A substation electromagnetic interference early warning and suppression regulation method and related device

Abstract

The application discloses a substation electromagnetic interference early warning and suppression regulation method and related equipment, first, the electromagnetic environment time domain sampling signal is acquired, the power spectrum is obtained through denoising reconstruction and frequency domain transformation, the main interference frequency band is identified and the frequency band boundary and energy accumulation intensity are extracted to construct the interference state sequence; then the state sequence in the historical window is input into the time sequence prediction model, the interference intensity and frequency band boundary of future multiple time windows are predicted, and early warning judgment is carried out according to the frequency domain overlap ratio of the predicted boundary and the current communication working frequency band; finally, the frequency band boundary prediction value is taken as a priori constraint, an MPC optimization problem is established under the constraints of available channels and transmission power, the optimal center frequency and transmission power sequence are solved by rolling, the first step control quantity is extracted from the optimization result and executed, the interference suppression dynamic adjustment is realized, the active avoidance, the coordinated regulation of the anti-interference performance and the energy efficiency are considered, and the stability and the spectrum utilization rate of the communication link are significantly improved.

Description

Technical Field

[0001] This application relates to the field of electromagnetic interference suppression technology, and more specifically, to a method and related equipment for electromagnetic interference early warning and suppression control in substations. Background Technology

[0002] Substations are core nodes in power systems. Internal high-voltage switch operations, partial discharges, and corona discharges can generate strong broadband electromagnetic transient interference, seriously threatening the reliability of communication links for wireless sensor networks, online monitoring devices, and inspection robots within the substation. With the advancement of smart substations and digital operation and maintenance, the real-time control and status awareness tasks undertaken by wireless communication are becoming increasingly critical, placing extremely high demands on electromagnetic interference suppression.

[0003] In existing technologies, fixed-channel communication is generally used in combination with passive anti-interference methods such as spread spectrum and error correction coding, or interference avoidance strategies based on spectrum sensing, that is, when the interference power in the working frequency band exceeds the threshold, it is switched to a preset backup channel; some systems introduce frequency hopping communication to obtain frequency diversity gain, but its frequency hopping pattern is usually pre-fixed, making it difficult to adaptively match the non-stationary time-varying characteristics of substation interference. The above-mentioned scheme has obvious defects: First, the decision-making relies solely on the current moment's interference snapshot, lacking the ability to predict future interference evolution trends. Under sudden interference such as transient pulses or periodic discharges, channel switching is delayed, which can easily lead to data packet loss or even communication interruption. Second, the control dimension is singular, mostly limited to switching channel frequencies, while ignoring the coordinated optimization of transmit power. It cannot balance communication quality and energy efficiency when the interference intensity changes dynamically. For example, it still transmits at high power under weak interference, resulting in energy waste, or fails to increase power in time to ensure link margin in residual interference. Third, channel and power selection are often based on heuristic rules or greedy search, without rolling time-domain optimization under multiple constraints such as available channel set, power limitations, and communication requirements, resulting in frequent invalid switching, low spectrum resource utilization, and insufficient link stability.

[0004] Therefore, there is an urgent need for an active suppression method that can accurately characterize the dynamic features of interference, predict its development trend, and coordinately optimize and control communication parameters. Summary of the Invention

[0005] This application provides a method and related equipment for early warning and suppression of electromagnetic interference in substations, which realizes the coordinated control from passive avoidance to active avoidance, taking into account both anti-interference performance and energy efficiency, and significantly improves the stability of communication links and spectrum utilization.

[0006] A method for early warning and suppression of electromagnetic interference in substations, comprising:

[0007] The time-domain sampling signal of the electromagnetic environment of the substation is acquired, and the time-domain sampling signal is denoised and reconstructed to obtain a denoised time-domain sequence;

[0008] The denoised time-domain sequence is subjected to frequency domain transformation to obtain a complex spectrum, and the power spectrum is calculated based on the complex spectrum;

[0009] Based on the power spectrum and the preset frequency band determination threshold, the main interference frequency bands are identified, the frequency band boundaries and the energy accumulation intensity within the frequency bands are extracted, and an interference state sequence is constructed from the frequency band boundaries and the energy accumulation intensity within the frequency bands.

[0010] The interference state sequence within the historical window is input into the time-series prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows.

[0011] Based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, an early warning determination is made and an early warning trigger result is output.

[0012] Using the predicted value of the frequency band boundary as a priori constraint, a model predictive control optimization problem is constructed. Under the constraints of available channel and power, the center frequency and transmit power of the communication link are optimized in a rolling manner. The first step of the control quantity is extracted from the optimal control sequence obtained by optimization and executed to achieve interference suppression adjustment of the communication link.

[0013] Optionally, the step of acquiring the time-domain sampling signal of the substation's electromagnetic environment and denoising and reconstructing the time-domain sampling signal to obtain a denoised time-domain sequence includes:

[0014] Continuous signals of the electromagnetic environment of the substation are collected within a fixed time window, and the continuous signals are discretized to form a time-domain sampling sequence.

[0015] The time-domain sampled sequence is denoised using a wavelet threshold denoising operator. The wavelet threshold denoising operator decomposes the signal into approximation coefficients and detail coefficients through wavelet decomposition, performs threshold quantization on the detail coefficients, and then outputs a denoised time-domain sequence through wavelet reconstruction.

[0016] Optionally, the step of identifying the main interference frequency bands based on the power spectrum and a preset frequency band determination threshold, and extracting the frequency band boundaries and energy accumulation intensity within the main interference frequency bands, includes:

[0017] The power value of each frequency point in the power spectrum is compared with the frequency band determination threshold. The minimum and maximum frequencies in the set of frequency points whose power values ​​exceed the frequency band determination threshold are respectively used as the lower limit and upper limit of the main interference frequency band.

[0018] When multiple discontinuous sets of frequency points exceeding the threshold are detected, the set with the largest frequency band width or the largest energy accumulation intensity within the frequency band is selected as the main interference frequency band.

[0019] The power values ​​of all frequency points within the main interference frequency band are summed to obtain the energy accumulation intensity within the frequency band.

[0020] Optionally, the step of determining the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and outputting the early warning trigger result, includes:

[0021] Based on the predicted frequency band boundary value and the upper and lower limits of the current communication link operating frequency band, the frequency domain overlap ratio is calculated. The frequency domain overlap ratio is the proportion of the intersection length of the predicted interference frequency band and the link operating frequency band to the bandwidth of the link operating frequency band.

[0022] When the frequency domain overlap ratio exceeds a preset overlap ratio threshold, and / or when the predicted interference intensity exceeds a preset intensity threshold, an early warning is triggered and an early warning trigger result is output, wherein the early warning trigger result includes progressive early warning levels or a single early warning signal.

[0023] Optionally, the step of using the frequency band boundary prediction value as a priori constraint to construct a model predictive control optimization problem, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints, includes:

[0024] In the prediction time domain, the center frequency and transmission power of the communication link at each prediction time are used as decision variables to construct a cost function, which includes a risk penalty term, a control quantity smoothing penalty term, and an energy consumption penalty term.

[0025] The risk penalty term is calculated based on the overlap ratio between the frequency band boundary prediction value and the link operating frequency band corresponding to the current decision variable, and is used to characterize the predicted interference risk to the communication link.

[0026] The control quantity smoothing penalty term is calculated based on the square of the L2 norm of the difference between control quantity vectors at adjacent time points, and is used to suppress frequent jumps in the center frequency and transmit power of the communication link.

[0027] The energy consumption penalty term is calculated based on the transmit power at the current moment and is used to characterize the energy consumption level of the link;

[0028] Under the constraints of available channels and power, find the optimal control sequence that minimizes the cost function;

[0029] The first step control quantity is extracted from the optimal control sequence as the execution control quantity at the current moment, and the process of constructing the cost function and solving the optimal control sequence is repeated in the next moment in a rolling time domain manner.

[0030] Optional, also includes:

[0031] When there are multiple sets of frequency points exceeding the threshold, the first interference frequency band corresponding to the main interference source and at least one second interference frequency band corresponding to the secondary interference source are identified from the multiple sets of frequency points exceeding the threshold based on the frequency band width and / or the energy accumulation intensity within the frequency band of each set of frequency points exceeding the threshold.

[0032] Extract the first interference state sequence of the first interference frequency band and the second interference state sequence of the second interference frequency band respectively;

[0033] The first interference state sequence and the second interference state sequence are respectively input into the time series prediction model to obtain the first prediction result and the second prediction result.

[0034] The overall overlap ratio is calculated based on the first prediction result and the second prediction result, and the early warning determination and interference suppression adjustment of the communication link are performed based on the overall overlap ratio.

[0035] Optionally, the cost function is expressed as:

[0036] ;

[0037] ;

[0038] Where J is the cost function, N p To predict the time domain length, f t+h and p t+h The link center frequency and transmit power at the h-th prediction time are respectively, u t+h Let h be the control vector at the h-th prediction time. As a risk penalty item, w B w U w P These are the weighting coefficients for the risk penalty, the control quantity smoothing penalty, and the energy consumption penalty, respectively.

[0039] A substation electromagnetic interference early warning and suppression control device includes:

[0040] The sampling and denoising unit is used to acquire the time-domain sampling signal of the electromagnetic environment of the substation, and to denoise and reconstruct the time-domain sampling signal to obtain a denoised time-domain sequence.

[0041] The frequency domain transformation unit is used to perform frequency domain transformation on the denoised time domain sequence to obtain a complex spectrum, and calculate the power spectrum based on the complex spectrum;

[0042] The state sequence unit is used to identify the main interference frequency bands based on the power spectrum and a preset frequency band determination threshold, extract the frequency band boundaries and the energy accumulation intensity within the frequency bands of the main interference frequency bands, and construct an interference state sequence from the frequency band boundaries and the energy accumulation intensity within the frequency bands.

[0043] The timing prediction unit is used to input the interference state sequence within the historical window into the timing prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows.

[0044] The early warning determination unit is used to determine the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and output the early warning trigger result.

[0045] The suppression and adjustment unit is used to construct a model predictive control optimization problem by taking the frequency band boundary prediction value as a priori constraint, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints. The first step control quantity is extracted from the optimal control sequence obtained by optimization and executed to realize the interference suppression and adjustment of the communication link.

[0046] A substation electromagnetic interference early warning and suppression control device includes a memory and a processor;

[0047] The memory is used to store programs;

[0048] The processor is used to execute the program to implement the various steps of the substation electromagnetic interference early warning and suppression control method as described in any of the above claims.

[0049] A readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the substation electromagnetic interference early warning and suppression control method as described in any of the preceding claims.

[0050] As can be seen from the above technical solution, the substation electromagnetic interference early warning and suppression control method and related equipment provided in this application first acquire the electromagnetic environment time-domain sampling signal, obtain the power spectrum through denoising reconstruction and frequency domain transformation, identify the main interference frequency bands, and extract their frequency band boundaries and energy accumulation intensity to construct an interference state sequence; then, input the state sequence within the historical window into the time-series prediction model to predict the interference intensity and frequency band boundaries of future multi-time windows, and make early warning judgments based on the frequency domain overlap ratio between the predicted boundaries and the current communication operating frequency band; finally, use the predicted value of the frequency band boundary as a priori constraint, establish a model predictive control optimization problem under the constraints of available channels and transmit power, solve the optimal center frequency and transmit power sequence in a rolling manner, extract the first step control quantity from the optimization result for execution, and realize dynamic adjustment of interference suppression. Compared with the prior art, its beneficial effects are:

[0051] First, by using the interference state sequence and time-series prediction model, the transient and periodic interference trends of substations can be perceived in advance. The early warning judgment provides ample prediction time for communication parameter adjustment, fundamentally solving the problem of switching lag and communication interruption caused by passive response based solely on the current interference, and enabling the control to shift from passive response to active avoidance.

[0052] Second, the center frequency and transmit power are incorporated into the model prediction control framework for joint optimization. While avoiding interference frequency bands, the power is adaptively adjusted according to the predicted intensity, overcoming the limitations of existing methods that only adjust the frequency or fix the power. When the predicted interference is weak, the power is automatically reduced to save energy, and when the interference is enhanced, the power is moderately increased to ensure link reliability, achieving the optimal synergy between anti-interference performance and energy efficiency.

[0053] Third, by employing a rolling time-domain optimization mechanism that executes only the first step of the control input, the method seeks the globally optimal solution while continuously satisfying available channel and power constraints. This effectively suppresses frequent channel jumps and suboptimal selections caused by threshold switching, significantly improving spectrum utilization and communication link stability. This method organically integrates prediction, early warning, and optimal control, providing a highly reliable and intelligent active interference suppression method for wireless communication in strong electromagnetic environments in substations. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0055] Figure 1 This is a flowchart of a substation electromagnetic interference early warning and suppression control method disclosed in an embodiment of this application;

[0056] Figure 2 This is a schematic diagram of a substation electromagnetic interference early warning and suppression control device disclosed in an embodiment of this application;

[0057] Figure 3 This is a hardware structure block diagram of a substation electromagnetic interference early warning and suppression control device disclosed in an embodiment of this application. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0059] This application can be used in a wide variety of general-purpose or special-purpose computing device environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor devices, distributed computing environments including any of the above devices, etc.

[0060] The following section introduces the solution proposed in this application. The technical solution is as follows, and details are provided below.

[0061] Figure 1 This is a flowchart of a substation electromagnetic interference early warning and suppression control method disclosed in an embodiment of this application.

[0062] like Figure 1 As shown, the method may include:

[0063] Step S1: Obtain the time-domain sampling signal of the electromagnetic environment of the substation, and denoise and reconstruct the time-domain sampling signal to obtain a denoised time-domain sequence.

[0064] Specifically, electromagnetic field probes or antennas with wideband response are deployed in electromagnetically sensitive locations such as the primary equipment area, switch room, and near secondary cabinets within the substation. These probes and antennas are used to continuously acquire time-domain waveforms of the spatial electromagnetic environment at a sampling rate capable of covering the main interference spectra such as partial discharge, switching operations, and corona discharge, obtaining the original time-domain sampled signal. This signal contains not only the target electromagnetic interference component but also environmental background noise, nearby communication signals, and random pulse spikes. To remove noise and preserve the transient time-domain form of the interference, the original time-domain sampled signal undergoes denoising and reconstruction processing. In one embodiment, a wavelet basis function similar to the interference pulse waveform is selected, and the signal is decomposed into multiple scales using wavelet transform. At each decomposition level, a threshold is adaptively determined based on the noise variance estimation results. The wavelet coefficients are then subjected to soft thresholding or hard thresholding to remove noise-corresponding coefficients. Finally, the signal is reconstructed using inverse wavelet transform to obtain the denoised time-domain sequence. In another implementation, empirical mode decomposition (EMD) is used to adaptively decompose the signal into a series of intrinsic mode functions (EMFs). By calculating the correlation between each mode and the original signal, noise-dominant low-order modes are identified and removed. Only the signal-dominant mode components are superimposed and reconstructed. Through this denoising and reconstruction process, random noise is effectively suppressed, while detailed features such as the pulse leading edge and oscillation envelope of the interfering signal are well preserved. The resulting denoised time-domain sequence has a significantly improved signal-to-interference-plus-noise ratio (SNR), providing a high-fidelity signal foundation for subsequent frequency domain analysis.

[0065] Step S2: Perform frequency domain transformation on the denoised time-domain sequence to obtain a complex spectrum, and calculate the power spectrum based on the complex spectrum.

[0066] Specifically, a time-domain window function is applied to the denoised time-domain sequence obtained in step S1. The window function type can be a Hanning window, a Hamming window, or a Blackman window to achieve a balance between frequency resolution and spectral leakage suppression. A Fast Fourier Transform is then performed on the windowed sequence to map the discrete time-domain signal to the frequency domain, generating a complex spectrum. This spectrum carries both amplitude and phase response information in real and imaginary parts at each frequency point. To intuitively characterize the distribution of electromagnetic interference energy along the frequency axis, its power spectrum is calculated based on the complex spectrum. A direct power spectrum estimation method is the periodogram method, which involves taking the square of the modulus of each point in the complex spectrum to obtain the power estimate for the corresponding frequency point. To reduce the statistical variance of the periodogram estimation and improve the smoothness of the power spectrum curve, a weighted overlapping average method is preferred: the denoised time-domain sequence is divided into multiple data segments of uniform length with partial overlap. After windowing each segment, the periodogram of each segment is calculated, and then the arithmetic mean of the periodograms of all segments is taken to obtain the average power spectrum. By reasonably setting the segment length, overlap ratio, and window function parameters, the spectrum leakage and variance estimation are effectively suppressed while meeting the required frequency resolution. This enables the power spectrum to clearly distinguish the main lobe and side lobe of multi-source narrowband interference and the spectral profile of broadband pulse interference in the electromagnetic environment of substations, providing a clear quantitative basis for interference frequency band identification.

[0067] Step S3: Based on the power spectrum and the preset frequency band determination threshold, identify the main interference frequency band, extract the frequency band boundary and the energy accumulation intensity within the frequency band of the main interference frequency band, and construct an interference state sequence from the frequency band boundary and the energy accumulation intensity within the frequency band.

[0068] Specifically, the frequency band determination threshold in this step is used to separate real electromagnetic interference from background noise. This threshold can be set based on the long-term statistical characteristics of the power spectrum during periods of operation without strong interference at the substation, using the noise floor mean superimposed with several times the standard deviation as a fixed threshold; alternatively, an adaptive threshold generation method based on local window sliding statistics can be used, where a local window slides across the power spectrum frequency by frequency point to estimate the noise power level in real time, dynamically generating a threshold that adjusts with changes in environmental noise. The amplitude of each frequency point in the power spectrum is compared point by point with the frequency band determination threshold, and frequency intervals whose amplitudes continuously exceed the threshold are marked as the initial interference frequency bands. To eliminate interval fragmentation and isolated spikes caused by instantaneous noise fluctuations, morphological closing operations are performed on the initial interference frequency bands, connecting adjacent intervals with very small spacing into continuous intervals and filling in internal micro-gaps, thereby filtering out spurious peaks with bandwidths lower than the preset minimum bandwidth, ultimately obtaining the main interference frequency bands with actual physical width. For each major interference frequency band, its lower and upper boundary frequencies are extracted as band boundaries. The power spectral amplitude within the band is then discretely accumulated or continuously integrated to obtain the band energy accumulation intensity, which characterizes the total interference energy of that band. The boundaries and energy intensities of all major interference frequency bands extracted at the same analysis time are organized into a multidimensional state vector according to a unified ascending frequency order rule. As the analysis window slides along the time axis, the state vectors at consecutive times are arranged sequentially, thus forming an interference state sequence that reflects the evolution of interference frequency migration, energy fluctuations, and the emergence of new interference.

[0069] Step S4: Input the interference state sequence in the historical window into the time series prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows.

[0070] Specifically, the time span is defined as a historical window containing a series of consecutive past interference state sequences. This window should be long enough to encompass the quasi-periodic characteristics and long-range dependencies of the interference events. The state vectors of the current moment and previous consecutive historical moments are extracted from the interference state sequence to form the input feature sequence, which is then fed into a pre-trained time-series prediction model. This model is responsible for extracting the temporal evolution patterns of the state sequence and generating future trend predictions. In one architecture, a Long Short-Term Memory (LSTM) network or a gated recurrent unit (GRU) network can be used, utilizing their gating mechanisms to selectively remember and forget historical information, capturing long-term dependent behaviors such as interference frequency band boundary drift and intensity gradual changes. In another architecture, a transformer model based on a self-attention mechanism can be used, synchronously modeling the correlation between different time steps through global self-attention calculations to adapt to complex scenarios such as interference jumps and multi-frequency band coupling. During the offline training phase, the massive interference state sequences accumulated from long-term operation at the substation are used, with the actual interference intensity and frequency band boundaries of future moments as labels for supervised training, allowing the model to internalize the dynamic evolution patterns of interference. During the online operation phase, the model performs rolling inference on the continuously sliding historical window, and outputs a sequence of predicted interference intensity values ​​for multiple consecutive time windows and the corresponding frequency band boundary prediction value sequence in a single forward calculation, providing ample forward-looking information for early warning judgment and forward-looking control.

[0071] Step S5: Based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, make an early warning determination and output the early warning trigger result.

[0072] Specifically, for a future time window predicted in step S4, the frequency range defined by the predicted frequency band boundary value corresponding to that time window is obtained. Simultaneously, the operating frequency band range occupied by the current communication link is determined, and the overlap width of the two ranges on the frequency axis is calculated as the proportion to the total width of the communication link's operating frequency band. This is used as the frequency domain overlap ratio to quantify the degree of spectral intrusion of the predicted interference into the communication channel. At the same time, the predicted interference intensity value for that time window is extracted from the prediction output, reflecting the energy level of the intrusive interference. The early warning determination adopts a dual-indicator joint decision-making mechanism of frequency domain overlap ratio and predicted interference intensity value: the frequency domain overlap ratio is compared with a preset frequency domain overlap early warning threshold, and the predicted interference intensity value is compared with a preset intensity early warning threshold. When the frequency domain overlap ratio exceeds its corresponding threshold and the predicted interference intensity value also exceeds its corresponding threshold, a significant interference threat is determined for that time window, and an early warning is triggered. To support differentiated response strategies, a multi-level threshold ladder can be further set. The early warning level is divided into multiple levels based on the magnitude by which the two indicators exceed the thresholds. For example, when the indicators exceed the first-level threshold, a low-level early warning is output; when they exceed higher thresholds, the early warning level is progressively increased. The early warning trigger results are output in the form of clear indicator signals and level codes, enabling the communication link control system to know in advance the degree of interference risk that it will face in a specific period of time, and reserving sufficient decision-making and action time for subsequent regulation.

[0073] Step S6: Using the predicted value of the frequency band boundary as a priori constraint, construct a model predictive control optimization problem, and perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints. Then, extract the first step control quantity from the optimal control sequence obtained by optimization and execute it to realize the interference suppression adjustment of the communication link.

[0074] Specifically, the predicted frequency band boundaries for multiple consecutive time windows generated in step S4 are considered as prior information of defined spectrum exclusion zones. This means that the center frequency of the communication link within the corresponding prediction time window must not fall within these frequency band intervals and their additional guard intervals. Using the center frequency and transmit power of the communication link as joint control variables, a prediction model describing the relationship between the receiver's signal-to-interference-plus-noise ratio (SIR) or bit error rate (BER) and the variables is established. Based on this, a finite-time domain model predictive control optimization problem is constructed. The optimization objective can be set as minimizing transmit power consumption under the hard constraint of ensuring communication quality meets the minimum SIR threshold, or maximizing channel capacity under the maximum allowable power. Simultaneously, a control increment penalty term can be introduced into the objective function to suppress frequent jumps in center frequency and transmit power. Strict constraints are stipulated: the center frequency must be selected from the pre-defined set of available channels and avoid all predicted interference frequency bands and guard intervals; the transmit power must not exceed the upper limit of the power amplifier's linear operation and the limits set by radio management regulations, and must not be lower than the minimum necessary power to ensure continuous link connectivity. In each control cycle, the current operating state of the communication link is acquired as feedback. Combined with the latest interference frequency band boundary prediction values, numerical optimization algorithms such as sequential quadratic programming or interior-point methods are used to solve the above optimization problem in a rolling manner, obtaining the optimal center frequency and transmit power control sequence for the next few steps. Only the first step of the control quantity in this sequence is immediately issued to the frequency synthesizer and adjustable gain amplifier and other actuators to update the parameters. In the next control cycle, the optimization problem is reconstructed and solved based on the newly acquired state feedback and the updated interference prediction, forming a closed-loop rolling control of prediction, optimization, execution, and feedback. This mechanism deeply integrates interference prediction and optimal control, and by proactively avoiding predicted interference frequency bands and coordinating the adjustment of center frequency and transmit power, it achieves a balanced optimization of active interference suppression and improved communication performance under the constraint of available resources.

[0075] As can be seen from the above technical solution, the substation electromagnetic interference early warning and suppression control method and related equipment provided in this application first acquire the electromagnetic environment time-domain sampling signal, obtain the power spectrum through denoising reconstruction and frequency domain transformation, identify the main interference frequency bands, and extract their frequency band boundaries and energy accumulation intensity to construct an interference state sequence; then, input the state sequence within the historical window into the time-series prediction model to predict the interference intensity and frequency band boundaries of future multi-time windows, and make early warning judgments based on the frequency domain overlap ratio between the predicted boundaries and the current communication operating frequency band; finally, use the predicted value of the frequency band boundary as a priori constraint, establish a model predictive control optimization problem under the constraints of available channels and transmit power, solve the optimal center frequency and transmit power sequence in a rolling manner, extract the first step control quantity from the optimization result for execution, and realize dynamic adjustment of interference suppression. Compared with the prior art, its beneficial effects are:

[0076] First, by using the interference state sequence and time-series prediction model, the transient and periodic interference trends of substations can be perceived in advance. The early warning judgment provides ample prediction time for communication parameter adjustment, fundamentally solving the problem of switching lag and communication interruption caused by passive response based solely on the current interference, and enabling the control to shift from passive response to active avoidance.

[0077] Second, the center frequency and transmit power are incorporated into the model prediction control framework for joint optimization. While avoiding interference frequency bands, the power is adaptively adjusted according to the predicted intensity, overcoming the limitations of existing methods that only adjust the frequency or fix the power. When the predicted interference is weak, the power is automatically reduced to save energy, and when the interference is enhanced, the power is moderately increased to ensure link reliability, achieving the optimal synergy between anti-interference performance and energy efficiency.

[0078] Third, by employing a rolling time-domain optimization mechanism that executes only the first step of the control input, the method seeks the globally optimal solution while continuously satisfying available channel and power constraints. This effectively suppresses frequent channel jumps and suboptimal selections caused by threshold switching, significantly improving spectrum utilization and communication link stability. This method organically integrates prediction, early warning, and optimal control, providing a highly reliable and intelligent active interference suppression method for wireless communication in strong electromagnetic environments in substations.

[0079] In some embodiments of this application, the process of step S1, obtaining the time-domain sampling signal of the substation electromagnetic environment, and denoising and reconstructing the time-domain sampling signal to obtain a denoised time-domain sequence, is described, and may specifically include:

[0080] Step S11: Collect continuous signals of the electromagnetic environment of the substation within a fixed time window, and discretize the continuous signals to form a time-domain sampling sequence.

[0081] Step S12: The time-domain sampling sequence is denoised using a wavelet threshold denoising operator. The wavelet threshold denoising operator decomposes the signal into approximation coefficients and detail coefficients through wavelet decomposition, performs threshold quantization on the detail coefficients, and then outputs the denoised time-domain sequence through wavelet reconstruction.

[0082] Specifically, electromagnetic environment signals are continuously acquired at designated monitoring points within the substation using electromagnetic induction probes or electric field antennas. The sampling system uses a front-end analog conditioning circuit to amplify the antenna-received signal with low noise and perform anti-aliasing filtering. Then, an analog-to-digital converter discretizes the continuous analog signal at equal intervals using a preset fixed sampling frequency, acquiring a set of discrete sampling points within a fixed-length time window to form a time-domain sampling sequence. The length of this time window is selected based on the transient duration and repetition period of the main interference sources within the substation, ensuring that each sampling sequence can completely capture the time-domain waveform characteristics of a single or multiple consecutive interference events, providing raw data with sufficient time resolution and amplitude accuracy for subsequent denoising and spectrum analysis.

[0083] The time-domain sampling sequence obtained in step S11 is input into the wavelet threshold denoising operator for processing. First, based on the waveform morphology characteristics of the interference pulses in the signal to be processed, a matching wavelet basis function is selected and the number of decomposition levels is determined. The time-domain sampling sequence is then decomposed into multiple scales using discrete wavelet transform. In each decomposition level, the signal is split into approximation coefficients reflecting the low-frequency profile of the signal and detail coefficients reflecting the high-frequency details of the signal. Noise energy is mainly distributed in the detail coefficients of each level. Second, threshold quantization is performed on the detail coefficients of each level: the noise standard deviation is estimated based on the statistical characteristics of the coefficients of that level, and a quantization threshold appropriate to the noise level is determined accordingly. Coefficients with absolute values ​​exceeding the threshold are retained or shrunk, while coefficients with absolute values ​​not exceeding the threshold are set to zero, thereby eliminating the coefficient components corresponding to noise. Finally, the processed detail coefficients of each level and the highest level approximation coefficients are subjected to inverse wavelet transform, reconstructing the original signal scale layer by layer, and outputting a denoised time-domain sequence with noise interference removed. This sequence effectively suppresses background noise and random glitches while preserving the rising edge steepness and oscillation details of the interference pulses, providing a high signal-to-noise ratio signal input for subsequent frequency domain transformation.

[0084] In some embodiments of this application, the process of identifying the main interference frequency bands and extracting the frequency band boundaries and energy accumulation intensity within the frequency bands based on the power spectrum and a preset frequency band determination threshold in step S3 is described, which may specifically include:

[0085] Step S31: Compare the power value of each frequency point in the power spectrum with the frequency band determination threshold, and take the minimum and maximum frequencies in the set of frequency points whose power values ​​exceed the frequency band determination threshold as the lower limit and upper limit of the main interference frequency band, respectively.

[0086] Step S32: When multiple discontinuous sets of frequency points exceeding the threshold are detected, select the set with the largest frequency band width or the largest energy accumulation intensity within the frequency band as the main interference frequency band.

[0087] Step S33 involves summing the power values ​​of all frequency points within the main interference frequency band to obtain the energy accumulation intensity within the frequency band.

[0088] Specifically, the power spectrum obtained in step S2 is scanned along the frequency axis point by point, and the power amplitude at each frequency point is compared with a preset frequency band determination threshold. Frequency points with power amplitudes greater than the threshold are marked as threshold-exceeding frequency points; frequency points with power amplitudes less than or equal to the threshold are determined as noise or background signals. Each connected set consisting of consecutive threshold-exceeding frequency points on the frequency axis corresponds to a candidate interference frequency band. For each candidate interference frequency band, the frequency values ​​of all frequency points in the set are traversed, and the frequency point with the smallest frequency value is retrieved, with its frequency value serving as the lower limit of the candidate interference frequency band; simultaneously, the frequency point with the largest frequency value is retrieved, with its frequency value serving as the upper limit of the candidate interference frequency band. Thus, the lower and upper limits of the frequency band together define the complete frequency range occupied by the candidate interference frequency band on the spectrum, achieving a quantitative description of the frequency range occupied by the interference.

[0089] In the electromagnetic environment of a real substation, multiple interference sources or high-order harmonics generated by the same interference source may occur simultaneously within the same power spectrum frame, resulting in a set of several unconnected super-threshold frequency points on the frequency axis. To determine the primary interference frequency band that poses the most significant threat to the communication link, this sub-step sets two optional selection criteria: one is based on the bandwidth, calculating the difference between the upper and lower limits of each candidate interference frequency band to obtain its bandwidth, and selecting the candidate interference frequency band with the largest bandwidth as the primary interference frequency band; the other is based on the energy accumulation intensity within the frequency band, pre-calculating the sum of the power values ​​of all frequency points within each candidate interference frequency band, and selecting the candidate interference frequency band with the largest energy accumulation intensity as the primary interference frequency band. When the candidate interference frequency bands pointed to by the two criteria are inconsistent, a comprehensive decision can be made by combining the location of the current operating frequency band of the communication link, and the interference frequency band that may cause the largest spectral intrusion to the communication link can be determined as the primary interference frequency band. Through the above selection strategy, it is ensured that the subsequent construction of the interference state vector focuses on the most threatening interference components, avoiding the dilution of state characteristics due to the introduction of secondary interference frequency bands.

[0090] Within the frequency range covered by the main interference band, starting from the frequency point corresponding to the lower limit of the band, the frequency axis is traversed point by point to the frequency point corresponding to the upper limit of the band, and the power amplitude corresponding to each frequency point on the power spectrum is read sequentially. All the read power amplitudes are arithmetically summed, and the sum is the cumulative energy intensity within the main interference band. This summation comprehensively reflects the total energy of all frequency components within the interference band, directly characterizing the overall emission intensity and potential hazard level of the interference band, and providing a key energy characteristic parameter parallel to the band boundary for constructing the interference state sequence.

[0091] Furthermore, this application may also include:

[0092] When there are multiple sets of frequency points exceeding the threshold, the first interference frequency band corresponding to the main interference source and at least one second interference frequency band corresponding to the secondary interference source are identified from the multiple sets of frequency points exceeding the threshold based on the frequency band width and / or the energy accumulation intensity within the frequency band of each set of frequency points exceeding the threshold.

[0093] Extract the first interference state sequence of the first interference frequency band and the second interference state sequence of the second interference frequency band respectively;

[0094] The first interference state sequence and the second interference state sequence are respectively input into the time series prediction model to obtain the first prediction result and the second prediction result.

[0095] The overall overlap ratio is calculated based on the first prediction result and the second prediction result, and the early warning determination and interference suppression adjustment of the communication link are performed based on the overall overlap ratio.

[0096] Specifically, for complex scenarios where multiple independent interference sources exist simultaneously in the electromagnetic environment of a substation, the aforementioned interference frequency band identification and control process is enhanced. When step S31 obtains multiple sets of unconnected super-threshold frequency points through threshold comparison, it indicates the presence of multiple interference sources in the current electromagnetic environment, either spatially or through different generation mechanisms. In this case, all candidate interference frequency bands are first ranked and classified based on at least one of the following indicators: frequency band width and energy accumulation intensity within each super-threshold frequency point set. The super-threshold frequency point set with the largest frequency band width or the highest energy accumulation intensity is identified as the first interference frequency band corresponding to the primary interference source. The remaining super-threshold frequency point sets that meet preset width or intensity thresholds are identified as the second interference frequency band corresponding to secondary interference sources, thus completing the hierarchical identification of interference sources. The interference state extraction operation in step S3 is performed on the first and second interference frequency bands respectively, independently acquiring their respective frequency band boundary sequences and energy accumulation intensity sequences within multiple consecutive time windows, forming the first interference state sequence and the second interference state sequence. These two types of sequences respectively characterize the independent frequency evolution trajectory and energy fluctuation law of the main interference source and the secondary interference source. Subsequently, the first and second interference state sequences are used as inputs and fed into the same or independently trained time series prediction model for forward prediction, corresponding to the first prediction result and the second prediction result containing the predicted values ​​of interference intensity and frequency band boundaries within multiple future time windows. In the early warning determination stage, the frequency domain overlap ratio between the predicted value of the frequency band boundary in the first prediction result and the current communication link operating frequency band, and the frequency domain overlap ratio between the predicted value of the frequency band boundary in the second prediction result and the current communication link operating frequency band are calculated respectively. The two are combined into a comprehensive overlap ratio according to a preset weight or by taking the union method. This comprehensive overlap ratio simultaneously reflects the joint encroachment situation of the main interference and the secondary interference on the communication link spectrum. The overall overlap ratio is compared with a preset overall overlap warning threshold. Simultaneously, the predicted intensity of each interference source can be used for auxiliary judgment, outputting a comprehensive warning trigger result including the warning level. During the interference suppression and control phase, the frequency band boundary prediction values ​​from the first and second prediction results are incorporated as prior constraints of the spectrum no-go zone into the constraints of the model predictive control optimization problem. Under available channel constraints and power constraints, the center frequency and transmit power of the communication link are solved in a rolling manner, enabling the optimized control strategy to simultaneously avoid the frequency bands occupied by both primary and secondary interference sources, achieving coordinated optimization of communication link reliability and transmission efficiency under multi-source interference environments.

[0097] In some embodiments of this application, the process of step S5, which involves determining the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and outputting the early warning trigger result, is described. Specifically, it may include:

[0098] Step S51: Calculate the frequency domain overlap ratio based on the predicted frequency band boundary value and the upper and lower limits of the current communication link operating frequency band. The frequency domain overlap ratio is the ratio of the intersection length of the predicted interference frequency band and the link operating frequency band to the bandwidth of the link operating frequency band.

[0099] Step S52: When the frequency domain overlap ratio exceeds a preset overlap ratio threshold, and / or when the predicted interference intensity exceeds a preset intensity threshold, an early warning is triggered and an early warning trigger result is output, wherein the early warning trigger result includes progressive early warning levels or a single early warning signal.

[0100] Specifically, firstly, the predicted frequency band boundary value corresponding to a specified future time window is extracted from the prediction results output in step S4. This boundary prediction value includes the lower and upper limits of the predicted interference frequency band, which together define a continuous predicted interference frequency interval on the frequency axis. Simultaneously, the operating frequency band parameters of the current communication link are obtained, including the lower and upper limits of its center frequency, which also constitute a continuous operating frequency interval. The predicted interference frequency interval and the operating frequency interval are compared on the frequency axis, and the intersection of the two intervals is obtained, i.e., the frequency range falling within both intervals. The frequency span length of this intersection is calculated as the intersection length between the predicted interference frequency band and the link's operating frequency band. Subsequently, the difference between the upper and lower limits of the communication link's operating frequency band is calculated to obtain the total bandwidth of the link's operating frequency band. The intersection length is divided by the total bandwidth of the link's operating frequency band, and the resulting ratio is the frequency domain overlap ratio. The value of this ratio ranges from zero to one. A larger value indicates a more severe spectral encroachment of the predicted interference band on the communication link. A value of zero indicates complete separation between the two in the frequency domain, while a value of one indicates that the predicted interference band completely covers the communication link's operating frequency band. This normalized ratio index enables an objective and quantitative evaluation of the degree of interference spectrum intrusion, eliminating evaluation bias caused by differences in absolute frequency spans and facilitating direct comparison with a uniformly preset threshold.

[0101] Based on the frequency domain overlap ratio calculated in step S51 and the interference intensity prediction value output in step S4, a dual-index joint early warning judgment is performed. Preset overlap ratio thresholds and preset intensity thresholds constitute two judgment thresholds in the spatial and energy domains, respectively. An early warning is triggered when the frequency domain overlap ratio exceeds the overlap ratio threshold and the interference intensity prediction value simultaneously exceeds the intensity threshold; or an early warning is triggered when either the frequency domain overlap ratio or the interference intensity prediction value exceeds its corresponding threshold. The specific joint logic can be configured according to the sensitivity of the substation communication link to spectrum encroachment and energy intrusion. The output of the early warning trigger result can take the form of progressively increasing early warning levels: multiple combinations of overlap ratio and intensity thresholds are set from low to high. When the prediction value only exceeds the lowest threshold combination, a low-level early warning is output, such as a warning level; when the prediction value further exceeds a higher threshold combination, the early warning level is progressively increased, for example, outputting a warning level and an emergency level sequentially. Different levels can correspond to different preset response actions and adjustments to the urgency. The output of the early warning trigger result can also take the form of a single early warning signal, i.e., without distinguishing levels, only outputting a valid early warning flag level or code when a significant interference threat is determined. The warning trigger result is transmitted to the communication link control unit in the form of a clear digital signal or status code, providing a direct trigger basis for the decision-making and activation of the control system.

[0102] In some embodiments of this application, the process of using the predicted value of the frequency band boundary as a priori constraint in step S6 to construct a model predictive control optimization problem, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints, is described. Specifically, it may include:

[0103] Step S61: In the prediction time domain, the center frequency and transmission power of the communication link at each prediction time are used as decision variables to construct a cost function, which includes a risk penalty term, a control quantity smoothing penalty term, and an energy consumption penalty term.

[0104] The risk penalty term is calculated based on the overlap ratio between the frequency band boundary prediction value and the link operating frequency band corresponding to the current decision variable, and is used to characterize the predicted interference risk to the communication link.

[0105] The control quantity smoothing penalty term is calculated based on the square of the L2 norm of the difference between control quantity vectors at adjacent time points, and is used to suppress frequent jumps in the center frequency and transmit power of the communication link.

[0106] The energy consumption penalty term is calculated based on the transmit power at the current moment and is used to characterize the energy consumption level of the link;

[0107] Step S62: Under the constraints of available channels and power, solve for the optimal control sequence that minimizes the cost function;

[0108] Step S63: Extract the first step control quantity from the optimal control sequence as the execution control quantity at the current moment, and repeat the process of constructing the cost function and solving the optimal control sequence in a rolling time domain manner at the next moment.

[0109] Specifically, within the prediction time domain spanning from the current moment to several future steps, the center frequency and transmit power of the communication link at each prediction moment are combined into a decision variable vector. All decision variable vectors at each prediction moment are arranged chronologically to form the control sequence to be optimized. The cost function consists of a weighted sum of three penalties. The first term is a risk penalty, constructed as follows: Based on the predicted frequency band boundaries obtained in step S4, the predicted interference frequency range for each moment is determined; based on the center frequency in the decision variables at that moment, the current operating frequency range occupied by the communication link is determined; the overlap width of the two ranges on the frequency axis is calculated as a proportion of the total bandwidth of the communication link's operating frequency band, yielding the frequency domain overlap ratio at that moment; this overlap ratio is converted into a risk cost value through a preset mapping function. The larger the overlap ratio, the higher the risk cost value, thus characterizing the degree of predicted interference risk faced by the communication link if the current decision variables are used at that moment. The second term is the control quantity smoothing penalty term, which is constructed by taking the difference of the decision variable vectors for each pair of adjacent time points in the prediction time domain, and calculating the sum of the squares of each component of the difference vector, i.e., the square of the L2 norm. This penalty term jointly measures the change amplitude of the center frequency and transmission power between adjacent time points; the larger the change amplitude, the higher the penalty value, thereby suppressing frequent or drastic jumps in the control quantity during the optimization process and ensuring the stable operation of the communication link. The third term is the energy consumption penalty term, which is constructed by taking the transmission power value in the decision variables at the current time point and converting it into an energy consumption cost value through linear or nonlinear mapping; the higher the transmission power, the greater the energy consumption cost, thereby guiding the optimization process to tend to choose a lower transmission power while meeting communication quality requirements, reducing the overall energy consumption of the system. The above three penalties are weighted and summed through preset weight coefficients. The weight coefficients determine the relative importance between the various optimization objectives and can be configured and adjusted according to the actual operating needs of the substation, so that the cost function comprehensively reflects the balance requirements of anti-interference performance, control stability, and energy efficiency.

[0110] Based on the cost function constructed in step S61, constraints are introduced and a finite-time domain optimization problem is solved. The constraints are divided into two categories: available channel constraints and power constraints. Available channel constraints stipulate that at each moment in the prediction time domain, the center frequency in the decision variables must be selected from a pre-defined set of available channel frequencies. This set is pre-allocated by the radio resource management strategy based on spectrum planning and in-site spectrum occupancy. Simultaneously, the communication link operating frequency band interval determined by this center frequency must not overlap with the predicted interference frequency band interval defined by the frequency band boundary prediction value obtained in step S4. In the case of multi-source interference scenarios, all identified interference frequency bands must be avoided. Furthermore, a pre-defined guard interval must be maintained between the interference frequency bands. Power constraints stipulate that at each moment in the prediction time domain, the transmit power in the decision variables must not exceed the upper limit of the linear operating area of ​​the transmit link power amplifier and the maximum equivalent isotropic radiated power allowed by radio management regulations, and must not be lower than the minimum transmit power required to maintain the minimum demodulation signal-to-noise ratio at the communication link receiver. Under the aforementioned constraints, numerical optimization methods are employed to minimize the cost function in the finite-time domain. For example, a sequential quadratic programming algorithm can be used to approximate the original problem as a quadratic programming subproblem at each iteration point, gradually approaching the optimal solution. Alternatively, the interior-point method can be used to transform the inequality constraints into penalty terms through a barrier function, then optimization can be performed within the feasible region. The solution process traverses the entire prediction time domain, ultimately outputting the optimal control sequence that minimizes the cost function globally or locally. This sequence contains the optimal center frequency and transmit power values ​​at each time step within the prediction time domain.

[0111] In the obtained optimal control sequence, only the first-step control quantity corresponding to the current control moment is extracted, namely the optimal center frequency value and optimal transmit power value at the first prediction moment, and this is used as the execution control quantity for the current control cycle. This execution control quantity is output to the frequency synthesizer and adjustable gain amplifier of the communication link in real time to complete the tuning of the center frequency and the setting of the transmit power, realizing the real-time update of communication parameters. After the execution of this control cycle is completed, the time advances by one control step, the system re-acquires the latest electromagnetic environment time-domain sampling signal, updates the interference state sequence, and obtains the updated future multi-time window frequency band boundary prediction value and interference intensity prediction value through the time-series prediction model. Using the current new communication link operating state as feedback, the cost function and constraints covering the new prediction time domain are reconstructed, the optimal control sequence is solved again, and the first-step control quantity is extracted and executed. The above process is repeated in each control cycle, forming a closed-loop optimization control based on the latest interference prediction information and link state feedback. This rolling time-domain execution mechanism ensures that each control decision is based on the latest prediction of the future interference situation, enabling timely response to dynamic changes in interference. By executing only the first step of the control, it effectively suppresses the cumulative effects of model mismatch and prediction errors, ensuring the continuous and reliable operation of the communication link and efficient utilization of resources in the electromagnetic interference environment of the substation.

[0112] The cost function is expressed as:

[0113] ;

[0114] ;

[0115] Where J is the cost function, N p To predict the time domain length, f t+h and p t+h The link center frequency and transmit power at the h-th prediction time are respectively, u t+h Let h be the control vector at the h-th prediction time. As a risk penalty item, w B w U w P These are the weighting coefficients for the risk penalty, the control quantity smoothing penalty, and the energy consumption penalty, respectively.

[0116] The following describes a substation electromagnetic interference early warning and suppression control device provided in the embodiments of this application. The substation electromagnetic interference early warning and suppression control device described below can be referred to in correspondence with the substation electromagnetic interference early warning and suppression control method described above.

[0117] See Figure 2 , Figure 2This is a schematic diagram of a substation electromagnetic interference early warning and suppression control device disclosed in an embodiment of this application.

[0118] like Figure 2 As shown, the substation electromagnetic interference early warning and suppression control device may include:

[0119] The sampling and denoising unit 110 is used to acquire the time-domain sampling signal of the electromagnetic environment of the substation, and to denoise and reconstruct the time-domain sampling signal to obtain a denoised time-domain sequence.

[0120] The frequency domain transformation unit 120 is used to perform frequency domain transformation on the denoised time domain sequence to obtain a complex spectrum, and calculate the power spectrum based on the complex spectrum;

[0121] The state sequence unit 130 is used to identify the main interference frequency band based on the power spectrum and a preset frequency band determination threshold, extract the frequency band boundary and the energy accumulation intensity within the frequency band of the main interference frequency band, and construct an interference state sequence from the frequency band boundary and the energy accumulation intensity within the frequency band.

[0122] The timing prediction unit 140 is used to input the interference state sequence within the historical window into the timing prediction model to predict the interference intensity prediction value and frequency band boundary prediction value for multiple future time windows.

[0123] The early warning determination unit 150 is used to determine the early warning based on the predicted value of the interference intensity and the frequency domain overlap ratio between the predicted value of the frequency band boundary and the current communication link operating frequency band, and output the early warning trigger result.

[0124] The suppression and adjustment unit 160 is used to construct a model predictive control optimization problem by taking the frequency band boundary prediction value as a priori constraint, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints. The first step control quantity is extracted from the optimal control sequence obtained by optimization and executed to realize the interference suppression and adjustment of the communication link.

[0125] As can be seen from the above technical solution, the substation electromagnetic interference early warning and suppression control method and related equipment provided in this application first acquire the electromagnetic environment time-domain sampling signal, obtain the power spectrum through denoising reconstruction and frequency domain transformation, identify the main interference frequency bands, and extract their frequency band boundaries and energy accumulation intensity to construct an interference state sequence; then, input the state sequence within the historical window into the time-series prediction model to predict the interference intensity and frequency band boundaries of future multi-time windows, and make early warning judgments based on the frequency domain overlap ratio between the predicted boundaries and the current communication operating frequency band; finally, use the predicted value of the frequency band boundary as a priori constraint, establish a model predictive control optimization problem under the constraints of available channels and transmit power, solve the optimal center frequency and transmit power sequence in a rolling manner, extract the first step control quantity from the optimization result for execution, and realize dynamic adjustment of interference suppression. Compared with the prior art, its beneficial effects are:

[0126] First, by using the interference state sequence and time-series prediction model, the transient and periodic interference trends of substations can be perceived in advance. The early warning judgment provides ample prediction time for communication parameter adjustment, fundamentally solving the problem of switching lag and communication interruption caused by passive response based solely on the current interference, and enabling the control to shift from passive response to active avoidance.

[0127] Second, the center frequency and transmit power are incorporated into the model prediction control framework for joint optimization. While avoiding interference frequency bands, the power is adaptively adjusted according to the predicted intensity, overcoming the limitations of existing methods that only adjust the frequency or fix the power. When the predicted interference is weak, the power is automatically reduced to save energy, and when the interference is enhanced, the power is moderately increased to ensure link reliability, achieving the optimal synergy between anti-interference performance and energy efficiency.

[0128] Third, by employing a rolling time-domain optimization mechanism that executes only the first step of the control input, the method seeks the globally optimal solution while continuously satisfying available channel and power constraints. This effectively suppresses frequent channel jumps and suboptimal selections caused by threshold switching, significantly improving spectrum utilization and communication link stability. This method organically integrates prediction, early warning, and optimal control, providing a highly reliable and intelligent active interference suppression method for wireless communication in strong electromagnetic environments in substations.

[0129] Optionally, the step of acquiring the time-domain sampling signal of the substation's electromagnetic environment and denoising and reconstructing the time-domain sampling signal to obtain a denoised time-domain sequence includes:

[0130] Continuous signals of the electromagnetic environment of the substation are collected within a fixed time window, and the continuous signals are discretized to form a time-domain sampling sequence.

[0131] The time-domain sampled sequence is denoised using a wavelet threshold denoising operator. The wavelet threshold denoising operator decomposes the signal into approximation coefficients and detail coefficients through wavelet decomposition, performs threshold quantization on the detail coefficients, and then outputs a denoised time-domain sequence through wavelet reconstruction.

[0132] Optionally, the step of identifying the main interference frequency bands based on the power spectrum and a preset frequency band determination threshold, and extracting the frequency band boundaries and energy accumulation intensity within the main interference frequency bands, includes:

[0133] The power value of each frequency point in the power spectrum is compared with the frequency band determination threshold. The minimum and maximum frequencies in the set of frequency points whose power values ​​exceed the frequency band determination threshold are respectively used as the lower limit and upper limit of the main interference frequency band.

[0134] When multiple discontinuous sets of frequency points exceeding the threshold are detected, the set with the largest frequency band width or the largest energy accumulation intensity within the frequency band is selected as the main interference frequency band.

[0135] The power values ​​of all frequency points within the main interference frequency band are summed to obtain the energy accumulation intensity within the frequency band.

[0136] Optionally, the step of determining the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and outputting the early warning trigger result, includes:

[0137] Based on the predicted frequency band boundary value and the upper and lower limits of the current communication link operating frequency band, the frequency domain overlap ratio is calculated. The frequency domain overlap ratio is the proportion of the intersection length of the predicted interference frequency band and the link operating frequency band to the bandwidth of the link operating frequency band.

[0138] When the frequency domain overlap ratio exceeds a preset overlap ratio threshold, and / or when the predicted interference intensity exceeds a preset intensity threshold, an early warning is triggered and an early warning trigger result is output, wherein the early warning trigger result includes progressive early warning levels or a single early warning signal.

[0139] Optionally, the step of using the predicted value of the frequency band boundary as a priori constraint to construct a model predictive control optimization problem, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints, includes:

[0140] In the prediction time domain, the center frequency and transmission power of the communication link at each prediction time are used as decision variables to construct a cost function, which includes a risk penalty term, a control quantity smoothing penalty term, and an energy consumption penalty term.

[0141] The risk penalty term is calculated based on the overlap ratio between the frequency band boundary prediction value and the link operating frequency band corresponding to the current decision variable, and is used to characterize the predicted interference risk to the communication link.

[0142] The control quantity smoothing penalty term is calculated based on the square of the L2 norm of the difference between control quantity vectors at adjacent time points, and is used to suppress frequent jumps in the center frequency and transmit power of the communication link.

[0143] The energy consumption penalty term is calculated based on the transmit power at the current moment and is used to characterize the energy consumption level of the link;

[0144] Under the constraints of available channels and power, find the optimal control sequence that minimizes the cost function;

[0145] The first step control quantity is extracted from the optimal control sequence as the execution control quantity at the current moment, and the process of constructing the cost function and solving the optimal control sequence is repeated in the next moment in a rolling time domain manner.

[0146] Optional, also includes:

[0147] When there are multiple sets of frequency points exceeding the threshold, the first interference frequency band corresponding to the main interference source and at least one second interference frequency band corresponding to the secondary interference source are identified from the multiple sets of frequency points exceeding the threshold based on the frequency band width and / or the energy accumulation intensity within the frequency band of each set of frequency points exceeding the threshold.

[0148] Extract the first interference state sequence of the first interference frequency band and the second interference state sequence of the second interference frequency band respectively;

[0149] The first interference state sequence and the second interference state sequence are respectively input into the time series prediction model to obtain the first prediction result and the second prediction result.

[0150] The overall overlap ratio is calculated based on the first prediction result and the second prediction result, and the early warning determination and interference suppression adjustment of the communication link are performed based on the overall overlap ratio.

[0151] Optionally, the cost function is expressed as:

[0152] ;

[0153] ;

[0154] Where J is the cost function, N p To predict the time domain length, f t+h and p t+h The link center frequency and transmit power at the h-th prediction time are respectively, u t+hLet h be the control vector at the h-th prediction time. As a risk penalty item, w B w U w P These are the weighting coefficients for the risk penalty, the control quantity smoothing penalty, and the energy consumption penalty, respectively.

[0155] The substation electromagnetic interference early warning and suppression control device provided in this application embodiment can be applied to substation electromagnetic interference early warning and suppression control equipment. Figure 3 The hardware structure block diagram of the electromagnetic interference early warning and suppression control equipment for substations is shown. (Refer to...) Figure 3 The hardware structure of the electromagnetic interference early warning and suppression control equipment for substations may include: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4.

[0156] In this embodiment of the application, the number of processor 1, communication interface 2, memory 3, and communication bus 4 is at least one, and processor 1, communication interface 2, and memory 3 communicate with each other through communication bus 4;

[0157] Processor 1 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0158] Memory 3 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device;

[0159] The memory stores a program, which the processor can call. The program is used for:

[0160] The time-domain sampling signal of the electromagnetic environment of the substation is acquired, and the time-domain sampling signal is denoised and reconstructed to obtain a denoised time-domain sequence;

[0161] The denoised time-domain sequence is subjected to frequency domain transformation to obtain a complex spectrum, and the power spectrum is calculated based on the complex spectrum;

[0162] Based on the power spectrum and the preset frequency band determination threshold, the main interference frequency bands are identified, the frequency band boundaries and the energy accumulation intensity within the frequency bands are extracted, and an interference state sequence is constructed from the frequency band boundaries and the energy accumulation intensity within the frequency bands.

[0163] The interference state sequence within the historical window is input into the time-series prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows.

[0164] Based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, an early warning determination is made and an early warning trigger result is output.

[0165] Using the predicted value of the frequency band boundary as a priori constraint, a model predictive control optimization problem is constructed. Under the constraints of available channel and power, the center frequency and transmit power of the communication link are optimized in a rolling manner. The first step of the control quantity is extracted from the optimal control sequence obtained by optimization and executed to achieve interference suppression adjustment of the communication link.

[0166] Optionally, the refined and extended functions of the program can be referred to the above description.

[0167] This application embodiment also provides a readable storage medium that can store a program suitable for execution by a processor, the program being used for:

[0168] The time-domain sampling signal of the electromagnetic environment of the substation is acquired, and the time-domain sampling signal is denoised and reconstructed to obtain a denoised time-domain sequence;

[0169] The denoised time-domain sequence is subjected to frequency domain transformation to obtain a complex spectrum, and the power spectrum is calculated based on the complex spectrum;

[0170] Based on the power spectrum and the preset frequency band determination threshold, the main interference frequency bands are identified, the frequency band boundaries and the energy accumulation intensity within the frequency bands are extracted, and an interference state sequence is constructed from the frequency band boundaries and the energy accumulation intensity within the frequency bands.

[0171] The interference state sequence within the historical window is input into the time-series prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows.

[0172] Based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, an early warning determination is made and an early warning trigger result is output.

[0173] Using the predicted value of the frequency band boundary as a priori constraint, a model predictive control optimization problem is constructed. Under the constraints of available channel and power, the center frequency and transmit power of the communication link are optimized in a rolling manner. The first step of the control quantity is extracted from the optimal control sequence obtained by optimization and executed to achieve interference suppression adjustment of the communication link.

[0174] Optionally, the refined and extended functions of the program can be referred to the above description.

[0175] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0176] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0177] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for early warning and suppression of electromagnetic interference in substations, characterized in that, include: The time-domain sampling signal of the electromagnetic environment of the substation is acquired, and the time-domain sampling signal is denoised and reconstructed to obtain a denoised time-domain sequence; The denoised time-domain sequence is subjected to frequency domain transformation to obtain a complex spectrum, and the power spectrum is calculated based on the complex spectrum; Based on the power spectrum and the preset frequency band determination threshold, the main interference frequency bands are identified, the frequency band boundaries and the energy accumulation intensity within the frequency bands are extracted, and an interference state sequence is constructed from the frequency band boundaries and the energy accumulation intensity within the frequency bands. The interference state sequence within the historical window is input into the time-series prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows. Based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, an early warning determination is made and an early warning trigger result is output. Using the predicted value of the frequency band boundary as a priori constraint, a model predictive control optimization problem is constructed. Under the constraints of available channel and power, the center frequency and transmit power of the communication link are optimized in a rolling manner. The first step of the control quantity is extracted from the optimal control sequence obtained by optimization and executed to achieve interference suppression adjustment of the communication link.

2. The method according to claim 1, characterized in that, The process of acquiring the time-domain sampling signal of the substation's electromagnetic environment, and denoising and reconstructing the time-domain sampling signal to obtain a denoised time-domain sequence includes: Continuous signals of the electromagnetic environment of the substation are collected within a fixed time window, and the continuous signals are discretized to form a time-domain sampling sequence. The time-domain sampled sequence is denoised using a wavelet threshold denoising operator. The wavelet threshold denoising operator decomposes the signal into approximation coefficients and detail coefficients through wavelet decomposition, performs threshold quantization on the detail coefficients, and then outputs a denoised time-domain sequence through wavelet reconstruction.

3. The method according to claim 1, characterized in that, The step of identifying the main interference frequency bands based on the power spectrum and a preset frequency band determination threshold, and extracting the frequency band boundaries and energy accumulation intensity within the main interference frequency bands, includes: The power value of each frequency point in the power spectrum is compared with the frequency band determination threshold. The minimum and maximum frequencies in the set of frequency points whose power values ​​exceed the frequency band determination threshold are respectively used as the lower limit and upper limit of the main interference frequency band. When multiple discontinuous sets of frequency points exceeding the threshold are detected, the set with the largest frequency band width or the largest energy accumulation intensity within the frequency band is selected as the main interference frequency band. The power values ​​of all frequency points within the main interference frequency band are summed to obtain the energy accumulation intensity within the frequency band.

4. The method according to claim 1, characterized in that, The step of determining the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and outputting the early warning trigger result, includes: Based on the predicted frequency band boundary value and the upper and lower limits of the current communication link operating frequency band, the frequency domain overlap ratio is calculated. The frequency domain overlap ratio is the proportion of the intersection length of the predicted interference frequency band and the link operating frequency band to the bandwidth of the link operating frequency band. When the frequency domain overlap ratio exceeds a preset overlap ratio threshold, and / or when the predicted interference intensity exceeds a preset intensity threshold, an early warning is triggered and an early warning trigger result is output, wherein the early warning trigger result includes progressive early warning levels or a single early warning signal.

5. The method according to claim 1, characterized in that, The step of using the predicted value of the frequency band boundary as a priori constraint to construct a model predictive control optimization problem, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints, includes: In the prediction time domain, the center frequency and transmission power of the communication link at each prediction time are used as decision variables to construct a cost function, which includes a risk penalty term, a control quantity smoothing penalty term, and an energy consumption penalty term. The risk penalty term is calculated based on the overlap ratio between the frequency band boundary prediction value and the link operating frequency band corresponding to the current decision variable, and is used to characterize the predicted interference risk to the communication link. The control quantity smoothing penalty term is calculated based on the square of the L2 norm of the difference between control quantity vectors at adjacent time points, and is used to suppress frequent jumps in the center frequency and transmit power of the communication link. The energy consumption penalty term is calculated based on the transmit power at the current moment and is used to characterize the energy consumption level of the link; Under the constraints of available channels and power, find the optimal control sequence that minimizes the cost function; The first step control quantity is extracted from the optimal control sequence as the execution control quantity at the current moment, and the process of constructing the cost function and solving the optimal control sequence is repeated in the next moment in a rolling time domain manner.

6. The method according to claim 3, characterized in that, Also includes: When there are multiple sets of frequency points exceeding the threshold, the first interference frequency band corresponding to the main interference source and at least one second interference frequency band corresponding to the secondary interference source are identified from the multiple sets of frequency points exceeding the threshold based on the frequency band width and / or the energy accumulation intensity within the frequency band of each set of frequency points exceeding the threshold. Extract the first interference state sequence of the first interference frequency band and the second interference state sequence of the second interference frequency band respectively; The first interference state sequence and the second interference state sequence are respectively input into the time series prediction model to obtain the first prediction result and the second prediction result. The overall overlap ratio is calculated based on the first prediction result and the second prediction result, and the early warning determination and interference suppression adjustment of the communication link are performed based on the overall overlap ratio.

7. The method according to claim 5, characterized in that, The cost function is expressed as: ； ； Where J is the cost function, N p To predict the time domain length, f t+h and p t+h The link center frequency and transmit power at the h-th prediction time are respectively, u t+h Let h be the control vector at the h-th prediction time. As a risk penalty item, w B w U w P These are the weighting coefficients for the risk penalty, the control quantity smoothing penalty, and the energy consumption penalty, respectively.

8. A substation electromagnetic interference early warning and suppression control device, characterized in that, include: The sampling and denoising unit is used to acquire the time-domain sampling signal of the electromagnetic environment of the substation, and to denoise and reconstruct the time-domain sampling signal to obtain a denoised time-domain sequence. The frequency domain transformation unit is used to perform frequency domain transformation on the denoised time domain sequence to obtain a complex spectrum, and calculate the power spectrum based on the complex spectrum; The state sequence unit is used to identify the main interference frequency bands based on the power spectrum and a preset frequency band determination threshold, extract the frequency band boundaries and the energy accumulation intensity within the frequency bands of the main interference frequency bands, and construct an interference state sequence from the frequency band boundaries and the energy accumulation intensity within the frequency bands. The timing prediction unit is used to input the interference state sequence within the historical window into the timing prediction model to predict the interference intensity and frequency band boundary values ​​for multiple future time windows. The early warning determination unit is used to determine the early warning based on the predicted interference intensity value and the frequency domain overlap ratio between the predicted frequency band boundary value and the current communication link operating frequency band, and output the early warning trigger result. The suppression and adjustment unit is used to construct a model predictive control optimization problem by taking the frequency band boundary prediction value as a priori constraint, and to perform rolling optimization of the center frequency and transmit power of the communication link under available channel constraints and power constraints. The first step control quantity is extracted from the optimal control sequence obtained by optimization and executed to realize the interference suppression and adjustment of the communication link.

9. A substation electromagnetic interference early warning and suppression control device, characterized in that, Including memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the substation electromagnetic interference early warning and suppression control method as described in any one of claims 1-7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements each step of the substation electromagnetic interference early warning and suppression control method as described in any one of claims 1-7.

Images