Electric energy metering box convenient for intelligent meter reading and meter reading method

Abstract

The invention discloses an electric energy metering box convenient for intelligent meter reading and a meter reading method, the electric energy metering box comprises an electric energy metering box outer shell, and the electric energy metering box outer shell is internally integrated with a multi-source acquisition module, a data collection module, a data collection module and a data collection module; the preprocessing module comprises a filtering subunit, a compression subunit and a format conversion subunit; the communication adjusting module comprises a communication monitoring unit, a prediction calculation unit and a communication control unit; the encryption transmission module comprises an encryption unit and a transmission unit; the data verification module is deployed in the data center server and comprises a decryption engine, a multi-layer verification engine and a log management engine; and the central processing module adopts a heterogeneous multi-core processor and comprises an ARM Cortex-A53 four-core processor and an FPGA (Field Programmable Gate Array) logic unit. The method solves the problems of unstable data transmission, low meter reading efficiency, poor data security and the like in the existing meter reading technology, and has higher creativity and practicability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system automation and metering, in particular to an electric energy metering box and meter reading method facilitating intelligent meter reading. BACKGROUND

[0002] In the prior art, with the rapid development of the power industry, the accuracy of electric energy metering and the efficiency of meter reading are increasingly valued. The traditional manual meter reading method has problems such as low efficiency, large error, high cost, etc., and has been gradually replaced by intelligent meter reading technology.

[0003] At present, the existing intelligent meter reading technology mainly adopts a single communication mode for data transmission, such as power line carrier communication or wireless communication. However, in a complex power environment, a single communication mode is easily disturbed, resulting in unstable data transmission, meter reading failure, etc. At the same time, the existing meter reading method has deficiencies in the comprehensiveness of data collection, often only collecting power consumption data of the electric energy meter, which is difficult to meet the needs of fine management of the power system. In addition, the security of data in the transmission process needs to be improved, and there is a risk of data leakage or tampering. SUMMARY

[0004] The purpose of the present application is to provide an electric energy metering box and meter reading method facilitating intelligent meter reading, which solves the problems of unstable data transmission, low meter reading efficiency, incomplete data collection, poor data security, etc. in the existing intelligent meter reading technology.

[0005] To achieve the above-mentioned purpose, the present application provides an electric energy metering box facilitating intelligent meter reading, comprising:

[0006] An electric energy metering box housing body, the electric energy metering box housing body is integrated with:

[0007] A multi-source collection module including a sensor array and a data aggregation subunit;

[0008] A preprocessing module including a filtering subunit, a compression subunit and a format conversion subunit;

[0009] A communication adjustment module including a communication monitoring unit, a prediction calculation unit and a communication control unit;

[0010] An encrypted transmission module including an encryption unit and a transmission unit;

[0011] A data verification module deployed in a data center server, including a decryption engine, a multi-layer verification engine and a log management engine;

[0012] A central processing module using a heterogeneous multi-core processor, including an ARM Cortex-A53 quad-core processor and an FPGA logic unit.

[0013] Preferably, the sensor array comprises a voltage sensor, a current sensor, a power factor sensor, a temperature and humidity integrated sensor, a state monitoring sensor, a harmonic sensor and a magnetic field sensor.

[0014] Preferably, the modules cooperate through the following modes: the multi-source acquisition module transmits raw data to the preprocessing module through an LVDS interface; the preprocessing module sends standardized data into the encrypted transmission module through a PCIe DMA; the central processing module configures module parameters through an SPI+GPIO control bus; and the communication adjustment module and the encrypted transmission module adjust the communication mode and the encryption parameters through a control line linkage.

[0015] The application also provides a meter reading method of the electric energy metering box for facilitating intelligent meter reading, comprising the following steps:

[0016] S1, configure a multi-source sensor combination to collect various parameters during the operation of the electric meter, use a dynamic perception scheduling mechanism to perform load grading, match a differentiated acquisition frequency, and ensure data continuity through a backup sensor self-diagnosis technology;

[0017] S2, use a self-adaptive intelligent filtering algorithm to eliminate high-frequency pulse interference and random noise, convert heterogeneous data into a JSON format and embed metadata, and reduce transmission volume through a lossless compression algorithm based on data correlation;

[0018] S3, construct an LSTM communication quality prediction model to predict the communication quality in the next 5 minutes, use a multi-index weighted decision algorithm to select the optimal communication mode, dynamically adjust the communication rate based on data priority, and reduce interference through a channel conflict avoidance mechanism;

[0019] S4, generate a dynamic key seed in combination with a timestamp, a device identifier and positioning information, encrypt data through an AES-256 algorithm, use a fragmentation transmission and verification mechanism, and construct a distributed communication node network to realize multi-hop transmission;

[0020] S5, the data center receives and decrypts the encrypted data, performs hierarchical verification on the decrypted data, implements a hierarchical retransmission strategy for abnormal data, and records a full life cycle log to realize data tracing.

[0021] Preferably, in S1, when a sensor fault is detected, the acquisition frequency of a backup sensor of the same type is automatically increased to 1.5 times that of the main sensor.

[0022] Preferably, in S2, for high-frequency pulse interference data, a wavelet transform filtering technology is used for processing; and for random noise data, a mean filtering method is used to eliminate the noise, and a dynamic threshold adjustment mechanism is used to avoid over-denoising.

[0023] Preferably, in S3, the core mathematical equation of the LSTM communication quality prediction model is as follows:

[0024] Forget gate: f t = σ(W f · [h t-1 , x t ]+b f ), where f t is the forget gate output, σ is the sigmoid activation function, W f is the forget gate weight matrix, h t-1 is the hidden layer state at the previous time, x t is the input communication parameter at the current time, and b f is the forget gate bias term.

[0025] Input gate: i t = σ(W i · [h t-1 , x t ]+b i ), where i t is the input gate output, W i is the input gate weight matrix, and b i is the input gate bias term. wherein is the candidate value vector, W C is the cell state update weight matrix, b C is the cell state update bias term, and tanh is the hyperbolic tangent activation function.

[0026] Cell state update: wherein C t is the cell state at the current time, * is element-wise multiplication, and C t-1 is the cell state at the previous time.

[0027] Output gate: o t = σ(W o · [h t-1 , x t ]+b o ), where o t is the output gate output, W o is the output gate weight matrix, b o is the output gate bias term, h t = o t *tanh(C t ), and h t is the hidden layer state at the current time.

[0028] Final prediction output: y t = W y · h t+b y wherein y t is the predicted future 5-minute communication quality evaluation value, W y is the weight matrix of the output layer, b y is the bias term of the output layer.

[0029] Preferably, the future 5-minute communication quality is predicted using the LSTM communication quality prediction model, specifically including the following steps:

[0030] The historical 6-month communication parameter data is normalized to convert the signal strength, bit error rate and channel occupancy rate to the [0, 1] interval, and an input feature sequence is constructed, and the corresponding label is the communication quality evaluation value after 5 minutes;

[0031] The sliding window method is used, and the input features of 30 consecutive time points are used as a sample to input the LSTM communication quality prediction model for training, the number of hidden layer nodes is set to 64, the number of iterations is set to 100, the learning rate is set to 0.001, and the Adam optimizer is used to minimize the mean square error loss function;

[0032] The communication adjustment module collects the current communication parameters every 10 seconds to construct the latest 30-time feature sequence to input the trained LSTM communication quality prediction model to obtain the predicted value of the future 5-minute communication quality; a preset minimum communication quality threshold is set, and when the predicted value is lower than the threshold, the switching process is started 10 seconds in advance, and the communication mode with the highest predicted evaluation value is selected for switching.

[0033] Preferably, in S4, a distributed communication node network is constructed with 3-5 nearby electric energy meters, and when the direct communication quality is poor, the optimal adjacent electric energy meter is selected as a relay node to realize multi-hop transmission.

[0034] Preferably, in S5, the hierarchical verification is divided into three levels in total, the first level is redundancy verification, the CRC32 redundancy check code added in the data transmission process is used to check the data integrity; the second level is cross verification, the calculated value is compared with the collected value by means of the physical correlation between the parameters of the electric energy meter, and if the deviation is more than 5%, it is determined as abnormal; the third level is historical data comparison, the current data is compared with the trained normal historical data distribution model by means of the machine learning anomaly detection model of the isolation forest algorithm, and the deviation range is marked as abnormal.

[0035] Therefore, the beneficial effects of the above-mentioned electric energy metering box and meter reading method for facilitating intelligent meter reading are:

[0036] (1) The present application adopts multi-dimensional data acquisition, which can not only obtain power consumption data, but also collect other related data such as voltage and current, providing comprehensive data support for the fine management of the power system.

[0037] (2) The application can automatically select the optimal communication mode and adjust the communication rate according to the communication environment through adaptive communication adjustment, thereby improving the stability and efficiency of data transmission.

[0038] (3) The application uses a dynamic encryption algorithm to encrypt and transmit data, thereby enhancing the security of data and effectively preventing data leakage and tampering.

[0039] (4) The intelligent data verification step of the application verifies data through multiple verification methods, thereby ensuring the accuracy and integrity of the copied data.

[0040] The technical solutions of the application will be further described in detail below with reference to the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a structural schematic diagram of an embodiment of an intelligent meter reading device of an electric energy metering box of the application;

[0042] Figure 2 is a step diagram of a meter reading method of an electric energy metering box convenient for intelligent meter reading of the application. DETAILED DESCRIPTION

[0043] The technical solutions of the application will be further described in detail below with reference to the accompanying drawings and examples.

[0044] Unless otherwise defined, the technical terms or scientific terms used in the application should be understood as the usual meanings understood by those skilled in the art to which the application belongs. The "first", "second" and similar words used in the application do not represent any order, number or importance, but are only used to distinguish different components. "Include" or "contain" and similar words mean that the elements or objects before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connected" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like only represent relative positional relationships, which may change accordingly when the absolute position of the described object changes.

[0045] Example 1

[0046] As shown in Figure 1 , the application provides an intelligent meter reading device of an electric energy metering box, which comprises an electric energy metering box shell, and the electric energy metering box shell is integrated with a multi-source collection module, a preprocessing module, a communication adjustment module, an encryption transmission module, a data verification module and a central processing module respectively.

[0047] The multi-source acquisition module is composed of a sensor array and a data aggregation subunit. In this embodiment, the sensor array includes: a voltage sensor (0-380V range, accuracy 0.2 level), a current sensor (0-100A range, accuracy 0.2 level), a power factor sensor (measurement range -1 to 1, accuracy 0.5 level), a temperature and humidity integrated sensor (temperature -40~85℃, humidity 0-100%RH, accuracy ±0.5℃ / ±2%RH), a state monitoring sensor (including a cover opening detection component and a pulse counting component), a harmonic sensor (capable of measuring 3-50 harmonics, accuracy 1 level), and a magnetic field sensor (measurement range 0-200mT, accuracy ±2%).

[0048] The sensor array is deployed according to function division. The power factor sensor cooperates with the voltage sensor and the current sensor to directly access the secondary circuit of the power meter inlet end (collects the voltage / current signal in the same path), and obtains the power factor parameter in real time through electrical connection. The state monitoring sensor is distributedly deployed, the cover opening detection component is installed at the hinge connection of the power meter shell (with a built-in magnetic control switch or a photoelectric sensor), and the pulse counting component is close to the pulse output terminal of the power meter (collects the pulse signal through photoelectric coupling), both of which are integrated on the internal PCB board of the meter and communicate with the data aggregation subunit through internal wiring. The temperature and humidity sensor and the magnetic field sensor are installed inside the meter shell, and the harmonic sensor is coupled to the power line through a mutual inductor.

[0049] The data aggregation subunit uses an STM32L431 microprocessor. According to the central processing module instruction, the sensor data is polled through the 485 bus, temporarily stored in the 256KB on-chip RAM, and triggered once every 1KB to push the data to the preprocessing module.

[0050] The preprocessing module includes a filtering subunit, a compression subunit, and a format conversion subunit. In this embodiment, the filtering subunit uses an FPGA chip (model XC7K325T) to integrate wavelet transform and mean filter algorithm hardware accelerator; the compression subunit is based on an ARM Cortex-M7 core (STM32H743) to run a lossless compression algorithm based on data correlation; the format conversion subunit realizes the standardized packaging of heterogeneous data (analog quantity, digital quantity, and pulse quantity) to JSON format through a special protocol chip, and embeds metadata such as acquisition timestamp (accuracy 1ms) and sensor ID. Each subunit realizes data interaction through an internal high-speed bus (AXI-Stream).

[0051] The FPGA of the filtering subunit performs real-time feature analysis on the input data, automatically switches the wavelet transform (high-frequency pulse interference) or mean filter (random noise) mode, and the processing delay is ≤1 ms; the compression subunit stores the difference of similar data (change rate <5%) for more than 3 groups in succession, and the compression ratio can reach 3:1; the format conversion subunit encapsulates the data according to the JSON schema specification, generates a standard frame containing 16-byte metadata+N-byte payload, and transmits it to the encrypted transmission module through the PCIe DMA.

[0052] The communication adjustment module is composed of a communication monitoring unit, a prediction calculation unit and a communication control unit. In this embodiment, the communication monitoring unit uses a radio frequency front-end chip (ADF7021) and a power line carrier module (PL360) to collect parameters such as signal strength (-110 to -30 dBm), bit error rate (0-1), channel occupancy rate (0-100%) of LoRa / NB-IoT / power line carrier in real time; the prediction calculation unit uses an edge AI chip (RK3588) to deploy an LSTM communication quality prediction model, supporting 5-minute-level communication quality prediction; the communication control unit realizes communication mode switching through a multi-channel switch chip (MAX4619), dynamically adjusts the communication rate (100 kbps-1 Mbps) combined with the PHY chip (DP83848), and integrates a channel conflict detection circuit (based on the CSMA / CA protocol).

[0053] The communication monitoring unit collects 3 groups of communication parameters per second, which are stored in the FIFO buffer after AD conversion; the RK3588 chip of the prediction calculation unit reads the buffer data every 10 seconds, inputs the LSTM model to calculate the communication quality score in the next 5 minutes; when the score of a certain communication mode is <50, the MAX4619 switch of the communication control unit switches to the backup channel, and at the same time, the transmission rate of the highest priority data is locked to 1 Mbps through the PHY chip.

[0054] The encrypted transmission module includes an encryption unit and a transmission unit. In this embodiment, the encryption unit uses a national secret chip (SM4 algorithm accelerator), combines the space-time information output by the real-time clock (RTC, accuracy 1 ppm) and the positioning module (dual-mode BDS B1I / GPS L1), generates a dynamic key seed, and realizes data encryption through the AES-256 algorithm; the transmission unit includes a radio frequency module (SX1278) and a power line carrier interface, supports data packet fragmentation (maximum 1024 bytes / piece) and fragmentation verification (CRC16), is connected with the encryption unit through the SPI interface, and receives the encrypted data frame.

[0055] The encryption unit obtains a timestamp from the RTC every 100 ms, generates a 128-bit key seed in combination with the longitude and latitude information (accurate to 0.1 m) of the Beidou module, and encrypts the preprocessed data through the AES-256 algorithm; the transmission unit sends the encrypted data in slices of 1024 bytes, each slice adding 8 bytes of sequence number + 2 bytes of CRC16 check, through the SX1278 module (LoRa mode), and triggers a backoff mechanism (backoff time 0-10 ms randomly) if a busy channel is detected.

[0056] The data verification module is deployed in a data center server (using an Intel Xeon Gold 6330 processor) and is composed of a decryption engine, a multi-layer verification engine, and a log management engine. In this embodiment, the decryption engine implements the dynamic key generation algorithm corresponding to the encryption unit through software; the multi-layer verification engine runs on a GPU cluster (NVIDIA A100) and performs parallel processing of CRC32 redundancy verification, physical formula cross verification (P = UIcosφ, etc.), and machine learning anomaly detection (based on the Isolation Forest algorithm); the log management engine uses a distributed database (MongoDB) to record operation logs throughout the life cycle of the data (collection-transmission-verification) and supports traceability queries based on timestamps and device IDs.

[0057] The decryption engine of the data center server reconstructs the key according to the space-time information in the data packet, and the decryption time is ≤50 ms; the multi-layer verification engine performs three-level verification in parallel on the GPU, and the single data packet processing time is ≤100 ms; the log management engine generates one log record (including device ID, timestamp, verification result) for every 1 data processed, stores it in the MongoDB cluster, and supports querying historical records for nearly 3 months according to the device ID.

[0058] The central processing module uses a heterogeneous multi-core processor (Zynq UltraScale+MPSoC) containing an ARM Cortex-A53 quad-core processor and an FPGA logic unit. It communicates with the multi-source collection module through an I2C bus to control the collection frequency schedule; it connects the preprocessing module and the encryption transmission module through a PCIe interface to realize data flow control; it receives state feedback from the communication adjustment module through an Ethernet interface and outputs communication switching instructions. It has a built-in real-time operating system (FreeRTOS) to realize the coordinated scheduling of various modules (scheduling period 10 ms).

[0059] The ARM core of the Zynq UltraScale+MPSoC runs the load level judgment algorithm, updates the collection frequency instruction every 500 ms, and sends it to the multi-source collection module through I2C; the FPGA logic unit realizes hardware-level synchronization of various modules (synchronization error <100 ns), and when a sensor fault signal is detected, it immediately triggers a backup sensor frequency boost instruction (1.5 times the current frequency).

[0060] The modules are coordinated through the following ways: the multi-source acquisition module transmits raw data to the preprocessing module through an LVDS interface; the standardized data processed by the preprocessing module is sent to the encrypted transmission module through a PCIe bus; the central processing module realizes parameter configuration and state monitoring of each module through a control bus (SPI+GPIO); and the communication adjustment module and the encrypted transmission module realize linkage adjustment of communication mode and encryption parameters through a special control line.

[0061] As shown in Figure 2 the embodiment, a meter reading method of an electric energy metering box facilitating intelligent meter reading is adopted, including the following steps:

[0062] S1, configure a multi-source sensor combination to collect various parameters during the operation of the electric meter, use a dynamic perception scheduling mechanism to classify loads, match differentiated collection frequencies, and use a backup sensor self-diagnosis technology to ensure data continuity.

[0063] Among them, the multi-source sensor combination includes voltage, current, power factor, temperature, humidity, state monitoring, harmonic and magnetic field sensors, and can collect multi-dimensional data such as power consumption, voltage, current, power factor, environmental parameters, equipment state, harmonic content and surrounding magnetic field strength.

[0064] Through the dynamic perception scheduling mechanism, the central processing module divides the load into three levels according to the real-time current and power values: peak (current ≥ 80% rated current or power ≥ 80% rated power), normal (current 30%-80% rated current and power 30%-80% rated power), and valley (current ≤ 30% rated current and power ≤ 30% rated power).

[0065] Match the differentiated collection frequency, the core sensor (voltage, current, etc.) 1-2 seconds / time at peak, and the environmental sensor 5 seconds / time; the core sensor 5-10 seconds / time and the environmental sensor 30 seconds / time at normal; the core sensor 30-60 seconds / time and the environmental sensor 180 seconds / time at valley.

[0066] Linkage sensor fault self-diagnosis technology, when detecting sensor failure, automatically increase the collection frequency of the same type of backup sensor to 1.5 times that of the main sensor, ensure the continuity and accuracy of data collection, and provide comprehensive data support for power system management.

[0067] S2, use an adaptive intelligent filtering algorithm to eliminate high-frequency pulse interference and random noise, convert heterogeneous data into JSON format and embed metadata, store only the difference value of continuous similar data through a lossless compression algorithm based on data correlation, thereby significantly reducing data transmission volume.

[0068] For high-frequency pulse interference data, wavelet transform filtering technology is used for processing; for random noise data, mean filtering method is used to eliminate, and through dynamic threshold adjustment mechanism to avoid excessive noise phenomenon. In the data format conversion link, the heterogeneous data is uniformly converted to JSON format by using standardized data packaging protocol, and the metadata information is embedded. In addition, with the help of lossless compression algorithm based on data correlation, the data transmission amount is significantly reduced.

[0069] S3, construct LSTM communication quality prediction model to predict future 5 minutes communication quality, adopt multi-index weighted decision algorithm to select optimal communication mode, dynamically adjust communication rate based on data priority, and reduce interference through channel conflict avoidance mechanism.

[0070] Among them, the core mathematical equation of LSTM communication quality prediction model is as follows:

[0071] Forget gate: f t =σ(W f ·[h t-1 ,x t ]+b f ), wherein f t is the output of the forget gate, σ is the sigmoid activation function, W f is the weight matrix of the forget gate, h t-1 is the hidden layer state at the last moment, x t is the input communication parameter at the current moment, and b f is the bias term of the forget gate.

[0072] Input gate: i t =σ(W i ·[h t-1 ,x t ]+b i ), wherein i t is the output of the input gate, W i is the weight matrix of the input gate, and b i is the bias term of the input gate; wherein, is the candidate value vector, W C is the cell state update weight matrix, b C is the cell state update bias term, and tanh is the hyperbolic tangent activation function.

[0073] Cell state update: wherein C t is the cell state at the current moment, * is element-level multiplication, and C t-1 is the cell state at the last moment.

[0074] Output gate: o t =σ(W o·[h t-1 , x t ]+b o ), where o t is the output gate output, W o is the weight matrix of the output gate, b o is the bias term of the output gate; h t =o t *tanh(C t ), where h t is the hidden layer state at the current moment.

[0075] Final prediction output: y t =W y ·h t +b y , where y t is the predicted communication quality evaluation value for the next 5 minutes, W y is the weight matrix of the output layer, b y is the bias term of the output layer.

[0076] Use the LSTM communication quality prediction model to predict the communication quality in the next 5 minutes. The specific steps include:

[0077] ① The communication parameter data of the past 6 months (collected every 10 seconds) was normalized, and the signal strength (-110dBm to -30dBm), bit error rate (0 to 1), and channel occupancy (0 to 100%) were converted to the interval [0, 1]. The input feature sequence was constructed, and the corresponding label was the communication quality evaluation value after 5 minutes (determined by expert scoring and actual transmission success rate).

[0078] ② Using the sliding window method, the input features of 30 consecutive moments are taken as a sample (time window length 300 seconds), and input into the LSTM communication quality prediction model for training. The number of hidden layer nodes is set to 64, the number of iterations is 100, the learning rate is 0.001, and the Adam optimizer is used to minimize the mean square error loss function.

[0079] ③ The communication adjustment module collects current communication parameters every 10 seconds, constructs the latest 30-minute feature sequences, and inputs them into the trained LSTM communication quality prediction model to obtain a five-minute communication quality forecast. Preset communication quality thresholds (60 and 50 points) are set. When the predicted value falls below 60, a communication mode switch warning is triggered. When the predicted value falls below 50, the switch process is initiated 10 seconds in advance, selecting the communication mode with the highest predicted evaluation value.

[0080] Multi-index weighted decision algorithm comprehensively considers signal strength (weight 40%), transmission delay (weight 30%), communication cost (weight 20%), and energy consumption level (weight 10%) and the like, and automatically selects the optimal communication mode. When dynamically adjusting the communication rate, a data priority classification mechanism is introduced, the power consumption data of the electric energy meter is defined as the highest priority, the voltage and current data are defined as the medium priority, and the temperature, humidity and other environmental data are defined as the low priority. According to the data priority and the real-time requirement, the communication rate is dynamically adjusted, the highest priority data uses the highest communication rate (1 Mbps), and the low priority data uses a lower communication rate (100 kbps), which reduces the overall energy consumption while ensuring the real-time transmission of key data.

[0081] The channel conflict avoidance mechanism automatically selects an idle channel for data transmission by monitoring the channel usage of surrounding communication devices, thereby reducing communication interference.

[0082] S4, a time-space double-factor dynamic encryption algorithm is adopted, a dynamic key seed is generated by combining a time stamp, a device identifier and positioning information, data is encrypted through an AES-256 algorithm, and it is ensured that each data packet key is unique and dynamically changes.

[0083] A fragmentation transmission and verification mechanism is adopted, data packets exceeding 1024 bytes are segmented into fragments with a check code and a serial number, the receiving end reassembles according to the serial number, and lost fragments only need to be retransmitted.

[0084] A distributed communication node network is constructed to realize multi-hop transmission, a distributed communication node network is constructed with 3-5 electric energy meters nearby, when the direct communication quality is poor, the optimal adjacent electric energy meter is automatically selected as a relay node to realize multi-hop transmission, and the coverage is expanded.

[0085] S5, the data center receives and decrypts the encrypted data, performs hierarchical verification on the decrypted data, implements a hierarchical retransmission strategy for abnormal data, records the whole life cycle log to realize data tracing, realizes the whole life cycle tracing of data, and facilitates fault troubleshooting and responsibility identification.

[0086] The decryption of the encrypted data is performed by a key dynamic generator, a decryption key is regenerated according to the time stamp, the unique device identifier and the spatial position information in the data packet, and the data is decrypted.

[0087] In the hierarchical verification, three levels are divided in total, the first level is redundancy verification, the data integrity is checked by adding CRC32 redundancy check code in the data transmission process; the second level is cross verification, the calculated value is compared with the collected value by means of the physical correlation between parameters of the electric energy meter, and the deviation is more than 5% to determine the abnormality; the third level is historical data comparison, the current data is compared with the trained normal historical data distribution model by using the machine learning anomaly detection model of the isolated forest algorithm, and the deviation range is marked as abnormal.

[0088] In the embodiment, for the data that fails to pass the verification, the retransmission instruction is sent once for slight abnormality (single verification item fails to pass); the retransmission instruction is sent three times continuously for serious abnormality (multiple verification items fail to pass or cannot be decrypted), and if it still fails to pass, the electric energy meter is marked as a fault state and the operation and maintenance personnel are notified.

[0089] Therefore, the electric energy metering box and the meter reading method convenient for intelligent meter reading have high creativity and practicality, and can solve the problems of unstable data transmission, low meter reading efficiency, poor data security and the like in the existing meter reading technology.

[0090] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application but not to limit them, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can still be modified or replaced by the equivalent, and these modifications or replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. An electric energy meter box that facilitates intelligent meter reading, characterized in that: include: The outer shell of the electric energy meter box is integrated with: Multi-source acquisition module: includes sensor array and data aggregation sub-unit; Preprocessing module: includes filtering subunit, compression subunit and format conversion subunit; Communication regulation module: includes communication monitoring unit, prediction calculation unit and communication control unit; Encrypted transmission module: includes encryption unit and transmission unit; The data verification module is deployed on the data center server and includes a decryption engine, a multi-layer verification engine, and a log management engine; The central processing module uses a heterogeneous multi-core processor, including an ARM Cortex-A53 quad-core processor and an FPGA logic unit.

2. The electric energy meter box for facilitating intelligent meter reading according to claim 1, characterized in that: The sensor array includes voltage sensors, current sensors, power factor sensors, temperature and humidity sensors, condition monitoring sensors, harmonic sensors and magnetic field sensors.

3. The electric energy meter box for facilitating intelligent meter reading according to claim 1, characterized in that: The modules collaborate with each other in the following ways: the multi-source acquisition module transmits the raw data to the pre-processing module via the LVDS interface; the pre-processing module sends the standardized data to the encryption transmission module via PCIe DMA; the central processing module configures the parameters of each module through the SPI+GPIO control bus; the communication adjustment module and the encryption transmission module adjust the communication mode and encryption parameters through the control line linkage.

4. A meter reading method for an electric energy meter box that facilitates intelligent meter reading, characterized in that: The following steps are involved: S1. Configure a multi-source sensor combination to collect various parameters during meter operation, use a dynamic sensing scheduling mechanism to classify loads, match differentiated collection frequencies, and ensure data continuity through backup sensor self-diagnosis technology; S2 uses an adaptive intelligent filtering algorithm to eliminate high-frequency pulse interference and random noise, converts heterogeneous data into JSON format and embeds metadata, and reduces transmission volume through a lossless compression algorithm based on data correlation; S3. Build an LSTM communication quality prediction model to predict communication quality in the next 5 minutes. Use a multi-index weighted decision algorithm to select the optimal communication method, dynamically adjust the communication rate based on data priority, and reduce interference through a channel conflict avoidance mechanism. S4. Generate a dynamic key seed by combining the timestamp, device identifier, and location information. Encrypt the data using the AES-256 algorithm, adopt a fragmented transmission and verification mechanism, and build a distributed communication node network to achieve multi-hop transmission. S5. The data center receives and decrypts the encrypted data, performs hierarchical verification on the decrypted data, implements a hierarchical retransmission strategy for abnormal data, and records full life cycle logs to achieve data traceability.

5. The meter reading method of an electric energy meter box that facilitates intelligent meter reading according to claim 4, characterized in that: In S1, when a sensor failure is detected, the acquisition frequency of the backup sensor of the same type is automatically increased to 1.5 times that of the main sensor.

6. The meter reading method of an electric energy meter box that facilitates intelligent meter reading according to claim 4, characterized in that: In S2, high-frequency pulse interference data is processed using wavelet transform filtering technology; random noise data is eliminated using mean filtering method, and excessive noise reduction is avoided through a dynamic threshold adjustment mechanism.

7. The meter reading method of an electric energy meter box that facilitates intelligent meter reading according to claim 4, characterized in that: In S3, the core mathematical equation of the LSTM communication quality prediction model is as follows: Forget gate: f t =σ(W f ·[h t-1 , x t ]+b f ), where f t is the output of the forget gate, σ is the sigmoid activation function, W f is the forget gate weight matrix, h t-1 is the hidden layer state at the previous moment, x t is the communication parameter input at the current moment, b f is the forget gate bias term; Input gate: i t =σ(W i ·[h t-1 , x t ]+b i ), where i t is the input gate output, W i is the input gate weight matrix, b i is the input gate bias term; in, is the candidate value vector, W C Update the weight matrix for the cell state, b C is the cell state update bias term, and tanh is the hyperbolic tangent activation function; Cell status update: Among them, C t is the cell state at the current moment, * is element-wise multiplication, C t-1 is the cell state at the previous moment; Output gate: o t =σ(W o ·[h t-1 , x t ]+b o ), where o t is the output gate output, W o is the weight matrix of the output gate, b o is the bias term of the output gate; h t =o t *tanh(C t ), where h t is the hidden layer state at the current moment; Final prediction output: y t =W y ·h t +b y , where y t is the predicted communication quality evaluation value for the next 5 minutes, W y is the weight matrix of the output layer, b y is the bias term of the output layer.

8. The meter reading method of an electric energy meter box that facilitates intelligent meter reading according to claim 7, characterized in that: Use the LSTM communication quality prediction model to predict the communication quality in the next 5 minutes. The specific steps include: The communication parameter data for the past six months is normalized, and the signal strength, bit error rate, and channel occupancy rate are converted to the [0, 1] interval to construct an input feature sequence. The corresponding label is the communication quality evaluation value after 5 minutes. Using the sliding window method, the input features of 30 consecutive moments are used as a sample and input into the LSTM communication quality prediction model for training. The number of hidden layer nodes is set to 64, the number of iterations is set to 100, the learning rate is set to 0.001, and the Adam optimizer is used to minimize the mean square error loss function. The communication adjustment module collects the current communication parameters every 10 seconds, constructs the latest 30-moment feature sequences and inputs them into the trained LSTM communication quality prediction model to obtain the communication quality prediction value for the next 5 minutes. It presets a minimum communication quality threshold. When the predicted value is lower than the threshold, the switching process is started 10 seconds in advance, and the communication method with the highest predicted evaluation value is selected for switching.

9. The meter reading method of an electric energy meter box facilitating intelligent meter reading according to claim 4, characterized in that: In S4, a distributed communication node network is built with 3-5 nearby electricity meters. When the direct communication quality is poor, the optimal adjacent electricity meter is automatically selected as the relay node to achieve multi-hop transmission.

10. The meter reading method of an electric energy meter box facilitating intelligent meter reading according to claim 4, characterized in that: In S5, the hierarchical verification is divided into three levels. The first level is redundancy verification, which uses the CRC32 redundancy check code added during data transmission to check data integrity. The second level is cross-verification, which uses the physical correlation between the various parameters of the electricity meter to compare the calculated value with the collected value. Any deviation exceeding 5% is considered an abnormality. The third level is historical data comparison. By using the machine learning anomaly detection model of the isolation forest algorithm, the current data is compared with the trained normal historical data distribution model. If it deviates from the range, it is marked as an anomaly.

Images