A hierarchical modeling method and device for power line carrier channel

Abstract

The present invention provides a hierarchical modeling method and device for power line carrier channels, the method comprising: collecting access characteristics and topology change rules of multiple terminals; calculating the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules; dividing the power line carrier channel into several layers based on the empirical value of the evaluation function; for each layer of power line carrier channels, modeling is performed by a pre-trained shared neural network, and inter-layer sharing of power line carrier channel parameters is achieved; the modeling result is used for the transmission of distribution network data. Compared with the prior art, the evaluation function is calculated by access characteristics and topology change rules, and hierarchical modeling is performed based on the evaluation function, and power line carrier channels with similar attributes are divided into the same layer, which can effectively improve the adaptability to actual power line carrier communication, and the constructed model has a large fitting space and high precision, and can support the real-time and reliable transmission of distribution network data.

Description

Technical Field

[0001] The present invention relates to the field of power line carrier communication, and in particular to a hierarchical modeling method and device for a power line carrier channel. Background Art

[0002] Power line carrier communication is an important communication technology for low-voltage distribution networks. It uses transmission lines as information transmission media and has the advantages of wide coverage and low communication costs. However, with the access of a high proportion of new energy, the communication environment and network topology of the distribution network are becoming increasingly complex, and its load characteristics are becoming increasingly difficult to analyze, making it difficult to construct a power line carrier channel model, which is quite different from the actual channel characteristics.

[0003] The existing power line carrier channel modeling method is based on the parameter matching method, or the power line carrier channel model is established through the transmission line theory, which requires assumptions about the communication environment and network topology information. However, the model constructed by this method has a large amount of calculation and is only applicable to power line carrier channels with relatively fixed network topology. In addition, this method directly constructs the communication model of the entire network, with a large fitting space and low model accuracy. It is easy to continue to have insufficient adaptability to actual power line carrier communications with a high proportion of new energy access, and it is difficult to support the real-time and reliable transmission of distribution network data. Therefore, there is an urgent need for a hierarchical modeling method for power line carrier channels with a high proportion of new energy access. Summary of the invention

[0004] The present invention provides a hierarchical modeling method and device for a power line carrier channel to solve the technical problem of how to improve the insufficient adaptability of actual power line carrier communication for high-proportion new energy access.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a hierarchical modeling method for a power line carrier channel, comprising:

[0006] Collect access characteristics and topology change rules of various terminals;

[0007] Calculating an evaluation function of a power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules;

[0008] Based on the evaluation function and the empirical value of the evaluation function, dividing the power line carrier channel into several layers;

[0009] For each layer of power line carrier channel, a pre-trained shared neural network is used to model it, and the inter-layer sharing of power line carrier channel parameters is achieved through shared weights; the modeling results are used for the transmission of distribution network data.

[0010] As a preferred solution, the access characteristics include voltage, current, impedance characteristics and operating status of the power line carrier channel where the terminal is located; the topology change rules include topology change frequency and the number of power line branches.

[0011] As a preferred solution, the evaluation function of the power line carrier channel corresponding to the terminal is calculated according to the access characteristics and the topology change rule, specifically:

[0012] The evaluation function θ is calculated according to the following formula:

[0013]

[0014] Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the power line branch, and Z j is the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

[0015] As a preferred solution, based on the evaluation function and the empirical value of the evaluation function, the power line carrier channel is divided into several layers, specifically:

[0016] Calculating the empirical value of the evaluation function according to historical data, and then calculating the minimum and maximum values ​​of the empirical value; wherein the historical data includes the historical data of the access characteristics and the historical data of the topology change law;

[0017] Determine the threshold of each layer of power line carrier channel according to the minimum value and the maximum value;

[0018] According to the evaluation function and the threshold, and at preset intervals, the power line carrier channel is divided into several layers.

[0019] As a preferred solution, the training method of the shared neural network includes:

[0020] The basic neural network defining the k-th power line carrier channel includes an input layer, a hidden layer and an output layer, and the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training;

[0021] The prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function, the basic neural network is trained, and weights are shared during the training process to obtain the trained shared neural network.

[0022] Correspondingly, an embodiment of the present invention further provides a hierarchical modeling device for a power line carrier channel, comprising a collection module, a calculation module, a hierarchical module and a modeling module; wherein:

[0023] The collection module is used to collect access characteristics and topology change rules of various terminals;

[0024] The calculation module is used to calculate the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules;

[0025] The stratification module is used to divide the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function;

[0026] The modeling module is used to model each layer of power line carrier channel through a pre-trained shared neural network, and realize inter-layer sharing of power line carrier channel parameters through shared weights; wherein the modeling results are used for the transmission of distribution network data.

[0027] As a preferred solution, the access characteristics include voltage, current, impedance characteristics and operating status of the power line carrier channel where the terminal is located; the topology change rules include topology change frequency and the number of power line branches.

[0028] As a preferred solution, the calculation module calculates the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules, specifically:

[0029] The calculation module calculates the evaluation function θ according to the following formula:

[0030]

[0031] Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the power line branch, and Z j is the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

[0032] As a preferred solution, the stratification module divides the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function, specifically:

[0033] The hierarchical module calculates the empirical value of the evaluation function according to the historical data, and then calculates the minimum and maximum values ​​of the empirical value; wherein the historical data includes the historical data of the access characteristics and the historical data of the topology change law;

[0034] Determine the threshold of each layer of power line carrier channel according to the minimum value and the maximum value;

[0035] According to the evaluation function and the threshold, and at preset intervals, the power line carrier channel is divided into several layers.

[0036] As a preferred solution, the hierarchical modeling device further includes a training module, and the training module is used to:

[0037] The basic neural network defining the k-th power line carrier channel includes an input layer, a hidden layer and an output layer, and the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training;

[0038] The prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function, the basic neural network is trained, and weights are shared during the training process to obtain the trained shared neural network.

[0039] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0040] The embodiment of the present invention provides a hierarchical modeling method and device for a power line carrier channel, the hierarchical modeling method comprising: collecting access characteristics and topology change rules of multiple terminals; calculating the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules; dividing the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function; for each layer of the power line carrier channel, modeling is performed by a pre-trained shared neural network, and inter-layer sharing of power line carrier channel parameters is achieved by sharing weights; wherein the modeling result is used for the transmission of distribution network data. Compared with the prior art, the evaluation function is calculated by access characteristics and topology change rules, and then hierarchical and modeled based on the evaluation function, that is, the power line carrier channels with similar attributes are divided into the same layer, which can effectively improve the adaptability of the actual power line carrier communication with a high proportion of new energy access, and the constructed model has a large fitting space and high model accuracy, which can effectively support the real-time and reliable transmission of distribution network data.

[0041] Furthermore, modeling is performed based on differentiated inter-layer shared neural networks and used for transmission in distribution networks to achieve inter-layer sharing of power line carrier channel parameters at each layer. Compared with the existing technology, this can effectively accelerate the convergence speed of the model and improve the accuracy of the channel model. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 : A flow chart of an embodiment of the hierarchical modeling method of the power line carrier channel provided by the present invention.

[0043] Figure 2 : A structural diagram of an embodiment of a hierarchical modeling device for a power line carrier channel provided by the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Embodiment one:

[0046] Please refer to Figure 1 , Figure 1 A hierarchical modeling method for a power line carrier channel provided by an embodiment of the present invention includes steps S1 to S4, wherein:

[0047] Step S1, collecting access characteristics and topology change rules of various terminals.

[0048] In this embodiment, the access characteristics and topology change laws of high-proportion new energy power terminals are mainly collected. Among them, the high-proportion new energy access characteristics can be described by including but not limited to the voltage, power, load impedance characteristics, and operating state (including grid-connected operating state and off-grid operating state) of the power line carrier channel where the terminal is located. The topology change law includes but is not limited to the power line index and topology change law, etc.

[0049] Step S2: calculating an evaluation function of a power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules.

[0050] As a preferred implementation, specifically: the evaluation function θ is calculated according to the following formula:

[0051]

[0052] Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the power line branch, and Z j is the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

[0053] Among them, the operating state S ＝ {S1, S2}. S1 represents the off-grid operation index, and S2 represents the grid-connected operation index.

[0054] Step S3: dividing the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function.

[0055] In this embodiment, a large amount of empirical information / historical data on high-proportion new energy access characteristics such as voltage, current, load impedance characteristics, operating status, etc. of the power line carrier channel, as well as empirical information / historical data on topological change laws such as the number of power line branches and topological change laws can be collected.

[0056] Then, the empirical value of the evaluation function is calculated based on the historical data, and the minimum value θ of the empirical value is further calculated. min and the maximum value θ max .

[0057] Then, the threshold of each layer of the power line carrier channel is determined according to the minimum value and the maximum value. For the power line carrier channel of the kth layer (a total of K layers), its threshold θ k It is expressed as:

[0058]

[0059] Furthermore, the preset interval may preferably be (θ max -θ min ) / (K-2).

[0060] Further, according to the evaluation function and the threshold value, and at a preset interval, the power line carrier channel is divided into several layers, and the power line carrier channel with the evaluation function θ is divided into the Fth layer. k (θ)∈{k1≤k≤K} layer, which can be specifically expressed as:

[0061]

[0062] It can be understood that for power line carrier channels with low voltage and current, small impedance, off-grid operation, small number of branches, and slow topology change frequency, due to their similar properties, their channel models are relatively simple and will be divided into k smaller layers. Conversely, for power line carrier channels with large voltage and current, small impedance, grid-connected operation, many branches, and fast topology change frequency, their channel models are complex and will be divided into k larger layers.

[0063] Step S4, for each layer of power line carrier channel, a corresponding model is respectively performed through a pre-trained shared neural network, and inter-layer sharing of power line carrier channel parameters is achieved through shared weights; wherein the modeling result is used for the transmission of distribution network data.

[0064] In this embodiment, the training method of the shared neural network includes:

[0065] A basic neural network adapted to the k-th power line carrier channel is defined, which includes three layers of network: input layer, hidden layer and output layer. And the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training.

[0066] At the same time, the basic neural network adapted to the k-th power line carrier channel needs to be trained T times in total, and its input vector is x k (t), whose output vector is y k (t). And, define its output vector as the (model) parameter used to describe the power line carrier channel of this layer.

[0067] As a further preferred solution, the input vector consists of two parts. The first part is specifically a training data sample consisting of indicators such as the power line length, the number of power line branches, the branch length, the load impedance characteristics, the power line type, and the power line noise intensity of the k-th layer power line carrier channel; the second part is the power line carrier channel (model) parameters y except for the k-th layer. k' (t), and the corresponding parameter sharing weight χ k,k' (t).

[0068] Furthermore, the prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function of the k-th power line carrier channel, which can be expressed as:

[0069]

[0070] in, is the objective function, is the theoretical model parameter of the k-th power line carrier channel, κ is the mean square error weight, Y k (t) is the parameter vector of the k-th power line carrier channel model at the end of the t-th training, which can be expressed as Y k (t) = {y k (1),y k (2),...,y k (t)}. Among them, ||.||2 represents the second norm, that is, the Euclidean distance, represents the mean square error.

[0071] Considering the multipath coupling between the power line carrier channels at each layer, as a further preferred implementation, the model parameter y of the k-th power line carrier channel is k (t) is shared with other layers to improve model accuracy and accelerate convergence. When the distance between the power line carrier channels of each layer is farther, the coupling degree is lower, and accordingly, the weight is smaller; on the other hand, the larger the model prediction error and mean square error of the power line carrier channels of each layer, the smaller the weight is. Therefore, the model parameter sharing weight χ between the power line carrier channel of the kth layer and the power line carrier channel of the k'th layer is k,k' (t) is expressed as:

[0072] Among them, χ0 is the preset model parameter sharing weight amplitude.

[0073] Then, the basic neural network is repeatedly optimized iteratively based on the above method to train the model parameters y of each layer of the power line carrier channel k (t), and update the corresponding shared neural network weights W k (t), until the maximum number of training times T is reached, the trained shared neural network is obtained.

[0074] In this embodiment, the trained shared neural network can be used to predict the parameters of each layer of the power line carrier model, providing a basis for implementing reliable transmission of distribution network data.

[0075] Accordingly, refer to Figure 2 The embodiment of the present invention also provides a hierarchical modeling device for a power line carrier channel, comprising a collection module 101, a calculation module 102, a hierarchical module 103 and a modeling module 104; wherein,

[0076] The collection module 101 is used to collect access characteristics and topology change rules of various terminals;

[0077] The calculation module 102 is used to calculate the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules;

[0078] The stratification module 103 is used to divide the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function;

[0079] The modeling module 104 is used to model each layer of the power line carrier channel through a pre-trained shared neural network, and realize inter-layer sharing of power line carrier channel parameters through shared weights; wherein the modeling results are used for the transmission of distribution network data.

[0080] As a preferred implementation, the access characteristics include voltage, current, impedance characteristics and operating status of the power line carrier channel where the terminal is located; the topology change rules include topology change frequency and the number of power line branches.

[0081] As a preferred implementation, the calculation module 102 calculates the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and the topology change rule, specifically:

[0082] The calculation module 102 calculates the evaluation function θ according to the following formula:

[0083]

[0084] Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the power line branch, and Z jis the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

[0085] As a preferred implementation, the stratification module 103 divides the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function, specifically:

[0086] The hierarchical module 103 calculates the empirical value of the evaluation function according to the historical data, and then calculates the minimum and maximum values ​​of the empirical value; wherein the historical data includes the historical data of the access characteristics and the historical data of the topology change law;

[0087] Determine the threshold of each layer of power line carrier channel according to the minimum value and the maximum value;

[0088] According to the evaluation function and the threshold, and at preset intervals, the power line carrier channel is divided into several layers.

[0089] As a preferred solution, the hierarchical modeling device further includes a training module, and the training module is used to:

[0090] The basic neural network defining the k-th power line carrier channel includes an input layer, a hidden layer and an output layer, and the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training;

[0091] The prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function, the basic neural network is trained, and weights are shared during the training process to obtain the trained shared neural network.

[0092] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0093] The embodiment of the present invention provides a hierarchical modeling method and device for a power line carrier channel, the hierarchical modeling method comprising: collecting access characteristics and topology change rules of multiple terminals; calculating the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules; dividing the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function; for each layer of the power line carrier channel, modeling is performed by a pre-trained shared neural network, and inter-layer sharing of power line carrier channel parameters is achieved by sharing weights; wherein the modeling result is used for the transmission of distribution network data. Compared with the prior art, the evaluation function is calculated by access characteristics and topology change rules, and then hierarchical and modeled based on the evaluation function, that is, the power line carrier channels with similar attributes are divided into the same layer, which can effectively improve the adaptability of the actual power line carrier communication with a high proportion of new energy access, and the constructed model has a large fitting space and high model accuracy, which can effectively support the real-time and reliable transmission of distribution network data.

[0094] Furthermore, modeling is performed based on differentiated inter-layer shared neural networks and used for transmission in distribution networks to achieve inter-layer sharing of power line carrier channel parameters at each layer. Compared with the existing technology, this can effectively accelerate the convergence speed of the model and improve the accuracy of the channel model.

[0095] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A hierarchical modeling method for a power line carrier channel, characterized in that: include: Collect access characteristics and topology change rules of various terminals; Calculating an evaluation function of a power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules; Based on the evaluation function and the empirical value of the evaluation function, dividing the power line carrier channel into several layers; For each layer of power line carrier channel, a pre-trained shared neural network is used to model it, and the power line carrier channel parameters are shared between layers through shared weights; the modeling results are used for the transmission of distribution network data; The access characteristics include the voltage, current, impedance characteristics and operating status of the power line carrier channel where the terminal is located; the topology change rules include the topology change frequency and the number of power line branches; The evaluation function of the power line carrier channel corresponding to the terminal is calculated according to the access characteristics and the topology change rule, specifically: The evaluation function θ is calculated according to the following formula: Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the number of power line branches, and Z j is the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

2. A hierarchical modeling method for a power line carrier channel as claimed in claim 1, characterized in that: The power line carrier channel is divided into several layers based on the evaluation function and the empirical value of the evaluation function, specifically: Calculating the empirical value of the evaluation function according to historical data, and then calculating the minimum and maximum values ​​of the empirical value; wherein the historical data includes the historical data of the access characteristics and the historical data of the topology change law; Determine the threshold of each layer of power line carrier channel according to the minimum value and the maximum value; According to the evaluation function and the threshold, and at preset intervals, the power line carrier channel is divided into several layers.

3. A hierarchical modeling method for a power line carrier channel as claimed in claim 1 or 2, characterized in that: The training method of the shared neural network includes: The basic neural network defining the k-th power line carrier channel includes an input layer, a hidden layer and an output layer, and the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training; The prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function, the basic neural network is trained, and weights are shared during the training process to obtain the trained shared neural network.

4. A hierarchical modeling device for a power line carrier channel, characterized in that: It includes acquisition module, calculation module, stratification module and modeling module; among them, The collection module is used to collect access characteristics and topology change rules of various terminals; The calculation module is used to calculate the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and topology change rules; The stratification module is used to divide the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function; The modeling module is used to model each layer of the power line carrier channel through a pre-trained shared neural network, and realize the inter-layer sharing of the power line carrier channel parameters through shared weights; wherein the modeling result is used for the transmission of distribution network data; The access characteristics include the voltage, current, impedance characteristics and operating status of the power line carrier channel where the terminal is located; the topology change rules include the topology change frequency and the number of power line branches; The calculation module calculates the evaluation function of the power line carrier channel corresponding to the terminal according to the access characteristics and the topology change rule, specifically: The calculation module calculates the evaluation function θ according to the following formula: Where V is the voltage of the power line carrier channel where the terminal is located, I is the current of the power line carrier channel where the terminal is located, L is the number of power line branches, and Z j is the impedance characteristic of the jth branch, S is the operating state, the operating state includes off-grid operating indicators and grid-connected operating indicators, f is the topology change frequency, α is the weight of the voltage of the power line carrier channel where the terminal is located, β is the weight of the current of the power line carrier channel where the terminal is located, λ is the weight of the impedance characteristic, μ is the weight of the operating state, γ is the weight of the number of power line branches, and ν is the weight of the topology change frequency.

5. A hierarchical modeling device for a power line carrier channel as claimed in claim 4, characterized in that: The stratification module divides the power line carrier channel into several layers based on the evaluation function and the empirical value of the evaluation function, specifically: The hierarchical module calculates the empirical value of the evaluation function according to the historical data, and then calculates the minimum and maximum values ​​of the empirical value; wherein the historical data includes the historical data of the access characteristics and the historical data of the topology change law; Determine the threshold of each layer of power line carrier channel according to the minimum value and the maximum value; According to the evaluation function and the threshold, and at preset intervals, the power line carrier channel is divided into several layers.

6. A hierarchical modeling device for a power line carrier channel as claimed in claim 4 or 5, characterized in that: The hierarchical modeling device further comprises a training module, wherein the training module is used for: The basic neural network defining the k-th power line carrier channel includes an input layer, a hidden layer and an output layer, and the weight matrix of the basic neural network is in, is the weight of the input layer, is the weight of the hidden layer, is the weight of the output layer, and t refers to the tth training; The prediction error and mean square error of the power line carrier channel parameters are minimized as the objective function, the basic neural network is trained, and weights are shared during the training process to obtain the trained shared neural network.

Images