Power line communication cross-transformer relay method and system

Abstract

The invention discloses a power line communication cross-transformer relay method and system. Relates to a signal processing method and device. The method aims to solve the problems of signal attenuation, distortion and interference after cross-transformer transmission of communication signals, and comprises the following steps: detecting to obtain a real-time state of a first channel at a moment t; taking the real-time state of the first channel at the t moment as input, and predicting through a state prediction model to obtain an estimated state of the first channel at the t + 1 moment; according to the estimation state of the first channel at the t + 1 moment, selecting a corresponding signal processing strategy at the t + 1 moment to process the communication signal; detecting to obtain the real-time state of the second channel at the t + 1 moment; the real-time state of the second channel at the t + 1 moment serves as input, and the estimation state of the second channel at the t + 2 moment is obtained through prediction of the state prediction model of the second channel; and according to the estimation state of the second channel at the t + 2 moment, selecting a corresponding signal processing strategy at the t + 2 moment to process the communication signal.

Description

Technical Field

[0001] The invention relates to a signal processing method and a device. Background Art

[0002] The power line network is not as stable and controllable as the existing communication environment, and it changes dynamically as the load of the power grid changes.

[0003] There are many types of loads connected in the power line network, especially nonlinear loads (such as inverters and rectifiers), which will generate harmonics and high-frequency interference. This interference will cause the frequency response of the channel characteristics to become complex and dynamically change. In power line communication across transformers, irregular attenuation and distortion occur to high-frequency communication signals, especially when the frequency is high and the transmission distance is long. This randomness and time-varying nature leads to high uncertainty in the power line channel characteristics. This causes signal attenuation, distortion and interference problems to occur in communication signals after transmission across transformers. Summary of the invention

[0004] The purpose of the present invention is to overcome the problems of signal attenuation, distortion and interference after communication signals are transmitted across transformers, and to provide a power line communication across transformer relay method and system.

[0005] The present invention provides a power line communication cross-transformer relay method, the method is as follows:

[0006] Detect and obtain the real-time status of the first channel at time t;

[0007] The first channel is a power line communication channel located on one side of the transformer;

[0008] The states include impedance characteristics, channel gain, and noise intensity;

[0009] The real-time state of the first channel at time t is used as input, and the estimated state of the first channel at time t+1 is predicted by the state prediction model of the first channel;

[0010] The channel state prediction model is a combination of LSTM long short-term memory network and Kalman filter;

[0011] The input end of LSTM is used as the input end of the channel state prediction model, the output end of LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model;

[0012] According to the estimated state of the first channel at time t+1, a corresponding signal processing strategy is selected at time t+1 to process the communication signal of the first channel;

[0013] Signal processing strategies include adjusting gain and frequency hopping;

[0014] Detect and obtain the real-time status of the second channel at time t+1;

[0015] The second channel is a power line communication channel located on the other side of the transformer; the first channel and the second channel are coupled through the transformer; the communication signal is transmitted through the first channel, the transformer and the second channel in sequence;

[0016] The real-time state of the second channel at time t+1 is used as input, and the estimated state of the second channel at time t+2 is predicted by the state prediction model of the second channel;

[0017] According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel.

[0018] The present invention also provides a power line communication inter-transformer relay system, comprising:

[0019] The first repeater is used to sample the communication signal of the first channel at time t to obtain the real-time state of the first channel at time t;

[0020] The states include impedance characteristics, channel gain, and noise intensity;

[0021] The real-time state of the first channel at time t is used as input, and the estimated state of the first channel at time t+1 is predicted by the state prediction model of the first channel;

[0022] The channel state prediction model is a combination of LSTM long short-term memory network and Kalman filter;

[0023] The input end of LSTM is used as the input end of the channel state prediction model, the output end of LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model;

[0024] The second repeater is used to select a corresponding signal processing strategy at time t+1 to process the communication signal of the first channel according to the estimated state of the first channel at time t+1;

[0025] Signal processing strategies include adjusting gain and frequency hopping;

[0026] At time t+1, the communication signal of the second channel is sampled to obtain the real-time state of the second channel at time t+1;

[0027] The first channel and the second channel are power line communication channels located at two sides of the transformer respectively and coupled by the transformer;

[0028] The real-time state of the second channel at time t+1 is used as input, and the estimated state of the second channel at time t+2 is predicted by the state prediction model of the second channel;

[0029] According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel.

[0030] The beneficial effects of the present invention are:

[0031] The power line communication cross-transformer relay method and system of the present invention can take into account both nonlinearity and time-varying properties.

[0032] LSTM processes nonlinear time series data, while the Kalman filter can adjust the model in real time and adapt to the rapid changes in the channel. It can capture the long-term trend of the power line communication channel and make accurate real-time corrections in the short term. It can make full use of historical data and real-time observation data to improve prediction accuracy and reduce errors. It is particularly suitable for complex, nonlinear and time-varying communication environments such as power line channels. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart of the power line communication cross-transformer relay method of the present invention;

[0034] Figure 2 It is a structural schematic diagram of the power line communication inter-transformer relay system of the present invention. DETAILED DESCRIPTION

[0035] 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.

[0036] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0037] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention. Specific implementation method 1

[0039] The power line communication cross-transformer relay method of this embodiment is as follows:

[0040] Detect and obtain the real-time status of the first channel at time t;

[0041] The first channel is a power line communication channel located on one side of the transformer;

[0042] The states include impedance characteristics, channel gain, and noise intensity;

[0043] The real-time state of the first channel at time t is used as input, and the estimated state of the first channel at time t+1 is predicted by the state prediction model of the first channel;

[0044] The channel state prediction model is a combination of LSTM long short-term memory network and Kalman filter;

[0045] The input end of LSTM is used as the input end of the channel state prediction model, the output end of LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model;

[0046] According to the estimated state of the first channel at time t+1, a corresponding signal processing strategy is selected at time t+1 to process the communication signal of the first channel;

[0047] Signal processing strategies include adjusting gain and frequency hopping;

[0048] Detect and obtain the real-time status of the second channel at time t+1;

[0049] The second channel is a power line communication channel located on the other side of the transformer; the first channel and the second channel are coupled through the transformer; the communication signal is transmitted through the first channel, the transformer and the second channel in sequence;

[0050] The real-time state of the second channel at time t+1 is used as input, and the estimated state of the second channel at time t+2 is predicted by the state prediction model of the second channel;

[0051] According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel.

[0052] Specifically, in this embodiment, the first channel and the second channel are power line communication channels located on both sides of the transformer and coupled by the transformer, wherein the communication signal passes through the first channel and the second channel in sequence, wherein the communication signal can be an uplink signal or a downlink signal.

[0053] The channel states of the first channel and the second channel can be characterized by the following parameters:

[0054] Impedance characteristic Z(t): The impedance of the power line changes with time. The transformer will cause discontinuous changes in impedance.

[0055] Channel gain G(t): indicates the attenuation of the communication signal during the transmission process.

[0056] Noise intensity N(t): The noise level in the power line, including background noise and impulse noise.

[0057] Signal bit error rate BER(t): reflects the current communication quality. Fluctuations in the bit error rate can indicate changes in channel status.

[0058] Signal power P(t): The change in signal power between the sending and receiving ends, used to measure signal strength and path loss.

[0059] Frequency selective attenuation F(t, f): The signal attenuation at different frequencies can reflect the changes in multipath effects.

[0060] Select and collect all or some of the above parameters, and use the input feature vector X of the LSTM in the channel state prediction model t , taking the impedance characteristics, channel gain and noise intensity as an example:

[0061] X t =[Z(t), G(t), N(t)]

[0062] The LSTM model is used to capture the long-term timing characteristics of the power line communication channel and output short-term prediction values ​​as the initial estimation input of the Kalman filter.

[0063] Input feature vector {X t-n , X t-n+1 , ..., X t}, where n is the length of the input sequence, representing the observed data at the past n moments. The channel state predicted by LSTM is The following forecast values ​​are included:

[0064] Predicted Impedance

[0065] Predicted Channel Gain

[0066] Predicted noise intensity

[0067] The output vector can be expressed as:

[0068]

[0069] The Kalman filter is used to correct the prediction results of LSTM in real time and deal with the rapid changes and mutations of the power line communication channel.

[0070] The state equation of the Kalman filter is as follows:

[0071] X t+1 =Ax t +w t

[0072] in:

[0073] x t=[Z(t), G(t), N(t)] T .

[0074] A is the state transfer matrix, which is initialized by the prediction value of LSTM.

[0075] is the process noise.

[0076] The Kalman filter observation equation is as follows:

[0077] z t =Hx t +V t

[0078] in:

[0079] z t =[Z obs (t),G obs (t),N obs (t),BER obs (t)] T is the real-time observation data, that is, the real-time status of the first / second channel at time t.

[0080] H is the observation matrix.

[0081] is the observation noise.

[0082] The Kalman filter update steps are as follows:

[0083] Prediction Steps:

[0084]

[0085] Update steps:

[0086]

[0087] Among them, K t+1 is the Kalman gain; P t is the state covariance matrix.

[0088] The predicted value of the above LSTM As the initial state prediction input of the Kalman filter

[0089] The Kalman filter is based on the real-time observation data z t Correct the prediction results of LSTM.

[0090] Final Output The corrected channel state estimation result, ie, the estimated state of the first / second channel at time t+1, is sent to the first repeater 1 or the second repeater 2 to select a corresponding signal processing strategy. Specific implementation method 2

[0092] This embodiment is a further description of the first embodiment, and this embodiment also includes:

[0093] After calculating the real-time state of the second channel at time t+1, the real-time state of the second channel at time t+1 is fed back to the state prediction model of the first channel;

[0094] Add the real-time state of the second channel at time t+1 to the input sequence of the LSTM at time t+1 in the state prediction model of the first channel;

[0095] And, the covariance matrix of the Kalman filter in the state prediction model of the first channel is adjusted according to the real-time state of the second channel at time t+1.

[0096] The other technical features of this embodiment are exactly the same as those of the first embodiment.

[0097] Specifically, this embodiment adopts a scheme in which repeaters are arranged at both ends of the transformer, that is, including a first repeater 1 located on the first channel and a second repeater 2 located on the second channel. If the communication signal is an uplink signal, the first repeater 1 is an uplink repeater and the second repeater 2 is a downlink repeater; if the communication signal is a downlink signal, the first repeater 1 is a downlink repeater and the second repeater 2 is an uplink repeater.

[0098] The channel state prediction model combining LSTM and Kalman filter can be applied to the repeaters at both ends, where:

[0099] The first repeater 1 is arranged at one side of the transformer and is mainly responsible for receiving signals from the first channel and sending them to the second repeater 2 after processing.

[0100] The first repeater 1 adjusts the communication signal processing strategy according to the prediction result, for example:

[0101] Gain Adjustment: If higher channel impedance or lower signal-to-noise ratio is predicted, increase the gain G(f).

[0102] Modulation mode switching: When the channel status is poor, switch to a more robust modulation mode (such as BPSK).

[0103] Frequency band selection: Select the best frequency band for transmission after frequency division and avoid high impedance frequency band.

[0104] The predicted and adjusted communication signal is sent from the first repeater 1 across the transformer to the second repeater 2 on the other side of the transformer.

[0105] The second repeater 2 is arranged on the other side of the transformer and is responsible for receiving the signal transmitted by the first repeater 1 and continuing to transmit it on the second channel. The second repeater 2 not only needs to monitor the channel status of the second channel, but also needs to evaluate the quality of the received communication signal. The second repeater 2 can use the actual measured channel status f t+1 The information is fed back to the first repeater 1 to update the channel state prediction model of the first repeater 1, thereby further improving the prediction accuracy and forming a closed-loop control.

[0106] Assume that the observation information of downlink feedback is f t+1 :

[0107] Adjust the LSTM input sequence: X LSTM ={[x t-n ,…,x t ,f t+1 ]}

[0108] The Kalman filter uses an adaptive Kalman filter, which can dynamically adjust the covariance matrix parameters and maintain good performance even in high noise environments.

[0109] Adjust the Kalman filter covariance matrix:

[0110] Among them, α is the adaptive adjustment coefficient, which controls the adjustment range; var(·) represents the variance.

[0111] In short, using the feedback information, the first repeater 1 can update the prediction model and improve the accuracy of the overall channel state prediction. The first repeater 1 and the second repeater 2 can dynamically adjust parameters such as frequency band selection, gain, modulation mode, etc. to adapt to changes in the power line environment and improve system communication performance. Specific implementation method three

[0113] This embodiment is a further explanation of the second embodiment. In this embodiment, the LSTM obtains the predicted state at time t+1 based on the real-time state prediction at time t, and uses the predicted state at time t+1 as the prior estimated state of the Kalman filter at time t+1;

[0114] The Kalman filter corrects the prior estimated state at time t+1 through the real-time state at time t to obtain the estimated state at time t+1.

[0115] The other technical features of this embodiment are exactly the same as those of the second embodiment. Specific implementation method four

[0117] This embodiment is a further description of the first, second or third embodiment, and this embodiment also includes:

[0118] Monitoring the dynamic indicators of the first channel and the second channel according to a set monitoring period;

[0119] Dynamic indicators include the standard deviation or rate of change of impedance, the standard deviation or rate of change of gain, the mean or variance of noise power, and the difference in bit error rate;

[0120] When the dynamic index meets the preset high dynamic range, the sampling period for detecting the real-time status is set to 1s to 30s;

[0121] When the dynamic index meets the preset medium dynamic range, the sampling period for detecting the real-time status is set to 10 minutes to 30 minutes;

[0122] When the dynamic index meets the preset low dynamic range, the sampling period for detecting the real-time status is set to 1h to 24h.

[0123] The other technical features of this embodiment are exactly the same as those of embodiments one, two or three.

[0124] Specifically, it is necessary to pre-set several common power line communication scenarios and formulate a suitable sampling interval strategy according to the channel change characteristics and differences in the communication environment.

[0125] Scenario A: Highly dynamic environment, characterized by frequent changes in power load and the presence of high-frequency interference sources (such as industrial equipment and power tools). This is often seen in industrial workshops, electric vehicle charging stations, and household high-power electrical appliances. Optimum sampling interval: 1-10 seconds.

[0126] Scenario B: Medium dynamic environment, characterized by a certain periodicity in load changes, but no significant mutations. Mainly affected by user electricity consumption behavior (such as residential areas). Mostly seen in residential areas, office buildings, shopping malls, etc.; optimal sampling interval: 10-15 minutes.

[0127] Scenario C: Low dynamic environment, characterized by stable load changes and obvious daily or weekly cycle trends. It is often seen in substation load monitoring and long-term stable industrial equipment operation environments. Optimum sampling interval: 1-24 hours.

[0128] To identify the above scenarios, the following key indicators can be selected:

[0129] Impedance change rate ΔZ: used to monitor load fluctuations and calculate the standard deviation or change rate of impedance:

[0130]

[0131] Signal gain fluctuation ΔG: used to detect signal attenuation changes and calculate the standard deviation or rate of change of gain:

[0132]

[0133] Noise power N(t): used to evaluate the stability of the noise environment and calculate the mean and variance of the noise power:

[0134]

[0135] Bit error rate change ΔBER: used to reflect the fluctuation of communication quality and calculate the change trend of bit error rate:

[0136] ΔBER=BER t -BER t-1

[0137] It can also include periodicity detection: using autocorrelation analysis to detect the periodicity of load and channel status, and determine whether there is an obvious daily or weekly cycle in the signal.

[0138] According to the changes in the monitored dynamic indicators, rule-based classification algorithms or machine learning models (such as decision trees or random forests) can be used to identify the current scenario and dynamically adjust the sampling interval:

[0139] Rule 1: If ΔZ, ΔG, and ΔBER are all high and the noise variance is large, it is considered a high-dynamic environment and a sampling interval of seconds (1-10 seconds) is selected.

[0140] Rule 2: If ΔZ and ΔG are low, the noise variance is moderate, and a significant daily cycle variation of the load is detected, it is considered to be a medium dynamic environment and a minute sampling interval (10-15 minutes) is selected.

[0141] Rule 3: If the indicator changes little and the autocorrelation analysis shows an obvious periodic trend (such as daily or weekly cycles), it is considered to be a low-dynamic environment and an hourly or daily sampling interval (1-24 hours) is selected.

[0142] At the same time, the sampling interval can also be adjusted dynamically according to the following process:

[0143] Initialization phase: When the system starts, sampling begins using the default sampling interval (such as 10 minutes) to collect initial data.

[0144] Monitoring phase: At regular intervals (such as every 30 minutes), the module calculates monitoring indicators (ΔZ, ΔG, ΔBER, Var(N), etc.).

[0145] Classification phase: Based on the calculated dynamic indicator changes, the current scenario is identified using decision rules or classification models.

[0146] Adjustment phase: According to the identified scene type, the sampling interval Δt and the LSTM input sequence length n are automatically adjusted:

[0147] If a highly dynamic environment is detected, shorten the sampling interval (1-10 seconds) and reduce the input sequence length (e.g., 10-50).

[0148] If a medium dynamic environment is detected, adjust the sampling interval to minutes (10-15 minutes) and the input sequence length to 50-100.

[0149] If a low dynamic environment is detected, the sampling interval is extended to the hourly or daily level (1-24 hours), and the input sequence length is 200-365.

[0150] Feedback mechanism: If the indicator change is found to increase significantly under the adjusted new interval, it will fall back to the previous sampling strategy to avoid system instability caused by over-adjustment.

[0151] Example: Assuming the system initially uses a sampling interval of 10 minutes:

[0152] When ΔZ>0.1, ΔG>0.1, bit error rate rise ΔBER>0.01 and noise power variance is high, the system identifies it as a high dynamic environment and adjusts the sampling interval to 5 seconds.

[0153] When it is detected that ΔZ<0.01, ΔG<0.01, the bit error rate is stable and there is an obvious daily cycle, the system identifies it as a low dynamic environment and adjusts the sampling interval to 1 hour. Specific implementation method five

[0155] This embodiment is a further description of the fourth embodiment. In this embodiment, the following is also included:

[0156] After sampling the communication signal of the first channel at time t, frequency division is performed on the communication signal of the first channel according to a preset frequency band;

[0157] Calculate and obtain the real-time status of the first channel corresponding to each frequency band at time t;

[0158] Score and sort each frequency band according to the real-time status of the first channel, and select the best N frequency bands, where N≥1;

[0159] At time t+1, the communication signal of the first channel is switched from the current frequency band to N frequency bands;

[0160] After sampling the communication signal of the second channel at time t+1, the communication signal of the second channel is divided into different frequencies according to a preset frequency band;

[0161] Calculate and obtain the real-time status of the second channel corresponding to each frequency band at time t+1;

[0162] Score and sort each frequency band according to the real-time status of the second channel, and select the best K frequency bands, K ≥ 1;

[0163] At time t+2, the communication signal of the second channel is switched from the current frequency band to K frequency bands.

[0164] The other technical features of this embodiment are exactly the same as those of embodiment five.

[0165] Specifically, the signal can be divided into three frequency bands: low frequency band f low , mid-band f mid , high frequency band f high .

[0166] The state prediction model is used to predict the channel state of each frequency band.

[0167] According to the prediction results, the signals in different frequency bands can be frequency-hopped. Frequency hopping means that if the channel prediction of a certain frequency band is not ideal (low signal-to-noise ratio, high impedance), you can choose to hop to other frequency bands to avoid interference.

[0168] However, in addition to frequency hopping processing, the following processing methods can also be included:

[0169] Gain adjustment, that is, for the frequency band with low signal-to-noise ratio (i.e. Where S th is the signal-to-noise ratio threshold), the gain can be increased.

[0170] For high signal-to-noise ratio frequency band Keep the gain as is or reduce it slightly to save power.

[0171] Dynamic frequency band selection, that is, based on the predicted impedance, select the frequency band with lower impedance for priority transmission, avoid the frequency band with high impedance and large interference, and improve communication efficiency.

[0172] Modulation mode adjustment, that is, based on the power prediction of the channel, if the predicted power is low, switching to a more interference-resistant modulation mode (for example, switching from QAM to BPSK). Specific implementation method 6

[0174] The power line communication inter-transformer relay system of this embodiment includes:

[0175] The first repeater 1 is used to sample the communication signal of the first channel at time t to obtain the real-time state of the first channel at time t;

[0176] The states include impedance characteristics, channel gain, and noise intensity;

[0177] The real-time state of the first channel at time t is used as input, and the estimated state of the first channel at time t+1 is predicted by the state prediction model of the first channel;

[0178] The channel state prediction model is a combination of LSTM long short-term memory network and Kalman filter;

[0179] The input end of LSTM is used as the input end of the channel state prediction model, the output end of LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model;

[0180] The second repeater 2 is used to select a corresponding signal processing strategy at time t+1 to process the communication signal of the first channel according to the estimated state of the first channel at time t+1;

[0181] Signal processing strategies include adjusting gain and frequency hopping;

[0182] At time t+1, the communication signal of the second channel is sampled to obtain the real-time state of the second channel at time t+1;

[0183] The first channel and the second channel are power line communication channels located at two sides of the transformer respectively and coupled by the transformer;

[0184] The real-time state of the second channel at time t+1 is used as input, and the estimated state of the second channel at time t+2 is predicted by the state prediction model of the second channel;

[0185] According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel. Specific implementation method seven

[0187] This embodiment is a further description of the sixth embodiment. In this embodiment, it is characterized in that:

[0188] The second repeater 2 is further configured to calculate the real-time state of the second channel at time t+1 and then feed back the real-time state of the second channel at time t+1 to the first repeater 1;

[0189] The first repeater 1 is further used to add the real-time state of the second channel at time t+1 to the input sequence at time t+1 of the LSTM in the state prediction model of the first channel;

[0190] And, the covariance matrix of the Kalman filter in the state prediction model of the first channel is adjusted according to the real-time state of the second channel at time t+1.

[0191] The other technical features of this embodiment are exactly the same as those of embodiment six. Specific implementation method eight

[0193] This embodiment is a further explanation of the seventh embodiment. In this embodiment, in the first repeater 1 and the second repeater 2, the LSTM obtains the predicted state at time t+1 according to the real-time state prediction at time t, and uses the predicted state at time t+1 as the prior estimated state of the Kalman filter at time t+1;

[0194] The Kalman filter corrects the prior estimated state at time t+1 through the real-time state at time t to obtain the estimated state at time t+1.

[0195] The other technical features of this embodiment are exactly the same as those of embodiment seven. Specific implementation method nine

[0197] This embodiment is a further description of the sixth, seventh or eighth embodiment. In this embodiment, a sampling scene dynamic monitor 3 is also included;

[0198] The sampling scene dynamic monitor 3 is used to monitor the dynamic indicators of the first channel and the second channel according to a set monitoring period, and send the dynamic indicators to the first repeater 1 and the second repeater 2;

[0199] Dynamic indicators include the standard deviation or rate of change of impedance, the standard deviation or rate of change of gain, the mean or variance of noise power, and the difference in bit error rate;

[0200] When the dynamic index meets the preset high dynamic range, the first repeater 1 and the second repeater 2 are further used to set the sampling period for detecting the real-time state to 1s to 30s;

[0201] When the dynamic index meets the preset medium dynamic range, the first repeater 1 and the second repeater 2 are also used to set the sampling period for detecting the real-time state to 10min to 30min;

[0202] When the dynamic index meets the preset low dynamic range, the first repeater 1 and the second repeater 2 are further used to set the sampling period for detecting the real-time state to 1 hour to 24 hours.

[0203] The other technical features of this embodiment are exactly the same as those of embodiments six, seven or eight. Specific implementation method 10

[0205] This embodiment is a further description of the ninth embodiment. In this embodiment, the first repeater 1 further includes a first frequency divider 4;

[0206] A first frequency divider 4 is used to divide the communication signal of the first channel according to a preset frequency band after sampling the communication signal of the first channel at time t;

[0207] The first repeater 1 is further used to calculate and obtain the real-time status of the first channel corresponding to each frequency band at time t;

[0208] Score and sort each frequency band according to the real-time status of the first channel, and select the best N frequency bands, where N≥1;

[0209] At time t+1, the communication signal of the first channel is switched from the current frequency band to N frequency bands;

[0210] The second repeater 2 also includes a second frequency divider 5;

[0211] The second frequency divider 5 is used to divide the communication signal of the second channel according to a preset frequency band after sampling the communication signal of the second channel at time t+1;

[0212] The second repeater 2 is further used to calculate and obtain the real-time status of the second channel corresponding to each frequency band at time t+1;

[0213] Score and sort each frequency band according to the real-time status of the second channel, and select the best K frequency bands, K ≥ 1;

[0214] At time t+2, the communication signal of the second channel is switched from the current frequency band to K frequency bands.

[0215] The other technical features of this embodiment are exactly the same as those of embodiment nine.

[0216] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the features of the various dependent claims and herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in conjunction with the individual embodiments may be used in other embodiments.

Claims

1. A power line communication cross-transformer relay method, characterized in that: Here’s how: Detect and obtain the real-time status of the first channel at time t; The first channel is a power line communication channel located on one side of the transformer; The states include impedance characteristics, channel gain, and noise intensity; Taking the real-time state of the first channel at time t as input, and predicting the estimated state of the first channel at time t+1 through the state prediction model of the first channel; The channel state prediction model is a combination model of LSTM long short-term memory network and Kalman filter; The input end of the LSTM is used as the input end of the channel state prediction model, the output end of the LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model; According to the estimated state of the first channel at time t+1, a corresponding signal processing strategy is selected at time t+1 to process the communication signal of the first channel; The signal processing strategy includes adjusting gain and frequency hopping; Detect and obtain the real-time status of the second channel at time t+1; The second channel is a power line communication channel located on the other side of the transformer; the first channel and the second channel are coupled through the transformer; the communication signal is transmitted through the first channel, the transformer and the second channel in sequence; Taking the real-time state of the second channel at time t+1 as input, and predicting the estimated state of the second channel at time t+2 through the state prediction model of the second channel; According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel.

2. The power line communication cross-transformer relay method according to claim 1, characterized in that: Also includes: After calculating the real-time state of the second channel at time t+1, the real-time state of the second channel at time t+1 is fed back to the state prediction model of the first channel; Adding the real-time state of the second channel at time t+1 to the input sequence of the LSTM at time t+1 in the state prediction model of the first channel; And, the covariance matrix of the Kalman filter in the state prediction model of the first channel is adjusted according to the real-time state of the second channel at time t+1.

3. The power line communication cross-transformer relay method according to claim 2, characterized in that: LSTM predicts the predicted state at time t+1 based on the real-time state at time t, and uses the predicted state at time t+1 as the prior estimated state of the Kalman filter at time t+1; The Kalman filter corrects the prior estimated state at time t+1 through the real-time state at time t to obtain the estimated state at time t+1.

4. The power line communication cross-transformer relay method according to claim 1, 2 or 3, characterized in that: Also includes: Monitoring the dynamic indicators of the first channel and the second channel according to a set monitoring period; The dynamic indicators include the standard deviation or change rate of impedance, the standard deviation or change rate of gain, the mean or variance of noise power, and the difference in bit error rate; When the dynamic index meets the preset high dynamic range, the sampling period for detecting the real-time state is set to 1s to 30s; When the dynamic index meets the preset medium dynamic range, the sampling period for detecting the real-time status is set to 10 minutes to 30 minutes; When the dynamic index meets the preset low dynamic range, the sampling period for detecting the real-time status is set to 1 hour to 24 hours.

5. The power line communication inter-transformer relay method according to claim 4, characterized in that: Also includes: After sampling the communication signal of the first channel at time t, frequency division is performed on the communication signal of the first channel according to a preset frequency band; Calculate and obtain the real-time status of the first channel corresponding to each frequency band at time t; Scoring and sorting each frequency band according to the real-time status of the first channel, and selecting the best N frequency bands, where N≥1; At time t+1, switching the communication signal of the first channel from the current frequency band to the N frequency bands; After sampling the communication signal of the second channel at time t+1, the communication signal of the second channel is divided into different frequencies according to a preset frequency band; Calculate and obtain the real-time status of the second channel corresponding to each frequency band at time t+1; Scoring and sorting each frequency band according to the real-time status of the second channel, and selecting the best K frequency bands, K≥1; At time t+2, the communication signal of the second channel is switched from the current frequency band to the K frequency bands.

6. A power line communication inter-transformer relay system, characterized in that: include: A first repeater (1) is used to sample the communication signal of the first channel at time t to obtain the real-time state of the first channel at time t; The states include impedance characteristics, channel gain, and noise intensity; Taking the real-time state of the first channel at time t as input, and predicting the estimated state of the first channel at time t+1 through the state prediction model of the first channel; The channel state prediction model is a combination model of LSTM long short-term memory network and Kalman filter; The input end of the LSTM is used as the input end of the channel state prediction model, the output end of the LSTM is used as the prior estimation information input end of the Kalman filter, and the correction information output end of the Kalman filter is used as the output end of the channel state prediction model; The second repeater (2) is used to select a corresponding signal processing strategy at time t+1 to process the communication signal of the first channel according to the estimated state of the first channel at time t+1; The signal processing strategy includes adjusting gain and frequency hopping; At time t+1, the communication signal of the second channel is sampled to obtain the real-time state of the second channel at time t+1; The first channel and the second channel are power line communication channels located at two sides of the transformer respectively and coupled by the transformer; Taking the real-time state of the second channel at time t+1 as input, and predicting the estimated state of the second channel at time t+2 through the state prediction model of the second channel; According to the estimated state of the second channel at time t+2, a corresponding signal processing strategy is selected at time t+2 to process the communication signal of the second channel.

7. The power line communication trans-transformer relay method according to claim 6, characterized in that: The second repeater (2) is further configured to, after calculating the real-time state of the second channel at time t+1, feed back the real-time state of the second channel at time t+1 to the first repeater (1); The first repeater (1) is further used to add the real-time state of the second channel at time t+1 to the input sequence of the LSTM at time t+1 in the state prediction model of the first channel; And, the covariance matrix of the Kalman filter in the state prediction model of the first channel is adjusted according to the real-time state of the second channel at time t+1.

8. The power line communication trans-transformer relay method according to claim 7, characterized in that: In the first repeater (1) and the second repeater (2), the LSTM predicts the predicted state at time t+1 based on the real-time state at time t, and uses the predicted state at time t+1 as the prior estimated state of the Kalman filter at time t+1; The Kalman filter corrects the prior estimated state at time t+1 through the real-time state at time t to obtain the estimated state at time t+1.

9. The power line communication trans-transformer relay method according to claim 6, 7 or 8, characterized in that: Also included is a sampling scene dynamics monitor (3); The sampling scene dynamic monitor (3) is used to monitor the dynamic indicators of the first channel and the second channel according to a set monitoring period, and send the dynamic indicators to the first repeater (1) and the second repeater (2); The dynamic indicators include the standard deviation or change rate of impedance, the standard deviation or change rate of gain, the mean or variance of noise power, and the difference in bit error rate; When the dynamic index meets the preset high dynamic range, the first repeater (1) and the second repeater (2) are further used to set the sampling period for detecting the real-time state to 1s to 30s; When the dynamic index meets the preset medium dynamic range, the first repeater (1) and the second repeater (2) are also used to set the sampling period for detecting the real-time state to 10 minutes to 30 minutes; When the dynamic index meets the preset low dynamic range, the first repeater (1) and the second repeater (2) are also used to set the sampling period for detecting the real-time state to 1 hour to 24 hours.

10. The power line communication trans-transformer relay method according to claim 9, characterized in that: The first repeater (1) further comprises a first frequency divider (4); The first frequency divider (4) is used to sample the communication signal of the first channel at time t and then divide the frequency of the communication signal of the first channel according to a preset frequency band; The first repeater (1) is also used to calculate and obtain the real-time status of the first channel corresponding to each frequency band at time t; Scoring and sorting each frequency band according to the real-time status of the first channel, and selecting the best N frequency bands, where N≥1; At time t+1, switching the communication signal of the first channel from the current frequency band to the N frequency bands; The second repeater (2) further comprises a second frequency divider (5); The second frequency divider (5) is used to sample the communication signal of the second channel at time t+1 and then divide the communication signal of the second channel according to a preset frequency band; The second repeater (2) is also used to calculate and obtain the real-time status of the second channel corresponding to each frequency band at time t+1; Scoring and sorting each frequency band according to the real-time status of the second channel, and selecting the best K frequency bands, K≥1; At time t+2, the communication signal of the second channel is switched from the current frequency band to the K frequency bands.

Images