Intelligent fusion terminal three-phase unbalance active compensation control system

Abstract

The present application relates to power distribution network power quality control and intelligent control technical field, specifically, it relates to a kind of intelligent fusion terminal three-phase imbalance active compensation control system.It includes: distribution network three-phase operating parameter synchronous acquisition unit;Three-phase imbalance pre-judgment type compensation strategy generation unit, time sequence pre-judgment-loss constraint joint optimization algorithm is handled to time sequence sequence of three-phase electric parameter time sequence alignment;Multi-type compensation device collaborative execution unit;Compensation effect closed-loop feedback check unit.The present application relies on time sequence rolling fitting to realize three-phase imbalance evolution trend early prediction, abandon traditional threshold post-trigger mode, solve the problem that governance response lags, transient load fluctuation is easy to cause power quality short time overrun;While introducing execution mechanism action loss quantization constraint to build joint optimization target, optimize equipment action frequency under the premise of guaranteeing imbalance governance standard, suppress device frequent switching aging, reduce the cost of operation and maintenance of substation.

Description

Technical Field

[0001] This invention relates to the field of power quality management and intelligent control technology for power distribution networks, specifically to an intelligent integrated terminal three-phase imbalance active compensation control system. Background Technology

[0002] Three-phase imbalance management in low-voltage distribution transformer areas is a core technological direction for improving power quality and refining line loss control in the construction of new power systems. Currently, the proportion of single-phase residential load and distributed photovoltaic access in low-voltage transformer areas continues to rise, further exacerbating three-phase imbalance conditions. Intelligent integrated terminals have become standard hardware in the centralized procurement of domestic power grid transformer areas. Relying on local edge computing capabilities to achieve power quality monitoring and equipment collaborative control, they represent the mainstream hardware architecture for transformer area management and play a crucial supporting role in ensuring power supply reliability and reducing transformer area operating losses.

[0003] Currently, the mainstream governance solutions implemented on a large scale in the industry are mainly divided into two categories. The first is the local threshold-triggered compensation solution. This solution takes the local monitoring and control device of the distribution transformer area as the core, collects the three-phase electrical parameters of the bus in real time, and triggers the switching of the phase-changing switch or the switching of the phase compensation branch according to the preset imbalance threshold. The solution has a simple logical architecture, low hardware cost, and strong operational reliability, and is the mainstream implementation solution for the three-phase imbalance transformation of existing distribution transformer areas in China. The second is the master station centralized dispatch compensation solution. This solution relies on the distribution network power consumption information collection master station to collect the operation data of multiple distribution transformer areas. After centralized analysis, the governance strategy is generated and then distributed to the terminal for execution. It can realize the global coordination of governance resources between distribution transformer areas at the county level and is suitable for large-scale distribution transformer area cluster management scenarios.

[0004] However, existing solutions still have common technical limitations in engineering implementation. First, existing governance strategies are all based on ex-post judgment triggering of real-time thresholds, lacking the ability to predict the evolution trend of imbalance. They are slow to respond to transient imbalance conditions caused by random load fluctuations, easily leading to short-term power quality exceeding limits, and are difficult to adapt to distribution transformer scenarios with strong load fluctuations. Second, existing solutions generally take achieving the imbalance standard as the single control objective, without quantitatively constraining the operating losses of compensation actuators. This easily leads to frequent switch operation during the governance process, accelerating the aging of switch contacts and compensation devices, and increasing the operation and maintenance costs of distribution transformers. As distribution network governance upgrades towards lean, long-life, and low-maintenance directions, existing solutions are no longer suitable for the development needs of efficient on-site governance at the edge. There is an urgent need to develop new compensation solutions that take into account both governance timeliness and equipment life constraints. Summary of the Invention

[0005] This invention addresses the technical problems of existing three-phase imbalance management technologies, such as reliance on post-event threshold determination, lack of ability to predict imbalance evolution trends, delayed response to transient conditions, single management objectives, lack of quantified constraints to compensate for actuator wear, and frequent equipment operation. It provides an intelligent fusion terminal three-phase imbalance active compensation control system. This system leverages the local edge computing capabilities of the intelligent fusion terminal to achieve advanced prediction and active compensation of three-phase imbalance states. Simultaneously, it incorporates actuator wear into the control objective system, reducing equipment wear and maintenance costs while ensuring effective power quality management, thus adapting to the application requirements of lean management in low-voltage distribution substations.

[0006] To address the aforementioned technical problems, the present invention aims to provide an intelligent fusion terminal three-phase imbalance active compensation control system, comprising: The distribution network three-phase operating parameter synchronous acquisition unit acquires the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution area, and performs filtering, calibration and timing alignment processing on the acquired electrical parameter data to obtain a timing-aligned three-phase electrical parameter timing sequence. The three-phase imbalance predictive compensation strategy generation unit is mounted on the edge computing main control chip of the intelligent fusion terminal. It uses a timing prediction-loss constraint joint optimization algorithm to process the timing sequence of the three-phase electrical parameters that are time-aligned, and generates an advanced active compensation control command that takes into account both minimum action loss and optimal compensation effect. It also receives the returned three-phase electrical parameter data to correct the prediction calculation parameters and loss constraint weights. The multi-type compensation device collaborative execution unit receives the advanced active compensation control command and drives the transformer area phase switching switch group to complete the phase-to-phase active load transfer and drives the phase compensation branch to complete the graded switching according to the advanced active compensation control command, and performs the three-phase imbalance active management operation. The compensation effect closed-loop feedback verification unit re-collects the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution substation within a preset period after the compensation is completed, and sends the re-collected three-phase electrical parameter data back to the three-phase imbalance predictive compensation strategy generation unit.

[0007] As a further improvement to this technical solution, the distribution network three-phase operating parameter synchronous acquisition unit includes a three-phase electrical parameter signal access module, a multi-channel synchronous sampling and conversion module, an electrical parameter filtering and calibration module, and a timing alignment output module, wherein: The three-phase electrical parameter signal access module receives the three-phase voltage analog signal and the three-phase current analog signal from the low-voltage bus side of the distribution substation. The amplitude adaptation and electrical isolation processing are completed by the front-end signal conditioning circuit, and the adapted analog electrical parameter signal is output to the multi-channel synchronous sampling conversion module. The multi-channel synchronous sampling and conversion module receives the adapted analog electrical parameter signal output by the three-phase electrical parameter signal access module, and completes synchronous analog-to-digital conversion using a multi-channel synchronous analog-to-digital sampling channel triggered by the same source clock to obtain the original three-phase voltage instantaneous value and three-phase current instantaneous value digital quantity and output them to the electrical parameter filtering and calibration module. The electrical parameter filtering calibration module receives the original three-phase voltage instantaneous value and three-phase current instantaneous value digital quantities output by the multi-channel synchronous sampling conversion module, and completes the filtering calibration process using the power frequency digital filtering algorithm and preset amplitude and phase calibration parameters to obtain the calibrated three-phase electrical parameter data and output it to the timing alignment output module. The timing alignment output module receives the calibrated three-phase electrical parameter data output by the electrical parameter filtering calibration module, completes the timing alignment processing of the three-phase data with the power frequency voltage zero crossing point as the reference, obtains the timing-aligned three-phase electrical parameter timing sequence, and outputs it to the three-phase unbalance predictive compensation strategy generation unit.

[0008] As a further improvement to this technical solution, the three-phase imbalance predictive compensation strategy generation unit includes a negative sequence component time-series prediction module, a loss constraint target construction module, a finite-step optimization solution module, and a control parameter adaptive correction module, wherein: The negative sequence component timing prediction module receives the timing sequence of the three-phase electrical parameters aligned with the timing sequence, calculates the evolution trend of the negative sequence component of the three-phase current recursively based on the timing rolling sampling window, and outputs the three-phase imbalance prediction state of the next cycle to the loss constraint target construction module. The loss constraint objective construction module receives the three-phase imbalance prediction state of the next cycle, quantifies the action loss of the compensation actuator into constraint factors to construct a joint optimization objective function, and outputs the joint optimization objective function to the finite step optimization solution module. The finite-step optimization solution module receives the joint optimization objective function, generates an advanced active compensation control command that balances minimum motion loss and optimal compensation effect through finite-step optimization solution, and outputs it outward. The adaptive correction module for control parameters receives the returned three-phase electrical parameter data, corrects the prediction calculation parameters and loss constraint weights based on the compensated operating data, and feeds back the corrected parameters to the negative sequence component timing prediction module and the loss constraint target construction module.

[0009] As a further improvement to this technical solution, the imbalance trend prediction process of the negative sequence component time series prediction module includes the following steps: S21.1 Receive the timing sequence of the three-phase electrical parameters aligned with the timing sequence, perform point-by-point recursive update on the timing rolling sampling window, and remove the earliest historical sampling point in the window for each new current sampling point to obtain the updated timing current sequence. S21.2 Calculation of the negative sequence components of the three-phase current at each sampling moment within the rolling sampling window based on the symmetrical component method. The effective value completes the sequence component decomposition of the three-phase current; S21.3, Negative-order components within a time-series rolling sampling window The negative order component of the effective value sequence is calculated using the least squares method for fitting. The gradient and rate of change of; S21.4. Predict the negative sequence component of the next sampling period recursively based on the gradient and rate of change. The effective value is used to obtain the predicted state of the three-phase imbalance in the next cycle.

[0010] As a further improvement to this technical solution, the joint optimization objective construction process of the loss constraint objective construction module includes the following steps: S22.1 Receive the three-phase imbalance prediction state for the next cycle, determine the equivalent life loss of a single mechanical action of the commutation switch and the equivalent electrical loss of a single switching of the phase compensation branch, and quantify the action loss of the compensation actuator corresponding to a single action. S22.2 Set the standard limit value for the three-phase imbalance after compensation as a constraint condition, and clarify the feasible domain of the optimization solution; S22.3. To achieve the optimal balance between imbalance suppression and total motion loss, a joint optimization objective function is constructed. The joint optimization objective function includes an imbalance suppression term and an action loss term, with weight coefficients set for each term to complete the bi-objective optimization modeling under loss constraints.

[0011] As a further improvement to this technical solution, the finite-step optimization solution module includes a discrete state construction submodule, a dynamic programming recursion submodule, and an optimal instruction output submodule, wherein: The discrete state construction submodule receives the joint optimization objective function, decomposes the commutation switch action combination and the phase compensation branch switching position into a finite discrete action state set, and outputs the discrete action state set to the dynamic programming recursive submodule. The dynamic programming recursive submodule receives a set of discrete action states, performs a finite-step recursive calculation according to the compensation stage, traverses all possible action combinations and calculates the corresponding joint optimization objective value, and outputs the objective calculation result of each action combination to the optimal instruction output submodule. The optimal instruction output submodule receives the target calculation results of each action combination, filters out the action combination with the smallest joint optimization target value, generates an advanced active compensation control instruction, and outputs it outward.

[0012] As a further improvement to this technical solution, the adaptive correction module for control parameters includes a compensation deviation calculation submodule, an operating parameter adjustment submodule, and a correction parameter feedback submodule, wherein: The compensation deviation calculation submodule receives the returned three-phase electrical parameter data, calculates the deviation between the actual three-phase imbalance and the expected imbalance after compensation, and outputs the deviation data to the operating parameter adjustment submodule. The operating parameter adjustment submodule receives deviation data and dynamically adjusts the window length parameter and loss constraint weight coefficient of the time-series rolling sampling window according to the magnitude and direction of the deviation, so as to obtain the corrected prediction calculation parameters and loss constraint weight. The correction parameter feedback submodule receives the corrected prediction calculation parameters and loss constraint weights, feeds back the corrected prediction calculation parameters to the negative sequence component time series prediction module, and feeds back the corrected loss constraint weights to the loss constraint target construction module.

[0013] As a further improvement to this technical solution, the multi-type compensation device collaborative execution unit includes a control command parsing and collaborative scheduling module, a transformer area phase-change switch group driving module, and a phase-compensation branch switching module, wherein: The control command parsing and collaborative scheduling module receives the advance active compensation control command, completes the command legality verification and execution timing allocation, and outputs the commutation control command to the transformer area commutation switch group drive module and the switching control command to the phase compensation branch switching module, respectively. The transformer area phase switching switch group drive module receives the phase switching control command and drives the transformer area phase switching switch group to perform phase switching operation to complete the phase-to-phase active load transfer. The phase compensation branch switching module receives the switching control command and drives the phase compensation branch to perform a graded switching operation to complete the active treatment of three-phase imbalance.

[0014] As a further improvement to this technical solution, the transformer substation phase-changing switch group drive module includes a switch status verification submodule, a zero-current switching execution submodule, and an action result feedback submodule, wherein: The switch status verification submodule receives the commutation control command, collects the current position status and branch current amplitude of the commutation switch group in the transformer area, and outputs the execution command to the zero current switching execution submodule after confirming that the safe switching conditions are met. The zero-current switching execution submodule receives the execution command and triggers the switch contacts to operate when the power frequency current crosses zero, thus completing the phase switching operation. The action result feedback submodule collects the switch position signal after switching and sends the switch action result back to the control command parsing and collaborative scheduling module.

[0015] As a further improvement to this technical solution, the compensation effect closed-loop feedback verification unit includes a re-sampling trigger control module, a three-phase electrical parameter re-acquisition module, and a re-sampling data feedback module, wherein: The re-sampling trigger control module receives the compensation execution completion signal, triggers the generation of a re-sampling command after a preset delay, and outputs the re-sampling command to the three-phase electrical parameter re-acquisition module. The three-phase electrical parameter re-acquisition module receives the re-acquisition command, re-acquisitions the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution substation, and outputs the re-acquisitioned three-phase electrical parameter data to the re-acquisition data feedback module. The re-sampling data backhaul module receives the three-phase electrical parameter data obtained from the re-sampling and backhauls the three-phase electrical parameter data to the three-phase imbalance predictive compensation strategy generation unit.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention constructs a three-phase electrical parameter timing prediction mechanism, relies on rolling timing fitting to predict the evolution trend of three-phase imbalance in advance, and replaces the traditional threshold-triggered governance mode. It effectively solves the industry problems of delayed response and poor adaptability to transient load fluctuations of conventional compensation schemes, eliminates the problem of short-term power quality exceeding limits, and improves the foresight and operating condition adaptability of three-phase imbalance governance in low-voltage distribution transformer areas. 2. This invention breaks through the technical limitations of traditional single governance objectives by establishing a joint optimization strategy with quantitative constraints on action loss. While ensuring the compliance of three-phase imbalance governance, it optimally controls the operation frequency and loss of phase switching switches and phase compensation branches, effectively suppressing frequent switching of compensation equipment, slowing down the aging rate of equipment, reducing the operation and maintenance costs of transformer equipment and the operation loss of distribution network, and adapting to the needs of lean and long-term governance of distribution network. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall system framework of the present invention. The meanings of the labels in the diagram are as follows: 1. Distribution network three-phase operating parameter synchronous acquisition unit; 11. Three-phase electrical parameter signal access module; 12. Multi-channel synchronous sampling and conversion module; 13. Electrical parameter filtering and calibration module; 14. Timing alignment output module; 2. Three-phase imbalance predictive compensation strategy generation unit; 21. Negative sequence component time-series prediction module; 22. Loss constraint target construction module; 23. Finite step optimization solution module; 24. Control parameter adaptive correction module; 3. Multi-type compensation device collaborative execution unit; 31. Control command parsing and collaborative scheduling module; 32. Transformer area phase switching switch group drive module; 33. Phase compensation branch switching module; 4. Compensation effect closed-loop feedback verification unit; 41. Re-sampling trigger control module; 42. Three-phase electrical parameter re-acquisition module; 43. Re-sampling data feedback module. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, this embodiment provides a smart fusion terminal three-phase imbalance active compensation control system. Using an edge computing main control chip deployed within the smart fusion terminal as the decision-making core, the system divides the three-phase imbalance management of low-voltage distribution substations into four stages: "synchronous perception - predictive optimization - collaborative execution - closed-loop verification," each undertaken by a separate functional unit. These four units are interconnected through an internal high-speed data bus and a real-time task scheduling mechanism: First, the distribution network three-phase operating parameter synchronous acquisition unit 1 provides three-phase fundamental electrical parameters with strictly aligned timing as the analysis benchmark; then, the three-phase imbalance predictive compensation strategy generation unit 2 generates advanced active compensation control commands that balance the management effect and device lifespan based on rolling prediction and loss constraints joint optimization; next, the multi-type compensation device collaborative execution unit 3 decomposes the commands into a commutation switch action sequence and phase compensation branch switching operations, driving field equipment to complete active load transfer and reactive power graded compensation; finally, the compensation effect closed-loop feedback verification unit 4 re-acquires the low-voltage bus side electrical parameters within a preset period after the management action is completed and transmits them back to the decision-making unit, forming a parameter self-correction closed loop.

[0020] This system is applicable to low-voltage distribution transformer areas where public distribution transformers are the main power supply source. Typical application scenarios include rural residential areas, urban-rural fringe areas, and small-scale industrial and commercial mixed areas. In these scenarios, the low-voltage busbar side of the distribution area is usually connected to several single-phase power supply branches. Due to factors such as asynchronous switching times of single-phase loads on the user side, seasonal load changes, and single-phase access of distributed photovoltaic systems, the amplitude and phase of the three-phase current often show significant differences, leading to excessive three-phase voltage imbalance on the busbar side. This results in increased additional losses in the distribution transformer, neutral point potential shift, and voltage exceeding limits for some end users. The intelligent fusion terminal in this embodiment can be installed in the integrated distribution cabinet or metering cabinet of the distribution transformer in the distribution area. Its external electrical interface is connected to the low-voltage busbar and field compensation devices through the secondary terminals of voltage transformers, secondary terminals of current transformers, and several digital I / O terminals. At the same time, it forms a local governance communication network with the distribution area's phase-changing switch group and phase compensation branch controller through RS485 or power line carrier communication interfaces. The terminal's internal main control chip uses an ARM Cortex-A series multi-core processor with a main frequency of no less than 1.2GHz. It integrates an FPU and a NEON single instruction multiple data acceleration unit on the chip, which can meet the requirements of real-time rolling window recursion, sequential component calculation and dynamic programming optimization algorithms for floating-point operation and parallel processing capabilities.

[0021] The distribution network three-phase operating parameter synchronous acquisition unit 1 acquires the instantaneous values ​​of three-phase voltage and three-phase current on the low-voltage bus side of the distribution substation, and performs filtering, calibration, and timing alignment processing on the acquired electrical parameter data to obtain a timing-aligned three-phase electrical parameter timing sequence. Specifically, the distribution network three-phase operating parameter synchronous acquisition unit 1 is used to obtain a high-precision, strictly timing-aligned sequence of three-phase voltage and current instantaneous values ​​from the low-voltage bus side of the distribution substation, providing the original data basis for subsequent three-phase imbalance prediction and compensation decisions. This unit consists of a three-phase electrical parameter signal input module 11, a multi-channel synchronous sampling and conversion module 12, an electrical parameter filtering and calibration module 13, and a timing alignment output module 14 cascaded together. The three-phase electrical parameter signal input module 11 receives the three-phase voltage analog signal and the three-phase current analog signal from the low-voltage bus side of the distribution substation. The amplitude adaptation and electrical isolation processing are completed by the front-end signal conditioning circuit, and the adapted analog electrical parameter signal is output to the multi-channel synchronous sampling conversion module 12. Specifically, the three-phase electrical parameter signal access module 11 is used to convert the high-voltage signal from the low-voltage bus side into a low-voltage signal suitable for analog-to-digital conversion, and to achieve electrical isolation between the primary equipment and the terminal circuit. In this embodiment, the three-phase electrical parameter signal access module 11 is configured with six independent analog input channels, corresponding to the three-phase voltage and three-phase current of A, B, and C, respectively. Each channel is equipped with a precision resistor voltage divider network or current sampling resistor at the front end to linearly map the rated input range (voltage 100V / 3, current 5A) to a ±2.5V peak range; a second-order active low-pass anti-aliasing filter is cascaded thereafter, with the cutoff frequency set to 2kHz to attenuate high-frequency components; the filter output is electrically isolated to a 3kV power frequency withstand voltage level via a linear optocoupler isolation circuit. The conditioned six analog signals are output to the multi-channel synchronous sampling conversion module 12 in a single-ended-to-ground manner. The above signal conditioning and isolation links are all mature technologies in this field, and the specific device selection and circuit topology will not be described in detail in this embodiment.

[0022] The multi-channel synchronous sampling and conversion module 12 receives the adapted analog electrical parameter signal output by the three-phase electrical parameter signal input module 11, and completes synchronous analog-to-digital conversion using a multi-channel synchronous analog-to-digital sampling channel triggered by the same source clock, obtaining the original three-phase voltage instantaneous value and the three-phase current instantaneous value digital quantity and outputting them to the electrical parameter filtering and calibration module 13. Specifically, the multi-channel synchronous sampling conversion module 12 performs synchronous analog-to-digital conversion on six analog signals, ensuring strict alignment of sampling times for each channel. The core of the multi-channel synchronous sampling conversion module 12 employs an 8-channel 16-bit successive approximation synchronous sampling ADC. All channels share a single sampling clock generated by frequency division of a temperature-compensated crystal oscillator, with clock jitter less than 50ps. The sampling time deviation between channels is controlled within 100ps, eliminating inter-channel phase errors caused by inconsistent sampling times at the hardware level. The sampling rate is fixed at 12.8kS / s, corresponding to uniform sampling of 256 points per power frequency cycle (20ms). After conversion, the ADC synchronously outputs the original three-phase voltage instantaneous value sequence through a parallel interface. , , Compared with the original three-phase current instantaneous value sequence ,in This is the sampling sequence number. The above digital quantity is sent to the electrical parameter filtering and calibration module 13 in binary original code format. The implementation of synchronous sampling architecture and multi-channel ADC is well known to those skilled in the art and will not be elaborated here.

[0023] The electrical parameter filtering calibration module 13 receives the original three-phase voltage instantaneous value and three-phase current instantaneous value digital quantity output by the multi-channel synchronous sampling conversion module 12, and completes the filtering calibration process by using the power frequency digital filtering algorithm and preset amplitude and phase calibration parameters to obtain the calibrated three-phase electrical parameter data and output it to the timing alignment output module 14. Specifically, the electrical parameter filtering calibration module 13 performs digital filtering and channel calibration processing on the raw instantaneous value sequence output by the multi-channel synchronous sampling conversion module 12. The filtering stage uses a fourth-order IIR bandpass digital filter with a passband center frequency of 50Hz and a -3dB bandwidth of 10Hz, effectively suppressing DC components, low-frequency drift, and interference from third and higher harmonics, while completely preserving the fundamental amplitude and phase information. The calibration stage calls the channel calibration parameter table pre-stored in the non-volatile memory for point-by-point correction: the amplitude calibration coefficient is obtained by segmented calibration of each channel across the full range, compensating for signal conditioning and ADC gain errors; the phase calibration coefficient is determined based on the measured phase frequency response offset of each channel at the 50Hz frequency point, compensating for transformer phase shift and filter group delay differences.

[0024] After calibration by the electrical parameter filtering calibration module 13, the comprehensive accuracy index of the instantaneous values ​​of the three-phase voltage and current is determined based on the engineering error propagation link. In this embodiment, the amplitude error of the front-end voltage / current transformer does not exceed ±0.1%, and the phase error does not exceed ±0.05; the amplitude error introduced by the signal conditioning and synchronous sampling link is approximately ±0.05%, and the phase deviation is approximately ±0.02; the passband fluctuation of the digital filtering link in the range of 50Hz ±0.5Hz is less than 0.01dB. The instantaneous values ​​of the calibrated three-phase voltage can be obtained by synthesizing the square root of each variance component. With the instantaneous value of three-phase current The overall amplitude error is better than ±0.2%, and the phase accuracy is better than ±0.1%. This accuracy level meets the requirements of the GB / T15543 standard for unbalance measurement accuracy, providing a reliable data foundation for subsequent unbalance calculation based on the symmetrical component method. The specific algorithm implementation for digital filtering and amplitude-phase calibration can refer to the commonly used power parameter measurement specifications in this field; the formulas for the electrical parameter filtering calibration module 13 are not detailed here. The calibrated data is then sent to the timing alignment output module 14.

[0025] The timing alignment output module 14 receives the calibrated three-phase electrical parameter data output by the electrical parameter filtering calibration module 13, completes the timing alignment processing of the three-phase data based on the zero-crossing point of the power frequency voltage, obtains the timing-aligned three-phase electrical parameter timing sequence, and outputs it to the three-phase unbalance predictive compensation strategy generation unit 2.

[0026] Specifically, the timing alignment output module 14 uses the positive zero-crossing point of the fundamental wave of phase A voltage as a unified reference to strictly align the calibrated six-channel instantaneous value sequence output by the electrical parameter filtering calibration module 13 on the time axis. The timing alignment output module 14 first detects... The sign of the time axis changes from negative to positive, and combined with linear interpolation, the sampling number corresponding to the zero-crossing point is accurately located as the reference time origin. Then, based on the fixed phase deviation of each channel relative to the A-phase voltage channel as specified by the factory, Lagrange quadratic interpolation is used to perform time shift resampling on the instantaneous value sequence of each channel, so that all six signals are synchronized to the same zero-crossing reference.

[0027] After alignment processing by timing alignment output module 14, the output includes a unified time tag. Instantaneous values ​​of three-phase voltage and instantaneous values ​​of three-phase current The structured time sequence has a constant time interval of 78.125 μs between adjacent records. This sequence eliminates the slight time deviation introduced by physical channel inconsistencies and filtering, accurately reflecting the instantaneous phase relationship of the three-phase electrical quantities on the low-voltage bus side. It provides a strictly synchronized time reference for the recursive calculation of the negative sequence component based on the symmetrical component method in the three-phase imbalance predictive compensation strategy generation unit 2. The time-aligned data is continuously transmitted to the three-phase imbalance predictive compensation strategy generation unit 2 through the high-speed data bus inside the fusion terminal.

[0028] The three-phase imbalance predictive compensation strategy generation unit 2 is mounted on the edge computing main control chip of the intelligent fusion terminal. It uses a timing prediction-loss constraint joint optimization algorithm to process the timing-aligned three-phase electrical parameter timing sequence, generating an advanced active compensation control command that balances minimum action loss and optimal compensation effect. It also receives the returned three-phase electrical parameter data to correct the prediction calculation parameters and loss constraint weights. Specifically, the three-phase imbalance predictive compensation strategy generation unit 2 receives the timing-aligned three-phase electrical parameter timing sequence output by the distribution network three-phase operation parameter synchronous acquisition unit 1, processes the timing sequence using the timing prediction-loss constraint joint optimization algorithm, generates an advanced active compensation control command that balances minimum action loss and optimal compensation effect, and receives the returned three-phase electrical parameter data from the compensation effect closed-loop feedback verification unit 4 to correct the prediction calculation parameters and loss constraint weights. The three-phase imbalance predictive compensation strategy generation unit 2 is composed of a negative sequence component timing prediction module 21, a loss constraint target construction module 22, a finite step optimization solution module 23, and a control parameter adaptive correction module 24, as follows: The negative-sequence component timing prediction module 21 receives the timing-aligned three-phase electrical parameter timing sequence, recursively calculates the evolution trend of the negative-sequence component of the three-phase current based on the timing rolling sampling window, and outputs the three-phase imbalance prediction state for the next cycle to the loss constraint target construction module 22. Specifically, unlike the passive response lag caused by the traditional scheme that only judges based on the instantaneous value of the imbalance at the current moment, this module achieves advanced prediction of the change trend of the negative-sequence component by introducing the timing rolling sampling window and least squares trend extrapolation, and sets up an intelligent trigger judgment mechanism to avoid malfunctions caused by short-term fluctuations. At the same time, this module performs prediction calculations with a fixed control cycle as the basic scheduling unit. The system control cycle is set to 200ms, corresponding to 10 power frequency cycles; the prediction result output, trigger judgment, and parameter correction are all based on the control cycle as a unified time reference. The processing flow of the negative-sequence component timing prediction module 21 specifically includes the following steps: S21.1 Receive the timing sequence of the three-phase electrical parameters aligned with the timing sequence, perform point-by-point recursive update on the timing rolling sampling window, and remove the earliest historical sampling point in the window for each new current sampling point to obtain the updated timing current sequence. In this step, the window length of the time-series rolling sampling window is defined as... One sampling point, The initial value is set to 256, corresponding to the number of sampling points within a complete power frequency cycle (20ms), and its value range is constrained as follows: and This ensures that at least one complete power frequency cycle is always covered within the window, guaranteeing the physical validity of the sequence component calculation. Let the sequence number of the latest sampling point be denoted as . Then the sequence of instantaneous three-phase current values ​​within the time-series rolling sampling window is: Each newly added current sampling point Simultaneously remove the earliest historical sampling points within the window. The window always retains the latest The instantaneous values ​​of the three-phase current corresponding to each sampling point are used to obtain the updated time-series current sequence. This rolling update mechanism ensures the continuous tracking capability of the negative-sequence component time-series prediction module 21 for load changes, while... The value can be dynamically adjusted by the adaptive correction module 24 of the control parameter according to the load fluctuation characteristics of the transformer area, so as to take into account both the response speed and smoothness of the prediction.

[0029] S21.2 Calculation of the negative sequence components of the three-phase current at each sampling moment within the rolling sampling window based on the symmetrical component method. The effective value completes the sequence component decomposition of the three-phase current; In this step, a sliding data window mechanism is used. For each sampling moment within the time-series rolling sampling window, a truncation length of [length missing] is extracted, with that moment as the endpoint. The historical current sampling sequence is used to extract the fundamental current phasor through windowed interpolation Discrete Fourier Transform (DFT). During the calculation, the power grid frequency is tracked in real time using a software phase-locked loop. A windowed interpolation algorithm is used to correct the amplitude and phase of the fixed-length DFT calculation results, suppressing spectral leakage errors caused by frequency fluctuations and ensuring the accuracy of fundamental phasor extraction. The calculated fundamental phasor is then used to calculate the effective value of the negative-sequence component at that sampling moment based on the symmetric component method. The results are stored in a historical buffer according to time sequence for use by the rolling sampling window. Taking phase A as an example, the real part of the fundamental current phasor... With the imaginary part Calculate using the following formula: ; ; in The number of sampling points within one power frequency cycle is taken in this embodiment. ; For phase A current at time... The instantaneous value, in A; The sampling point offset from the current moment backwards is a dimensionless integer. The fundamental current phasor of phase A is synthesized from its real and imaginary parts. The fundamental current phasors of phase B and phase C The same principle applies.

[0030] It should be noted that, Fixed data window length for DFT computation, versus dynamically adjustable time-series rolling sampling window length. Functionally independent. The DFT calculation requires... Each sampling point is taken from a historical data buffer ending at the current sampling time. This buffer is maintained independently and has a fixed length. When the time-series rolling sampling window length When adjusted by the adaptive correction module 24 of the control parameters, Keeping 256 constant, the DFT calculation is unaffected. The impact of changes. Time-series rolling sampling window. Used only for trend fitting in step S21.3, its lower limit of value range is constrained. The upper limit constraint is This is to ensure a balance between the statistical significance of the fitted samples and real-time performance.

[0031] After obtaining the three-phase fundamental current phasors, the negative-sequence component phasors are calculated using the symmetrical component method. Let the rotating operator... Then the negative sequence current phasor for: ; The effective value of the negative sequence component at that sampling moment is obtained by taking the magnitude of the negative sequence current phasor. :

[0032] All inside the window The above calculation is performed sequentially on each sampling point to obtain the sequence of effective values ​​of the negative-order components corresponding to the time-series rolling sampling window. ,in For the position index within the window, This sequence fully depicts the trajectory of changes in the degree of three-phase imbalance within the current time window.

[0033] Simultaneously, based on the latest set of DFT calculation results within the current window, the effective value of the current actual negative sequence component is calculated from the three-phase fundamental current phasor at the current moment using the symmetrical component method. and the effective value of the positive sequence component This leads to the current actual three-phase current imbalance. : ; Used in the subsequent intelligent triggering and judgment process, along with the predicted value. Make a joint judgment.

[0034] S21.3, Negative-order components within a time-series rolling sampling window The negative order component of the effective value sequence is calculated using the least squares method for fitting. The gradient and rate of change of; In this step, this embodiment uses a second-order polynomial model to analyze the effective value sequence of the negative order components. Perform fitting. Use the relative time of the sampling points within the window. As the independent variable, To use the sampling period Normalized integer index, Dimensionless; The sampling period is [value]. The fitted model is: ; in The unit is A, which means The reference value of the negative-order component at time; The unit is A, which represents the change of the negative order component in each sampling period, i.e., the first-order gradient. The unit is A, which represents the rate of change of the gradient of the negative order component, i.e., the second-order rate of change.

[0035] Utilize the window Data from each sampling point ,in Construct a system of normal equations: ; Solving the above system of equations using the Choreski decomposition method yields the coefficients. , , The optimal unbiased estimate. First-order gradient. Reflects the current increasing or decreasing trend and rate of change of the negative-order components, the second-order rate of change. This reflects whether the changing trend itself is accelerating or decelerating, and the two together constitute a quantitative description of the unbalanced evolution trend. Compared with existing techniques that only use the first-order difference method, this module introduces a second-order term, which can effectively capture the nonlinear changing characteristics of the load and significantly improve the prediction accuracy.

[0036] S21.4. Predict the negative sequence component of the next sampling period recursively based on the gradient and rate of change. The effective value is used to obtain the predicted state of the three-phase imbalance in the next cycle.

[0037] In this step, "next sampling period" specifically refers to the sampling time corresponding to one power frequency cycle offset from the end time of the current window, and the offset amount is... Each sampling interval; the prediction calculation is performed according to the system control cycle. In each control cycle, a trend fitting and extrapolation calculation is completed, and a prediction state is output for compensation decision-making.

[0038] Based on the coefficients obtained from the fitting in step S21.3 , , The effective value of the negative sequence component after one power frequency cycle is predicted using a second-order Taylor recursive formula. Let the normalized time corresponding to the end of the current window be... The normalized time increment corresponding to the predicted target time is (That is, the number of sampling points corresponding to one power frequency cycle). The predicted value is: ; If the predicted calculation results Then force set Since the effective value of the negative-order component is physically non-negative, the fitting extrapolation may produce negative values ​​under extreme trends. This abnormal situation is uniformly treated as a zero value.

[0039] Predict the effective value of the negative sequence component in the next cycle. Then, the predicted three-phase current imbalance needs to be calculated based on the effective value of the positive sequence component. Positive sequence current phasor Simultaneous calculation using the symmetric component method: ; The effective value of the positive-order component of the modulus value .

[0040] This embodiment employs engineering simplification: In a single-frequency cycle short-step prediction scenario, the change in the total load of the transformer area (positive-sequence component) is much smaller than the relative change in the unbalance (negative-sequence component), therefore the positive-sequence component uses the measured value of the current cycle. This simplification is applicable to stable operating conditions with a load change rate of less than 10% / s, and can significantly reduce the computational load while ensuring prediction accuracy; if a sudden load jump occurs, it switches to a direct decision-making mode based on measured values.

[0041] Then predict the three-phase current imbalance. for: ; in It is a dimensionless percentage, representing the ratio of the predicted negative-sequence current to the positive-sequence current in the next cycle.

[0042] This prediction of three-phase current imbalance This refers to the three-phase imbalance prediction state for the next cycle output by the negative sequence component timing prediction module 21. The above-mentioned second-order polynomial extrapolation prediction method is suitable for operating conditions with stable load changes. However, when sudden changes occur in the transformer area, such as a step switching of a single-phase high-power load, the prediction results may have significant deviations. Therefore, a load change identification and degradation processing mechanism is set up: calculating the difference between the effective value of the negative sequence component in the current cycle and that in the previous cycle. The sudden change threshold is set at 5% to 10% of the rated current of the transformer area, with a typical value of 8% of the rated current of the transformer area in typical projects; if If the load exceeds the preset threshold for sudden change, it is determined to be a sudden change in load condition. The prediction result is then invalidated, and the current actual imbalance is used directly as the decision-making basis. The system switches to the compensation strategy generation mode based on measured values ​​to avoid decision-making errors caused by prediction failure. After the sudden change in load condition ends, the system automatically reverts to the prediction-based compensation mode.

[0043] Meanwhile, the negative sequence component timing prediction module 21 is equipped with an intelligent triggering judgment step: With preset time-segmented dynamic thresholds The comparison is performed, and the current actual imbalance calculated in step S21.2 is also considered. Perform a joint determination. The determination rule is: if the following conditions are met... If the current imbalance is too high, it indicates that the imbalance in the next cycle is expected to exceed the limit, triggering the compensation strategy generation process and outputting the predicted state to the loss constraint target construction module 22; if the current actual imbalance is too high... But the predicted value This indicates that the imbalance will naturally fall below the threshold in the next cycle, suppressing this trigger and avoiding unnecessary compensation actions caused by transient fluctuations such as short-term start-stop of single-phase high-power loads.

[0044] Meanwhile, a fallback trigger mechanism is added: if the actual imbalance exceeds the standard but the prediction does not exceed the standard for three consecutive control cycles and the trigger is suppressed, or if the actual imbalance exceeds the threshold for 10 consecutive seconds, a compensation strategy generation process will be forcibly triggered to ensure that the system will inevitably perform governance actions when the imbalance exceeds the standard for a long period of time, thus avoiding governance failure caused by prediction bias.

[0045] Among them, time-based dynamic thresholds Based on historical operating data of the distribution area, thresholds are set differently according to peak, valley, and normal periods: the threshold is set at 5% for peak load periods (e.g., 08:00 to 12:00, 17:00 to 21:00), 3% for valley load periods (e.g., 00:00 to 06:00), and 4% for other normal periods. This time-based differentiated threshold strategy is one of the improvements of this module compared to the fixed threshold scheme, achieving an adaptive balance between governance effectiveness and equipment lifespan during different time periods.

[0046] The loss constraint target construction module 22 receives the predicted state of the three-phase imbalance for the next cycle, quantifies the operating loss of the compensation actuator as a constraint factor to construct a joint optimization objective function, and outputs the joint optimization objective function to the finite-step optimization solution module 23. Specifically, unlike traditional schemes that only minimize the imbalance as a single optimization objective, the core improvement of this module lies in explicitly incorporating the equipment operating loss into the objective function, constructing a joint optimization model that takes into account both the mitigation effect and the device lifespan. The processing flow of the loss constraint target construction module 22 specifically includes the following steps: S22.1 Receive the three-phase imbalance prediction state for the next cycle, determine the equivalent life loss of a single mechanical action of the commutation switch and the equivalent electrical loss of a single switching of the phase compensation branch, and quantify the action loss of the compensation actuator corresponding to a single action. In this step, the phase-change switch group in the transformer substation adopts a permanent magnet vacuum contactor type phase-change switch. The actual service life of the phase-change switch is affected by both the number of mechanical operations and the electrical corrosion of the contacts during load switching, among which the mechanical life is more important. 1×10 5 Next, electrical life Approximately 1 / 5 of the mechanical life, or 2 × 10⁻⁶. 4 This system ensures the commutator operates at the zero-crossing point of the power frequency current through a zero-current switching execution submodule (see Multi-type Compensation Device Cooperative Execution Unit 3), significantly reducing contact electrical erosion. Therefore, this embodiment calculates the loss equivalent primarily based on mechanical life, while reserving an electrical life correction interface for on-site calibration according to actual load conditions. The overall purchase cost of a single commutator switch... If the cost is 800 yuan, then the equivalent life loss of a single mechanical action of the commutator switch is... for: ; Substituting the numerical values, we get Yuan / time indicates the equivalent economic cost incurred due to the mechanical lifespan of the commutator for each phase switching operation. Yuan.

[0047] Furthermore, the phase-compensation branch uses parallel capacitor banks switched via AC contactors, and the equivalent electrical loss of a single switching operation is... Taking into account both the contactor contact erosion loss and the capacitor charging and discharging transient loss, this embodiment specifies the rated number of electrical operations for a single AC contactor. 5×10 4 Second, comprehensive cost The cost is 300 yuan; the transient energy loss during a single capacitor switching operation. It is 20J, and the unit price of electricity is... Take 1.67 × 10 -7 Yuan / J. Then Calculate using the following formula: ; Substituting into the calculation, we get Yuan / time indicates that the equivalent economic cost of each switching action performed by the phase compensation branch is 0.006 yuan.

[0048] Let the total number of operations of the commutator switch group in the current control cycle be . The total number of switching operations for the phase-compensation branch is Then the total amount of compensation for the action loss of the actuator. Expressed as: ; in The unit is yuan. For dimensionless degrees, The number is a dimensionless number.

[0049] S22.2 Set the standard limit value for the three-phase imbalance after compensation as a constraint condition, and clarify the feasible domain of the optimization solution; In this step, this embodiment sets the compensated three-phase current imbalance degree according to the provisions of the current national standard GB / T15543 "Power Quality Three-Phase Voltage Imbalance". It should meet the following requirements: This constraint serves as a hard constraint for joint optimization. Any combination of actions that causes the predicted imbalance after compensation to exceed this limit is deemed an infeasible solution and excluded from the feasible region of the optimization solution, thereby clarifying the range of the feasible region of the optimization solution.

[0050] S22.3. To achieve the optimal balance between imbalance suppression and total motion loss, a joint optimization objective function is constructed. The joint optimization objective function includes an imbalance suppression term and an action loss term. Weight coefficients are set for each term to complete the bi-objective optimization modeling under loss constraints.

[0051] In this step, this embodiment uses a normalized weighted summation method to construct a joint optimization objective function to eliminate the inconsistency in the dimensions of imbalance and motion loss. Joint optimization objective function Defined as: ; in To predict the three-phase current imbalance after compensation, a dimensionless percentage is used. As the normalized reference value for imbalance, this embodiment takes... That is, the permissible limit specified in the national standard; The total amount of compensation for the movement losses of the actuators is expressed in yuan. As the reference value for normalizing motion loss, this embodiment takes Yuan, corresponding to the maximum expected motion loss under the worst operating conditions (based on all). All phase-change switches operate simultaneously. (Estimation of extreme cases where component phase compensation branches are switched on and off simultaneously). The weight coefficient for the imbalance suppression term is dimensionless. The two weighting coefficients are dimensionless and satisfy the normalization constraint: .

[0052] Furthermore, normalization processing makes and All are converted to dimensionless ratios, objective function This is a dimensionless comprehensive evaluation value. The initial default value is set to... , This corresponds to the preference for "prioritizing governance effectiveness"; in actual operation, the adaptive correction module 24 dynamically adjusts the control parameters based on the compensation effect. Unlike conventional single-objective optimization that only pursues the lowest imbalance while ignoring the acceleration losses caused by frequent equipment movements, this joint optimization objective function introduces a movement loss term and sets adjustable weights, allowing the control strategy to flexibly trade off the imbalance suppression effect against the frequency of equipment movements. This is a significant algorithmic improvement of this module compared to existing technologies. The completed joint optimization objective function... Output to the finite-step optimization solution module 23.

[0053] The finite-step optimization solution module 23 receives the joint optimization objective function, generates an advanced active compensation control command that balances minimum motion loss and optimal compensation effect through finite-step optimization, and outputs it outward. Specifically, unlike traditional continuous optimization methods that rely on iterative solvers and may converge to local optima, this module discretizes the continuous control space of the commutation switch and compensation branch into a finite number of action combinations, and uses finite-step dynamic programming for global traversal optimization to ensure a deterministic global optimal solution is obtained under the constraints of embedded chip computing resources. This is the core improvement of the finite-step optimization solution module 23. The finite-step optimization solution module 23 consists of a discrete state construction submodule, a dynamic programming recursion submodule, and an optimal command output submodule, wherein: The discrete state construction submodule receives the joint optimization objective function, decomposes the commutation switch action combination and the phase compensation branch switching position into a finite discrete action state set, and outputs the discrete action state set to the dynamic programming recursive submodule. Specifically, the configuration of the transformer substation Each phase-change switch has three target phase options: "Connect to phase A", "Connect to phase B", and "Connect to phase C". (Note: The last part is a repetition of the first part and can be left as is.) The currently connected phases of the phase-switching switch are respectively The target is different after its action. The optional set is When the target phase is the same as the current phase, the switch action count is incremented. When the target phase is different from the current phase, the switch action count is increased. .all The operating states of the commutation switch constitute the commutation switch operating combination vector. The total number of theoretical combinations is 3 M Seeds. Meanwhile, the transformer area is equipped with... Each phase compensation branch has two discrete states: "engaged" and "disengaged." The action state vector is... ,in 0 represents withdrawal, 1 represents investment, and the theoretical total number of combinations is 2. K kind.

[0054] Simultaneously, the discrete state construction submodule needs to perform feasibility screening on all theoretical combinations, based on the imbalance constraints set in step S22.2. In this embodiment, it is composed of action combinations. Predicting Imbalance After Calculation and Compensation The method is as follows; First, based on the combination of commutation switch actions Determine the current distribution after load redistribution for each single-phase branch. Assume the active and reactive power vectors of each single-phase branch are known (uploaded to the intelligent fusion terminal by the acquisition units of each commutation switch branch via the local communication network). The calculation uses a constant current engineering approximation model, where the current amplitude and phase of a single branch remain unchanged before and after commutation. After commutation, the total current phasor of each phase is obtained by directly summing the phasors of the branch currents switched to the corresponding phase according to Kirchhoff's current law. This model is applicable to conventional distribution area conditions where low-voltage bus voltage fluctuations do not exceed ±5%. It is a general engineering approximation in the field of three-phase unbalanced commutation management in distribution networks, and its calculation accuracy meets engineering application requirements. Let the fundamental current phasors of phases A, B, and C after commutation be... Its value is obtained by re-accumulating the branch loads connected to each phase switch according to the phase after switching.

[0055] Secondly, based on the phase compensation branch switching vector Determine the reactive power compensation for each phase. Let the first phase be... The single-group capacity of the phase compensation branch is (Unit: kvar) If the connected phases are fixed as A, B, or C, then the compensated reactive power of each phase after switching is the sum of the capacities of all connected groups. The changes in reactive power of each phase after compensation further correct the imaginary parts of the current phasors of each phase, yielding the compensated three-phase fundamental current phasors. .

[0056] Finally, the phasors of the compensated three-phase fundamental currents Calculate the effective value of the negative-order component using the symmetric component method described in step S21.2. and the effective value of the positive sequence component After compensation, the three-phase current imbalance is predicted. : ; Satisfying all theoretical combinations The combination of these elements is preserved to form a finite set of discrete action states: ; Each element in This represents a feasible scheme for the combined operation of a commutator switch and a compensation branch, with a total number of elements. It is a finite positive integer.

[0057] In particular, if after feasibility screening If no combination of actions can make the compensated imbalance meet the standard limit, then the following exception handling strategy is activated: the constraint condition is changed from... Gradually relaxed to 5%, 6%, and so on, until... Not empty, take the relaxed constraint The smallest combination is taken as the approximate optimal solution; if no feasible solution is found after relaxing the limit to 6%, the current state of each compensation device remains unchanged, no action commands are output, and a "compensation unreachable" event is recorded in the terminal event log. This exception handling strategy ensures that the system will not fall into an unsolvable deadlock under extreme load conditions.

[0058] Finally, the set of discrete action states Output to the dynamic programming recursion submodule.

[0059] The dynamic programming recursive submodule receives a set of discrete action states, performs a finite-step recursive calculation according to the compensation stage, traverses all possible action combinations and calculates the corresponding joint optimization objective value, and outputs the objective calculation result of each action combination to the optimal instruction output submodule. Specifically, the dynamic programming recursive submodule receives a set of discrete action states. The calculation is performed in a finite number of steps according to the compensation phase, traversing all possible action combinations and calculating the corresponding joint optimization objective value.

[0060] This embodiment divides a complete compensation decision-making process into two recursive stages. The first stage is the commutation switch operation stage, whose state transition only involves... The phase adjustment of the commutator has an operating cost equal to the number of commutator operations. and The product; the second stage is the phase-compensation branch switching stage, which further implements the intermediate load distribution obtained after the phase commutation in the first stage. The switching operation of the component phase compensation branch has an action cost of the number of switching operations. and The product of.

[0061] The solution steps for the dynamic programming recursive submodule are as follows. First, iterate through all feasible commutation switch action combinations in the first stage, calculate the intermediate state after commutation for each combination, i.e., the three-phase current distribution after the load redistribution in each phase, and calculate the local action cost based on this. Then, based on each intermediate state in the first stage, the process is recursively applied to the second stage, traversing all switching combinations of the phase compensation branches, and calculating the predicted three-phase current imbalance after compensation using the method described in the discrete state construction submodule. and total operating cost Ultimately and Substitute the values ​​into the joint optimization objective function to calculate the objective function value corresponding to the complete action combination. The traversal process covers All Each action combination is recorded. corresponding The value is output to the optimal instruction output submodule. Because... Given a finite discrete set and a fixed number of two recursive stages, the above dynamic programming process is deterministically completed within a finite number of steps, without relying on iterative convergence conditions, and can be solved within the real-time task cycle of an embedded edge computing chip.

[0062] The optimal instruction output submodule receives the target calculation results of each action combination, selects the action combination with the smallest joint optimization target value, generates an advanced active compensation control instruction, and outputs it outward.

[0063] Specifically, the optimal instruction output submodule receives the target calculation results of each action combination and selects the action combination with the smallest joint optimization target value. ,Right now: ; If there are multiple action combinations that make If both values ​​are minimized, then the total number of actions is selected. The fewest combinations as This is to further reduce the frequency of equipment operation. According to Generate an advance active compensation control command. The command message includes the following fields: the target connected phase of each commutator. The timing markers for the operation of each phase-switching switch (the identifier of the switch to be operated and its switching time window), and the switching status of each phase compensation branch. The instruction message is output in structured message form through the high-speed data bus inside the intelligent fusion terminal to the collaborative execution unit 3 of the multi-type compensation device.

[0064] The adaptive control parameter correction module 24 receives the returned three-phase electrical parameter data, corrects the prediction calculation parameters and loss constraint weights based on the compensated operating data, and feeds the corrected parameters back to the negative sequence component timing prediction module 21 and the loss constraint target construction module 22. Specifically, unlike the conventional scheme where control parameters remain constant, the adaptive control parameter correction module 24 calculates the compensation deviation online and dynamically adjusts the prediction window length and optimization weights accordingly, enabling the entire control system to adapt to changes in the load characteristics of the transformer area. The adaptive control parameter correction module 24 consists of a compensation deviation calculation submodule, an operating parameter adjustment submodule, and a correction parameter feedback submodule, wherein: The compensation deviation calculation submodule receives the returned three-phase electrical parameter data, calculates the deviation between the actual three-phase imbalance and the expected imbalance after compensation, and outputs the deviation data to the operating parameter adjustment submodule. Specifically, the compensation deviation calculation submodule receives the three-phase electrical parameter data returned by the compensation effect closed-loop feedback verification unit 4. The returned data is after a preset delay following the completion of the compensation process. (After the system enters a new steady state) The sequence of instantaneous three-phase current values ​​is resampled, and the compensation deviation calculation submodule calculates the actual three-phase current imbalance after compensation using the symmetrical component method described in step S21.2. And the expected imbalance predicted by the negative sequence component timing prediction module 21 in the previous control cycle. Compare the results and calculate the deviation: ; The unit is a dimensionless percentage. This indicates that the forecast was overly optimistic, and the actual imbalance was higher than the forecast value. This indicates that the prediction was conservative, and the actual imbalance was lower than the predicted value. The magnitude of the deviation reflects the accuracy of the prediction. Output to the runtime parameter adjustment submodule.

[0065] The operation parameter adjustment submodule receives deviation data and dynamically adjusts the window length parameter and loss constraint weight coefficient of the time-series rolling sampling window according to the magnitude and direction of the deviation, so as to obtain the corrected prediction calculation parameters and loss constraint weight. Specifically, the operating parameter adjustment submodule receives the deviation amount. The window length parameter of the time-series rolling sampling window is dynamically adjusted independently based on the magnitude and direction of the deviation. Weighting coefficients for loss constraints Window length adjustment and weight adjustment run in parallel, independent of each other, and are triggered by different statistical characteristics.

[0066] Time-series rolling sampling window length parameter The adjustment rules are as follows. Define the deviation tolerance threshold. Number of consecutive observation periods If continuous Within each control cycle All greater than This triggers a window length adjustment. The adjustment direction is: if... exist If the value remains positive for a given period (indicating a persistently optimistic forecast, suggesting drastic load changes and an underfitting model), then the window length should be reduced. To improve response speed; if exist If the value remains negative for an extended period (indicating a consistently conservative forecast, suggesting a relatively gradual load change and an overfitting model), then increase the window length. To enhance smoothness, adjust the step size each time. With 32 sampling points, the adjusted window length is limited to the range [256, 512]. The minimum value is 256 (corresponding to 1 power frequency cycle, 20ms), and the maximum value is 512 (corresponding to 2 power frequency cycles, 40ms). The dynamic adjustment of the window length can adaptively match the load fluctuation characteristics of different transformer areas and even different time periods of the same transformer area. This is a direct improvement of the fixed window length scheme in this module.

[0067] Loss constraint weighting coefficient The adjustment rules are as follows. Define the cumulative integral of the absolute deviation: ; in For the most recent The deviation sequence for each control cycle. Using absolute deviation accumulation avoids the decrease in adjustment sensitivity caused by the mutual cancellation of positive and negative deviations. Simultaneously, this... Within a cycle Number of positive values The number of times the sum is negative .

[0068] like Exceeding the cumulative threshold and This indicates that the governance effect continues to fall short of expectations and the forecast remains overly optimistic. Therefore, the value of ω1 needs to be increased to strengthen the optimization weight of the imbalance suppression term. The adjustment amount is... ,Right now , If Σ Exceeding the cumulative threshold and This indicates that the governance effect continues to exceed expectations and the forecast was conservative, so it is appropriate to increase [the risk / reward]. Values ​​are used to reduce the frequency of equipment operation, i.e. ;like This indicates that the prediction accuracy is within an acceptable range, and the weighting coefficients remain unchanged.

[0069] Through the above dynamic adjustments, the window length... The response is to the degree of volatility in short-term forecast deviations, and the weighting ratio. The response is to the direction of long-term systemic deviations. The two mechanisms complement and coordinate with each other, enabling the control strategy to switch smoothly and continuously between "prioritizing governance effectiveness" and "prioritizing equipment lifespan".

[0070] Finally, the parameter adjustment submodule outputs the corrected prediction calculation parameters. Weighting coefficients for loss constraints To the parameter feedback submodule.

[0071] The parameter correction feedback submodule receives the corrected prediction calculation parameters and loss constraint weights, feeds back the corrected prediction calculation parameters to the negative sequence component time series prediction module 21, and feeds back the corrected loss constraint weights to the loss constraint target construction module 22.

[0072] Specifically, the parameter correction feedback submodule receives the corrected prediction calculation parameters. Weighting coefficients for loss constraints ,Will Feedback is sent to the negative-sequence component timing prediction module 21 to update the window length parameter of the timing rolling sampling window in step S21.1; Feedback is sent to the loss constraint objective construction module 22 to update the weight coefficients of the joint optimization objective function in step S22.3.

[0073] Furthermore, parameter updates employ a shared memory marking mechanism: a parameter shared area is allocated in the intelligent fusion terminal's memory. The negative-order component timing prediction module 21 and the loss constraint target construction module 22 read this shared area at the beginning of each control cycle to obtain the latest parameters. After completing parameter adjustments, the parameter correction feedback submodule writes the new parameters to the shared area and increments the parameter version number. If the new parameter value is the same as the current value, no write operation is performed to avoid unnecessary memory refreshes.

[0074] Through the aforementioned closed-loop self-correction mechanism, the three-phase imbalance predictive compensation strategy generation unit 2 can continuously monitor the compensation effect and adaptively adjust the core control parameters online, enabling the system to maintain stable governance effects and reasonable equipment operation frequency under different load compositions, different seasons, and different operating periods. This achieves a technical upgrade of three-phase imbalance governance from "open-loop control" to "closed-loop self-optimization".

[0075] The multi-type compensation device collaborative execution unit 3 receives proactive compensation control commands and, based on these commands, drives the transformer area phase-change switch group to complete phase-to-phase active load transfer and drives the phase-by-phase compensation branches to complete tiered switching, thus performing proactive three-phase imbalance mitigation operations. Specifically, it is responsible for converting the proactive compensation control commands generated by the three-phase imbalance predictive compensation strategy generation unit 2 into a deterministic operation sequence for the field compensation devices. The multi-type compensation device collaborative execution unit 3 consists of a control command parsing and collaborative scheduling module 31, a transformer area phase-change switch group driving module 32, and a phase-by-phase compensation branch switching module 33. The specific implementation methods of each module are described below.

[0076] The control command parsing and collaborative scheduling module 31 receives the advance active compensation control command, completes the command legality verification and execution timing allocation, and outputs the commutation control command to the transformer area commutation switch group drive module 32 and the switching control command to the phase compensation branch switching module 33, respectively. Specifically, the proactive compensation control command message is a structured data frame, including a frame header, command type flag, target phase field of the commutator switch, switching status field of the phase compensation branch, timestamp, and cyclic redundancy check code. After receiving the message, the control command parsing and coordinated scheduling module 31 first performs cyclic redundancy check. If the check passes, it parses the contents of each field; if the check fails, it discards the frame and records the check error event in the system log, waiting for the next cycle command.

[0077] After the instruction validity verification is passed, the control instruction parsing and coordinated scheduling module 31 performs execution timing allocation. The phase switching of the commutator and the switching of the phase compensation branch are executed sequentially: the commutator action is executed first, and after the feedback confirmation of the commutator action result, the switching of the phase compensation branch is executed. This timing allocation strategy ensures that the load distribution of each phase is determined and the compensation calculation basis is accurate when the phase compensation branch is switched. After the timing allocation is completed, the control instruction parsing and coordinated scheduling module 31 encapsulates the target phase and action timing information of the commutator into a commutator control instruction and outputs it to the transformer area commutator group drive module 32; it also encapsulates the phase compensation branch switching status into a switching control instruction and outputs it to the phase compensation branch switching module 33.

[0078] The transformer substation phase-change switch group drive module 32 receives the phase-change control command and drives the transformer substation phase-change switch group to perform phase switching operations, completing the inter-phase active load transfer; specifically, it includes a switch status verification submodule, a zero-current switching execution submodule, and an action result feedback submodule, wherein: The switch status verification submodule receives the commutation control command, collects the current position status and branch current amplitude of the commutation switch group in the transformer area, and outputs the execution command to the zero current switching execution submodule after confirming that the safe switching conditions are met. Specifically, the switch status verification submodule receives the commutation control command, collects the current position status and branch current amplitude of the commutation switch group in the transformer area, and outputs an execution command to the zero-current switching execution submodule after confirming that the safe switching conditions are met. In this embodiment, the commutation switch group includes The permanent magnet vacuum contactor type phase-commutation switch is connected in series with each single-phase power supply branch.

[0079] Furthermore, the switch status verification submodule detects the current connected phase of each commutation switch through the switch auxiliary contact and compares it one by one with the target phase in the commutation control command: if the current phase of a switch is consistent with the target phase, then the switch is not included in the current action sequence; if they are inconsistent, then the switch is marked as a switch to be switched.

[0080] For each switch to be switched, the switch status verification submodule further reads the secondary current amplitude of the current transformer in its branch. If the branch current amplitude exceeds the safe switching threshold (set to 80% of the rated current of the switching switch in this embodiment), the switch is determined not to meet the safe switching conditions, and its processing is postponed to the next control cycle. The postponement event is reported to the superior module. Only when all switches to be switched meet the safe switching conditions is the execution command output to the zero-current switching execution submodule.

[0081] The zero-current switching execution submodule receives the execution command and triggers the switch contacts to operate when the power frequency current crosses zero, thus completing the phase switching operation. Specifically, the zero-current switching execution submodule receives the execution command and triggers the switch contacts to operate when the power frequency current crosses zero, completing the phase switching operation. The zero-current switching execution submodule continuously monitors the instantaneous current value of the branch where the switch to be switched is located. When it detects that the instantaneous current value crosses the zero-value range (in this embodiment, [-0.05A, 0.05A]), it outputs a contact action trigger signal to the drive coil of the commutation switch. The switch contacts complete the opening and closing actions within this zero-current window, minimizing the erosion of the contacts by the electric arc. After issuing the trigger signal, the zero-current switching execution submodule starts a timeout timer. If no confirmation signal of switch operation completion is received within 100ms, the switching is deemed a failure, the fault event is recorded, and the failure status is sent back to the control command parsing and coordinated scheduling module 31.

[0082] The action result feedback submodule collects the switch position signal after switching and sends the switch action result back to the control command parsing and coordinated scheduling module 31.

[0083] Specifically, the action result feedback submodule collects the switch position signal after switching and sends the switch action result back to the control command parsing and coordinated scheduling module 31. After each switch action is completed, the action result feedback submodule reads the status of its auxiliary contacts, confirms that the actual connected phase of the switch is consistent with the target phase, and summarizes the action results (success / failure / delay) of each switch into an action result message, which is then sent back to the control command parsing and coordinated scheduling module 31.

[0084] Furthermore, the control command parsing and coordinated scheduling module 31 determines whether to continue the phase-by-phase compensation branch switching operation based on the action results: if all phase switching switches operate successfully, the switching control command is output to the phase-by-phase compensation branch switching module 33 according to the original timing sequence; if there is a switching failure or delay, the actual active load distribution of each phase is recalculated based on the actual connected phase of each switch fed back by the action results, and then the reactive power compensation demand and switching level of the phase-by-phase compensation branch are adjusted accordingly. After generating the updated switching command, it is output to ensure that the compensation amount matches the actual load distribution.

[0085] Phase compensation branch switching module 33 receives switching control commands and drives the phase compensation branch to perform graded switching operations to cooperate in completing the active treatment of three-phase imbalance.

[0086] Specifically, the configuration of the transformer substation Each phase-compensation branch is connected in parallel between the three-phase low-voltage busbars (A, B, and C) and the neutral line. Each phase-compensation branch consists of a set of fixed-capacity parallel capacitors connected in series with an AC contactor. The switching control command includes the target switching state for each phase-compensation branch. (0 for exit, 1 for engage). The phase compensation branch switching module 33 analyzes the instructions channel by channel. For compensation branches that require a change in switching status, it outputs contactor coil drive signals through the digital I / O interface to perform the engagement or disengagement operation.

[0087] To prevent inrush current impact caused by residual capacitor voltage, the phase compensation branch switching module 33, upon receiving an exit command, first disconnects the contactor of that branch and starts a discharge timer. Under the action of the discharge circuit, the residual capacitor voltage drops below the safety threshold through the discharge resistor within a preset discharge time (30 seconds in this embodiment) before the branch is allowed to re-engage. This discharge interlocking mechanism is implemented internally as a state machine, with each compensation branch switching between three states: "engaged," "exited," and "discharge waiting." The state transition logic is triggered by the switching control command. After completing all switching operations, the phase compensation branch switching module 33 transmits the actual switching status of each branch back to the control command parsing and coordinated scheduling module 31 for reference in the decision calculation of the next control cycle.

[0088] The compensation effect closed-loop feedback verification unit 4 re-acquires the instantaneous values ​​of the three-phase voltage and three-phase current on the low-voltage bus side of the distribution substation within a preset period after the compensation is completed, and transmits the re-acquired three-phase electrical parameter data back to the three-phase imbalance predictive compensation strategy generation unit 2. Specifically, this module is responsible for re-acquiring the three-phase electrical parameters on the low-voltage bus side of the distribution substation after the compensation is completed, and transmitting the re-acquired data back to the three-phase imbalance predictive compensation strategy generation unit 2, providing measured feedback basis for adaptive correction of control parameters. The compensation effect closed-loop feedback verification unit 4 is composed of a re-acquisition trigger control module 41, a three-phase electrical parameter re-acquisition module 42, and a re-acquisition data transmission module 43 in sequence. The specific implementation methods of each module are described below.

[0089] The re-sampling trigger control module 41 receives the compensation execution completion signal, triggers the generation of a re-sampling command after a preset delay, and outputs the re-sampling command to the three-phase electrical parameter re-acquisition module 42. Specifically, the compensation execution completion signal is issued by the control command parsing and coordinated scheduling module 31 after confirming that the feedback of the phase switching action result and the phase compensation branch switching operation have been completed, and that each actuator has entered a steady state. This signal is a digital level transition signal, which is connected to the re-sampling trigger control module 41 through the GPIO interface inside the fusion terminal.

[0090] After receiving the compensation completion signal, the resampling trigger control module 41 starts a delay timer with a preset delay duration. The delay duration is set based on the following engineering considerations: after the phase switching switch is switched and the capacitor is switched on, the three-phase voltage and current on the low-voltage bus side need to undergo a brief electromagnetic transient process, and at the same time, the parallel capacitor discharge circuit also needs a certain amount of time to reach a steady state; The delay ensures that the system has entered a new steady-state operating condition by the time of re-sampling, and the re-sampling data can accurately reflect the three-phase imbalance level after compensation, avoiding interference from transient processes on the accuracy of the feedback data. After the delay timer expires, the re-sampling trigger control module 41 generates a re-sampling command and outputs it to the three-phase electrical parameter re-acquisition module 42.

[0091] The three-phase electrical parameter re-acquisition module 42 receives the re-acquisition command, re-acquisitions the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution substation, and outputs the re-acquisitioned three-phase electrical parameter data to the re-acquisition data feedback module 43. Specifically, the three-phase electrical parameter re-acquisition module 42 reuses the hardware acquisition link of the distribution network three-phase operating parameter synchronous acquisition unit 1, namely, the front-end acquisition channel composed of the three-phase electrical parameter signal access module 11, the multi-channel synchronous sampling conversion module 12, the electrical parameter filtering and calibration module 13, and the timing alignment output module 14. The re-acquisition command triggers this acquisition link to execute a complete data acquisition process: after signal conditioning and isolation, the six analog signals are converted to digital by the multi-channel synchronous sampling ADC at a sampling rate of 12.8 kS / s for one power frequency cycle (256 sampling points); after filtering by a fourth-order IIR bandpass digital filter and channel amplitude and phase calibration, timing alignment is completed with the positive zero-crossing point of the fundamental wave of phase A voltage as the reference, finally obtaining the re-acquisitioned structured three-phase electrical parameter timing sequence, which includes a unified time label and the instantaneous values ​​of the three-phase voltage. With the instantaneous value of three-phase current .

[0092] Finally, after the three-phase electrical parameter re-acquisition module 42 completes the re-acquisition, it temporarily stores the re-acquisitioned three-phase electrical parameter data in the fusion terminal memory in the form of a buffer, and outputs a data ready signal to the re-acquisition data feedback module 43.

[0093] The re-sampling data transmission module 43 receives the three-phase electrical parameter data obtained from the re-sampling and transmits the three-phase electrical parameter data back to the three-phase imbalance prediction compensation strategy generation unit 2.

[0094] Specifically, the re-sampled data backhaul is implemented using a shared memory mechanism within the fusion terminal. The re-sampled data backhaul module 43 writes the re-sampled three-phase electrical parameter timing sequence into a preset shared memory address area according to a prescribed data structure (time tag and instantaneous value sequence of three-phase voltage and current). After writing, it increments the data version number and sets the data ready flag. The control parameter adaptive correction module 24 in the three-phase imbalance predictive compensation strategy generation unit 2 polls the data ready flag in each control cycle. Once the flag is detected as set, it reads the re-sampled data from the shared memory area and resets the flag.

[0095] Through the shared memory backhaul method described above, the resampled data does not need to go through serial communication or network protocol stacks, and the latency can be controlled at the microsecond level, meeting the real-time requirements of online self-correction of control parameters for feedback data. After the three-phase imbalance predictive compensation strategy generation unit 2 acquires the resampled data, the compensation deviation calculation submodule calculates the actual three-phase current imbalance after compensation based on the symmetrical component method. This leads to the adaptive correction of the prediction parameters and loss constraint weights, completing a full closed loop from perception, decision-making, execution to feedback verification.

[0096] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0097] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A smart fusion terminal three-phase imbalance active compensation control system, characterized in that, include: Distribution network three-phase operating parameter synchronous acquisition unit (1), the distribution network three-phase operating parameter synchronous acquisition unit (1) acquires the instantaneous values ​​of three-phase voltage and three-phase current on the low-voltage bus side of the distribution area, and performs filtering calibration and timing alignment processing on the acquired electrical parameter data to obtain the timing-aligned three-phase electrical parameter timing sequence; The three-phase imbalance predictive compensation strategy generation unit (2) is mounted on the edge computing main control chip of the intelligent fusion terminal. It uses the timing prediction-loss constraint joint optimization algorithm to process the timing sequence of the three-phase electrical parameters aligned with the timing, and generates an advanced active compensation control command that takes into account both the minimum action loss and the optimal compensation effect. It also receives the three-phase electrical parameter data back to correct the prediction calculation parameters and loss constraint weights. Multi-type compensation device collaborative execution unit (3) receives the advanced active compensation control command, and drives the transformer area phase switching switch group to complete the phase-to-phase active load transfer and drives the phase compensation branch to complete the graded switching according to the advanced active compensation control command, and performs the three-phase imbalance active treatment operation. The compensation effect closed-loop feedback verification unit (4) re-collects the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution substation within a preset period after the compensation is completed, and transmits the re-collected three-phase electrical parameter data back to the three-phase unbalance prediction compensation strategy generation unit (2).

2. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 1, characterized in that, The three-phase operating parameter synchronous acquisition unit (1) of the distribution network includes a three-phase electrical parameter signal access module (11), a multi-channel synchronous sampling and conversion module (12), an electrical parameter filtering and calibration module (13), and a timing alignment output module (14), wherein: The three-phase electrical parameter signal access module (11) accesses the three-phase voltage analog signal and the three-phase current analog signal on the low-voltage bus side of the distribution area, completes amplitude adaptation and electrical isolation processing through the front-end signal conditioning circuit, and outputs the adapted analog electrical parameter signal to the multi-channel synchronous sampling conversion module (12). The multi-channel synchronous sampling conversion module (12) receives the adapted analog electrical parameter signal output by the three-phase electrical parameter signal access module (11), and completes synchronous analog-to-digital conversion using a multi-channel synchronous analog-to-digital sampling channel triggered by the same source clock, to obtain the original three-phase voltage instantaneous value and the three-phase current instantaneous value digital quantity and output them to the electrical parameter filtering and calibration module (13). The electrical parameter filtering calibration module (13) receives the original three-phase voltage instantaneous value and three-phase current instantaneous value digital quantity output by the multi-channel synchronous sampling conversion module (12), and completes the filtering calibration process by using the power frequency digital filtering algorithm and preset amplitude and phase calibration parameters to obtain the calibrated three-phase electrical parameter data and output it to the timing alignment output module (14). The timing alignment output module (14) receives the calibrated three-phase electrical parameter data output by the electrical parameter filtering calibration module (13), completes the timing alignment processing of the three-phase data based on the zero-crossing point of the power frequency voltage, obtains the timing-aligned three-phase electrical parameter timing sequence, and outputs it to the three-phase unbalanced predictive compensation strategy generation unit (2).

3. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 1, characterized in that, The three-phase imbalance predictive compensation strategy generation unit (2) includes a negative sequence component timing prediction module (21), a loss constraint target construction module (22), a finite step optimization solution module (23), and a control parameter adaptive correction module (24), wherein: The negative sequence component timing prediction module (21) receives the timing sequence of the three-phase electrical parameters aligned with the timing sequence, calculates the evolution trend of the negative sequence component of the three-phase current based on the timing rolling sampling window, and outputs the three-phase imbalance prediction state of the next cycle to the loss constraint target construction module (22). The loss constraint target construction module (22) receives the three-phase imbalance prediction state of the next cycle, quantifies the loss of the compensation actuator into a constraint factor to construct a joint optimization objective function, and outputs the joint optimization objective function to the finite step optimization solution module (23). The finite step optimization solution module (23) receives the joint optimization objective function, generates an advanced active compensation control command that takes into account both minimum motion loss and optimal compensation effect through finite step optimization solution, and outputs it outward; The control parameter adaptive correction module (24) receives the three-phase electrical parameter data transmitted back, corrects the prediction calculation parameters and loss constraint weights according to the compensated operating data, and feeds back the corrected parameters to the negative sequence component timing prediction module (21) and the loss constraint target construction module (22).

4. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 3, characterized in that, The imbalance trend prediction process of the negative sequence component time series prediction module (21) includes the following steps: S21.1 Receive the timing sequence of the three-phase electrical parameters aligned with the timing sequence, perform point-by-point recursive update on the timing rolling sampling window, and remove the earliest historical sampling point in the window for each new current sampling point to obtain the updated timing current sequence. S21.2 Calculation of the negative sequence components of the three-phase current at each sampling moment within the rolling sampling window based on the symmetrical component method. The effective value completes the sequence component decomposition of the three-phase current; S21.3, Negative-order components within a time-series rolling sampling window The negative order component of the effective value sequence is calculated using the least squares method for fitting. The gradient and rate of change of; S21.

4. Predict the negative sequence component of the next sampling period recursively based on the gradient and rate of change. The effective value is used to obtain the predicted state of the three-phase imbalance in the next cycle.

5. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 3, characterized in that, The joint optimization objective construction process of the loss constraint objective construction module (22) includes the following steps: S22.1 Receive the three-phase imbalance prediction state for the next cycle, determine the equivalent life loss of a single mechanical action of the commutation switch and the equivalent electrical loss of a single switching of the phase compensation branch, and quantify the action loss of the compensation actuator corresponding to a single action. S22.2 Set the standard limit value for the three-phase imbalance after compensation as a constraint condition, and clarify the feasible domain of the optimization solution; S22.

3. To achieve the optimal balance between imbalance suppression and total motion loss, a joint optimization objective function is constructed. The joint optimization objective function includes an imbalance suppression term and an action loss term, with weight coefficients set for each term to complete the bi-objective optimization modeling under loss constraints.

6. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 3, characterized in that, The finite-step optimization solution module (23) includes a discrete state construction submodule, a dynamic programming recursion submodule, and an optimal instruction output submodule, wherein: The discrete state construction submodule receives the joint optimization objective function, decomposes the commutation switch action combination and the phase compensation branch switching position into a finite discrete action state set, and outputs the discrete action state set to the dynamic programming recursive submodule. The dynamic programming recursive submodule receives a set of discrete action states, performs a finite-step recursive calculation according to the compensation stage, traverses all possible action combinations and calculates the corresponding joint optimization objective value, and outputs the objective calculation result of each action combination to the optimal instruction output submodule. The optimal instruction output submodule receives the target calculation results of each action combination, filters out the action combination with the smallest joint optimization target value, generates an advanced active compensation control instruction, and outputs it outward.

7. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 3, characterized in that, The adaptive correction module for control parameters (24) includes a compensation deviation calculation submodule, an operating parameter adjustment submodule, and a correction parameter feedback submodule, wherein: The compensation deviation calculation submodule receives the returned three-phase electrical parameter data, calculates the deviation between the actual three-phase imbalance and the expected imbalance after compensation, and outputs the deviation data to the operating parameter adjustment submodule. The operating parameter adjustment submodule receives deviation data and dynamically adjusts the window length parameter and loss constraint weight coefficient of the time-series rolling sampling window according to the magnitude and direction of the deviation, so as to obtain the corrected prediction calculation parameters and loss constraint weight. The correction parameter feedback submodule receives the corrected prediction calculation parameters and loss constraint weights, feeds back the corrected prediction calculation parameters to the negative sequence component time series prediction module (21), and feeds back the corrected loss constraint weights to the loss constraint target construction module (22).

8. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 1, characterized in that, The multi-type compensation device collaborative execution unit (3) includes a control command parsing and collaborative scheduling module (31), a transformer area phase switching group drive module (32), and a phase compensation branch switching module (33), wherein: The control command parsing and collaborative scheduling module (31) receives the advance active compensation control command, completes the command legality verification and execution timing allocation, and outputs the commutation control command to the transformer area commutation switch group drive module (32) and the switching control command to the phase compensation branch switching module (33). The transformer area phase switching switch group drive module (32) receives the phase switching control command and drives the transformer area phase switching switch group to perform phase switching operation to complete the phase-to-phase active load transfer. The phase compensation branch switching module (33) receives the switching control command and drives the phase compensation branch to perform a graded switching operation to complete the active treatment of three-phase imbalance.

9. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 8, characterized in that, The transformer substation phase-change switch group drive module (32) includes a switch status verification submodule, a zero-current switching execution submodule, and an action result feedback submodule, wherein: The switch status verification submodule receives the commutation control command, collects the current position status and branch current amplitude of the commutation switch group in the transformer area, and outputs the execution command to the zero current switching execution submodule after confirming that the safe switching conditions are met. The zero-current switching execution submodule receives the execution command and triggers the switch contacts to operate when the power frequency current crosses zero, thus completing the phase switching operation. The action result feedback submodule collects the switch position signal after switching and sends the switch action result back to the control command parsing and collaborative scheduling module (31).

10. The intelligent fusion terminal three-phase imbalance active compensation control system according to claim 1, characterized in that, The compensation effect closed-loop feedback verification unit (4) includes a re-sampling trigger control module (41), a three-phase electrical parameter re-acquisition module (42), and a re-sampling data feedback module (43), wherein: The re-sampling trigger control module (41) receives the compensation execution completion signal, triggers the generation of a re-sampling command after a preset delay, and outputs the re-sampling command to the three-phase electrical parameter re-acquisition module (42). The three-phase electrical parameter re-acquisition module (42) receives the re-acquisition command, re-acquisitions the instantaneous values ​​of the three-phase voltage and the three-phase current on the low-voltage bus side of the distribution area, and outputs the re-acquisitioned three-phase electrical parameter data to the re-acquisition data feedback module (43). The re-sampling data feedback module (43) receives the three-phase electrical parameter data obtained from the re-sampling and feeds the three-phase electrical parameter data back to the three-phase imbalance prediction compensation strategy generation unit (2).

Images