WO2020248946A1 - 一种识别电器的方法及装置 - Google Patents

一种识别电器的方法及装置 Download PDF

Info

Publication number
WO2020248946A1
WO2020248946A1 PCT/CN2020/094991 CN2020094991W WO2020248946A1 WO 2020248946 A1 WO2020248946 A1 WO 2020248946A1 CN 2020094991 W CN2020094991 W CN 2020094991W WO 2020248946 A1 WO2020248946 A1 WO 2020248946A1
Authority
WO
WIPO (PCT)
Prior art keywords
noise signal
segments
data
electrical appliance
plc device
Prior art date
Application number
PCT/CN2020/094991
Other languages
English (en)
French (fr)
Inventor
宋碧薇
张�浩
谢于明
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP20821932.9A priority Critical patent/EP3968530A4/en
Publication of WO2020248946A1 publication Critical patent/WO2020248946A1/zh
Priority to US17/547,037 priority patent/US11658702B2/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B3/00—Line transmission systems
    • H04B3/54—Systems for transmission via power distribution lines
    • H04B3/542—Systems for transmission via power distribution lines the information being in digital form
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B3/00—Line transmission systems
    • H04B3/02—Details
    • H04B3/46—Monitoring; Testing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B3/00—Line transmission systems
    • H04B3/54—Systems for transmission via power distribution lines
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D2204/00—Indexing scheme relating to details of tariff-metering apparatus
    • G01D2204/20—Monitoring; Controlling
    • G01D2204/24—Identification of individual loads, e.g. by analysing current/voltage waveforms

Definitions

  • This application relates to the field of electric power technology, and in particular to a method and device for identifying electrical appliances.
  • Power Line Communication refers to the modulation of analog signals or digital signals on the power line for transmission through carrier waves, without the need to re-establish the network, as long as there are wires to transmit data, it is a unique communication method for power systems.
  • PLC Power Line Communication
  • the use of existing low-voltage power lines as the transmission medium for communication has the advantages of simple networking, no need for rewiring, and relatively low cost.
  • the low-voltage power line itself is designed to transmit power to the electrical appliances in the circuit, not for communication.
  • the electrical appliances in the circuit will produce electromagnetic interference (ie, noise signals) during use, which will affect the communication quality of the PLC equipment. Therefore, identifying the strong noisy electrical appliances in the circuit helps to reduce the maintenance cost of the PLC equipment when it is interfered.
  • the current electrical appliance identification method is mainly based on smart meter data, that is, by monitoring the smart meter and recording the electrical load data in the user's home to identify the working electrical appliance. For example, based on the four features in the smart meter data: active power P, reactive power Q, harmonic component h, harmonic magnitude mh, all possible feature combinations are stored in advance, and the feature combinations to be identified are compared item by item to identify Electrical appliances.
  • active power P active power P
  • reactive power Q reactive power Q
  • harmonic component h harmonic magnitude mh
  • all possible feature combinations are stored in advance, and the feature combinations to be identified are compared item by item to identify Electrical appliances.
  • the number of appliances in the user's home increases with the above method, the number of feature combinations increases exponentially.
  • the embodiments of the present application provide a method and device for identifying electrical appliances, so as to realize the identification of electrical appliances simply and efficiently.
  • an embodiment of the present application provides a method for identifying electrical appliances, including: a PLC device acquires a noise signal in a circuit; the PLC device acquires first data based on the noise signal, and the first data is used to describe the The time-frequency characteristics of the noise signal; the PLC device obtains the electrical appliance identification result corresponding to the noise signal based on the electrical appliance identification model and the first data; wherein the electrical appliance identification model is based on including at least one known electrical appliance The signal of the noise signal is obtained.
  • the PLC device acquires the noise signal in the circuit, acquires the first data based on the noise signal, and further acquires the electrical appliance identification model and the first data based on the noise signal.
  • the identification result of the electrical appliance corresponding to the noise signal does not need to use additional equipment such as smart meters, and directly uses the PLC equipment to collect noise signals and extract the time-frequency characteristics of the noise signals as the first data.
  • electrical appliances with serious interference to the line can be identified, so that The method is simple and convenient to reduce the maintenance cost of PLC equipment when it is interfered.
  • the PLC device sends the first data to a server, and the server stores the electrical appliance identification model; the PLC device receives the electrical appliance identification result corresponding to the noise signal from the server.
  • the server can use the received first data sent by each PLC device as a training sample to continuously improve and revise the electrical appliance identification model, thereby increasing the accuracy of the electrical appliance identification results obtained based on the electrical appliance identification mode.
  • the PLC device when the PLC device acquires the first data based on the noise signal, the PLC device may segment the noise signal, and acquire the first data based on the segmented noise signal.
  • the signal length of the noise signal is at least one AC cycle
  • the noise signal includes N sampling points, and N is a positive integer
  • the PLC device obtains the first signal based on the segmented noise signal
  • the data can divide the N sampling points into M segments, extract time-domain features for each of the M segments, and obtain 1 M-dimensional data, where M is a positive integer and M ⁇ N; and
  • the N sampling points are divided into K fragments, L fragments are selected from the K fragments, and frequency domain features are extracted for each of the L fragments to obtain L M-dimensional data, and both L and K Is a positive integer, L ⁇ K ⁇ N;
  • the PLC device uses the 1 M-dimensional data and the L M-dimensional data as the first data.
  • the time-frequency characteristics of the noise signal can be extracted simply and efficiently.
  • the PLC device extracts time-domain features for each of the M segments, which may mean that the PLC device calculates the maximum value for each of the M segments, Or mean, or quantile.
  • the time-domain characteristics of the noise signal can be extracted simply and efficiently.
  • the L fragments include at least one of the fragments where the sampling points include peaks, the sampling points include the troughs, and the sampling points include the zero points.
  • the PLC device extracts frequency domain features for at least one of the above three segments, and can extract the main frequency domain features in the entire AC cycle, that is, to achieve Use fewer segments to get more frequency domain features.
  • the PLC device extracts frequency domain features for each of the L segments, which may mean that the PLC device calculates the power spectral density for each of the L segments , Or energy spectral density, or spectral density.
  • the frequency domain characteristics of the noise signal can be extracted simply and efficiently.
  • the PLC device After the PLC device obtains the electrical appliance identification result corresponding to the noise signal, the PLC device sends the electrical appliance identification result to the terminal device.
  • the user can turn off the electric appliance indicated by the electric appliance identification result or lower the power of the electric appliance indicated by the electric appliance identification result based on the electric appliance identification result.
  • an embodiment of the present application provides a device for identifying electrical appliances.
  • the device may be a PLC device or a chip in the PLC device.
  • the device may include a processing unit, a sending unit, and a receiving unit.
  • the processing unit may be a processor, and the sending unit and receiving unit may be transceivers;
  • the PLC device may also include a storage unit, and the storage unit may be a memory; the storage unit is used to store instructions
  • the processing unit executes the instructions stored in the storage unit, so that the PLC device executes the first aspect or any one of the possible design methods in the first aspect.
  • the processing unit may be a processor, and the sending unit and receiving unit may be input/output interfaces, pins or circuits, etc.; the processing unit executes instructions stored in the storage unit to The chip executes the first aspect or any one of the possible design methods in the first aspect.
  • the storage unit is used to store instructions.
  • the storage unit can be a storage unit in the chip (for example, a register, cache, etc.), or a storage unit in the PLC device located outside the chip (for example, a read-only memory, Random access memory, etc.).
  • an embodiment of the present application also provides a computer-readable storage medium that stores a computer program, and when the computer program runs on a computer, the computer executes the method of the first aspect.
  • the embodiments of the present application also provide a computer program product containing a program, which when running on a computer, causes the computer to execute the method of the first aspect.
  • FIG. 1 is a schematic diagram of the distribution of household electrical appliances and PLC equipment in this application;
  • FIG. 2 is a schematic diagram of the interaction between the PLC device and the server in this application;
  • FIG. 3 is an overview flowchart of the method for identifying electrical appliances in this application.
  • Figure 4(a) is a schematic diagram of a noise signal with a signal length of one AC cycle in this application;
  • Figure 4(b) is a schematic diagram of extracting time-domain features from noise signals in this application.
  • Figure 4(c) is a schematic diagram of extracting frequency domain features from noise signals in this application.
  • Figure 5 is a schematic diagram of establishing an electrical appliance identification model in this application.
  • Figure 6 is a schematic diagram of model training using a neural network model in this application.
  • Figure 7 is one of the schematic structural diagrams of a device in this application.
  • Figure 8 is the second structural diagram of a device in this application.
  • Figure 9 is a schematic structural diagram of a PLC device in this application.
  • FIG. 1 it is a schematic diagram of the distribution of household electrical appliances and PLC equipment.
  • the PLC device shown in Figure 1 includes a mother router and 3 child routers.
  • the mother router and 3 child routers are in an electric meter loop.
  • the 3 child routers are distributed in 3 rooms respectively.
  • the mother router is connected to the network through a network cable.
  • the network transmission function of PLC equipment is very strong, but it is susceptible to interference from noise signals generated by surrounding electrical appliances. Electrical appliances such as televisions, air conditioners, computers, refrigerators, and washing machines in Figure 1 may all interfere with surrounding PLC equipment.
  • FIG. 1 is only an example, and should not be construed as a limitation on the scope of implementation of the present application.
  • PLC equipment refers to equipment that uses power lines as communication carriers to convert any power socket under an electric meter loop into a network interface, plug and play, and can be connected to the network without additional wiring.
  • the PLC device may be a power cat, a child-mother router, etc.
  • Figure 2 this is a schematic diagram of the interaction between the PLC device and the server in this application.
  • the server can be used to save the electrical appliance recognition model, and can also be used to train the electrical appliance recognition model.
  • the server can be a cloud analyzer, that is, a computing unit deployed in the cloud.
  • the embodiments of the present application provide a method for identifying electrical appliances, which is used to easily and efficiently recognize electrical appliances, thereby reducing the maintenance cost of PLC equipment when it is interfered.
  • the method includes:
  • Step 300 The PLC device obtains the noise signal in the circuit.
  • the PLC device processes the signal transmitted in the circuit to extract the noise signal.
  • the signal length of the noise signal is at least one alternating current cycle
  • the noise signal includes N sampling points
  • N is a positive integer.
  • the signal length of the noise signal collected by the PLC device can be one or more complete alternating current cycles.
  • the noise signal The signal length is 20ms, as shown in Figure 4(a).
  • the PLC device may periodically obtain the noise signal in the circuit, or when the PLC device detects that the network communication quality is poor, it may obtain the noise signal in the circuit.
  • Step 310 The PLC device obtains the first data based on the noise signal.
  • the PLC device can segment the noise signal to obtain the first data, that is, compress the noise signal to a certain extent.
  • the PLC device can divide N sampling points into M segments, extract time domain features for each of the M segments, and obtain 1 M-dimensional data, that is, M time-domain feature information , M is a positive integer, M ⁇ N.
  • the PLC device can also divide N sampling points into K segments, select L segments from K segments, extract frequency domain features for each of the L segments, and obtain L M-dimensional data, both L and K It is a positive integer, L ⁇ K ⁇ N. Based on this, the PLC device obtains L+1 M-dimensional data as the first data.
  • the PLC device first obtains L results after extracting frequency-domain features for each of the L segments, in order to be consistent with extracting time-domain features for each of the M segments
  • the obtained 1 M-dimensional data has the same dimensionality.
  • the L results can be transformed into L M-dimensional data to meet the needs of subsequent calculations. The specific mathematical transformation method will not be omitted here. Repeat.
  • extracting time-domain features for each of the M segments may refer to calculating the maximum value, or the average value, or the quantile for each of the M segments.
  • Extracting the frequency domain feature data for each of the L segments may refer to calculating the power spectral density, or energy spectral density, or spectral density for each of the L segments.
  • the L segments here may include at least one of the segments in which the sampling points include peaks, the sampling points include segments in the troughs, and the sampling points include segments in which the zero point is included.
  • the PLC device extracts frequency domain features for the above three segments, and can extract the main frequency domain features in the entire AC power cycle, that is, achieve more frequency domain features with fewer segments.
  • the noise signal is segmented and time-frequency characteristics are extracted to obtain the first data.
  • time-frequency characteristics are extracted to obtain the first data.
  • Step 320 The PLC device obtains an electrical appliance identification result corresponding to the noise signal based on the electrical appliance identification model and the first data; wherein the electrical appliance identification model is obtained based on a signal including the noise signal of at least one known electrical appliance. It should be understood that the PLC device can save the electrical appliance identification model, or other devices can save the electrical appliance identification model. In a possible design, the PLC device sends the first data to the server, and the server stores the appliance identification model. The server obtains the appliance identification result corresponding to the noise signal based on the appliance identification model and the first data, and sends the noise signal corresponding to the PLC device Electric appliance identification result.
  • the advantage of the server storing the electrical appliance identification model is that the server can use the received first data sent by each PLC device as a training sample to continuously improve and revise the electrical appliance identification model, thereby increasing the number of electrical appliance identification results obtained based on the appliance identification mode accuracy.
  • multiple sub-PLC devices and one parent PLC device are installed in one indoor space, where the parent PLC device and multiple sub-PLC devices are in an electric meter loop, and each room can install at least one sub-PLC device.
  • the child PLC device can acquire the noise signal in the circuit and acquire the first data based on the noise signal.
  • the child PLC device transmits the first data to the parent PLC device, and the parent PLC device uploads the received first data to the server.
  • the child PLC device may acquire the noise signal in the circuit and transmit the noise signal to the parent PLC device, and the parent PLC device acquires the first data based on the noise signal, and uploads the first data to the server.
  • the parent PLC device may obtain the noise signal in the circuit, obtain the first data based on the noise signal, and upload the first data to the server.
  • the PLC device after the PLC device obtains the electrical appliance identification result corresponding to the noise signal, the PLC device sends the electrical appliance identification result to the terminal device.
  • the server after the server obtains the electric appliance identification result corresponding to the noise signal based on the electric appliance identification model and the first data, the server sends the electric appliance identification result corresponding to the noise signal to the PLC device and the terminal device.
  • the terminal device here refers to the terminal device associated with the PLC device. Therefore, after the terminal device receives the electrical appliance identification result, in order to ensure the normal operation of the PLC device, it can turn off the electrical appliance indicated by the electrical appliance identification result or lower the power of the electrical appliance indicated by the electrical appliance identification result.
  • the processor may be a processor in a server or a processor in other devices. As shown in Figure 5, the following methods can be used but not limited to the establishment of an electrical appliance identification model:
  • Step 1 The processor uses a large number of signals including noise signals of at least one electrical appliance as an original training set. Specifically, each signal is cut to a fixed length according to the alternating current cycle. For example, the signal length of each signal is at least one alternating current cycle and includes N sampling points. For a low-voltage power distribution network with a power frequency of 50 Hz, the signal of each signal The length is 20ms.
  • the signal including the noise signal of at least one known electrical appliance may refer to a single noise signal of a known electrical appliance or a mixed signal including the noise signal of at least one known electrical appliance, for example, the single noise of electrical appliance A Signal, or a mixed signal including the noise signal of the appliance A, or a mixed signal of the noise signal of the appliance A and the noise signal of the appliance B, or a noise signal of the appliance A, the noise signal of the appliance B, and the noise signal of the appliance C Mixed signal.
  • the processor may further divide the known electrical appliances into strong interference electrical appliances and non-strong interference electrical appliances according to the type and brand of the electrical appliances in advance, for example: Dyson hair dryer, Siemens washing machine are strong interference electrical appliances, Lenovo computer It is a non-strong interference electrical appliance.
  • the processor may use a large number of signals including at least one kind of strong interference electrical noise signal as the original training set, so as to reduce the training complexity of the electrical appliance identification model.
  • Step 2 The processor obtains the first data of each signal for each signal in the original training set to form a target training set.
  • the processor may divide the N sampling points included in the signal into M segments for each signal in the original training set, and extract time-domain features for each of the M segments to obtain 1 M-dimensional data , M is a positive integer, M ⁇ N.
  • the processor can also divide the N sampling points into K fragments, select L fragments from the K fragments, extract frequency domain features for each of the L fragments, and obtain L M-dimensional data, both L and K It is a positive integer, L ⁇ K ⁇ N. Based on this, the processor can obtain L+1 M-dimensional data as the first data of the signal, and the processor can obtain the first data of each signal for each signal in the original training set as the target training set.
  • each signal in the original training set includes 4,000,000 sampling points.
  • For any one of the signals first divide it into 1600 segments, and obtain the maximum smoothing of each segment (for example, Find the maximum value, or mean value, or quantile, etc.) to obtain data with a dimension of 1600, that is, 1600 time-domain feature information.
  • the 4,000,000 sampling points are roughly divided into 40 segments again, and the sample points including the peaks, the sampling points including the troughs, and the sampling points including the zero point are selected from them, a total of 3 segments, and the power spectral density is calculated separately , Obtain 3 frequency domain feature information, and further adjust the window so that the dimension of the frequency domain feature data is the same as the dimension of the time domain feature data, so as to obtain 3 data with a dimension of 1600, and the processor obtains the signal The first data.
  • the first data of the signal includes 4 data with dimensions of 1600, corresponding to 1600 time domain feature information and 4800 frequency domain feature information.
  • the processor obtains a target training set, and each first data in the target training set includes 4 data with dimensions of 1600.
  • Step 3 The processor performs model training based on the target training set to obtain an electrical appliance recognition model.
  • the processor uses a supervised machine learning method to perform further feature extraction and learning based on the target training set to train an electrical appliance recognition model.
  • the supervised machine learning method may be a neural network mathematical model, and specifically may be a convolutional neural network (convolutional neural network, CNN) implementation model.
  • the processor may input each first data in the target training set in the form of multiple channels into a neural network model containing 3 convolutional layers and two fully connected layers for model training, as shown in Fig. 6 shown.
  • the specific model training process may include but not limited to the following steps:
  • Step 301 The processor performs one-hot encoding on the electrical appliance label.
  • the electrical appliance label is Midea humidifier, and the corresponding training label is [1 0 0 0 0]; the electrical appliance label is OnePlus, and the corresponding training label is [0 1 0 0 0]; the electrical appliance label is Feike hair dryer , The corresponding training label is [0 0 1 0]; the electrical label is Siemens washing machine, and the corresponding training label is [0 0 0 1 0]; the electrical label is OP lamp, and the corresponding training label is [0 0 0 0 1 ];
  • the electrical labels are OnePlus mobile phones and Siemens washing machines, and the corresponding training labels are [0 1 0 1 0].
  • Step 302 The processor randomly initializes the coefficients of each layer of the neural network.
  • Step 303 The processor obtains the output value by propagating the input first data through the convolutional layer and the fully connected layer, and uses the sigmiod function to activate the output value, that is, the output value is mapped to the interval [0,1] to obtain the output Probability value, that is, the probability value of each electrical appliance.
  • Step 304 The processor calculates the error between the output probability value and the electrical appliance label; specifically, the above error can be obtained by a loss function (Margin Loss).
  • a loss function Margin Loss
  • Step 305 If the error is higher than the specified threshold, the processor may update the network weight
  • Step 306 The processor continuously iteratively updates the network weight through step 304 until the error is lower than the specified threshold, and the training ends.
  • the loss function calculates the error of each category separately, and when the output probability value of the correct category (the noise signal of the appliance is indeed included in the mixed signal) is less than 0.9, the square of the part less than 0.9 is taken as the category
  • the output probability value of the error classification (the noise signal of the appliance is not in the mixed signal) is greater than 0.1
  • the square of the part greater than 0.1 is regarded as the error of this type, and then the two types of errors are multiplied by different weights, and it can be Obtain the total error between the output probability value and the appliance label, that is, the sum of the error of the correct classification and the error of the error classification.
  • the first data input in a certain iteration is the first data of the mixed signal containing the noise signal of the Midea humidifier, and the mixed signal does not include the noise signal of a mobile phone, the noise signal of the Flying Branch hair dryer, the Siemens washing machine
  • the output probability value is [0.8 0.1 0 0.2 0]
  • the appearance probability of Midea humidifier is 0.8
  • the electrical label for this training is [1 0 0 0 0 0].
  • the probability of occurrence of OnePlus mobile phone is 0.1
  • the probability of occurrence of Flying Branch hair dryer is 0
  • the probability of occurrence of Siemens washing machine is 0.2
  • the probability of occurrence of Op table lamp is 0.
  • 0.8 is the output probability value of correct classification
  • the electrical appliance identification model obtained by the above method can not only have better anti-interference ability against signal attenuation and time change, but also can identify multiple strong interfering electrical appliances in the mixed noise by learning multiple electrical appliance tags.
  • the model can be deployed on the server.
  • the PLC device can obtain the first data corresponding to the noise signal, and send the first data to the server.
  • the server obtains the electrical appliance identification result based on the electrical appliance identification model and the The recognition result is fed back to the PLC device.
  • the identification result of the electrical appliance corresponding to the noise signal indicates at least one electrical appliance or indicates that there is no strong interfering electrical appliance.
  • the server uses the first data as input data, and obtains the output probability value based on the electrical appliance identification model, and feeds it back as the electrical appliance identification result, or the server uses the identification of the electrical appliance indicated by the output probability value as the electrical appliance identification result Give feedback.
  • the server can directly feed the output probability value as an electrical appliance identification result to the PLC device, or the server can analyze the output probability value and obtain that the electrical appliance indicated by the output probability value is beautiful Humidifiers and Flying Branch hair dryers, and then feed back the logos of Midea humidifiers and Flying Branch hair dryers as electrical identification results to the PLC equipment.
  • the server uses the first data as input data to obtain the output probability value based on the electrical appliance identification model. If the output probability values are all less than the preset threshold, the server indicates that the electrical appliance identification result fed back to the PLC device is not There are strong interference appliances. For example, assuming the output probability value is [0.11 0.18 0.02 0.05 0] and the preset threshold value is 0.2, the electrical appliance identification result fed back by the server to the PLC device indicates that there is no strong interfering electrical appliance.
  • the PLC device acquires the noise signal in the circuit, acquires the first data based on the noise signal, and further acquires the electrical appliance identification model and the first data based on the noise signal.
  • the identification result of the electrical appliance corresponding to the noise signal does not need to use additional equipment such as smart meters, and directly uses the PLC equipment to collect noise signals and extract the time-frequency characteristics of the noise signals as the first data.
  • electrical appliances with serious interference to the line can be identified, so that The method is simple and convenient to reduce the maintenance cost of PLC equipment when it is interfered.
  • each solution of the communication method provided in the embodiments of the present application is introduced from the perspective of the PLC device itself and the interaction between the PLC device and the server.
  • the PLC device and the server include hardware structures and/or software modules corresponding to the respective functions.
  • the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed by hardware or computer software-driven hardware depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.
  • an embodiment of the present application further provides an apparatus 700 including a transceiver unit 702 and a processing unit 701. It should be understood that the apparatus described in Fig. 7 may be a PLC device in the embodiment corresponding to Figs. 3 to 6.
  • the apparatus 700 is used to implement the function of the PLC device in the above method.
  • the device can be a PLC device or a device in a PLC device.
  • the processing unit 701 obtains the noise signal in the circuit
  • the processing unit 701 obtains first data based on the noise signal, and the first data is used to describe the time-frequency characteristics of the noise signal;
  • the processing unit is further configured to obtain an electrical appliance identification result corresponding to the noise signal based on the electrical appliance identification model and the first data; wherein the electrical appliance identification model is based on a noise signal including at least one known electrical appliance Signal.
  • the processing unit 701 may call the transceiver unit 702 to send the first data to a server, where the server stores the electrical appliance identification model; and receive the electrical appliance identification result corresponding to the noise signal from the server.
  • the processing unit 701 and the transceiver unit 702 refer to the record in the above method embodiment.
  • the division of modules in the embodiments of the present application is illustrative, and is only a logical function division. In actual implementation, there may be other division methods.
  • the functional modules in the various embodiments of the present application may be integrated into one process. In the device, it can also exist alone physically, or two or more modules can be integrated into one module.
  • the above-mentioned integrated modules can be implemented in the form of hardware or software functional modules.
  • the device may be a chip system.
  • the chip system may be composed of chips, or may include chips and other discrete devices.
  • the device includes a processor and an interface, and the interface may be an input/output interface.
  • the processor completes the function of the aforementioned processing unit 701
  • the interface completes the function of the aforementioned transceiver unit 702.
  • the device may also include a memory, where the memory is used to store a program that can be run on the processor, and the processor implements the method of the embodiment shown in FIG. 3 when the program is executed.
  • an embodiment of the present application further provides an apparatus 800.
  • the device described in FIG. 8 may be the PLC device in the embodiment corresponding to FIG. 3 to FIG. 6.
  • the device 800 includes: a communication interface 801, at least one processor 802, and at least one memory 803.
  • the communication interface 801 is used to communicate with other devices (for example, a server) through a transmission medium.
  • the memory 803 is used to store computer programs.
  • the processor 802 calls the computer program stored in the memory 803, and transmits and receives data through the communication interface 801 to implement the method of the embodiment shown in FIG. 3.
  • the memory 803 is used to store a computer program; the processor 802 calls the computer program stored in the memory 803, and executes the method executed by the PLC device in the foregoing embodiment through the communication interface 801.
  • the communication interface 801 may be a transceiver, a circuit, a bus, a module, or other types of communication interfaces.
  • the processor 802 may be a general-purpose processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the embodiments of the present application The disclosed methods, steps and logic block diagrams.
  • the general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application may be directly embodied as being executed and completed by a hardware processor, or executed and completed by a combination of hardware and software modules in the processor.
  • the memory 803 may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), etc., and may also be a volatile memory, such as random access memory (random access memory). -access memory, RAM).
  • the memory is any other medium that can be used to carry or store desired program codes in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.
  • the memory in the embodiments of the present application may also be a circuit or any other device capable of realizing a storage function.
  • the memory 803 and the processor 802 are coupled.
  • the coupling in the embodiments of the present application is an interval coupling or a communication connection between devices, units or modules, and may be in electrical, mechanical or other forms, and is used for information exchange between devices, units or modules.
  • the memory 803 may also be located outside the apparatus 800.
  • the processor 802 may cooperate with the memory 803 to operate.
  • the processor 802 may execute program instructions stored in the memory 803.
  • At least one of the at least one memory 803 may also be included in the processor 802.
  • the embodiment of the present application does not limit the connection medium between the aforementioned communication interface 801, the processor 802, and the memory 803.
  • the memory 803, the processor 802, and the communication interface 801 may be connected by a bus, and the bus may be divided into an address bus, a data bus, and a control bus.
  • the apparatus in the embodiment shown in FIG. 7 may be implemented by the apparatus 800 shown in FIG. 8.
  • the processing unit 701 may be implemented by the processor 802
  • the transceiver unit 702 may be implemented by the communication interface 801.
  • an embodiment of the present application further provides a PLC device, including a slave machine 100 and a parent machine 200.
  • a PLC device including a slave machine 100 and a parent machine 200.
  • the PLC device described in FIG. 9 may be the PLC device in the embodiments corresponding to FIGS. 3 to 6.
  • the slave machine 100 includes a communication interface 101, a modem 102, and a processor 103.
  • the slave machine 100 may also include a memory 104.
  • the host computer 200 includes a communication interface 201, a modem 202, and a processor 203.
  • the host computer 200 may also include a memory 204.
  • the parent machine 200 is connected to the network access device through a network cable, and the parent machine 200 and the child machine 100 are connected through a power line, and both belong to the same meter loop.
  • the processor 103 and the processor 203 may be a central processing unit (CPU), a hardware chip, or any combination thereof, and may implement or execute the methods and steps disclosed in the embodiments corresponding to FIGS. 3 to 6 And logical block diagram.
  • the apparatus in the embodiment shown in FIG. 7 may be implemented by the PLC device shown in FIG. 9.
  • the processing unit 701 may be implemented by the modem 102, the processor 103, the modem 202, and the processor 203, and the transceiver unit 702 may be implemented by the communication interface 101 and the communication interface 201.
  • the modem 102 in the slave machine 100 modulates the uplink data of the user equipment received through the communication interface 101 to obtain a modulated signal.
  • the processor 103 loads the modulated signal on a current and transmits it to the master machine 200 through the power line.
  • the processor 203 in the master machine extracts the modulated signal from the signal received from the slave machine 100, then demodulates the uplink data through the modem 202, and sends it to the network access device through the network cable.
  • the modem 202 in the master computer 200 modulates the downlink data received through the communication interface 201 and sent to the user equipment to obtain a modulated signal.
  • the processor 203 loads the modulated signal on the current and transmits it to the slave via the power line.
  • Machine 100 The processor 103 in the slave machine 100 extracts the modulated signal from the signal received from the master machine, then decodes the downlink data through the modem 102, and transmits the downlink data to the user equipment through the communication interface 101.
  • the processor 103 in the slave machine 100 includes a noise signal acquisition module, and the noise signal acquisition module is used to acquire the noise signal in the circuit.
  • the noise signal acquisition module may be implemented in the form of hardware.
  • the noise signal acquisition module may include hardware such as a digital-to-analog converter, an analog-to-digital converter, and an analog front end, or it may be implemented in a combination of hardware and software functional modules.
  • the processor 203 in the main machine 200 includes a noise signal analysis module, and the noise signal analysis module can obtain the first data based on the noise signal collected by the slave machine 100.
  • the noise signal analysis module can be implemented in the form of hardware, software function modules, or a combination of hardware and software function modules.
  • the processor 103 in the slave unit 100 includes a noise signal acquisition module and a noise signal analysis module.
  • the noise signal acquisition module can acquire the noise signal in the circuit, and the noise signal analysis module can acquire the first data based on the noise signal. And the first data is transmitted to the parent machine 200.
  • the processor 203 in the host computer 200 includes a noise signal acquisition module and a noise signal analysis module.
  • the noise signal acquisition module can acquire the noise signal in the circuit, and the noise signal analysis module can acquire the first data based on the noise signal.
  • the noise signal acquisition module may be integrated inside the processor 103 (or processor 203), or located outside the processor 103 (or processor 203) as a separate chip.
  • the noise signal analysis module can be integrated inside the processor 103 (or processor 203), or located outside the processor 103 (or processor 203) as a separate chip.
  • the noise signal acquisition module and the noise signal acquisition module may also be integrated in a chip located outside the processor 103 (or processor 203).
  • the embodiment of the present application also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program runs on a computer, the computer executes the method of the embodiment shown in FIG. 3.
  • the methods provided in the embodiments of the present application may be implemented in whole or in part by software, hardware, firmware, or any combination thereof.
  • software When implemented by software, it can be implemented in the form of a computer program product in whole or in part.
  • the computer program product includes one or more computer instructions.
  • the computer may be a general-purpose computer, a dedicated computer, a computer network, network equipment, user equipment, or other programmable devices.
  • the computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, server, or data center.
  • the computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center integrated with one or more available media.
  • the usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a digital video disc (digital video disc, DVD for short)), or a semiconductor medium (for example, a solid state disk Solid State Disk SSD), etc.

Abstract

一种识别电器的方法及装置，该方法包括：PLC设备获取电路中的噪声信号；PLC设备基于噪声信号获取第一数据，第一数据用于描述噪声信号的时频特征；PLC设备基于电器识别模型和第一数据，获得噪声信号对应的电器识别结果；其中，电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。因此，无需借助智能电表等额外设备，直接利用PLC设备采集噪声信号并提取噪声信号中的时频特征作为第一数据，并基于电器识别模型识别对线路干扰严重的电器。

Description

一种识别电器的方法及装置
相关申请的交叉引用
本申请要求在2019年06月11日提交中国专利局、申请号为201910502300.2、申请名称为“一种识别电器的方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及电力技术领域，尤其涉及一种识别电器的方法及装置。
背景技术
电力线通信(Power Line Communication，PLC)是指通过载波方式将模拟信号或数字信号调制在电力线上进行传输，不需要重新架设网络，只要有电线就能进行数据传递，是电力系统特有的通信方式。在低压配电网中，利用已有低压电力线作为传输介质进行通信，具有组网简单、不需要重新布线且成本相对较低的优点。
然而，低压电力线本身是为电路中的电器传送电能设计的，而非为通信设计的。电路中的电器在使用过程中会产生电磁干扰(即噪声信号)对PLC设备的通信质量产生影响。因此，识别电路中的强噪声电器有助于降低PLC设备在受到干扰时的维护成本。
当前的电器识别方法主要是基于智能电表数据，即通过监控智能电表记录用户家中的电器负载数据来识别正在工作的电器。例如，基于智能电表数据中的四个特征：有功功率P、无功功率Q、谐波分量h、谐波大小mh，事先储存所有可能的特征组合，对待识别特征组合逐项进行对比进而识别出电器。但是，采用上述方法当用户家中电器数增多时，特征组合数呈指数增长。
发明内容
本申请实施例提供一种识别电器的方法及装置，用以简便高效实现识别电器。
第一方面，本申请实施例提供一种识别电器的方法，包括：PLC设备获取电路中的噪声信号；所述PLC设备基于所述噪声信号获取第一数据，所述第一数据用于描述所述噪声信号的时频特征；所述PLC设备基于电器识别模型和所述第一数据，获得所述噪声信号对应的电器识别结果；其中，所述电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。
相较于现有技术基于智能电表数据识别电器的方法，在本申请实施例中，PLC设备获取电路中的噪声信号，基于噪声信号获取第一数据，进一步基于电器识别模型和第一数据，获得噪声信号对应的电器识别结果。因此，本申请实施例无需借助智能电表等额外设备，直接利用PLC设备采集噪声信号并提取噪声信号中的时频特征作为第一数据，基于电器识别模型能够识别对线路干扰严重的电器，进而可以实现降低PLC设备在受到干扰时的维护成本，方法简便。
在一种可能的设计中，所述PLC设备向服务器发送所述第一数据，所述服务器存储所述电器识别模型；所述PLC设备从所述服务器接收所述噪声信号对应的电器识别结果。
采用上述设计，服务器可以将接收的各个PLC设备发送的第一数据，作为训练样本，不断完善和修正电器识别模型，进而可以增加基于电器识别模式得到的电器识别结果的准确性。
在一种可能的设计中，在所述PLC设备基于所述噪声信号获取第一数据时，PLC设备可以将所述噪声信号分段，基于分段后的噪声信号获取第一数据。
采用上述设计可以避免上传服务器的数据量过大。
在一种可能的设计中，所述噪声信号的信号长度为至少一个交流电周期，所述噪声信号包括N个采样点，N为正整数；所述PLC设备基于分段后的噪声信号获取第一数据可以将所述N个采样点分为M个片段，针对所述M个片段中的每个片段提取时域特征，获得1个M维数据，M为正整数，M≤N；以及将所述N个采样点分为K个片段，从所述K个片段中选取L个片段，针对所述L个片段中的每个片段提取频域特征，获得L个M维数据，L和K均为正整数，L≤K≤N；所述PLC设备将所述1个M维数据和所述L个M维数据作为第一数据。
采用上述设计可以简便高效地提取噪声信号的时频特征。
在一种可能的设计中，所述PLC设备针对所述M个片段中的每个片段提取时域特征，可以是指所述PLC设备针对所述M个片段中的每个片段计算最大值、或均值、或分位数。
采用上述设计可以简便高效地提取噪声信号的时域特征。
在一种可能的设计中，L个片段包括采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段中的至少一个。
采用上述设计，由于噪声信号的信号长度为至少一个交流电周期，PLC设备针对上述3个片段中的至少一个片段提取频域特征，可以实现对整个交流电周期中的主要频域特征进行提取，即实现利用较少的片段获得较多的频域特征。
在一种可能的设计中，所述PLC设备针对所述L个片段中的每个片段提取频域特征，可以是指所述PLC设备针对所述L个片段中的每个片段计算功率谱密度、或能量谱密度、或频谱密度。
采用上述设计可以简便高效地提取噪声信号的频域特征。
在一种可能的设计中，在所述PLC设备获得所述噪声信号对应的电器识别结果之后，所述PLC设备向终端设备发送所述电器识别结果。
采用上述设计，为了保证PLC设备的正常工作，用户可以基于上述电器识别结果关闭电器识别结果指示的电器或者调低电器识别结果指示的电器的功率。
第二方面，本申请实施例提供一种识别电器的装置，该装置可以是PLC设备，也可以是PLC设备内的芯片。该装置可以包括处理单元、发送单元和接收单元。当该装置是PLC设备时，该处理单元可以是处理器，该发送单元和接收单元可以是收发器；该PLC设备还可以包括存储单元，该存储单元可以是存储器；该存储单元用于存储指令，该处理单元执行该存储单元所存储的指令，以使该PLC设备执行第一方面或第一方面任意一种可能的设计中的方法。当该装置是PLC设备内的芯片时，该处理单元可以是处理器，该发送单元和接收单元可以是输入/输出接口、管脚或电路等；该处理单元执行存储单元所存储的指令，以使该芯片执行第一方面或第一方面任意一种可能的设计中的方法。该存储单元用于存储指令，该存储单元可以是该芯片内的存储单元(例如，寄存器、缓存等)，也可以是该PLC设备内的位于该芯片外部的存储单元(例如，只读存储器、随机存取存储器等)。
第三方面，本申请实施例还提供一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序，当该计算机程序在计算机上运行时，使得计算机执行上述第一方面的方法。
第四方面，本申请实施例还提供一种包含程序的计算机程序产品，当其在计算机上运行时，使得计算机执行上述第一方面的方法。
附图说明
图1为本申请中一种家庭电器及PLC设备的分布示意图；
图2为本申请中PLC设备与服务器的交互示意图；
图3为本申请中识别电器的方法的概述流程图；
图4(a)为本申请中信号长度为一个交流电周期的噪声信号的示意图；
图4(b)为本申请中针对噪声信号提取时域特征的示意图；
图4(c)为本申请中针对噪声信号提取频域特征的示意图；
图5为本申请中建立电器识别模型的示意图；
图6为本申请中采用神经网络模型进行模型训练的示意图；
图7为本申请中一种装置的结构示意图之一；
图8为本申请中一种装置的结构示意图之二；
图9为本申请中一种PLC设备的结构示意图。
具体实施方式
为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述。
本申请实施例可以应用于家庭PLC设备干扰源定位场景。如图1所示，为一种家庭电器及PLC设备的分布示意图。图1所示的PLC设备包括一个母路由器和3个子路由器，母路由器和3个子路由器在一个电表回路中，3个子路由器分别分布于3个房间，母路由器通过网线连接网络。PLC设备的网络传输功能很强，但是很容易受到周围电器产生的噪声信号的干扰，图1中的电视机、空调、电脑、冰箱、洗衣机等电器都可能对周围的PLC设备产生干扰。应理解的是，图1所示的示意图仅为举例说明，并不应该理解成对本申请实施范围的限定。
在本申请实施例中，PLC设备是指利用电力线作为通信载体，将一个电表回路下的任何一个电源插座转换为网络接口，即插即用，无须另外布线，就可以实现接入网络的设备。例如，PLC设备可以为电力猫、子母路由器等。如图2所示，为本申请中PLC设备与服务器的交互示意图。其中，服务器可以用于保存电器识别模型，还可以用于训练电器识别模型，例如，服务器可以为云端分析器，即部署在云端的计算单元。
应理解的是，不同品牌不同类别的电器在电路中产生的噪声信号不同，因此，可以通过识别噪声信号达到识别电器的目的。但是，由于电器噪声信号类型多样且线路、空开、桥接等对噪声信号衰减均有不同程度的影响，因此，噪声信号有较强的时变性，难以通过一般的人工方法进行特征提取。基于此，本申请实施例提供一种识别电器的方法，用以简便高效实现识别电器，进而可以实现降低PLC设备在受到干扰时的维护成本。参阅图3所 示，该方法包括：
步骤300：PLC设备获取电路中的噪声信号。
具体的，PLC设备对在电路中传输的信号进行处理，提取出噪声信号。
在一种可能的设计中，噪声信号的信号长度为至少一个交流电周期，噪声信号包括N个采样点，N为正整数。例如，对于工频为50Hz的低压配电网，PLC设备采集的噪声信号的信号长度可以为一个或多个完整交流电周期，其中，当噪声信号的信号长度为一个交流电周期时，则该噪声信号的信号长度为20ms，如图4(a)所示。
示例性地，PLC设备可以周期性获取电路中的噪声信号，或者，PLC设备检测到网络通信质量较差时，获取电路中的噪声信号。
步骤310：PLC设备基于噪声信号获取第一数据。
若PLC设备直接上传采集的噪声信号至服务器，数据量较大，因此，PLC设备可以将噪声信号分段，获取第一数据，即对噪声信号进行一定压缩。
在一种可能的设计中，PLC设备可以将N个采样点分为M个片段，针对M个片段中的每个片段提取时域特征，获得1个M维数据，即M个时域特征信息，M为正整数，M＜N。PLC设备还可以将N个采样点分为K个片段，从K个片段中选取L个片段，针对L个片段中的每个片段提取频域特征，获得L个M维数据，L和K均为正整数，L≤K＜N。基于此，PLC设备获得L+1个M维数据作为第一数据。应理解的是，当L≠M时，PLC设备在针对L个片段中的每个片段提取频域特征后，首先得到L个结果，为了与针对M个片段中的每个片段提取时域特征获得的1个M维数据的维数相同，采用本领域技术人员熟知的数学变换方法，可以将L个结果变换为L个M维数据，以满足后续计算需要，具体数学变换方法此处不再赘述。
进一步地，针对M个片段中的每个片段提取时域特征可以是指针对M个片段中的每个片段计算最大值、或均值、或分位数等。针对L个片段中的每个片段提取频域特征数据可以是指针对L个片段中的每个片段计算功率谱密度、或能量谱密度、或者频谱密度等。这里的L个片段可以包括采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段中的至少一个。作为一个可选的实施例，当L＝3，且这3个片段分别为采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段时，由于噪声信号的信号长度为至少一个交流电周期，PLC设备针对上述3个片段提取频域特征，就可以实现对整个交流电周期中的主要频域特征进行提取，即实现利用较少的片段获得较多的频域特征。
作为一个可选的实施例，对噪声信号进行细分段，得到M个片段，M＜N，针对每个片段提取时域特征，获得1个M维数据，如图4(b)所示；对噪声信号进行粗分段，得到K个片段，K＜M＜N，并针对K个片段中的L个片段提取频域特征，例如，当L＝3，且这3个片段分别为采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段，针对这3个片段分别计算功率谱密度，获得L个M维数据，如图4(c)所示，将获得的L+1个M维数据，作为第一数据。
应理解的是，这里将噪声信号进行分段并提取时频特征，获得第一数据，可以有多种实现方式，此处仅为举例，不作为本申请的限定。
步骤320：PLC设备基于电器识别模型和第一数据，获得噪声信号对应的电器识别结果；其中，电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。应理解的是，PLC设备可以保存电器识别模型，也可以由其他设备保存电器识别模型。在一种可 能的设计中，PLC设备向服务器发送第一数据，服务器存储电器识别模型，服务器基于电器识别模型和第一数据获得噪声信号对应的电器识别结果，并向PLC设备发送噪声信号对应的电器识别结果。其中，服务器存储电器识别模型的好处在于，服务器可以将接收的各个PLC设备发送的第一数据，作为训练样本，不断完善和修正电器识别模型，进而可以增加基于电器识别模式得到的电器识别结果的准确性。
示例性地，一个室内空间安装有多个子PLC设备和一个母PLC设备，其中，母PLC设备和多个子PLC设备在一个电表回路中，每个房间可安装至少一个子PLC设备。子PLC设备可以获取电路中的噪声信号，并基于噪声信号获取第一数据，子PLC设备将第一数据传输至母PLC设备，由母PLC设备将接收到的第一数据上传至服务器。或者，子PLC设备可以获取电路中的噪声信号，并将噪声信号传输至母PLC设备，由母PLC设备基于噪声信号获取第一数据，并将第一数据上传至服务器。又或者，母PLC设备可以获取电路中的噪声信号，并基于噪声信号获取第一数据，并将第一数据上传至服务器。
此外，在一种可能的设计中，在PLC设备获得噪声信号对应的电器识别结果之后，PLC设备向终端设备发送电器识别结果。在另一种可能的设计中，在服务器基于电器识别模型和第一数据获得噪声信号对应的电器识别结果后，服务器向PLC设备和终端设备发送噪声信号对应的电器识别结果。这里的终端设备是指与PLC设备关联的终端设备。因此，终端设备在接收到电器识别结果后，为了保证PLC设备的正常工作，可以关闭电器识别结果指示的电器或者调低电器识别结果指示的电器的功率。
应理解的是，以下仅以处理器为例，说明建立电器识别模型的过程，该处理器可以为服务器中的处理器，或者其他设备中的处理器。如图5所示，建立电器识别模型可以采用但不限于以下方法：
步骤1：处理器将海量包括至少一种电器的噪声信号的信号作为原始训练集合。具体的，每个信号按交流电周期截取为固定长度，例如，每个信号的信号长度为至少一个交流电周期，包括N个采样点，对于工频为50Hz的低压配电网，每个信号的信号长度为20ms。其中，包括至少一种已知电器的噪声信号的信号可以是指一种已知电器的单独的噪声信号或者包括至少一种已知电器的噪声信号的混合信号，例如，电器A的单独的噪声信号，或者包括电器A的噪声信号的混合信号，或者包括电器A的噪声信号和电器B的噪声信号的混合信号，或者包括电器A的噪声信号、电器B的噪声信号和电器C的噪声信号的混合信号。作为一个可选的实施例，处理器还可事先根据电器的类型和品牌将各个已知电器划分为强干扰电器和非强干扰电器，例如：戴森吹风机、西门子洗衣机为强干扰电器，联想电脑为非强干扰电器。在建立电器识别模型时，处理器可将海量包括至少一种强干扰电器的噪声信号的信号作为原始训练集合，以降低电器识别模型的训练复杂度。
步骤2：处理器针对原始训练集合中每个信号获取每个信号的第一数据，构成目标训练集合。
具体的，处理器可以针对原始训练集合中的每个信号将该信号包括的N个采样点分为M个片段，针对M个片段中的每个片段提取时域特征，获得1个M维数据，M为正整数，M＜N。处理器还可以将N个采样点分为K个片段，从K个片段中选取L个片段，针对L个片段中的每个片段提取频域特征，获得L个M维数据，L和K均为正整数，L≤K＜N。基于此，处理器可以获得L+1个M维数据作为该信号的第一数据，进而处理器可以针对原始训练集合中的每个信号得到每个信号的第一数据，作为目标训练集合。
作为一个可选的实施例，假设原始训练集合中的每个信号包括4000000个采样点，针对其中任一个信号，首先将其细均分为1600段，通过对每个片段求最大平滑(例如，求最大值、或均值、或分位数等)得到维度为1600的数据，即1600个时域特征信息。然后，将这4000000个采样点重新粗均分为40段，并从中选取采样点包括波峰的片段、采样点包括波谷的片段和采样点包括零点的片段，共3个片段，分别求功率谱密度，得到3个频域特征信息，进一步通过调整分窗使得到的频域特征数据的维数与时域特征数据的维数相同，从而得到3个维度为1600的数据，处理器得到该信号的第一数据，该信号的第一数据包括4个维度为1600的数据，对应1600个时域特征信息和4800个频域特征信息。进一步，采用上述方法，处理器得到目标训练集合，目标训练集合中的每个第一数据包括4个维度为1600的数据。
步骤3：处理器基于目标训练集合进行模型训练得到电器识别模型。
示例性地，处理器基于目标训练集合利用有监督的机器学习方法进行进一步的特征提取和学习从而训练出电器识别模型。其中，有监督的机器学习方法可以为神经网络数学模型，具体可以为卷积神经网络(convolutional neural network，CNN)实现模型。作为一个可选的实施例，处理器可以将目标训练集合中的每个第一数据以多通道的形式输入包含3个卷积层和两个全连接层的神经网络模型进行模型训练，如图6所示。
其中，具体模型训练过程可以包括但不限于以下步骤：
步骤301：处理器对电器标签进行one-hot编码。示例性地，电器标签为美的加湿器，相应的训练标签为[1 0 0 0 0]；电器标签为一加手机，相应的训练标签为[0 1 0 0 0]；电器标签为飞科吹风机，相应的训练标签为[0 0 1 0 0]；电器标签为西门子洗衣机，相应的训练标签为[0 0 0 1 0]；电器标签为欧普台灯，相应的训练标签为[0 0 0 0 1]；电器标签为一加手机和西门子洗衣机，相应的训练标签为[0 1 0 1 0]。
步骤302：处理器对神经网络每一层的系数进行随机初始化。
步骤303：处理器对输入的第一数据经过卷积层、全连接层的传播得到输出值，并使用sigmiod函数对输出值进行激活，即将输出值映射到[0,1]区间内，得到输出概率值，即每种电器出现的概率值。
步骤304：处理器求出输出概率值与电器标签之间的误差；具体的，上述误差可以由损失函数(Margin Loss)得到。
步骤305：若误差高于指定阈值，处理器可以更新网络权重；
步骤306：处理器通过步骤304不断迭代更新网络权重直至误差低于指定阈值，结束训练。
示例性地，损失函数对每种分类的误差单独进行计算，当正确分类(该电器的噪声信号确实包含在混合信号中)的输出概率值小于0.9时，将小于0.9的部分的平方作为该类的误差；当错误分类(该电器的噪声信号不在混合信号中)的输出概率值大于0.1时，大于0.1的部分的平方作为该类的误差，然后对两类误差乘以不同的权重，就可以得到输出概率值与电器标签之间的总误差，即正确分类的误差和错误分类的误差之和。例如，若某次迭代中输入的第一数据为包含美的加湿器的噪声信号的混合信号的第一数据，且该混合信号不包括一加手机的噪声信号、飞科吹风机的噪声信号、西门子洗衣机的噪声信号和欧普台灯的噪声信号，则该次训练的电器标签为[1 0 0 0 0],若输出概率值为[0.8 0.1 0 0.2 0]，则表示美的加湿器的出现概率为0.8，一加手机的出现概率为0.1，飞科吹风机的出现概率为0，西 门子洗衣机的出现概率为0.2,欧普台灯的出现概率为0。其中，0.8为正确分类的输出概率值，其他为错误分类的输出概率值。由于此时存在正确分类的输出概率值小于0.9，错误分类的概率值大于0.1，假设正确分类的权值为1，错误分类的权值为0.5，则该次训练的误差为(0.9-0.8) 2+0.5(0.2-0.1) 2＝0.015。其中，(0.9-0.8) 2为正确分类的误差，(0.2-0.1) 2为错误分类的误差。此时，若指定阈值为0.01，则处理器需要不断迭代更新网络权重直至误差低于指定阈值。
采用上述方法得到的电器识别模型不仅可以对信号的衰减和时变有较好的抗干扰性，且通过对多电器标签的学习，可以识别混合噪声中的多个强干扰电器。
进一步地，在电器识别模型训练完成后，可以将该模型部署在服务器上。PLC设备可以针对采集到的噪声信号，获取该噪声信号对应的第一数据，并将该第一数据发送至服务器，服务器基于电器识别模型和该第一数据，获得电器识别结果，并将该电器识别结果反馈至PLC设备。
在一种可能的设计中，噪声信号对应的电器识别结果指示至少一种电器或者指示不存在强干扰电器。
作为一个可选的实施例，服务器将该第一数据作为输入数据，基于电器识别模型得到输出概率值，作为电器识别结果进行反馈，或者，服务器将输出概率值指示的电器的标识作为电器识别结果进行反馈。例如，假设输出概率值为[0.91 0.18 0.92 0.05 0]，服务器可以直接将该输出概率值作为电器识别结果反馈给PLC设备，或者，服务器基于输出概率值解析得到该输出概率值指示的电器为美的加湿器和飞科吹风机，进而将美的加湿器的标识和飞科吹风机的标识作为电器识别结果反馈给PLC设备。
作为另一个可选的实施例，服务器将该第一数据作为输入数据，基于电器识别模型得到输出概率值，若输出概率值均小于预设阈值，则服务器向PLC设备反馈的电器识别结果指示不存在强干扰电器。例如，假设输出概率值为[0.11 0.18 0.02 0.05 0]，预设阈值为0.2，服务器向PLC设备反馈的电器识别结果指示不存在强干扰电器。
相较于现有技术基于智能电表数据识别电器的方法，在本申请实施例中，PLC设备获取电路中的噪声信号，基于噪声信号获取第一数据，进一步基于电器识别模型和第一数据，获得噪声信号对应的电器识别结果。因此，本申请实施例无需借助智能电表等额外设备，直接利用PLC设备采集噪声信号并提取噪声信号中的时频特征作为第一数据，基于电器识别模型能够识别对线路干扰严重的电器，进而可以实现降低PLC设备在受到干扰时的维护成本，方法简便。
上述本申请提供的实施例中，分别从PLC设备本身、以及从PLC设备与服务器之间交互的角度对本申请实施例提供的通信方法的各方案进行了介绍。可以理解的是，PLC设备和服务器，为了实现上述功能，其包含了执行各个功能相应的硬件结构和/或软件模块。本领域技术人员应该很容易意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，本申请能够以硬件或硬件和计算机软件的结合形式来实现。某个功能究竟以硬件还是计算机软件驱动硬件的方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
与上述构思相同，如图7所示，本申请实施例还提供一种装置700，该装置700包括收发单元702和处理单元701。应理解的是，图7所述的装置可以为图3至图6对应的实 施例中的PLC设备。
一示例中，装置700用于实现上述方法中PLC设备的功能。该装置可以是PLC设备，也可以是PLC设备中的装置。
其中，处理单元701获取电路中的噪声信号；
所述处理单元701基于所述噪声信号获取第一数据，所述第一数据用于描述所述噪声信号的时频特征；
所述处理单元，还用于基于电器识别模型和所述第一数据，获得所述噪声信号对应的电器识别结果；其中，所述电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。其中，所述处理单元701可以调用收发单元702，用于向服务器发送所述第一数据，所述服务器存储所述电器识别模型；以及从所述服务器接收所述噪声信号对应的电器识别结果。
关于处理单元701、收发单元702的具体执行过程，可参见上方法实施例中的记载。本申请实施例中对模块的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，另外，在本申请各个实施例中的各功能模块可以集成在一个处理器中，也可以是单独物理存在，也可以两个或两个以上模块集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。
作为另一种可选的变形，该装置可以为芯片系统。本申请实施例中，芯片系统可以由芯片构成，也可以包含芯片和其他分立器件。示例性地，该装置包括处理器和接口，该接口可以为输入/输出接口。其中，处理器完成上述处理单元701的功能，接口完成上述收发单元702的功能。该装置还可以包括存储器，存储器用于存储可在处理器上运行的程序，处理器执行该程序时实现上述如图3所示实施例的方法。
与上述构思相同，如图8所示，本申请实施例还提供一种装置800。应理解的是，图8所述的装置可以为图3至图6对应的实施例中的PLC设备。该装置800中包括：通信接口801、至少一个处理器802、至少一个存储器803。通信接口801，用于通过传输介质和其它设备(例如，服务器)进行通信。存储器803，用于存储计算机程序。处理器802调用存储器803存储的计算机程序，通过通信接口801收发数据实现上述如图3所示实施例的方法。
示例性地，当该装置为PLC设备时，存储器803用于存储计算机程序；处理器802调用存储器803存储的计算机程序，通过通信接口801执行上述实施例中PLC设备执行的方法。
在本申请实施例中，通信接口801可以是收发器、电路、总线、模块或其它类型的通信接口。处理器802可以是通用处理器、数字信号处理器、专用集成电路、现场可编程门阵列或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件，可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件处理器执行完成，或者用处理器中的硬件及软件模块组合执行完成。存储器803可以是非易失性存储器，比如硬盘(hard disk drive，HDD)或固态硬盘(solid-state drive，SSD)等，还可以是易失性存储器(volatile memory)，例如随机存取存储器(random-access memory，RAM)。存储器是能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质，但不限于此。本申请实施例中的存储器还可以是电路或者 其它任意能够实现存储功能的装置。存储器803和处理器802耦合。本申请实施例中的耦合是装置、单元或模块之间的间隔耦合或通信连接，可以是电性，机械或其它的形式，用于装置、单元或模块之间的信息交互。作为另一种实现，存储器803还可以位于装置800之外。处理器802可以和存储器803协同操作。处理器802可以执行存储器803中存储的程序指令。所述至少一个存储器803中的至少一个也可以包括于处理器802中。本申请实施例中不限定上述通信接口801、处理器802以及存储器803之间的连接介质。例如，本申请实施例在图8中以存储器803、处理器802以及通信接口801之间可以通过总线连接，所述总线可以分为地址总线、数据总线、控制总线等。
可以理解的，上述图7所示实施例中的装置可以以图8所示的装置800实现。具体的，处理单元701可以由处理器802实现，收发单元702可以由通信接口801实现。
示例性地，如图9所示，本申请实施例还提供一种PLC设备，包括子机100和母机200。应理解的是，图9所述的PLC设备可以为图3至图6对应的实施例中的PLC设备。
其中，子机100包括通信接口101、调制解调器102、处理器103，可选的，子机100还可以包括存储器104。母机200包括通信接口201、调制解调器202、处理器203，可选的，母机200还可以包括存储器204。母机200通过网线连接网络接入设备，母机200与子机100通过电力线相连，同属于一个电表回路。其中，处理器103和处理器203可以是中央处理器(central processing unit，CPU)，硬件芯片或者其任意组合，可以实现或者执行图3至图6对应的实施例中的公开的各方法、步骤及逻辑框图。
可以理解的，上述图7所示实施例中的装置可以以图9所示的PLC设备实现。具体的，处理单元701可以由调制解调器102、处理器103、调制解调器202、处理器203实现，收发单元702可以由通信接口101和通信接口201实现。
子机100中的调制解调器102对通过通信接口101接收到的用户设备的上行数据进行调制，得到调制后的信号，处理器103将调制后的信号加载于电流，通过电力线传输至母机200。母机中的处理器203在从子机100接收到的信号中提取出调制后的信号，再通过调制解调器202解调出上行数据，通过网线发往网络接入设备。同理，母机200中的调制解调器202将通过通信接口201接收到的发往用户设备的下行数据进行调制，得到调制后的信号，处理器203将调制后的信号加载于电流，通过电力线传输至子机100。子机100中的处理器103在从母机接收到的信号中提取出调制后的信号，再通过调制解调器解102调出下行数据，并通过通信接口101将下行数据传输至用户设备。
在一示例中，如图9所示，子机100中的处理器103包括噪声信号获取模块，噪声信号获取模块用于获取电路中的噪声信号。噪声信号获取模块具体可以采用硬件的形式实现，例如，噪声信号获取模块可以包括数模转换器、模数转换器以及模拟前端等硬件，也可以采用硬件和软件功能模块的相结合的形式实现。母机200中的处理器203包括噪声信号分析模块，噪声信号分析模块可以基于子机100采集的噪声信号获取第一数据。噪声信号分析模块具体可以采用硬件的形式实现，也可以采用软件功能模块的形式实现，还可以采用硬件和软件功能模块的相结合的形式实现。
在一示例中，子机100中的处理器103包括噪声信号获取模块和噪声信号分析模块，噪声信号采集模块可以获取电路中的噪声信号，噪声信号分析模块可以基于该噪声信号获取第一数据，并将第一数据传输至母机200。
在一示例中，母机200中的处理器203包括噪声信号获取模块和噪声信号分析模块， 噪声信号采集模块可以获取电路中的噪声信号，噪声信号分析模块可以基于该噪声信号获取第一数据。
应理解的是，噪声信号采集模块可以集成于处理器103(或处理器203)内部，或者作为一个单独的芯片位于处理器103(或处理器203)外部。噪声信号分析模块可以集成于处理器103(或处理器203)内部，或者作为一个单独的芯片位于处理器103(或处理器203)外部。噪声信号采集模块和噪声信号采集模块还可以集成于一个芯片位于处理器103(或处理器203)外部。
本申请实施例还提供一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序，当该计算机程序在计算机上运行时，使得计算机执行上述如图3所示实施例的方法。
本申请实施例提供的方法中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本发明实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、网络设备、用户设备或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(digital subscriber line，简称DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机可以存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质(例如，软盘、硬盘、磁带)、光介质(例如，数字视频光盘(digital video disc，简称DVD))、或者半导体介质(例如,固态硬盘Solid State Disk SSD)等。
以上所述，以上实施例仅用以对本申请的技术方案进行了详细介绍，但以上实施例的说明只是用于帮助理解本发明实施例的方法，不应理解为对本发明实施例的限制。本技术领域的技术人员可轻易想到的变化或替换，都应涵盖在本发明实施例的保护范围之内。

Claims (17)

  1. 一种识别电器的方法，其特征在于，包括：
    电力线通信PLC设备获取电路中的噪声信号；
    所述PLC设备基于所述噪声信号获取第一数据，所述第一数据用于描述所述噪声信号的时频特征；
    所述PLC设备基于电器识别模型和所述第一数据，获得所述噪声信号对应的电器识别结果；其中，所述电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。
  2. 如权利要求1所述的方法，其特征在于，所述PLC设备基于电器识别模型和所述第一数据，获得所述噪声信号对应的电器识别结果，包括：
    所述PLC设备向服务器发送所述第一数据，所述服务器存储所述电器识别模型；
    所述PLC设备从所述服务器接收所述噪声信号对应的电器识别结果。
  3. 如权利要求1或2所述的方法，其特征在于，所述PLC设备基于所述噪声信号获取所述第一数据，包括：
    所述PLC设备将所述噪声信号分段，基于分段后的噪声信号获取第一数据。
  4. 如权利要求3所述的方法，其特征在于，所述噪声信号的信号长度为至少一个交流电周期，所述噪声信号包括N个采样点，N为正整数；
    所述PLC设备将所述噪声信号分段，基于分段后的噪声信号获取第一数据，包括：
    所述PLC设备将所述N个采样点分为M个片段，针对所述M个片段中的每个片段提取时域特征，获得1个M维数据，M为正整数，M≤N；
    所述PLC设备将所述N个采样点分为K个片段，从所述K个片段中选取L个片段，针对所述L个片段中的每个片段提取频域特征，获得L个M维数据，L和K均为正整数，L≤K≤N；
    所述PLC设备将所述1个M维数据和所述L个M维数据作为第一数据。
  5. 如权利要求4所述的方法，其特征在于，所述PLC设备针对所述M个片段中的每个片段提取时域特征，包括：
    所述PLC设备针对所述M个片段中的每个片段计算最大值、或均值、或分位数。
  6. 如权利要求4或5所述的方法，其特征在于，L个片段包括采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段中的至少一个。
  7. 如权利要求4-6任一项所述的方法，其特征在于，所述PLC设备针对所述L个片段中的每个片段提取频域特征，包括：
    所述PLC设备针对所述L个片段中的每个片段计算功率谱密度、或能量谱密度、或频谱密度。
  8. 如权利要求1-7任一项所述的方法，其特征在于，所述方法还包括：
    在所述PLC设备获得所述噪声信号对应的电器识别结果之后，所述PLC设备向终端设备发送所述电器识别结果。
  9. 一种识别电器的装置，其特征在于，包括：
    处理单元，用于获取电路中的噪声信号；
    所述处理单元，用于基于所述噪声信号获取第一数据，所述第一数据用于描述所述噪声信号的时频特征；
    所述处理单元，还用于基于电器识别模型和所述第一数据，获得所述噪声信号对应的电器识别结果；其中，所述电器识别模型是基于包括至少一种已知电器的噪声信号的信号得到的。
  10. 如权利要求9所述的装置，其特征在于，所述装置还包括：
    发送单元，用于向服务器发送所述第一数据，所述服务器存储所述电器识别模型；
    接收单元，用于从所述服务器接收所述噪声信号对应的电器识别结果。
  11. 如权利要求9或10所述的装置，其特征在于，所述处理单元，用于将所述噪声信号分段，基于分段后的噪声信号获取第一数据。
  12. 如权利要求11所述的装置，其特征在于，所述噪声信号的信号长度为至少一个交流电周期，所述噪声信号包括N个采样点，N为正整数；
    所述处理单元，用于将所述N个采样点分为M个片段，针对所述M个片段中的每个片段提取时域特征，获得1个M维数据，M为正整数，M≤N；将所述N个采样点分为K个片段，从所述K个片段中选取L个片段，针对所述L个片段中的每个片段提取频域特征，获得L个M维数据，L和K均为正整数，L≤K≤N；将所述1个M维数据和所述L个M维数据作为第一数据。
  13. 如权利要求12所述的装置，其特征在于，所述处理单元，用于针对所述M个片段中的每个片段计算最大值、或均值、或分位数。
  14. 如权利要求12或13所述的装置，其特征在于，L个片段包括采样点包括波峰的片段，采样点包括波谷的片段，采样点包括零点的片段中的至少一个。
  15. 如权利要求12-14任一项所述的装置，其特征在于，所述处理单元，用于针对所述L个片段中的每个片段计算功率谱密度、或能量谱密度、或频谱密度。
  16. 如权利要求9-15任一项所述的装置，其特征在于，所述发送单元，还用于在获得所述噪声信号对应的电器识别结果之后，向终端设备发送所述电器识别结果。
  17. 一种计算机存储介质，其特征在于，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令在被所述计算机调用时用于使所述计算机执行权利要求1～8中的任一项所述的方法。
PCT/CN2020/094991 2019-06-11 2020-06-08 一种识别电器的方法及装置 WO2020248946A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP20821932.9A EP3968530A4 (en) 2019-06-11 2020-06-08 METHOD AND DEVICE FOR IDENTIFYING AN ELECTRICAL DEVICE
US17/547,037 US11658702B2 (en) 2019-06-11 2021-12-09 Electric appliance identification method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910502300.2A CN112152667A (zh) 2019-06-11 2019-06-11 一种识别电器的方法及装置
CN201910502300.2 2019-06-11

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/547,037 Continuation US11658702B2 (en) 2019-06-11 2021-12-09 Electric appliance identification method and apparatus

Publications (1)

Publication Number Publication Date
WO2020248946A1 true WO2020248946A1 (zh) 2020-12-17

Family

ID=73780667

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/094991 WO2020248946A1 (zh) 2019-06-11 2020-06-08 一种识别电器的方法及装置

Country Status (4)

Country Link
US (1) US11658702B2 (zh)
EP (1) EP3968530A4 (zh)
CN (1) CN112152667A (zh)
WO (1) WO2020248946A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101368985A (zh) * 2007-08-14 2009-02-18 通用电气公司 认知式电功率表
US20090072985A1 (en) * 2007-09-18 2009-03-19 Georgia Tech Research Corporation Detecting actuation of electrical devices using electrical noise over a power line
CN101772901A (zh) * 2007-08-01 2010-07-07 索尼株式会社 用于在电力线信道上传输信号的方法和电力线通信调制解调器
CN102193041A (zh) * 2010-01-25 2011-09-21 索尼公司 用于识别连接到电网的电气设备的方法、系统和传感器

Family Cites Families (30)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5812437A (ja) * 1981-07-15 1983-01-24 Matsushita Electric Works Ltd 電力線搬送装置
US6329905B1 (en) * 2000-03-23 2001-12-11 Thalia Products, Inc. Power line communication system and method
US7030734B2 (en) * 2002-01-02 2006-04-18 International Business Machines Corporation Verification of proper coupling of power line communication units to power circuits
JP5075664B2 (ja) * 2008-02-15 2012-11-21 株式会社東芝 音声対話装置及び支援方法
CA2758270C (en) * 2009-04-10 2016-01-05 Lg Electronics Inc. Home appliance and home appliance system
US8156055B2 (en) * 2009-05-04 2012-04-10 ThinkEco, Inc. System and method for utility usage, monitoring and management
US20120093240A1 (en) * 2010-01-18 2012-04-19 Atheros Communications, Inc. Interference detection in a powerline communication network
US20120134395A1 (en) * 2010-11-30 2012-05-31 Texas Instruments Incorporated Power Line Communications (PLC) Across Different Voltage Domains Using Multiple Frequency Subbands
CN102347811B (zh) * 2011-11-04 2013-10-23 兆讯恒达微电子技术(北京)有限公司 一种电力线通信信道脉冲噪声检测方法
KR20130090978A (ko) * 2012-02-07 2013-08-16 삼성전자주식회사 전력선 통신 장치 및 방법, 이를 이용한 부하 전력 모니터링 장치 및 방법
AU2013251524B2 (en) 2012-04-25 2016-05-12 Bidgely Inc. Energy disaggregation techniques for low resolution whole-house energy consumption data
EP2893605B1 (en) * 2012-09-06 2019-04-24 Auckland UniServices Limited Local demand side power management for electric utility networks
US9344151B2 (en) * 2012-11-21 2016-05-17 Facebook, Inc. Power line communication monitor
US9647722B2 (en) * 2013-03-15 2017-05-09 Dockon Ag Adaptive line equalizer for improving data communication over less than perfect power lines or transmission lines
US9465059B2 (en) * 2013-03-28 2016-10-11 Sony Corporation Method, sensor and system for analyzing appliances in a power line network
US20160028437A1 (en) * 2014-07-22 2016-01-28 Hyperion Energy Group Llc Methods, systems, and apparatus for the monitoring, controlling, and communicating of lighting systems
KR101599107B1 (ko) * 2013-08-07 2016-03-02 설병화 DP(Data Plus)방식을 이용한 전력선의 데이터 송수신 방법
CN103544953B (zh) * 2013-10-24 2016-01-20 哈尔滨师范大学 一种基于背景噪声最小统计量特征的声音环境识别方法
US9615270B2 (en) * 2013-11-04 2017-04-04 State Grid Corporation Of China Intelligent power utilization communication product detecting system
KR102188090B1 (ko) * 2013-12-11 2020-12-04 엘지전자 주식회사 스마트 가전제품, 그 작동방법 및 스마트 가전제품을 이용한 음성인식 시스템
CN105098806B (zh) * 2014-05-21 2018-05-04 华为技术有限公司 电能调度方法和装置、电源管理方法和装置
CN104270206B (zh) * 2014-09-25 2017-12-05 国家电网公司 一种用于用电信息采集系统的干扰类型辨识装置及方法
CN106054636B (zh) * 2016-06-27 2019-03-19 国网河南省电力公司南阳供电公司 基于载波技术的家用电器智能控制系统
KR102502574B1 (ko) * 2016-08-12 2023-02-22 삼성전자주식회사 오디오 신호를 처리하기 위한 장치 및 방법
CN106409289B (zh) * 2016-09-23 2019-06-28 合肥美的智能科技有限公司 语音识别的环境自适应方法、语音识别装置和家用电器
CN108073856A (zh) * 2016-11-14 2018-05-25 华为技术有限公司 噪音信号的识别方法及装置
CN106940806B (zh) * 2017-03-07 2020-06-02 中国海洋大学 一种识别电器的用电状态的方法和系统
CN107102196A (zh) * 2017-05-10 2017-08-29 中国海洋大学 一种复杂电力信号的噪声滤除装置及方法
CN107368805A (zh) * 2017-07-17 2017-11-21 深圳森阳环保材料科技有限公司 一种基于智能家居系统电器识别的遥控器
CN108880622B (zh) * 2018-09-20 2021-03-19 广东石油化工学院 一种电力线通信系统中脉冲噪声的识别方法及系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101772901A (zh) * 2007-08-01 2010-07-07 索尼株式会社 用于在电力线信道上传输信号的方法和电力线通信调制解调器
CN101368985A (zh) * 2007-08-14 2009-02-18 通用电气公司 认知式电功率表
US20090072985A1 (en) * 2007-09-18 2009-03-19 Georgia Tech Research Corporation Detecting actuation of electrical devices using electrical noise over a power line
CN102193041A (zh) * 2010-01-25 2011-09-21 索尼公司 用于识别连接到电网的电气设备的方法、系统和传感器

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3968530A4 *

Also Published As

Publication number Publication date
EP3968530A4 (en) 2022-07-20
US20220103203A1 (en) 2022-03-31
EP3968530A1 (en) 2022-03-16
CN112152667A (zh) 2020-12-29
US11658702B2 (en) 2023-05-23

Similar Documents

Publication Publication Date Title
US11616845B2 (en) Electrical meter for determining a power main of a smart plug
WO2017211288A1 (zh) 非侵入式在线实时电力负荷识别方法及识别系统
Liu et al. Mobile-edge computing framework with data compression for wireless network in energy internet
US20120093240A1 (en) Interference detection in a powerline communication network
CN106462905B (zh) 用于识别电力客户的负荷波动性的系统、方法和装置及有形计算机可读介质
CN102882556B (zh) 一种用电信息的采集方法、装置和系统
TW201806421A (zh) 功率分配方法以及使用所述方法的基地台
CN111459135A (zh) 基于物联网的智能家居故障状态追溯方法及中控中心
WO2020248946A1 (zh) 一种识别电器的方法及装置
CN110166892A (zh) 一种带万能红外控制功能的智能音箱系统及运作方法
CN115175347A (zh) 一种无线携能通信网络资源分配优化方法
CN112782506A (zh) 一种电力物联网领域多通道窄带无线自主网系统及其通信方法
US20150220124A1 (en) Power connectivity monitoring for computing systems
WO2020082214A1 (zh) 一种抄表方法、装置及系统
CN110944378A (zh) 5g移动通信场景下d2d通信的noma功率分配方法
CN105005245A (zh) 数据处理方法
Aggarwal et al. Optical Wireless Channel Characterization Using Discrete And Stationary Wavelet Transform
CN220629347U (zh) 基于arm数据传输的智能家居系统
CN113691966B (zh) 基于信能同传的音频播放方法、系统、设备及存储介质
Fu et al. Implementation of centralized cooperative spectrum sensing based on USRP
CN203708276U (zh) 一种集成Zigbee的无线MoCA系统
CN105187458A (zh) 一种基于硬件本地认证请求的WiFi芯片认证系统及认证方法
CN117614484A (zh) 一种宽带载波通信异常处理方法及系统
CN117749837A (zh) 一种用电信息采集终端信息交互方法
CN115622617A (zh) 一种基于noma的认知反向散射通信鲁棒传输方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20821932

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020821932

Country of ref document: EP

Effective date: 20211209