A variable pressure regulation loss compensation control method and system

Abstract

The application discloses a variable-voltage regulation loss compensation control method and system, relates to the variable-voltage control technical field of a power system, and has the technical features that by matching the fluctuation propagation characteristics of different loss links, different collection intervals are set, shorter intervals are used when collecting loss data with a faster response speed, and transient fluctuation information in a load mutation process can be completely captured; longer intervals are used when collecting slowly-changing loss data such as environment correction loss, and the generation of redundant data is effectively reduced. Secondly, the variable-voltage loss is no longer regarded as a static value generated at the same time, but the propagation path between different losses is identified, the delay time and phase shift of the loss propagation in the equipment are obtained by inversion, the compensation signal is calibrated on this basis, the output compensation signal can be aligned with the loss fluctuation actually arriving at the compensation point in time and phase, and the problems of compensation lag and overcompensation existing in the traditional compensation mode are solved.

Description

Technical Field

[0001] This invention relates to the field of power system transformer control technology, and in particular to a transformer regulation loss compensation control method and system. Background Technology

[0002] As a core node in the power transmission and distribution network, the energy consumption of transformer regulation systems directly affects the power supply efficiency of the entire power system. In large-scale power consumption scenarios such as industrial production and commercial power supply, the cumulative losses generated by the long-term continuous operation of multiple transformer devices have become an indispensable part of the power system's energy consumption. Therefore, improving the loss compensation accuracy of transformer regulation systems has always been a key focus of the industry. Currently, the industry generally adopts a technical approach that combines offline calibration with real-time compensation, as follows: First, in a standard laboratory environment, the static loss characteristics of the transformer equipment under different load rates and ambient temperatures are tested to generate corresponding loss-load comparison data tables. During actual operation, the basic operating parameters such as input and output voltage and current of the equipment are acquired at fixed and uniform acquisition intervals. The theoretical total loss value under the current operating conditions is calculated by looking up the table, and then a single-amplitude compensation signal is output to the actuator for adjustment.

[0003] However, during the implementation of the above technical solution, at least the following technical problems were discovered: Firstly, for electromagnetic losses such as winding copper losses directly caused by current changes, their fluctuation response speed is extremely fast. Fixed acquisition intervals often cannot fully capture the high-frequency transient changes during load changes, leading to systematic deviations in loss calculations. On the other hand, for slow-varying losses such as environmental correction losses caused by heat exchange between equipment and the environment, excessively high acquisition frequencies will generate a large amount of redundant data, occupying the system's limited computing and storage resources.

[0004] Secondly, different types of losses occur in different parts of the equipment. Their propagation process in structures such as the core, windings, and shell has obvious time delays and phase distortions. The compensation signals output by existing technologies do not take these propagation characteristics into account, which leads to the compensation signal and the actual loss fluctuations acting on the system being out of sync in time and phase. This inevitably results in compensation lag or overcompensation, making it difficult to achieve the expected compensation effect. Therefore, we propose a transformer regulation loss compensation control method and system. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a variable voltage regulation loss compensation control method and system, which solves the technical problems of high-frequency loss information loss and asynchronous compensation in existing technologies.

[0006] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A method for compensating losses in transformer regulation, comprising the following steps: Based on the fluctuation propagation characteristics of different loss links within the transformer regulation system, differentiated acquisition intervals are set to collect multi-dimensional operating data and generate a continuous time series with unified timestamps. The instantaneous loss value of each stage is calculated step by step based on the continuous time series. High-frequency fluctuation characteristic components covering the main energy range of transient loss fluctuations are extracted from each loss series simultaneously to generate a loss fluctuation characteristic time series. Using the time series of the winding copper loss fluctuation characteristics with the fastest response speed as the source fluctuation sequence, the propagation path of the loss fluctuation of each other link relative to the source sequence is identified, and the three types of propagation parameters corresponding to each path are obtained by inversion: propagation delay time, fluctuation attenuation coefficient and phase offset. Based on the three types of propagation parameters for all propagation paths, a three-dimensional loss fluctuation compensation matrix containing time, amplitude, and phase dimensions is constructed. Based on the three-dimensional loss fluctuation compensation matrix, the basic compensation amount of each link is calculated and the corresponding time and phase information is bound to generate dynamic compensation parameters for each link. Apply phase pre-calibration processing, which is opposite to the propagation phase offset, to the dynamic compensation parameters of each sub-stage; According to the order of propagation delay time of each propagation path from smallest to largest, the phase pre-calibrated compensation control signals are output to the transformer regulation system in sequence; Within a preset fixed time interval after all compensation control signals have fully taken effect, multi-dimensional operating data is re-acquired and the actual total loss value is calculated. The actual total loss value is compared with the preset target loss value to obtain the loss deviation value. It is then determined whether the loss deviation value is less than the preset allowable deviation threshold. If it is less than the preset allowable deviation threshold, the current three-dimensional loss fluctuation compensation matrix and phase pre-calibration compensation strategy are maintained. Otherwise, the process returns to re-identify the loss fluctuation propagation path and invert the propagation parameters.

[0007] Preferably, the step of setting differentiated acquisition intervals based on loss fluctuation propagation characteristics is as follows: Based on the electromagnetic wave propagation characteristics corresponding to winding copper losses, a first type of acquisition interval is set to acquire instantaneous electrical operating parameters of the transformer regulation system. To address the propagation characteristics of hysteresis fluctuations corresponding to iron core losses, a second type of acquisition interval is set to acquire instantaneous load operating parameters of the transformer regulation system. To address the propagation characteristics of heat exchange fluctuations corresponding to environmental correction losses, a third type of acquisition interval is set to collect environmental impact parameters of the transformer regulation system. To address the propagation characteristics of structural aging fluctuations corresponding to additional losses, a fourth type of acquisition interval is set to collect equipment status parameters of the transformer regulation system. All collected data are globally aligned using timestamps of uniform precision to generate continuous time series data for each dimension.

[0008] Preferably, the step of extracting high-frequency fluctuation feature components is as follows: The instantaneous values ​​of winding copper loss, core iron loss, additional loss and environmental correction loss are calculated at each time step to generate the continuous loss time series corresponding to each link. High-pass filtering with uniform parameters is used to extract high-frequency fluctuation components covering the main energy range of transient loss fluctuations from the loss time series of each stage. The extracted high-frequency fluctuation components of each stage are re-aligned according to the timestamps to generate a time series of loss fluctuation characteristics for each stage.

[0009] Preferably, the steps for retrieving the inversion propagation parameters are as follows: The time series of winding copper loss fluctuation characteristics within a preset time period is taken as the source fluctuation series, and the time series of loss fluctuation characteristics of other links within the same time period are taken as the target fluctuation series. Cross-correlation analysis was performed on the source wave sequence and each target wave sequence to calculate the cross-correlation coefficient at different time offsets; The time offset corresponding to the maximum cross-correlation coefficient is taken as the propagation delay time of the loss fluctuation in this link, and the amplitude ratio of the source sequence to the corresponding target sequence is taken as the fluctuation attenuation coefficient of the propagation path. Spectral analysis is performed on the source wave sequence and the target wave sequence at the corresponding maximum cross-correlation coefficient, and the phase difference of the main frequency component is extracted as the propagation phase offset.

[0010] Preferably, the step of constructing the three-dimensional loss fluctuation compensation matrix is ​​as follows: A two-dimensional basic matrix is ​​constructed, with the matrix rows corresponding to four loss elements: winding copper loss, core iron loss, additional loss, and environmental correction loss, and the matrix columns corresponding to three types of propagation parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. Set the propagation parameters corresponding to the winding copper loss as the reference values ​​and fill them into the first row of the matrix; The propagation parameters corresponding to core iron loss, additional loss, and environmental correction loss are sequentially filled into the corresponding column positions of the corresponding row of the matrix; All parameters within the matrix are normalized, and parameters of different dimensions are uniformly mapped to a preset numerical range to obtain a three-dimensional loss fluctuation compensation matrix containing spatiotemporal phase information.

[0011] Preferably, the step of generating the phase pre-calibration compensation control signal is as follows: Extract the propagation delay time, fluctuation attenuation coefficient and phase offset corresponding to each link from the three-dimensional loss fluctuation compensation matrix; By combining the instantaneous loss value of each link at the current moment with the corresponding fluctuation attenuation coefficient, the basic compensation amount of each link is calculated; The basic compensation amount is bound to the corresponding propagation delay time and phase offset to generate dynamic compensation parameters for each stage with time-series phase information. Phase modulation is applied to the dynamic compensation parameter of each sub-stage, and the modulation angle is equal in magnitude and opposite in direction to the propagation phase offset corresponding to that parameter; The pre-calibrated digital compensation parameters are converted into analog control signals recognizable by the transformer regulation system, sorted by propagation delay time from smallest to largest, and a compensation signal output queue is generated.

[0012] A transformer regulation loss compensation control system, the control system comprising: The loss fluctuation adaptation acquisition module sets differentiated acquisition intervals based on the fluctuation propagation characteristics of different loss stages, collects multi-dimensional operating data, and generates a continuous time series with unified timestamps. The loss fluctuation feature extraction module calculates the instantaneous loss value of each stage based on the continuous time series, extracts high-frequency fluctuation feature components covering the main energy range of transient loss fluctuation, and generates a loss fluctuation feature time series. The propagation parameter inversion module uses the time series of the winding copper loss fluctuation characteristics as the source fluctuation sequence to identify the propagation path of each other and invert three types of propagation parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. The three-dimensional compensation matrix construction module constructs a three-dimensional loss fluctuation compensation matrix containing time, amplitude, and phase dimensions based on the three types of propagation parameters for all propagation paths. The phase pre-calibration compensation module calculates the dynamic compensation parameters of each component based on the three-dimensional compensation matrix and applies a phase pre-calibration process that is opposite to the propagation phase offset. The timing output module outputs the phase pre-calibrated compensation control signals to the transformer regulation system in ascending order of propagation delay time; The compensation effect verification module re-collects multi-dimensional operating data and calculates the actual total loss value within a preset fixed time interval after all compensation control signals have fully taken effect. The deviation judgment and feedback module compares the actual total loss value with the preset target loss value to obtain the loss deviation value, judges whether the deviation is within the allowable range, and maintains the current strategy or triggers the propagation parameters to be re-inverted based on the judgment result.

[0013] Preferably, the loss fluctuation adaptation acquisition module includes: The acquisition interval configuration unit sets four different acquisition intervals according to the fluctuation propagation characteristics of different loss links; The multi-dimensional data acquisition unit collects instantaneous electrical operation parameters, instantaneous load operation parameters, environmental impact parameters, and equipment status parameters according to the corresponding acquisition intervals. The time series alignment unit globally aligns all collected data according to timestamps of uniform precision, generating continuous time series of data in each dimension.

[0014] Preferably, the propagation parameter inversion module includes: The sequence extraction unit extracts the source fluctuation sequence within a preset time period and the target fluctuation sequences within the same time period. The cross-correlation analysis unit calculates the cross-correlation coefficients between the source sequence and each target sequence at different time offsets; The propagation parameter calculation unit determines the propagation delay time based on the maximum cross-correlation coefficient, the fluctuation attenuation coefficient based on the sequence amplitude ratio, and the phase offset based on the phase difference of the main frequency components.

[0015] Preferably, the phase pre-calibration compensation module includes: The parameter extraction unit extracts three types of propagation parameters corresponding to each stage from the three-dimensional loss fluctuation compensation matrix. The basic compensation calculation unit calculates the basic compensation amount for each link by combining the instantaneous loss value and fluctuation attenuation coefficient of each link at the current moment. The phase pre-calibration unit binds the basic compensation amount to the timing phase information and applies a phase modulation that is opposite to the propagation phase offset. The signal conversion and sorting unit converts the phase-precalibrated digital compensation parameters into analog control signals and sorts them according to the propagation delay time to generate an output queue.

[0016] (III) Beneficial Effects By setting differentiated acquisition intervals to match the fluctuation propagation characteristics of different loss links, a shorter interval is used when acquiring loss data with a fast response speed, such as winding copper loss, which can completely capture transient fluctuation information during load change. A longer interval is used when acquiring loss data with a slow change, such as environmental correction loss, which effectively reduces the generation of redundant data and allows system resources to be concentrated on processing high-frequency loss information that has a greater impact on the compensation effect. Secondly, transformer loss is no longer regarded as a static value generated at the same moment. Instead, by identifying the propagation path between different losses, the delay time and phase offset of loss propagation inside the equipment are obtained. Based on this, the compensation signal is pre-calibrated in phase, so that the output compensation signal can be aligned with the actual loss fluctuation reaching the compensation point in time and phase, solving the problems of compensation lag and overcompensation in traditional compensation methods. Furthermore, by constructing a three-dimensional loss fluctuation compensation matrix that includes three types of information: time, amplitude, and phase, a more comprehensive and structured description of loss fluctuations in different stages can be achieved. This provides an accurate basis for calculating compensation parameters for each stage, avoiding the problem of insufficient accuracy caused by single total loss compensation. Moreover, the propagation parameters and compensation strategies can be dynamically adjusted according to the compensation effect during actual operation, ensuring that the system maintains stable compensation performance under different operating conditions such as load fluctuations and environmental changes. Attached Figure Description

[0017] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0018] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a trend diagram of the total transformer loss in this invention. Detailed Implementation

[0019] This application provides a transformer regulation loss compensation control method and system to solve the technical problems of high-frequency loss information loss and compensation asynchrony in the prior art.

[0020] Example 1: As Figure 1 As shown, taking the S11 type oil-immersed distribution transformer with a capacity of 10kV / 0.4kV and 1000kVA in an industrial plant as an example, the transformer regulation loss compensation control system adopted consists of a data acquisition unit, a main controller, a compensation execution mechanism, and a human-machine interaction unit. The data acquisition unit installs voltage transformers and current transformers on the high-voltage and low-voltage sides of the transformer, respectively. During installation, the transformers are ensured to be coaxial with the busbar and have uniform air gaps to avoid additional errors. Six resistance temperature sensors are pre-embedded at the ends (50mm from the upper end face) and middle of the three-phase windings, as well as at the upper and lower yokes of the core. After pre-embedding, they undergo vacuum impregnation to ensure tight contact and reliable insulation with the measured parts. One set of ambient temperature and humidity sensors is installed on the southwest-facing and northeast-shaded sides of the transformer casing, 1.5m above the ground to avoid direct sunlight and rain. An altitude sensor is installed inside the control cabinet, away from heat-generating components. All sensors have built-in hardware timestamp modules, directly adding GPS-synchronized timestamps during data acquisition instead of being added by the main controller upon reception, eliminating the impact of RS485 bus transmission delay on time accuracy. All signal cables use shielded twisted-pair cables with single-ended shielding. Grounding with a grounding resistance of less than 4Ω is required. The cable must be laid separately from the power cable at a distance of at least 50cm and shielded by galvanized steel conduit. The RS485 bus uses a daisy-chain topology, with a 120Ω matching resistor connected to the terminal and gold-plated terminals crimped with conductive paste to ensure reliable contact. Sensor calibration employs the standard current transformer method to calibrate voltage and current transformers. Five calibration points are selected within the 0–120% rated current range to measure ratio and phase errors, with a calibration accuracy of at least 0.1%. The temperature sensor is calibrated using a constant temperature bath at five points: 0℃, 25℃, 50℃, 75℃, and 100℃. A standard temperature and humidity chamber and a barometer are used to calibrate the environmental and altitude sensors, respectively. During implementation, a 5% amplitude 50Hz power frequency interference was observed in the current transformer data. This was found to be caused by electromagnetic coupling from the power cable and resolved by adding a second-order Butterworth low-pass filter and re-laying the shielded steel conduit.

[0021] The main controller uses an industrial-grade microcontroller with a built-in high-precision temperature-compensated crystal oscillator and global time synchronization via a GPS module. The power input terminal is connected to an EMI power filter, a varistor, and a TVS diode to suppress power grid conducted interference and operational overvoltage. The GPS antenna is installed on the top of the control cabinet in an unobstructed position to ensure reception of at least 4 satellite signals and synchronizes the local clock every 10 minutes.

[0022] The compensation actuator adopts a thyristor zero-crossing switching reactive power compensation device with a compensation capacity of 0-100kVar divided into 10 groups of 10kVar self-healing low-voltage parallel capacitors. A series current-limiting reactor suppresses the inrush current. It can be used in two modes: phase compensation and three-phase common compensation. It can connect the capacitor when the voltage crosses zero and disconnect the capacitor when the current crosses zero, limiting the inrush current to within 1.5 times the rated current and avoiding the generation of operating overvoltage.

[0023] The human-machine interface unit is equipped with an industrial resistive touch screen to display operating status, real-time loss data, compensation parameters and fault information. It supports online parameter setting and historical data query. The historical data can be stored in a cyclic overwrite mode to save at least one year of minute-level operating data and all fault event records.

[0024] During use, sensor calibration and parameter self-test are first completed. Calibration includes calibration of the ratio error and phase error of voltage and current transformers, and calibration of the zero point and range of temperature sensors. Self-test includes sensor communication status, main controller memory status, compensation device switching status, etc. After confirming that all equipment is working properly, the factory-preset basic parameters are automatically loaded, including core loss coefficients k1 and k2, additional loss coefficient reference values ​​k30 and k40, environmental correction coefficient reference values ​​a0, b0, and c0, as well as operating parameters such as filter cutoff frequency and allowable deviation threshold, and then the automatic compensation control process is entered.

[0025] The main controller first automatically configures the acquisition intervals for four types of data based on the wave propagation characteristics of different loss stages. Existing technologies generally use a uniform acquisition interval, which cannot simultaneously capture high-frequency loss information and make rational resource utilization. However, based on the physical nature of loss, we treat different types of loss as waves propagating at different speeds within the equipment, and match the corresponding acquisition interval according to the wave propagation speed: For electromagnetic waves corresponding to winding copper losses, their propagation speed in the copper conductor is close to the speed of light, and the current change response time during load abrupt changes is in the millisecond range. Therefore, a 1-millisecond acquisition interval is set to continuously acquire the instantaneous values ​​of the three-phase input voltage, output voltage, input current, and output current of the high-voltage and low-voltage sides (A, B, and C phases), with each sampling point simultaneously recording the corresponding GPS timestamp; For hysteresis waves corresponding to iron core losses, their propagation speed in the silicon steel laminations is approximately 3000 m / s, and changes in magnetic flux density require magnetic domain rotation and wall displacement processes, with a response time in the tens of milliseconds range. A 10-millisecond acquisition interval is set to continuously collect three-phase power factor, real-time load power, and load power change rate. For heat exchange fluctuations corresponding to environmental correction losses, which are conducted through air and transformer oil at a speed of approximately 1 m / s, and whose temperature changes are a slow thermal equilibrium process with a response time in the hundreds of milliseconds, a 100-millisecond acquisition interval is set to continuously collect ambient temperature, ambient humidity, and altitude of the installation site. For structural aging fluctuations corresponding to additional losses, which are caused by fatigue accumulation and insulation deterioration from long-term equipment operation, with a change cycle in the days or even months, a 1-minute acquisition interval is set to continuously collect the equipment's cumulative operating time, real-time three-phase winding temperature, real-time core temperature, and insulation aging level. The insulation aging level is updated quarterly by default through offline detection of transformer oil dielectric loss value and furfural content. When an abnormal increase in winding temperature or a sudden increase in losses is detected, a temporary detection request will be triggered to remind maintenance personnel to update.

[0026] The collected raw data first undergoes second-order Butterworth hardware filtering and moving average software denoising to remove 50Hz power frequency interference, high-frequency white noise, and impulse noise. Then, the time series alignment unit performs linear interpolation on the data at different sampling rates with a 1-millisecond time scale. For load data sampled for 10 milliseconds, 9 linear interpolation points are inserted between adjacent sampling points; for environmental data sampled for 100 milliseconds, 99 interpolation points are inserted; and for equipment status data sampled for 1 minute, 59,999 interpolation points are inserted. Finally, all data are unified onto the same time axis to generate a continuous time series of data in various dimensions with a timestamp accuracy of 1 millisecond.

[0027] Based on the aligned continuous time series, the main controller calculates the instantaneous loss values ​​of the four components at each time step. The winding copper loss is calculated according to Joule's law, with the formula being P. cu (t)=I(t) 2 ×R(t), where P cu I(t) is the instantaneous value of the winding copper loss at time t, in watts; I(t) is the instantaneous value of the phase current at time t, in amperes; R(t) is the phase winding resistance at time t, in ohms. Since the resistance of copper changes linearly with temperature, R(t) = R 20 ×[1+α×(T(t)-20)],R 20 The DC resistance value of the winding at 20℃ is provided by the transformer factory test report. α is the standard value of the temperature coefficient of resistance of copper, 0.00393 / ℃. T(t) is the average value of the temperature sensors at the end and middle of the phase winding at time t. By correcting the winding resistance in real time, the accuracy is improved compared with the calculation method of fixed resistance value in the existing technology.

[0028] Iron core losses consist of two parts: hysteresis loss and eddy current loss, denoted by the formula P. fe (t)=k1×f(t)×B(t) 2 +k2×f(t) 2 ×B(t) 2 In the formula, P fe (t) represents the instantaneous value of the core iron loss at time t, in watts. k1 and k2 are the hysteresis loss coefficient and eddy current loss coefficient, respectively, which are determined by the core material and lamination process and calibrated through transformer no-load test. f(t) is the power supply frequency at time t, obtained by voltage zero-crossing detection, in Hertz. B(t) is the core magnetic flux density at time t, calculated according to Faraday's law of electromagnetic induction, in Tesla. B(t) = U(t) / (4.44 × f(t) × N × S), where U(t) is the instantaneous value of the phase voltage at time t, N is the number of winding turns, and S is the core cross-sectional area, all of which are fixed parameters of the transformer.

[0029] Additional losses include winding stray losses, core stray losses, and dielectric losses. Their magnitude is directly proportional to the winding copper losses and core iron losses, and the proportionality coefficient increases with the degree of equipment aging. The formula is P. add (t)=k3(t)×P cu (t)+k4(t)×P fe (t), where P add (t) represents the instantaneous value of the additional loss at time t, in watts. k3(t) and k4(t) are the copper loss additional loss coefficient and iron loss additional loss coefficient at time t, respectively. The main controller obtains the additional loss coefficient by querying a pre-stored reference table based on the current cumulative running time of the equipment and the insulation aging level. The reference table is generated by fitting the long-term aging test data of the same type of transformer and is divided into 5 aging levels. The higher the level, the greater the additional loss coefficient.

[0030] Environmentally corrected losses are caused by the effects of ambient temperature, humidity, and altitude on heat dissipation efficiency. When the ambient temperature, humidity, or altitude increases, the transformer's heat dissipation capacity decreases, and the actual losses will be higher than the theoretical losses under standard conditions. The formula is P. env (t)=k5(t)×(P cu (t)+P fe (t)), where P env (t) represents the instantaneous value of the environmental correction loss at time t, in watts (P). cu (t)+P fe (t) represents the base loss value at time t, and k5(t) represents the environmental correction coefficient at time t, calculated using the environmental correction formula: k5(t) = a × (T) env (t)-T0)+b×H env (t)+c×H alt (t), T env (t) represents the average ambient temperature on the sunny and shady sides, T0 is the standard ambient temperature of 20℃, and H env (t) represents the ambient relative humidity, H alt (t) represents the altitude of the installation location, and a, b, and c are the temperature correction factor, humidity correction factor, and altitude correction factor, respectively.

[0031] After calculating the continuous loss time series of each stage, an adaptive sliding window high-pass filtering algorithm is used to extract high-frequency fluctuation components. The sliding window size is set to 50 milliseconds, the step size is 1 millisecond, and the filter cutoff frequency is set to 10 Hz. This frequency band covers the main energy range of transient loss fluctuations caused by load changes such as motor starting, equipment switching, and electric arc furnace operation in industrial scenarios. Components below 10 Hz are mainly slow load changes and DC bias, which have little impact on transient compensation. The algorithm automatically adjusts the filter coefficient according to the signal-to-noise ratio (SNR): when the SNR is greater than 30 dB, the filter coefficient is set to 0.1 to retain more high-frequency details; when the SNR is between 10 and 30 dB, the filter coefficient is set to 0.3 to balance denoising effect and detail preservation; and when the SNR is less than 10 dB, the filter coefficient is set to 0.5 to enhance denoising capability. Finally, the extracted high-frequency fluctuation components of each stage are re-aligned according to the timestamp to generate a loss fluctuation feature time series for subsequent propagation parameter inversion.

[0032] The winding copper loss fluctuation characteristic time series, which has the fastest response speed, is used as the source fluctuation series. This is because the winding copper loss is directly generated by the current and is the fastest responding of all electromagnetic losses, able to reflect load changes earliest, making it the most suitable time reference. The module extracts the source fluctuation series data within the most recent 2 seconds for analysis, and simultaneously extracts the loss fluctuation characteristic time series of core iron loss, additional loss, and environmental correction loss within the same time period as the target fluctuation series. The selection of a 2-second extraction time is optimized. Too short a time will result in insufficient sample size and reduced analysis accuracy, while too long a time will introduce too much historical data and affect real-time performance. For high-frequency fluctuations above 10Hz, 2 seconds containing 2000 sampling points is sufficient for accurate signal analysis.

[0033] First, all sequences are normalized to map their amplitudes to the 0-1 interval, eliminating the influence of amplitude differences between different loss stages on the analysis results. The normalization formula is x'(t)=(x(t)-x min ) / (x max -x min ), where x'(t) is the normalized sequence value, x(t) is the original sequence value, and x min and x max These are the minimum and maximum values ​​of the sequence, respectively.

[0034] Then, a fast cross-correlation analysis is performed on the source sequence and each target sequence, using FFT to reduce the computational complexity from O(N) to O(N). 2 The computation time is reduced to O(NlogN), and for a sequence of length 2000, the computation time is shortened from approximately 200 milliseconds to approximately 10 milliseconds. The formula for the cross-correlation function is... Where R(τ) is the cross-correlation coefficient at a time offset of τ, x(t) is the normalized source fluctuation sequence, y(t) is the normalized target fluctuation sequence, N is the sequence length, and τ is the time offset. The offset range is set to 0 to 100 milliseconds with a step size of 1 millisecond. This is because for a 10kV distribution transformer, the maximum propagation delay of loss fluctuations inside the equipment will not exceed 100 milliseconds. This setting can significantly reduce the amount of calculation while ensuring accuracy.

[0035] The cross-correlation coefficient ranges from -1 to 1. The larger the absolute value, the stronger the correlation between the two sequences. The time offset corresponding to the maximum value of the cross-correlation coefficient is taken as the propagation delay time of the loss fluctuation relative to the winding copper loss. This method can accurately measure the time difference between two fluctuation signals with an accuracy of 1 millisecond. It is a loss propagation characteristic measurement method that has never been used in the existing technology. The ratio of the amplitude of the source sequence at the maximum value of the cross-correlation coefficient to the amplitude of the corresponding target sequence at the same time is taken as the fluctuation attenuation coefficient of the propagation path. The larger the attenuation coefficient, the smaller the energy loss of the loss fluctuation during the propagation process.

[0036] Finally, a 2048-point Fast Fourier Transform is performed on the source sequence and the target sequence with the corresponding maximum cross-correlation coefficient. A Hanning window is added to reduce spectral leakage, and the phase spectra of the two sequences are obtained. The frequency component with the largest amplitude integer multiple of the power frequency is extracted as the main frequency. The phase difference between the two sequences at the main frequency is calculated as the phase offset of the propagation path, which ranges from -π to π. This reflects the phase change of loss fluctuations during propagation. Through the above process, the propagation delay time, fluctuation attenuation coefficient, and phase offset corresponding to the three links of core iron loss, additional loss, and environmental correction loss are obtained respectively. These three types of parameters completely describe the propagation characteristics of loss fluctuations inside the equipment.

[0037] The module establishes a 4x3 two-dimensional basic matrix. The four rows correspond to four loss components: winding copper loss, core iron loss, additional loss, and environmental correction loss. The three columns correspond to three types of parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. It is important to note that although this matrix is ​​physically two-dimensional, it contains information in three independent dimensions: time, amplitude, and phase. Therefore, it is called a three-dimensional loss fluctuation compensation matrix, which is fundamentally different from the one-dimensional compensation parameter table in the existing technology that only contains amplitude information. The parameters corresponding to winding copper loss are set as the baseline values, i.e., propagation delay time 0, fluctuation attenuation coefficient 1, and phase offset 0, and filled into the first row of the matrix. Then, the parameters of the other three components are filled into the corresponding columns of the corresponding rows.

[0038] Since the three types of parameters have different dimensions and value ranges, normalization processing is required to facilitate subsequent compensation calculations: the propagation delay time is divided by the maximum value among all delay times to obtain the normalized propagation delay time, with a value range of 0-1; the fluctuation attenuation coefficient is divided by the maximum value among all attenuation coefficients to obtain the normalized fluctuation attenuation coefficient, with a value range of 0-1; the phase offset is mapped from the interval [-π,π] to the interval [0,1] through a linear transformation, with the formula φ'=(φ+π) / (2π); after the normalization processing, a three-dimensional loss fluctuation compensation matrix containing three types of spatiotemporal information—time, amplitude, and phase—is obtained. A double buffering mechanism is used when updating the matrix to avoid data inconsistency caused by the compensation calculation module reading the data during the update process. This matrix is ​​the core of the entire compensation system, and all subsequent compensation parameter calculations are based on it.

[0039] The phase pre-calibration compensation module extracts the normalized parameters of each stage from the three-dimensional matrix and restores them to actual physical quantities: actual propagation delay time = normalized propagation delay time × maximum propagation delay time, actual fluctuation attenuation coefficient = normalized fluctuation attenuation coefficient × maximum fluctuation attenuation coefficient, actual phase offset = normalized phase offset × 2π - π; combining the instantaneous loss value of each stage at the current moment and the actual fluctuation attenuation coefficient, the basic compensation amount of each stage is calculated, and the formula is C. i (t)=P i (t)×α i , where C i (t) represents the basic compensation amount of the i-th element at time t, P i (t) represents the instantaneous loss value of the i-th element at time t, α i Let be the actual fluctuation attenuation coefficient of the i-th stage. The basic compensation amount is adjusted according to the attenuation coefficient to compensate for the energy loss of the loss fluctuation during propagation, ensuring that the amplitude of the compensation signal when it reaches the point of loss generation is equal to the actual loss amplitude. The basic compensation amount is bound to the corresponding actual propagation delay time and actual phase offset to generate dynamic compensation parameters for each stage with timing phase information. In order to counteract the phase distortion of the loss fluctuation during propagation, phase pre-calibration processing is applied to each compensation parameter. This is a key technology to solve the compensation lag problem. In the existing technology, there is a fixed phase difference between the compensation signal and the loss signal, which causes the phase of the loss signal to change when the compensation signal arrives, resulting in overcompensation or undercompensation. However, by applying a phase modulation to the compensation signal that is equal in magnitude and opposite in direction to the propagation phase offset, the phase of the compensation signal when it reaches the point of loss generation is completely opposite to the phase of the actual loss fluctuation, thereby achieving maximum cancellation.

[0040] In practice, the dominant frequency ω of the loss fluctuation is first determined by performing a Fourier transform on the source wave sequence. This frequency is typically 50Hz (the power frequency) or an integer multiple thereof. Then, the phase modulation coefficient is calculated based on the phase offset, and the base compensation is sinusoidally modulated using the formula C. i '(t)=C i (t)×cos(ωt-φ i ), where C i '(t) represents the pre-calibration compensation amount of the i-th element at time t, and C i (t) represents the basic compensation amount of the i-th element at time t, ω represents the main frequency of loss fluctuation, and φ i Let t be the propagation phase offset of the i-th stage, and t be the current time.

[0041] All dynamic compensation parameters for each stage are sorted in ascending order of propagation delay time to generate a compensation signal output queue. The shorter the propagation delay time of a stage, the earlier its compensation signal output time. This ensures that all compensation signals arrive at their respective loss generation points simultaneously for synchronous compensation. For example, if the propagation delay time of winding copper loss is 0 milliseconds, core iron loss is 20 milliseconds, additional loss is 50 milliseconds, and environmental correction loss is 80 milliseconds, then the output order is: winding copper loss compensation signal → core iron loss compensation signal → additional loss compensation signal → environmental correction loss compensation signal. Each signal is separated by the corresponding propagation delay time. At the same time, the response time of the compensation device is measured in advance and added to the propagation delay time to correct the output time of the compensation signal in advance and eliminate the influence of the actuator response delay.

[0042] Then, the phase-precalibrated digital compensation value is converted into a 12-bit analog control signal of 0-10V and output to the thyristor-switched reactive power compensation device. The compensation device calculates the number of capacitor banks to be switched on based on the received analog control signal and automatically switches the corresponding capacitor banks using zero-crossing switching technology to achieve dynamic reactive power compensation. It is important to note the physical mechanism by which reactive power compensation reduces transformer active power losses: the copper loss of a transformer is proportional to the square of the current. Reducing reactive power leads to a decrease in total current, thus significantly reducing copper losses. Simultaneously, the reduction in reactive power increases the magnetic flux of the iron core. With a slight decrease in density, iron losses will also decrease accordingly. For a 1000kVA transformer, when the power factor increases from 0.85 to 0.95, copper losses can be reduced by about 20%, resulting in significant energy savings. During the output process, the main controller monitors the operating status of the compensation device in real time, including the switching status of each capacitor bank, output current, voltage, temperature, etc. If a switching failure, overcurrent, overvoltage, or overheating fault is detected, the output compensation signal will be stopped immediately and an audible and visual alarm will be issued. At the same time, information such as the fault occurrence time, fault type, and operating parameters at the time of the fault will be stored in the historical database for subsequent analysis and processing.

[0043] After all compensation control signals have been fully output and twice the longest propagation delay time has elapsed, ensuring that all compensation signals have propagated and taken effect, and have entered a stable phase, the compensation effect verification module re-collects multi-dimensional operating data to avoid inaccurate verification results due to data collection during the transition process. The module then calculates the current actual total loss value using the same method as before. The actual total loss value is compared with the preset target loss value to calculate the loss deviation. The preset target loss value, calculated based on the transformer's rated parameters and the current load rate, represents the theoretically achievable minimum loss value under this operating condition. The preset allowable deviation threshold is set to 1% of the rated loss. If the loss deviation value is less than the allowable threshold, it indicates... The current compensation effect meets the requirements. Maintain the current three-dimensional compensation matrix and compensation strategy, and continue to perform periodic compensation control with a period of 1 second. If the loss deviation value is greater than or equal to the allowable threshold, it means that the current propagation parameters can no longer accurately reflect the actual operating status. This may be due to accelerated equipment aging, significant changes in environmental conditions, or changes in load characteristics. Automatically return to the propagation parameter inversion step, re-extract the loss fluctuation characteristic time series of the most recent 2 seconds, calculate new propagation parameters and update the three-dimensional compensation matrix, and then generate compensation parameters based on the new matrix for compensation until the loss deviation value is less than the allowable threshold. This mechanism can ensure that the optimal compensation effect is maintained throughout the entire life cycle without manual intervention.

[0044] Routine maintenance includes daily checks of operating status and fault alarm information, weekly cleaning of dust inside the control cabinet and tightening of wiring terminals, monthly backup of system parameters and historical data, quarterly calibration of sensors and system parameters, and annual comprehensive equipment overhaul including transformer oil quality testing and insulation resistance testing. For dry-type transformers, due to their different heat dissipation characteristics compared to oil-immersed transformers (using air self-cooling or wind cooling), the environmental correction loss acquisition interval is adjusted to 50 milliseconds, the filter cutoff frequency is increased to 15Hz, and the environmental correction formula is modified to adapt to their heat dissipation characteristics. For prefabricated substations, due to limited space and severe electromagnetic interference, shielding measures are increased by using higher-grade shielded cables, and the main controller is installed away from the transformer. For photovoltaic grid-connected transformers, due to frequent fluctuations in photovoltaic output power, the update interval of propagation parameters is shortened to 2 seconds, and a photovoltaic power prediction function is added to generate compensation parameters in advance. For explosion-proof transformers in mines, explosion-proof sensors and controllers that meet explosion-proof standards are used, and a gas concentration monitoring function is added; when the gas concentration exceeds the standard, the compensation device automatically stops operating. Furthermore, a health status assessment model can be established by long-term monitoring of the changing trends of propagation parameters, such as... Figure 2As shown, for example, short circuits between winding turns can shorten the propagation delay time of copper loss fluctuations, loose iron core can increase the attenuation coefficient of iron loss fluctuations, and insulation deterioration can cause characteristic changes in the phase shift of additional losses. By identifying these characteristic changes, potential faults can be detected 3 to 6 months in advance, enabling predictive maintenance of transformer equipment and significantly reducing equipment downtime and maintenance costs.

[0045] Example 2: Taking Example 1 as an example, a transformer regulation loss compensation control system is provided. The control system includes: The loss fluctuation adaptation acquisition module sets differentiated acquisition intervals based on the fluctuation propagation characteristics of different loss stages, collects multi-dimensional operational data, and generates continuous time series with unified timestamps. The loss fluctuation adaptation acquisition module includes: The acquisition interval configuration unit sets four different acquisition intervals according to the fluctuation propagation characteristics of different loss links; The multi-dimensional data acquisition unit collects instantaneous electrical operation parameters, instantaneous load operation parameters, environmental impact parameters, and equipment status parameters according to the corresponding acquisition intervals; the time series alignment unit globally aligns all acquired data with timestamps of uniform precision to generate continuous time series of data for each dimension; the loss fluctuation feature extraction module calculates the instantaneous loss value of each link based on the continuous time series, extracts high-frequency fluctuation feature components covering the main energy range of transient loss fluctuations, and generates a loss fluctuation feature time series. The propagation parameter inversion module uses the time series of winding copper loss fluctuation characteristics as the source fluctuation sequence, identifies the propagation paths of other links, and inverts three types of propagation parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. The propagation parameter inversion module includes: a sequence extraction unit, which extracts the source fluctuation sequence and each target fluctuation sequence within a preset time period; a cross-correlation analysis unit, which calculates the cross-correlation coefficient between the source sequence and each target sequence at different time offsets; and a propagation parameter solution unit, which determines the propagation delay time based on the maximum value of the cross-correlation coefficient, determines the fluctuation attenuation coefficient based on the sequence amplitude ratio, and determines the phase offset based on the phase difference of the main frequency components. The three-dimensional compensation matrix construction module constructs a three-dimensional loss fluctuation compensation matrix containing time, amplitude, and phase dimensions based on the three types of propagation parameters for all propagation paths. The phase pre-calibration compensation module calculates the dynamic compensation parameters of each stage based on a three-dimensional compensation matrix and applies phase pre-calibration processing opposite to the propagation phase offset. The phase pre-calibration compensation module includes: a parameter extraction unit that extracts the three types of propagation parameters corresponding to each stage from the three-dimensional loss fluctuation compensation matrix; a basic compensation calculation unit that calculates the basic compensation amount for each stage by combining the instantaneous loss value and fluctuation attenuation coefficient of each stage at the current moment; a phase pre-calibration unit that binds the basic compensation amount with the timing phase information and applies phase modulation opposite to the propagation phase offset; and a signal conversion and sorting unit that converts the phase-precalibrated digital compensation parameters into analog control signals and sorts them according to the propagation delay time to generate an output queue. The timing output module outputs the phase pre-calibrated compensation control signals to the transformer regulation system in ascending order of propagation delay time; The compensation effect verification module re-collects multi-dimensional operating data and calculates the actual total loss value within a preset fixed time interval after all compensation control signals have fully taken effect. The deviation judgment and feedback module compares the actual total loss value with the preset target loss value to obtain the loss deviation value, judges whether the deviation is within the allowable range, and maintains the current strategy or triggers the propagation parameters to be re-inverted based on the judgment result.

[0046] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for compensating and controlling losses in transformer regulation, characterized in that, The control method involves the following steps: Based on the fluctuation propagation characteristics of different loss links within the transformer regulation system, differentiated acquisition intervals are set to collect multi-dimensional operating data and generate a continuous time series with unified timestamps. The instantaneous loss value of each stage is calculated step by step based on the continuous time series. High-frequency fluctuation characteristic components covering the main energy range of transient loss fluctuations are extracted from each loss series simultaneously to generate a loss fluctuation characteristic time series. Using the time series of the winding copper loss fluctuation characteristics with the fastest response speed as the source fluctuation sequence, the propagation path of the loss fluctuation of each other link relative to the source sequence is identified, and the three types of propagation parameters corresponding to each path are obtained by inversion: propagation delay time, fluctuation attenuation coefficient and phase offset. Based on the three types of propagation parameters for all propagation paths, a three-dimensional loss fluctuation compensation matrix containing time, amplitude, and phase dimensions is constructed. Based on the three-dimensional loss fluctuation compensation matrix, the basic compensation amount of each link is calculated and the corresponding time and phase information is bound to generate dynamic compensation parameters for each link. Apply phase pre-calibration processing, which is opposite to the propagation phase offset, to the dynamic compensation parameters of each sub-stage; According to the order of propagation delay time of each propagation path from smallest to largest, the phase pre-calibrated compensation control signals are output to the transformer regulation system in sequence; Within a preset fixed time interval after all compensation control signals have fully taken effect, multi-dimensional operating data is re-acquired and the actual total loss value is calculated. The actual total loss value is compared with the preset target loss value to obtain the loss deviation value. It is then determined whether the loss deviation value is less than the preset allowable deviation threshold. If it is less than the preset allowable deviation threshold, the current three-dimensional loss fluctuation compensation matrix and phase pre-calibration compensation strategy are maintained. Otherwise, the process returns to re-identify the loss fluctuation propagation path and invert the propagation parameters.

2. The transformer regulation loss compensation control method according to claim 1, characterized in that, The steps for setting differentiated acquisition intervals based on the propagation characteristics of loss fluctuations are as follows: Based on the electromagnetic wave propagation characteristics corresponding to winding copper losses, a first type of acquisition interval is set to acquire instantaneous electrical operating parameters of the transformer regulation system. To address the propagation characteristics of hysteresis fluctuations corresponding to iron core losses, a second type of acquisition interval is set to acquire instantaneous load operating parameters of the transformer regulation system. To address the propagation characteristics of heat exchange fluctuations corresponding to environmental correction losses, a third type of acquisition interval is set to collect environmental impact parameters of the transformer regulation system. To address the propagation characteristics of structural aging fluctuations corresponding to additional losses, a fourth type of acquisition interval is set to collect equipment status parameters of the transformer regulation system. All collected data are globally aligned using timestamps of uniform precision to generate continuous time series data for each dimension.

3. The transformer regulation loss compensation control method according to claim 1, characterized in that, The steps for extracting high-frequency fluctuation feature components are as follows: The instantaneous values ​​of winding copper loss, core iron loss, additional loss and environmental correction loss are calculated at each time step to generate the continuous loss time series corresponding to each link. High-pass filtering with uniform parameters is used to extract high-frequency fluctuation components covering the main energy range of transient loss fluctuations from the loss time series of each stage. The extracted high-frequency fluctuation components of each stage are re-aligned according to the timestamps to generate a time series of loss fluctuation characteristics for each stage.

4. The transformer regulation loss compensation control method according to claim 1, characterized in that, The steps for retrieving the inversion propagation parameters are as follows: The time series of winding copper loss fluctuation characteristics within a preset time period is taken as the source fluctuation series, and the time series of loss fluctuation characteristics of other links within the same time period are taken as the target fluctuation series. Cross-correlation analysis was performed on the source wave sequence and each target wave sequence to calculate the cross-correlation coefficient at different time offsets; The time offset corresponding to the maximum cross-correlation coefficient is taken as the propagation delay time of the loss fluctuation in this link, and the amplitude ratio of the source sequence to the corresponding target sequence is taken as the fluctuation attenuation coefficient of the propagation path. Spectral analysis is performed on the source wave sequence and the target wave sequence at the corresponding maximum cross-correlation coefficient, and the phase difference of the main frequency component is extracted as the propagation phase offset.

5. The transformer regulation loss compensation control method according to claim 1, characterized in that, The steps for constructing the three-dimensional loss fluctuation compensation matrix are as follows: A two-dimensional basic matrix is ​​constructed, with the matrix rows corresponding to four loss elements: winding copper loss, core iron loss, additional loss, and environmental correction loss, and the matrix columns corresponding to three types of propagation parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. Set the propagation parameters corresponding to the winding copper loss as the reference values ​​and fill them into the first row of the matrix; The propagation parameters corresponding to core iron loss, additional loss, and environmental correction loss are sequentially filled into the corresponding column positions of the corresponding row of the matrix; All parameters within the matrix are normalized, and parameters of different dimensions are uniformly mapped to a preset numerical range to obtain a three-dimensional loss fluctuation compensation matrix containing spatiotemporal phase information.

6. The transformer regulation loss compensation control method according to claim 1, characterized in that, The steps for generating the phase pre-calibration compensation control signal are as follows: Extract the propagation delay time, fluctuation attenuation coefficient and phase offset corresponding to each link from the three-dimensional loss fluctuation compensation matrix; By combining the instantaneous loss value of each link at the current moment with the corresponding fluctuation attenuation coefficient, the basic compensation amount of each link is calculated; The basic compensation amount is bound to the corresponding propagation delay time and phase offset to generate dynamic compensation parameters for each stage with time-series phase information. Phase modulation is applied to the dynamic compensation parameter of each sub-stage, and the modulation angle is equal in magnitude and opposite in direction to the propagation phase offset corresponding to that parameter; The pre-calibrated digital compensation parameters are converted into analog control signals recognizable by the transformer regulation system, sorted by propagation delay time from smallest to largest, and a compensation signal output queue is generated.

7. A transformer regulation loss compensation control system, characterized in that, The control system includes: The loss fluctuation adaptation acquisition module sets differentiated acquisition intervals based on the fluctuation propagation characteristics of different loss stages, collects multi-dimensional operating data, and generates a continuous time series with unified timestamps. The loss fluctuation feature extraction module calculates the instantaneous loss value of each stage based on the continuous time series, extracts high-frequency fluctuation feature components covering the main energy range of transient loss fluctuation, and generates a loss fluctuation feature time series. The propagation parameter inversion module uses the time series of the winding copper loss fluctuation characteristics as the source fluctuation sequence to identify the propagation path of each other and invert three types of propagation parameters: propagation delay time, fluctuation attenuation coefficient, and phase offset. The three-dimensional compensation matrix construction module constructs a three-dimensional loss fluctuation compensation matrix containing time, amplitude, and phase dimensions based on the three types of propagation parameters for all propagation paths. The phase pre-calibration compensation module calculates the dynamic compensation parameters of each component based on the three-dimensional compensation matrix and applies a phase pre-calibration process that is opposite to the propagation phase offset. The timing output module outputs the phase pre-calibrated compensation control signals to the transformer regulation system in ascending order of propagation delay time; The compensation effect verification module re-collects multi-dimensional operating data and calculates the actual total loss value within a preset fixed time interval after all compensation control signals have fully taken effect. The deviation judgment and feedback module compares the actual total loss value with the preset target loss value to obtain the loss deviation value, judges whether the deviation is within the allowable range, and maintains the current strategy or triggers the propagation parameters to be re-inverted based on the judgment result.

8. A transformer regulation loss compensation control system according to claim 7, characterized in that, The loss fluctuation adaptation and acquisition module includes: The acquisition interval configuration unit sets four different acquisition intervals according to the fluctuation propagation characteristics of different loss links; The multi-dimensional data acquisition unit collects instantaneous electrical operation parameters, instantaneous load operation parameters, environmental impact parameters, and equipment status parameters according to the corresponding acquisition intervals. The time series alignment unit globally aligns all collected data according to timestamps of uniform precision, generating continuous time series of data in each dimension.

9. A transformer regulation loss compensation control system according to claim 7, characterized in that, The propagation parameter inversion module includes: The sequence extraction unit extracts the source fluctuation sequence within a preset time period and the target fluctuation sequences within the same time period. The cross-correlation analysis unit calculates the cross-correlation coefficients between the source sequence and each target sequence at different time offsets; The propagation parameter calculation unit determines the propagation delay time based on the maximum cross-correlation coefficient, the fluctuation attenuation coefficient based on the sequence amplitude ratio, and the phase offset based on the phase difference of the main frequency components.

10. A transformer regulation loss compensation control system according to claim 7, characterized in that, The phase pre-calibration compensation module includes: The parameter extraction unit extracts three types of propagation parameters corresponding to each stage from the three-dimensional loss fluctuation compensation matrix. The basic compensation calculation unit calculates the basic compensation amount for each link by combining the instantaneous loss value and fluctuation attenuation coefficient of each link at the current moment. The phase pre-calibration unit binds the basic compensation amount to the timing phase information and applies a phase modulation that is opposite to the propagation phase offset. The signal conversion and sorting unit converts the phase-precalibrated digital compensation parameters into analog control signals and sorts them according to the propagation delay time to generate an output queue.

Images