Power grid information transmission method and system based on Internet of Things

Abstract

The present application relates to the technical field of power grid information transmission, and specifically to a power grid information transmission method and system based on the Internet of Things, the method comprising: determining the optimized weight of each neuron in each training period of each neural network; based on the data symbol in each training period and the optimized weight of the winning neuron, determining the channel interference degree when each subcarrier receives each data symbol, and combining the initial learning rate, determining the optimized learning rate of the winning neuron and the optimized learning rate of all non-winning neurons in each training period of each neural network; determining the channel gain estimation value when each subcarrier receives each data symbol, and transmitting the power grid information. The present application adopts a blind channel estimation method of a self-organizing competitive neural network to perform channel estimation on different subcarriers respectively, thereby improving the reliability of power grid information transmission.

Description

Technical Field

[0001] The present application relates to the technical field of power grid information transmission, and specifically to a power grid information transmission method and system based on the Internet of Things. Background Art

[0002] With the rapid development of Internet of Things technology, the integration of power system and Internet of Things is becoming closer. The introduction of Internet of Things technology in smart grid can collect grid information through a series of sensors such as temperature, humidity, vibration, etc. installed on smart grid equipment and lines, and transmit it to the monitoring center through object-to-object connection, so as to monitor the operation of smart grid in real time and respond promptly when the grid fails to ensure the safe operation of the power system.

[0003] Power line carrier communication is a communication method that uses power lines to achieve data transmission and information exchange. It is combined with the OFDM communication system to realize power grid information transmission based on the Internet of Things. Channel estimation is an important part of power line carrier communication, and its performance directly affects the reliability and effectiveness of the entire communication system. At present, the traditional blind channel estimation algorithm cannot adapt to the time-varying characteristics of the power line carrier communication channel and the interference of impulse noise in the channel. The channel estimation performance obtained is poor, which reduces the reliability of power grid information transmission. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a power grid information transmission method and system based on the Internet of Things. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for transmitting power grid information based on the Internet of Things, the method comprising the following steps:

[0006] All power grid information of the Internet of Things is obtained, and the power grid information is converted into multiple symbols by analog-to-digital conversion and constellation mapping. The transmission channel is divided into multiple subcarriers by orthogonal frequency division multiplexing technology, and the symbols are transmitted through the subcarriers and the symbols obtained at the receiving end are recorded as data symbols. Based on all the data symbols received by each subcarrier, the neural network is trained to obtain the optimized weights and optimized learning rates of each neuron in the neural network when each subcarrier receives each data symbol. The training process is as follows:

[0007] Each subcarrier corresponds to a neural network. Each time each subcarrier receives a data symbol, its neural network is trained once. A preset training period corresponds to a data symbol. In each training period, the initial weights and initial learning rates of all neurons are respectively the optimized weights and optimized learning rates at the end of the previous training period. According to the initial weights of each neuron and its distance from the data symbol, the competition distance of each neuron in each training period is determined, and the winning and non-winning neurons in each training period are obtained based on the competition distance.

[0008] In each training period, the initial learning rate of each neuron is used as the weighted value of the data symbol and its conjugate difference, and the weighted result is fused with the initial weight of the corresponding neuron to obtain the optimized weight of each neuron in each training period of each neural network;

[0009] Based on the data symbols in each training period and the optimized weights of the winning neurons, the channel interference degree when each subcarrier receives each data symbol is determined, and combined with the initial learning rate, the optimized learning rate of the winning neurons in each training period of each neural network is determined; based on the competition distance of all neurons in each training period and the initial learning rate of each non-winning neuron, the optimized learning rate of each non-winning neuron in each training period of each neural network is determined;

[0010] Based on the optimized weights of all neurons in each training period, the estimated channel gain value when each subcarrier receives each data symbol is determined to transmit the power grid information.

[0011] Preferably, the competition distance of each neuron in each training period is the product of the modulus of the initial weight of each neuron in each training period and the distance.

[0012] Preferably, the method for determining the winning and non-winning neurons in each training period of each neural network is:

[0013] The neuron with the smallest competition distance in each period of each neural network is regarded as the winning neuron, and the remaining neurons are non-winning neurons.

[0014] Preferably, the expression of the optimized weight of each neuron in each training period of each neural network is:

[0015] ; In the formula, represents the optimized weight of neuron n in the kth training period of neural network w; represents the initial weight of neuron n in the kth training period of neural network w; represents the initial learning rate of the winning neuron in the kth training period of the neural network w; represents the initial learning rate of the non-winning neuron n in the kth training period of the neural network w; , They respectively represent the data symbol and the conjugate of the data symbol in the kth training period of the neural network w.

[0016] Preferably, the channel interference degree when each subcarrier receives each data symbol is a modulus value of the ratio of each data symbol received by each subcarrier to the optimized weight of the winning neuron in the training period in which it is located.

[0017] Preferably, the optimized learning rate of the winning neuron in each training period of each neural network is the product of the initial learning rate of the winning neuron in each training period of each neural network and the channel interference degree of the data symbol.

[0018] Preferably, the method for determining the optimized learning rate of each non-winning neuron in each training period of each neural network is:

[0019] The ratio of the competitive distance of each non-winning neuron to the competitive distance of the winning neuron in each training period of each neural network is analyzed, and the optimized learning rate of each non-winning neuron in each training period of each neural network is the product of the initial learning rate of each non-winning neuron in each training period of each neural network and the ratio.

[0020] Preferably, the expression of the channel gain estimation value when each subcarrier receives each data symbol is: ; In the formula, represents the channel gain estimation value when the s-th subcarrier receives data symbol m; represents the optimized weight of neuron n in the training period where the data symbol m received by the s-th subcarrier is located; j represents an imaginary unit; Represents the number of all neurons in the neural network for the s-th subcarrier.

[0021] Preferably, the transmitting of the power grid information includes:

[0022] The equalized value of the data symbol m received by the sth subcarrier The expression is: ; In the formula, represents the data symbol m received by the s-th subcarrier;

[0023] The equalization value of each data symbol received by all subcarriers is demodulated to obtain a digital signal, and all digital signals are converted into digital-to-analog signals to obtain the transmitted power grid information.

[0024] In a second aspect, an embodiment of the present application further provides an Internet of Things-based power grid information transmission system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned Internet of Things-based power grid information transmission methods when executing the computer program.

[0025] This application has at least the following beneficial effects:

[0026] The present application analyzes the initial weights and initial learning rates of all neurons in each training period of the neural network of each subcarrier, which are the optimized weights and optimized learning rates at the end of the previous training period, respectively. According to the initial weights of each neuron and its distance from the data symbol, the competition distance of each neuron in each training period is determined. The beneficial effect of the present application is that by adjusting the role of different weights in the neuron weight sequence in calculating the neuron competition distance, the influence of impulse noise on subsequent channel estimation is reduced to avoid an increase in the bit error rate; and based on the competition distance, the winning and non-winning neurons in each training period are obtained; based on the data symbol in each training period, the initial weights of each neuron and initial learning rate, determine the optimized weights and optimized learning rates of each neuron in each training period of each neural network, and its beneficial effect is that by accumulating the changes in channel interference in a short period of time, the degree of noise influence on the transmission of the next data symbol is judged, and the accuracy of the calculation of neuron learning rate and the adaptability of different neurons in the neural network to channel changes are improved; the present application is based on the optimized weights of all neurons in each training period, determines the channel gain estimation value when each subcarrier receives each data symbol, and equalizes the data symbol based on the channel gain estimation value, demodulates the equalized data symbol, and obtains the power grid information after transmission, and its beneficial effect is that the reliability of data transmission is improved. The present application adopts the blind channel estimation method of the self-organizing competitive neural network, performs channel estimation on different subcarriers respectively, improves the accuracy of the neural network in channel estimation, and improves the reliability of power grid information transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 A flowchart of a method for transmitting power grid information based on the Internet of Things provided in accordance with an embodiment of the present application;

[0029] Figure 2A schematic diagram of a neural network training process provided for one embodiment of the present application;

[0030] Figure 3 A diagram of the channel estimation optimization effect provided for an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0031] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of the power grid information transmission method and system based on the Internet of Things proposed in the present application are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0032] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0033] The specific scheme of the power grid information transmission method and system based on the Internet of Things provided by the present application is described in detail below with reference to the accompanying drawings.

[0034] See also Figure 1 , which shows a flowchart of a method for transmitting power grid information based on the Internet of Things provided by an embodiment of the present application, the method comprising the following steps:

[0035] Step S1: Obtain all power grid information of the Internet of Things, convert the power grid information into multiple symbols using analog-to-digital conversion and constellation mapping, divide the transmission channel into multiple subcarriers using orthogonal frequency division multiplexing technology, transmit the symbols through the subcarriers and obtain the symbols at the receiving end, which are recorded as data symbols.

[0036] The close integration of IoT technology and smart grid can obtain various grid information in the power system in real time, such as: grid operation status, grid parameters, environmental information and fault data. Sensors are installed on the IoT perception layer to collect various grid information of grid equipment within a preset time period, and the collected various grid information is converted into digital signals suitable for transmission on the communication channel through sampling, quantization, encoding and other operations. Furthermore, the digital signal is mapped into multiple symbols using the constellation mapping of QPSK orthogonal phase shift keying modulation. In QPSK orthogonal phase shift keying modulation, each modulation symbol carries 2 bits of information, and each combination of 2 bits is converted into a complex signal.

[0037] It should be noted that there are many mapping methods. In this embodiment, the QPSK quadrature phase shift keying modulation mapping is: 00—1, 01—j, 10—-1, 11—-j. Each symbol after mapping is a complex number. The implementer can also map 00 to 1+j, 01 to -1+j, 10 to -1-j, and 11 to 1-j. The implementer can set the mapping rules according to the specific situation, and this embodiment does not impose any special restrictions.

[0038] Therefore, in this embodiment, the digital signal is mapped by QPSK constellation to obtain the transmission symbol .

[0039] Furthermore, OFDM orthogonal frequency division multiplexing is used to divide the communication channel frequency band into 256 subcarriers. After the FFT fast Fourier transform at the receiving end of the subcarrier, the symbol received by the subcarrier is obtained and recorded as a data symbol. The data symbol is different from the symbol sent by the sending end of the subcarrier, and the data symbol contains additive noise and multiplicative noise in the frequency band where the subcarrier is located.

[0040] Among them, QPSK quadrature phase shift keying modulation and its constellation diagram, and OFDM orthogonal frequency division multiplexing are all well-known technologies in the field of signal processing, and their specific principles and processes are not repeated here.

[0041] Step S2: Every time each subcarrier receives a data symbol, the self-organizing competitive neural network of each subcarrier is trained once. A preset training period corresponds to one data symbol. In each training period, the initial weights and initial learning rates of all neurons are the optimized weights and optimized learning rates at the end of the previous training period, respectively. According to the initial weights of each neuron and its distance from the data symbol, the competition distance of each neuron in each training period is determined, and the winning and non-winning neurons in each training period are obtained based on the competition distance.

[0042] In the coherent OFDM communication system used in power line carrier communication, the modulation of the transmitted symbols adopts a high-order modulation method. Therefore, the receiving end of the monitoring center needs to obtain accurate state information of the power line carrier communication channel in order to correctly demodulate the received data symbols.

[0043] Since there are many branches in the power grid, the signal is reflected at the node, so that it cannot be directly transmitted from the transmitter to the receiver on the power line. Therefore, power line carrier communication has multipath propagation characteristics, which causes frequency selective fading of the transmitted signal. At the same time, the continuous access and disconnection of different loads in the power grid causes different channel attenuation conditions at different times, resulting in different fading conditions of subcarriers in different frequency bands at different times.

[0044] After the data symbols obtained by QPSK modulation are transmitted through the power grid, their amplitude and phase will change in real time, resulting in a more dispersed distribution of the data symbols in the constellation diagram, deviating from the initial symbol center in the constellation diagram. The initial symbol center of the constellation diagram is .

[0045] A self-organizing competitive neural network is used to estimate the channels of different subcarriers of power line carrier communication. Since data symbols are mainly divided into four types, the number of neurons of the self-organizing competitive neural network is set to 4. As the initial weight of the self-organizing competitive neural network of each subcarrier, each subcarrier has a neural network composed of 4 neurons. When a data symbol is received, the neural network training begins, specifically:

[0046] Each subcarrier corresponds to a neural network, and each time each subcarrier receives a data symbol, its neural network is trained once, a preset training period corresponds to a data symbol, and each training of the neural network corresponds to a data symbol. The preset training period is set to the time when each subcarrier receives a data symbol and its self-organizing competitive neural network starts training to the time when the weights and learning rates of all neurons in the self-organizing competitive neural network are updated once. Among them, the neural network completes one training in each training period, and the weights and learning rates of the neurons in the neural network are also updated after each training is completed. Among them, it should be noted that the neural network in this embodiment adopts a self-organizing competitive neural network, and the following neural networks all represent self-organizing competitive neural networks.

[0047] In the neural network of each subcarrier, for each training period, the initial weights and initial learning rates of all neurons are the optimized weights and optimized learning rates at the end of the previous training period.

[0048] Furthermore, the competitive distance of each neuron in each training period is determined based on the initial weight of each neuron and its distance from the data symbol. The competitive distance of each neuron in each training period is determined by multiplying the modulus of the initial weight of each neuron in each training period by the distance.

[0049] According to the competitive distance of each neuron in each training period, it can be understood that the larger the initial weight of the neuron, the larger the modulus value, the greater the distance between the neuron and the data symbol, the larger the competitive distance of each neuron, and the less likely it is to be selected as the winning neuron; conversely, the smaller the modulus value of the initial weight of the neuron, the greater the distance between the neuron and the data symbol, and the more likely it is to be selected as the winning neuron.

[0050] Further, the winning and non-winning neurons in each training period are obtained based on the competition distance, specifically:

[0051] The neuron with the smallest competition distance in each period of each neural network is regarded as the winning neuron, and the remaining neurons are non-winning neurons.

[0052] Step S3: Based on the data symbols in each training period, the initial weights of each neuron and the initial learning rate, determine the optimized weights of each neuron in each training period; based on the data symbols in each training period and the optimized weights of the winning neurons, determine the channel interference degree when each subcarrier receives each data symbol, and determine the initial learning rate of the winning neuron in each training period of each neural network in combination with the initial learning rate; based on the competition distance of all neurons in each training period and the optimized learning rate of each non-winning neuron, determine the optimized learning rate of each non-winning neuron in each training period of each neural network.

[0053] In order to make the neural network better adapt to the input power grid information data, the characteristics of the data symbols received by the neural network are analyzed, and similar data symbols are matched to maintain the overall dynamic balance of the neural network. At the same time, it is ensured that when the channel interference is large, the neurons can adjust the weights faster to adapt to the rapid changes of the channel. And by adjusting the learning rate of the neural network, the neural network can adapt to the changes of the input data faster and improve the accuracy of the channel state estimation.

[0054] Thus, based on the data symbols in each training period, the initial weights of each neuron and the initial learning rate, the optimized weights of each neuron in each training period are determined; based on the data symbols in each training period and the optimized weights of the winning neurons, the channel interference degree when each subcarrier receives each data symbol is determined, and combined with the initial learning rate, the optimized learning rate of the winning neuron in each training period of each neural network is determined; based on the competition distance of all neurons in each training period and the optimized learning rate of the winning neuron, the optimized learning rate of each non-winning neuron in each training period of each neural network is determined, specifically:

[0055] (1) Based on the data symbols, the initial weights of each neuron and the initial learning rate in each training period, the optimized weights of each neuron in each training period are determined.

[0056] The expression of the optimized weight of neuron n in the kth training period of neural network w is: ; In the formula, represents the optimized weight of neuron n in the kth training period of neural network w; represents the initial weight of neuron n in the kth training period of neural network w; represents the initial learning rate of the winning neuron in the kth training period of the neural network w; represents the initial learning rate of the non-winning neuron n in the kth training period of the neural network w; , They represent the data symbol and the conjugate of the data symbol in the kth training period of the neural network w, respectively. Among them, after the subcarrier receives the first data symbol, the initial learning rate of all neurons is set to 1 when its self-organizing competitive neural network starts training.

[0057] (2) Further, based on the data symbols in each training period and the optimized weights of the winning neurons, the channel interference degree when each subcarrier receives each data symbol is determined, specifically:

[0058] The channel interference degree when each subcarrier receives each data symbol is the modulus value of the ratio of each data symbol received by each subcarrier to the optimized weight of the winning neuron in the training period.

[0059] According to the channel interference degree when each subcarrier receives each data symbol, it can be understood that when the modulus value of the ratio of the data symbol of the subcarrier to the weight of the winning neuron in its training period is larger, the interference degree of noise on the channel is greater and the channel interference degree is greater; conversely, when the modulus value of the ratio of the data symbol of the subcarrier to the weight of the winning neuron in its training period is smaller, the interference degree of noise on the channel is smaller and the channel interference degree is smaller.

[0060] (3) Further, based on the channel interference degree when each subcarrier receives each data symbol and in combination with the initial learning rate, the optimized learning rate of the winning neuron in each training period of each neural network is determined, specifically:

[0061] The optimized learning rate of the winning neuron in each training period of each neural network is the product of the initial learning rate of the winning neuron in each training period of each neural network and the channel interference degree of the data symbol.

[0062] (4) Further, the ratio of the competitive distance of each non-winning neuron to the competitive distance of the winning neuron in each training period of each neural network is analyzed, and the optimized learning rate of each non-winning neuron in each training period of each neural network is the product of the initial learning rate of each non-winning neuron in each training period of each neural network and the ratio.

[0063] By analyzing the difference in the competition distances of the winning and non-winning neurons of different subcarriers in the current channel of the power line carrier communication system, the optimized learning rate of the next non-winning neuron is updated. When the competition distance of the non-winning neuron is closer to the competition distance of the winning neuron, the distribution of the data symbol reception value in the constellation diagram is more discrete, and the calculated optimized learning rate of the non-winning neuron is larger to adapt to the severe noise in the channel. In addition, when the optimized learning rate of the winning neuron is larger, the channel changes faster. At the same time, a larger optimized learning rate of the non-winning neuron is set to improve the adaptability of different neurons in the neural network to channel changes and improve the channel estimation performance.

[0064] Preferably, the neural network training process diagram provided in this embodiment is as follows Figure 2 shown.

[0065] Step S4: Based on the weights of all neurons in each training period, the estimated channel gain when each subcarrier receives each data symbol is determined, and the power grid information is transmitted.

[0066] According to the different neuron weights in the neural network when the monitoring center receives each data symbol reception value, the channel gain of different subcarriers in the current power line carrier communication channel is estimated:

[0067] The channel gain estimate when the s-th subcarrier receives each data symbol m The expression is:

[0068] ; In the formula, represents the optimized weight of neuron n in the training period where the data symbol m received by the s-th subcarrier is located; j represents an imaginary unit; Represents the number of all neurons in the neural network for the s-th subcarrier.

[0069] The blind channel estimation method of the self-organizing competitive neural network is used to perform channel estimation on different subcarriers respectively, reducing the influence of the selective frequency fading of the power line carrier communication channel on the channel estimation. At the same time, by adjusting the calculation of different neuron learning rates, the performance of the neural network for channel estimation is improved, and its ability to resist nonlinear effects and impulse noise interference in the power line carrier communication process is enhanced.

[0070] Preferably, the channel estimation optimization effect diagram provided in this embodiment is as follows: Figure 3 As shown in the figure, the horizontal axis represents the signal-to-noise ratio in dB, the vertical axis represents the bit error rate, the dotted line represents the channel estimation curve with a fixed learning rate, and the solid line represents the channel estimation curve after optimizing the learning rate of the neural network in this embodiment.

[0071] Further, according to the channel gain estimation value when each subcarrier receives each data symbol and the data symbol received by the subcarrier, the equalization value of the data symbol received by the subcarrier is determined, specifically:

[0072] The equalized value of the data symbol m received by the sth subcarrier The expression is: ; In the formula, Represents the data symbol m received by the s-th subcarrier.

[0073] Furthermore, the equalization value of each data symbol received by all subcarriers is demodulated to obtain digital signals, and all digital signals are converted from digital to analog to obtain power grid information.

[0074] Among them, digital-to-analog conversion is a well-known technology in signal processing, and the specific principle and process will not be repeated here.

[0075] So far, this embodiment uses the blind channel estimation method of the self-organizing competitive neural network to perform channel estimation on different subcarriers respectively, thereby reducing the influence of the selective frequency fading of the power line carrier communication channel on the channel estimation. At the same time, by adjusting the calculation of the learning rate of different neurons, the performance of the neural network for channel estimation is improved, and its ability to resist nonlinear effects and impulse noise interference in the power line carrier communication process is enhanced, thereby improving the reliability of power grid information transmission.

[0076] Based on the same inventive concept as the above method, an embodiment of the present application also provides an Internet of Things-based power grid information transmission system, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned Internet of Things-based power grid information transmission methods are implemented.

[0077] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0078] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0079] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A power grid information transmission method based on the Internet of Things, characterized in that: The method comprises the following steps: All power grid information of the Internet of Things is obtained, and the power grid information is converted into multiple symbols by analog-to-digital conversion and constellation mapping. The transmission channel is divided into multiple subcarriers by orthogonal frequency division multiplexing technology, and the symbols are transmitted through the subcarriers and the symbols obtained at the receiving end are recorded as data symbols. Based on all the data symbols received by each subcarrier, the neural network is trained to obtain the optimized weights and optimized learning rates of each neuron in the neural network when each subcarrier receives each data symbol. The training process is as follows: Each subcarrier corresponds to a neural network. Each time each subcarrier receives a data symbol, its neural network is trained once. A preset training period corresponds to a data symbol. In each training period, the initial weights and initial learning rates of all neurons are respectively the optimized weights and optimized learning rates at the end of the previous training period. According to the initial weights of each neuron and its distance from the data symbol, the competition distance of each neuron in each training period is determined, and the winning and non-winning neurons in each training period are obtained based on the competition distance. In each training period, the initial learning rate of each neuron is used as the weighted value of the data symbol and its conjugate difference, and the weighted result is fused with the initial weight of the corresponding neuron to obtain the optimized weight of each neuron in each training period of each neural network; Based on the data symbols in each training period and the optimized weights of the winning neurons, the channel interference degree when each subcarrier receives each data symbol is determined, and combined with the initial learning rate, the optimized learning rate of the winning neurons in each training period of each neural network is determined; based on the competition distance of all neurons in each training period and the initial learning rate of each non-winning neuron, the optimized learning rate of each non-winning neuron in each training period of each neural network is determined; Based on the optimized weights of all neurons in each training period, the estimated channel gain of each subcarrier when receiving each data symbol is determined to transmit the power grid information; The competition distance of each neuron in each training period is the product of the modulus of the initial weight of each neuron in each training period and the distance; The method for determining the winning and non-winning neurons in each training period of each neural network is as follows: The neuron with the smallest competition distance in each period of each neural network is regarded as the winning neuron, and the remaining neurons are non-winning neurons; The expression of the optimized weight of each neuron in each training period of each neural network is: ; In the formula, represents the optimized weight of neuron n in the kth training period of neural network w; represents the initial weight of neuron n in the kth training period of neural network w; represents the initial learning rate of the winning neuron in the kth training period of the neural network w; represents the initial learning rate of the non-winning neuron n in the kth training period of the neural network w; , denote the data symbol and the conjugate of the data symbol in the kth training period of the neural network w respectively; The channel interference degree when each subcarrier receives each data symbol is the modulus value of the ratio of each data symbol received by each subcarrier to the optimized weight of the winning neuron in the training period in which it is located; The optimized learning rate of the winning neuron in each training period of each neural network is the product of the initial learning rate of the winning neuron in each training period of each neural network and the channel interference degree of the data symbol; The method for determining the optimized learning rate of each non-winning neuron in each training period of each neural network is as follows: The ratio of the competitive distance of each non-winning neuron to the competitive distance of the winning neuron in each training period of each neural network is analyzed, and the optimized learning rate of each non-winning neuron in each training period of each neural network is the product of the initial learning rate of each non-winning neuron in each training period of each neural network and the ratio.

2. The method for transmitting power grid information based on the Internet of Things according to claim 1, characterized in that: The expression of the channel gain estimation value when each subcarrier receives each data symbol is: ; In the formula, represents the channel gain estimation value when the s-th subcarrier receives data symbol m; represents the optimized weight of neuron n in the training period where the data symbol m received by the s-th subcarrier is located; j represents an imaginary unit; Represents the number of all neurons in the neural network for the s-th subcarrier.

3. The method for transmitting power grid information based on the Internet of Things according to claim 2, characterized in that: The transmitting of the power grid information includes: The equalized value of the data symbol m received by the sth subcarrier The expression is: ; In the formula, represents the data symbol m received by the s-th subcarrier; The equalization value of each data symbol received by all subcarriers is demodulated to obtain a digital signal, and all digital signals are converted into digital-to-analog signals to obtain the transmitted power grid information.

4. A power grid information transmission system based on the Internet of Things, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the power grid information transmission method based on the Internet of Things are implemented as described in any one of claims 1-3.

Images