents the channel attention weight of the c-th channel; This is the activation function.

4. The intelligent analysis method for extracting information from smart meters as described in claim 1, characterized in that, The accuracy assessment is as follows: ; The confidence level verification is as follows: ; in, The pulse constant of a standard meter. The pulse accuracy of the meter under test; The pulse constant of the meter under test; denoted as the critical value of the t-distribution, where n is the amount of historical data on the pulse constant of the meter under test; The confidence level of the data displayed by the meter under test; The historical average value of the pulse constant of the meter under test. This represents the historical standard deviation of the pulse constant of the meter under test.

5. A smart meter information extraction and intelligent analysis system, characterized in that, The method for intelligent analysis of smart meter information extraction as described in any one of claims 1-4 includes: The image acquisition module is configured to extract external information of the meter from the acquired image of the meter under test; The information acquisition module is configured to acquire the internal information of the meter under test when it is in use. The internal information of the meter includes pulse signals, current signals and voltage signals. The accuracy assessment module is configured to standardize the pulse signal and perform accuracy assessment and confidence verification on the meter under test based on the obtained standardized pulse constant. The remaining life assessment module is configured to construct a feature matrix based on the current and voltage signals of the meter under test and historical fault data, score the health of the meter under test based on the feature matrix, and predict the remaining life based on the health score of the meter under test at continuous time points. The monitoring module is configured to perform monitoring and analysis of the meter under test based on the extracted external information of the meter, the results of accuracy assessment and confidence verification, as well as the health score and remaining life.

6. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-4.

8. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method described in any one of claims 1-4.

Images