An intelligent disaster prevention and energy-saving socket and its management system based on load identification algorithm

Abstract

A smart disaster prevention and energy-saving socket based on a load identification algorithm and a management system thereof, comprising a socket upper cover, a plug guide groove and a retaining frame, a jack reed, a power-on indicator light, a socket bottom cover and a PCB circuit board, each jack corresponds to a power-on indicator light, and the PCB circuit board comprises an MCU controller, a WIFI module, a USB serial port, an NFC module, an electric energy metering module, a relay, a terminal block, a fuse, an AC‑DC converter and a power supply module. The present invention provides a smart disaster prevention and energy-saving socket based on a load identification algorithm and a management system thereof, which uses a WeChat applet to perform online management of electrical appliances, embeds an NFC module into a self-designed socket circuit, and can provide services such as identification of electrical appliances as soon as they are inserted, conversion of ordinary appliances to smart appliance management as soon as an NFC tag is attached, denial of access to prohibited electrical appliances in specific scenarios, and remote management of appliances.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent sockets, and in particular to an intelligent disaster prevention and energy-saving socket based on a load identification algorithm and a management system thereof. Background Art

[0002] Electrical appliances have become essential household equipment in every family. Sockets are necessary devices for connecting electrical appliances to power. If they can be controlled and managed intelligently, then the management of sockets can be used to achieve intelligent control of ordinary household appliances to a large extent, so as to achieve simple management of the working status of electrical appliances, safe electricity use, and reduce the waste of electricity resources caused by electrical appliances in standby mode.

[0003] Li Miao and other scholars proposed a WiFi-based smart socket controlled by WeChat platform, and Zhou Jingjing and others proposed a power-off delay control method for smart sockets in the IoT environment, using AD converters to detect the power of smart sockets and feed back the power consumption information of the electrical equipment connected to the smart sockets to users in real time. Users can turn on and off the sockets at any time to achieve smart socket control, thereby achieving energy saving. Few scholars have involved smart sockets based on load identification, and the main function of smart sockets currently studied by scholars is the power-off function, but they have ignored the realization of functions such as identification and management of various electrical appliances.

[0004] Existing smart socket solutions generally have functions such as remote control of socket power on and off, total power consumption statistics, and timer switches, but they are unable to identify specific electrical appliances connected to specific sockets and manage the power consumption of individual electrical appliances. To achieve the management of ordinary individual electrical appliances, you can only replace smart homes, such as replacing ordinary electric fans with a series of smart products such as "Mijia Electric Fans", which means that they are not completely "smart". In terms of safety, ordinary smart sockets only have simple overload protection and other functions, and cannot meet special safety needs in special scenarios. Nowadays, fires caused by lithium battery charging are everywhere in campuses and communities. Existing load identification function equipment such as products of foreign companies such as Sense use active learning labeling technology. Furniture small appliances usually take months or even years to identify, and there are shortcomings such as poor recognition accuracy of appliances of the same model and the inability to use active learning technology for real-time learning prediction of items that do not exist in the background tag library. Summary of the invention

[0005] In order to overcome the shortcomings of existing smart sockets such as the inability to identify inserted appliances, inconvenience in managing the access of high-risk appliances according to scenarios, inability to intelligently manage ordinary appliances, and insufficient energy-saving display functions such as electricity charges and power, the present invention provides an intelligent disaster prevention and energy-saving socket based on a load identification algorithm and its management system, uses a WeChat applet to manage electrical appliances online, embeds an NFC module into a self-designed socket circuit, and designs a dual-verification intelligent labeling technology for electrical appliances based on NFC tags and load identification algorithms to solve the problems that are difficult to solve by algorithms such as the extremely slow recognition speed of small appliances using active learning algorithms, the inability to predict new data in real time, and the recognition of appliances of the same model. Services such as identification of electrical appliances as soon as they are inserted, conversion of ordinary appliances to smart appliance management as soon as an NFC tag is affixed can be provided, access denial of prohibited appliances in specific scenarios, and remote management of appliances can be provided.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] The invention discloses an intelligent disaster prevention and energy-saving socket based on a load identification algorithm, comprising a socket upper cover, a plug guide groove and a retaining frame, a socket reed, a power-on indicator light, a socket bottom cover plate and a PCB circuit board, each socket corresponds to a power-on indicator light, the PCB circuit board comprises an MCU controller, a WIFI module, a USB serial port, an NFC module, an energy metering module, a relay, a terminal block, a fuse, an AC-DC converter and a power module, the MCU controller adopts an ESP32-S module, and the ESP32-S module integrates an antenna switch, a radio frequency Balun, a power amplifier, a low noise amplifier, a filter and a power supply tube. The NFC module includes an NFC reader / writer chip and an NFC antenna, the NFC antenna is connected to the NFC reader / writer chip, the NFC reader / writer chip is connected to the ESP32-S module, the electric energy metering module adopts the HLW8012 chip, the 2nd and 3rd pins of the HLW8012 chip are the current differential signal input terminals, the 4th pin is the voltage sampling signal input terminal, the 6th and 7th pins of the HLW8012 chip are respectively connected to the 13th and 16th pins of the ESP32-S module, and the ESP32-S module calculates the power, current and voltage values ​​by measuring the period of the high-frequency pulses output by the 6th and 7th pins of the HLW8012 chip;

[0008] The power module is connected to 220V AC through the terminal block. After the current flows through the fuse, it is converted into voltage through the AC-DC converter to power other circuit modules. The ESP32-S module is connected to the user terminal through a relay and to the cloud through the WIFI module. An NFC tag is attached to the plug end of the electrical equipment. The data information in the NFC tag matches the relevant parameters of the home appliance.

[0009] Furthermore, the NFC reader / writer chip adopts the PN5321A3HN chip, and the 27th and 28th pins of the PN5321A3HN chip are respectively connected to the 36th and 33th pins of the ESP32-S module, and the PN5321A3HN chip communicates with the ESP32-S module through I2C; the USB serial port adopts the CH340C chip, and the 2nd and 3rd pins of the CH340C chip are respectively connected to the U0TXD and U0RXD pins of the ESP32-S module.

[0010] An intelligent disaster prevention and energy-saving socket management system based on a load identification algorithm, the system architecture includes a data acquisition and computing architecture, a cloud and a terminal, the data acquisition and computing architecture includes a sensor layer and an edge gateway layer, the cloud includes a cloud server and a GPU server, the terminal includes an intelligent disaster prevention and energy-saving socket and a user terminal, and the user interacts with the sensor layer through the user terminal;

[0011] The MCU controller and WIFI module constitute the edge gateway layer, and the power metering module constitutes the sensor layer. The sensor layer collects data and uploads it to the edge gateway. The edge gateway runs the activation monitoring algorithm to obtain valid data and transmits it to the GPU server interface layer. The server interface layer stores the effective data of the electrical appliance in the MySQL database and the Redis database and uses the label-consistent convolutional transform neural network model for load identification and uploads it to the cloud server. The cloud server is responsible for the deployment of the label-consistent convolutional transform neural network model and various functional interfaces, providing users with electrical appliance identification, real-time warning, intelligent analysis of electricity consumption data, and electricity consumption data callback viewing services, and also provides App and WeChat applets to interact with user terminals.

[0012] The beneficial effects of the present invention are mainly manifested in: the data acquisition and computing architecture adopts a sensor layer, an edge gateway layer, and a cloud layer, which ensures the quality of transmitted data and saves bandwidth; a convolutional transform network is used to perform load identification on electrical appliances connected to the socket, which can intelligently identify the inserted appliances and perform intelligent management of ordinary appliances; the socket has an embedded NFC module that can read and write electrical appliance information, and save the information on the NFC tag of the plug, solving the problems of extremely slow recognition of small appliances that are difficult to solve by algorithms, inability to predict new data in real time, and difficulty in identifying appliances of the same model; compared with ordinary smart sockets, the intelligent disaster prevention and energy-saving socket based on the load identification algorithm additionally provides an electrical appliance load identification function, a customized access denial service for special electrical appliances in specific scenarios, and a series of services derived therefrom. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is an exploded view of the intelligent disaster prevention and energy-saving socket of the present invention.

[0014] Figure 2 This is a view of a common household appliance plug after NFC is attached in the present invention.

[0015] Figure 3 This is the hardware architecture diagram of the PCB circuit board.

[0016] Figure 4 This is the circuit schematic diagram of the Power Supply module in the present invention.

[0017] Figure 5 This is a schematic diagram of the USB module circuit in the present invention.

[0018] Figure 6 This is the circuit schematic diagram of the WIFI&MCU module in the present invention.

[0019] Figure 7 FIG. 4 is a schematic diagram of a Power Monitoring circuit in the present invention.

[0020] Figure 8 This is a schematic diagram of the NFC circuit in the present invention.

[0021] Fig. 9 It is the circuit principle diagram of the relay RELAY module in the present invention.

[0022] Fig.10 This is the architecture diagram of the smart socket system.

[0023] Fig.11 The intelligent marking technology of the present invention uses a flow chart. DETAILED DESCRIPTION

[0024] The present invention will be further described below in conjunction with the accompanying drawings.

[0025] Reference Figures 1 to 11, an intelligent disaster prevention and energy-saving socket based on load identification algorithm, including a socket cover 1, a plug guide groove and a holder 2, a socket reed 3, a power-on indicator light 4, a socket bottom cover 5 and a PCB circuit board 6, each socket corresponds to a power-on indicator light, the PCB circuit board 6 includes an MCU controller, a WIFI module, a USB serial port, an NFC module, an energy metering module, a relay, a terminal block, a fuse, an AC-DC converter and a power module, the MCU controller adopts an ESP32-S module, and the ESP32-S module integrates an antenna switch, a radio frequency Balun, a power amplifier, a low noise amplifier, a filter and a and a power management module, the NFC module includes an NFC reader / writer chip and an NFC antenna, the NFC antenna is connected to the NFC reader / writer chip, the NFC reader / writer chip is connected to the ESP32-S module, the electric energy metering module adopts the HLW8012 chip, the 2nd and 3rd pins of the HLW8012 chip are the current differential signal input terminals, the 4th pin is the voltage sampling signal input terminal, the 6th and 7th pins of the HLW8012 chip are respectively connected to the 13th and 16th pins of the ESP32-S module, and the ESP32-S module calculates the power, current and voltage values ​​by measuring the period of the high-frequency pulses output by the 6th and 7th pins of the HLW8012 chip;

[0026] The power module is connected to 220V AC through the terminal block. After the current flows through the fuse, it is converted into voltage through the AC-DC converter to power other circuit modules. The ESP32-S module is connected to the user terminal through a relay and to the cloud through the WIFI module. An NFC tag is attached to the plug end of the electrical equipment. The data information in the NFC tag matches the relevant parameters of the home appliance.

[0027] Furthermore, the NFC reader / writer chip adopts the PN5321A3HN chip, and the 27th and 28th pins of the PN5321A3HN chip are respectively connected to the 36th and 33th pins of the ESP32-S module, and the PN5321A3HN chip communicates with the ESP32-S module through I2C; the USB serial port adopts the CH340C chip, and the 2nd and 3rd pins of the CH340C chip are respectively connected to the U0TXD and U0RXD pins of the ESP32-S module.

[0028] An intelligent disaster prevention and energy-saving socket management system based on a load identification algorithm, the system architecture includes a data acquisition and computing architecture, a cloud and a terminal, the data acquisition and computing architecture includes a sensor layer and an edge gateway layer, the cloud includes a cloud server and a GPU server, the terminal includes an intelligent disaster prevention and energy-saving socket and a user terminal, and the user interacts with the sensor layer through the user terminal;

[0029] The MCU controller and WIFI module constitute the edge gateway layer, and the power metering module constitutes the sensor layer. The sensor layer collects data and uploads it to the edge gateway. The edge gateway runs the activation monitoring algorithm to obtain valid data and transmits it to the GPU server interface layer. The server interface layer stores the effective data of the electrical appliance in the MySQL database and the Redis database and uses the label-consistent convolutional transform neural network model for load identification and uploads it to the cloud server. The cloud server is responsible for the deployment of the label-consistent convolutional transform neural network model and various functional interfaces, providing users with electrical appliance identification, real-time warning, intelligent analysis of electricity consumption data, and electricity consumption data callback viewing services, and also provides App and WeChat applets to interact with user terminals.

[0030] like Figure 1 As shown, the intelligent disaster prevention and energy-saving socket based on the load identification algorithm consists of six parts, a socket cover 1 (including several sockets and wire plug holes), a plug guide groove and a retaining frame 2, a socket spring 3, a power-on indicator light 4, a socket bottom cover 5, and a PCB circuit board 6.

[0031] like Figure 2 As shown, both the three-plug and the two-plug can be affixed with prepared readable and writable NFC tags. The NFC tag is affixed to the plug end of the electrical device. The data information in the NFC tag matches the relevant parameters of the home appliance (can be set on the mobile terminal). Before inserting the home appliance, the user brings the NFC tag close to the smart socket. The NFC reader and writer module embedded in the socket reads the information in the coil and uploads the corresponding data to the cloud through the WIFI module. During the use of home appliances, the power metering module will read the effective values ​​of current and voltage and provide temperature detection. The corresponding data is also transmitted to the cloud in real time through the WIFI module to detect the use of home appliances. When the cloud detects that the current curve is abnormal, it sends an instruction to the MCU controller to control the relay to disconnect and avoid danger. Users can also control the on and off of the relay on the mobile terminal to control the home appliance. The NFC tag is pasted in the following position: Figure 2 .

[0032] like Figure 3 The PCB circuit board shown is composed of six parts: power supply module, USB module, WIFI&MCU module, NFC module, relay module, and power monitoring module.

[0033] like Figure 4As shown in the figure, the power module is connected to 220V AC through the terminal block. After the current flows through the ceramic fuse, it is converted to 5V DC by the AC-DC converter for the relay circuit and HLW8012 chip. Then, the 5V DC is converted to 3.3V DC by the LOD (low dropout regulator) regulator for the NFC circuit, USB download circuit and ESP32-S module.

[0034] like Figure 5 As shown in the figure, the USB circuit mainly realizes the download function of the program. When the download line is connected, the USB bus is converted into a serial port through the CH340C chip. Pins 2 and 3 of the CH340C chip are connected to the U0TXD and U0RXD pins of the ESP32-S module to realize convenient burning of the program.

[0035] like Figure 6 As shown, the WIFI & MCU part is mainly composed of ESP32-S module, which integrates antenna switch, RF Balun, power amplifier, low noise amplifier, filter and power management module. Wi-Fi supports a wide range of communication connections and also supports direct connection to the Internet through a router; while Bluetooth allows users to connect to mobile phones or broadcast BLEBeacon for signal detection. While having excellent connectivity and ease of use, it also has a certain performance CPU and integrates a lot of peripheral interfaces, SPI, IIC, IIS, AD, DA, PWM, IR, UART, CAN, which can fully meet the control and calculation requirements of the present invention.

[0036] like Figure 7 As shown, the main function of some circuits of the electric energy metering module is to realize the collection of parameters related to electric energy. In the present invention, the sampling circuit of the HLW8012 chip adopts a resistor sampling method. Pins 2 and 3 of the HLW8012 chip are current differential signal input terminals, and pin 4 is a voltage sampling signal input terminal. Pins 6 and 7 of the HLW8012 chip are respectively connected to pins 13 and 16 of the ESP32-S module. The high-frequency pulse output from pin 6 of the HLW8012 chip indicates the power and calculates the electric energy; a square wave with a duty cycle of 1:1 is output. The high-frequency pulse output from pin 7 of the HLW8012 chip indicates the effective value of the current or voltage, SEL selects; a square wave with a duty cycle of 1:1 is output. The interface between the ESP32-S module and the HLW8012 chip calculates the power, current, and voltage values ​​by measuring the period of the high-frequency pulse output from pins 6 and 7 of the HLW8012 chip.

[0037] like Figure 8As shown in the figure, the NFC module is the part that performs the NFC tag reading and writing functions. The external antenna is connected to the TX1, TX2 and RX pins of the PN5321A3HN chip after a series of impedance matching. The 27 and 28 pins of the PN5321A3HN chip are connected to the 36 and 33 pins of the ESP32-S module respectively. The PN5321A3HN chip communicates with the ESP32-S module via I2C.

[0038] like Fig. 9 As shown in the figure, the relay RELAY module is the part that controls the on-off of the circuit. When using the smart socket, the user can send instructions to the ESP32-S module through the mobile phone or WeChat applet to control the on-off of the circuit through the relay. The GPIO12 of the ESP32-S module generates a small current, which controls the electromagnetic coil through the MOS tube, thereby controlling the on-off of the circuit. When GPIO12 is low frequency, the circuit is turned on, and when GPIO12 is high frequency, the circuit is disconnected.

[0039] like Fig.10 As shown, the sensing layer is mainly composed of the HLW8012 chip, which continuously collects power data and uploads it to the edge gateway layer. The edge gateway layer is composed of an edge computing group with lightweight computing capabilities. The edge computing group of the present invention is embedded in the socket module and integrated in the ESP32-S module. It is responsible for activation detection and extraction of sensor data using the activation monitoring algorithm, screening out invalid disturbance data, and outputting valid data to upload to the cloud layer. In terms of the selection of the database, the present invention selects the MySQL relational database to store sensor data and selects the Redis non-relational database as a buffer. The database will continuously receive and store the power data transmitted by the intelligent terminal and save it, while providing a training set for the load identification model.

[0040] Under low-frequency sampling, the step-like changes brought about by the change of load state can be characterized as a sudden change in the discreteness of data points, which can be identified by the coefficient of variation and the load operation law. The sliding window algorithm of CNN can extract representative activation curve features, and combined with manually defined features, it can more accurately describe the characteristics of the activation curve.

[0041] The activation monitoring algorithm is described in detail as follows:

[0042] In actual situations, the WIFI & MCU module will continuously receive the data collected by the Power Monitoring module in real time, so the new n-point active data is taken as the calculation window P t , use the intermediate time t to mark the window:

[0043]

[0044] The intuitive manifestation of the change in load state is that the power curve undergoes a step-like change, and the discreteness of the sample data also increases accordingly. Therefore, the local standard deviation and local average amplitude of the sample can be calculated to detect potential activation events. After continuously sliding the window, the load state change points can be screened out, and the positive and negative classifications of large differential values ​​can be used to classify them into candidate switching time sets. The candidate step point set has been generated, and the elements in the above two sets will be further paired to obtain the first and last moments of the activation event for the following conditional screening:

[0045] ① The load should have a small activation time ω under normal conditions and should satisfy the small activation interval δ

[0046] ②The power during activation should not be less than the initial background power

[0047] ③The background power before and after activation should be similar

[0048] ④The power fluctuations caused by the start and end of activation should be similar

[0049] Traverse the candidate class step point set and apply the above rules, record the activation start and end timestamps t that meet the conditions on ,t off , if no valid activation is found, continue to wait and analyze new uploaded data.

[0050] This example is a row of sockets that can be plugged into existing power strips to achieve convenience and intelligence. The row modular design also gives users more flexible choices. Paste an NFC tag on the plug end of an ordinary electrical device. The information in the NFC tag matches the relevant parameters of the electrical appliance (which can be set on the mobile terminal). Before inserting the electrical appliance, the user brings the NFC tag close to the socket. The socket reads the information in the coil and uploads the corresponding data to the cloud through the ESP32-S module. During the use of electrical appliances, the energy metering module reads the effective value of current and voltage and provides temperature detection. The corresponding data is also transmitted to the cloud in real time through the ESP32-S module to detect the use of electrical appliances. When the cloud detects that the current curve is abnormal, it sends a command to the MCU controller to control the relay to disconnect and avoid danger. Users can also control the on and off of the relay on the mobile terminal to achieve control of electrical appliances.

[0051] The energy metering module uses the HLW8012 chip, which has two programmable gain amplifiers (PGA) and analog-to-digital converters (ADC) inside. It performs analog-to-digital conversion on current and voltage sampling signals to obtain digital signals. The chip calculates the active power value, current effective value, and voltage effective value. After the frequency conversion module, the HLW8012 chip converts the active power value, current effective value, and voltage effective value into square wave pulse output (duty cycle 1:1). The size of each value is proportional to the size of the frequency and inversely proportional to the size of the cycle. The active power, power, voltage effective value, current effective value, and total power consumption indicators obtained by the HLW8012 chip are used as the main classification basis. Human behavior pattern data such as time intervals (morning, noon, afternoon, and evening, lighting appliances and battery chargers are generally turned on at night) and other multi-modal data are used to assist classification to improve its recognition accuracy.

[0052] Convolutional Transform Learning (CTL) is a recently proposed unsupervised framework for feature generation based on learned convolutions. CTL can be seen as the analytical counterpart of another emerging representation learning paradigm, namely Convolutional Dictionary Learning (CDL), where the main idea is to learn convolutional filters that operate on data to produce features that account for data reliability constraints. In this project, a supervised convolutional transform formulation is proposed to solve the multi-label classification problem. In multi-class classification, there is a one-to-one mapping between samples and classes, while in multi-label classification, a sample can correspond to multiple classes at the same time (one-to-many mapping).

[0053] CTL learns a set of filters (t m )1≤m≤m, to generate a set of features According to the original research conversion learning, a sparsity penalty is imposed on the features; consistent with the research on CNN, non-negativity constraints (rectangular linear units) are also imposed on the features. Then, the convolution filters and coefficients learned from the data are trained. This is expressed as a minimization problem, and the present invention adopts a minimization method after adding label consistency. With the help of approximate operators with contraction properties, it can be proved that the generated sequence converges to a local solution. The training phase ends here. The testing phase includes applying the update rule of the feature matrix X to the test data. Therefore, the generated features are projected onto the label space. In fact, the generated label mapping may not be binary, but it is real-valued. Therefore, the present invention will use a threshold of 0.5 to determine the activity class.

[0054] It can analyze and return the results of electrical load identification, accurately provide the category of electrical appliances, and feedback to users for double-verification electrical appliance intelligent labeling. In addition, this algorithm has the characteristics of high recognition accuracy, fast recognition response speed, and strong stability.

[0055] As a control module, the relay module has an interactive relationship between the control system (input circuit) and the controlled system (output circuit). When using a smart socket, users can send instructions to the embedded MCU controller through a mobile phone or WeChat applet, so that the MCU controller generates a small current, which then controls the large current on the socket through the relay. A relay is an electrical control device that controls a large current through a small current, and is well reflected in a smart socket.

[0056] The NFC module consists of an NFC reader / writer chip and an NFC antenna embedded in a PCB board. NFC is a safe and reliable short-range private communication method. The NFC communication setting procedure is simple and the communication establishment time is very short, only about 0.1s. Therefore, the present invention chooses to use the NFC reading mode on the smart socket to read the electrical appliance information.

[0057] The present invention selects to use the ESP32-S module on the MCU controller. The ESP32-S module is a universal WiFi-BT-BLE MCU module. The ESP32-S module is based on the ESP32-S0WD single-core chip design, and adopts an ultra-small 802.11b / g / nWi-Fi+BT SoC module and a low-power dual-core 32-bit CPU. It can be used as an application processor with a main frequency of up to 240MHz and a computing power of up to 600DMIPS. It integrates the functions of antenna, RF balun, power amplifier, receiving low-noise amplifier, filter, power management module, etc., and only requires very few peripheral devices to achieve powerful processing performance, reliable security performance and Wi-Fi & Bluetooth functions.

[0058] The use process of the dual-verification smart labeling technology for electrical appliances based on NFC tags and load identification algorithms is as follows: the user uses a specially made readable and writable NFC tag and sticks the NFC tag in the middle of the electrical appliance plug (both three-head sockets and two-head sockets are universal). When a new electrical appliance is used, the user clicks on the new electrical appliance in the applet. By turning on the electrical appliance, the socket collects the useful power changes and transmits them to the ESP32 chip for activation processing. When the activation function is satisfied, the data is uploaded to the cloud server through the WIFI module. Mysql records the electrical characteristics of the electrical appliance. The background uses the convolutional transform neural network to predict the type of electrical appliance, and reminds the user to enter the confirmation information on the applet side. After the user enters the basic information and parameters of the new electrical appliance in the cloud, the information is written into the Mysql table to establish a user tag library, and the information is fed back to the ESP32 chip. The ESP32 chip exchanges the returned data to the built-in micro NFC reader-writer chip of the socket. The NFC reader-writer chip is responsible for writing the electrical appliance data into the NFC tag. When an old appliance is plugged into a socket, the socket's built-in micro NFC antenna reads the tag data, and the ESP32-S module transmits the appliance's power consumption information and tag data to the cloud via WIFI. The cloud compares the appliance tag data and the appliance's power consumption information through the tag library. If they match, the information will continue to be recorded. If they do not match, it means that the tag information is incorrect or the appliance information does not exist in the tag library. If there is no information, the convolutional neural network is used to re-predict the type of appliance and return the prediction result so that the user can re-write the information and overwrite the tag. If there is no match, the tag library information shall prevail. After generating an appliance portrait in the cloud and matching it with the socket model, you can experience the functions of managing appliances by socket, viewing appliance power consumption and basic information with one click on the mini program.

[0059] This socket can recognize that the lithium battery is charging and automatically refuse to power on. It should be noted that the scene manager can set the specific scene in the applet to refuse which electrical appliances to connect, ensuring the flexibility of the application. The background big data statistical mining module can compare the power consumption and other parameters of specific electrical appliances with similar electrical appliances, and summarize the energy-saving report to the user. At the same time, it can remotely power off unnecessary electrical appliances to achieve energy saving and safety.

[0060] If the type of electrical appliance is identified as a prohibited appliance set by the administrator, the background sends a cut-off command to the ESP32-S module to control the relay to implement the power-off operation, thereby realizing the operation of refusing access to the specified type of electrical appliances.

[0061] The present invention uses a load identification method based on a convolutional transformation network with label consistency to roughly classify electrical appliances. It can identify and monitor the power consumption of heat-prone and dangerous electrical appliances such as resistive appliances. When the connected electrical appliance is identified as a high-risk appliance in the scenario, the connection is automatically rejected. When the user connects to the electrical appliance for the first time on the mini program side, the user can be prompted to label and name it according to the identified electrical appliance. The next time the appliance is inserted, the convolutional transformation network model is automatically used in conjunction with NFC and the electrical appliance information stored in the tag to perform two-way verification to achieve automatic identification. After identifying the prohibited electrical appliances published by the scene administrator, the MCU controller is controlled to refuse access.

[0062] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the invention patent concept and are for illustrative purposes only. The protection scope of the invention patent should not be considered to be limited to the specific forms described in this embodiment. The protection scope of the invention patent also extends to the equivalent technical means that ordinary technicians in this field can think of based on the invention patent concept.

Claims

1. An intelligent disaster prevention and energy-saving socket management system based on load identification algorithm, characterized in that: The system includes a data acquisition computing architecture, a cloud and a terminal. The data acquisition computing architecture includes a sensor layer and an edge gateway layer. The cloud includes a cloud server and a GPU server. The terminal includes an intelligent disaster prevention and energy-saving socket and a user terminal. The user interacts with the sensor layer through the user terminal. The intelligent disaster prevention and energy-saving socket comprises a socket upper cover, a plug guide groove and a retaining frame, a socket reed, a power-on indicator light, a socket bottom cover and a PCB circuit board, each socket corresponds to a power-on indicator light, and the PCB circuit board comprises an MCU controller, a WIFI module, a USB serial port, an NFC module, an electric energy metering module, a relay, a terminal block, a fuse, an AC-DC converter and a power module. The MCU controller adopts an ESP32-S module, and the ESP32-S module integrates an antenna switch, a radio frequency Balun, a power amplifier, a low noise amplifier, a filter and a power management module. The FC module includes an NFC reader / writer chip and an NFC antenna, the NFC antenna is connected to the NFC reader / writer chip, the NFC reader / writer chip is connected to the ESP32-S module, the electric energy metering module adopts the HLW8012 chip, the 2nd and 3rd pins of the HLW8012 chip are the current differential signal input terminals, the 4th pin is the voltage sampling signal input terminal, the 6th and 7th pins of the HLW8012 chip are respectively connected to the 13th and 16th pins of the ESP32-S module, and the ESP32-S module calculates the power, current and voltage values ​​by measuring the period of the high-frequency pulses output by the 6th and 7th pins of the HLW8012 chip; The power module is connected to 220V AC through the terminal block. After the current flows through the fuse, it is converted into voltage through the AC-DC converter to power other circuit modules. The ESP32-S module is connected to the user terminal through a relay and to the cloud through the WIFI module. The plug end of the electrical equipment is attached with an NFC tag. The data information in the NFC tag matches the relevant parameters of the home appliance. The MCU controller and WIFI module constitute the edge gateway layer, and the power metering module constitutes the sensor layer. The sensor layer collects data and uploads it to the edge gateway. The edge gateway runs the activation monitoring algorithm to obtain valid data and transmits it to the GPU server interface layer. The server interface layer stores the effective data of the electrical appliance in the MySQL database and the Redis database and uses the label-consistent convolutional transform neural network model for load identification and uploads it to the cloud server. The cloud server is responsible for the deployment of the label-consistent convolutional transform neural network model and various functional interfaces, providing users with electrical appliance identification, real-time warning, intelligent analysis of electricity consumption data, and electricity consumption data callback viewing services, and also provides App and WeChat applets to interact with user terminals.

2. The intelligent disaster prevention and energy-saving socket management system based on load identification algorithm as claimed in claim 1, characterized in that: The NFC reader / writer chip adopts the PN5321A3HN chip, the 27th and 28th pins of the PN5321A3HN chip are respectively connected to the 36th and 33th pins of the ESP32-S module, and the PN5321A3HN chip communicates with the ESP32-S module via I2C; The USB serial port uses the CH340C chip, and the pins 2 and 3 of the CH340C chip are connected to the U0TXD and U0RXD pins of the ESP32-S module respectively.

Images