Multi-time scale reactive power fusion resource coordination control method and system based on phase modifier

Abstract

The invention discloses a multi-time-scale reactive power fusion resource coordination control method and system based on a phase modifier, particularly relates to the field of voltage stabilization and reactive power compensation of a power system, and is used for solving the problems of unstable state judgment, device output swing and secondary disturbance caused by a traditional fixed threshold criterion. The method comprises the steps of generating a time sequence data frame by collecting bus voltage and phasor as well as a device state and a limit value, performing phase angle consistency judgment on the time sequence data frame, calculating a phase angle double-window stability degree and a backing response residual error ratio, and generating a handover credible coefficient output state label object by adopting sequential backing dominant learning. When the current state is a transient state, a bottom instruction object is generated to specify rapid reactive power resource execution, when the current state is converted into a steady state, a relay sequence object is generated to issue a phase modifier takeover target and a rapid reactive power resource exit target, deviation is verified in an execution period, an update object is calculated once again, smooth relay and ordered labor division are realized, and voltage quality and operation efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of voltage stability and reactive power compensation in power systems, and more specifically, to a multi-time-scale reactive power fusion resource coordination control method and system based on synchronous condensers. Background Technology

[0002] In hierarchical control with voltage and reactive power as the core objectives, synchronous condensers, static var compensators (SVCs), and grid-connected inverters share the responsibility of stabilizing voltage and distributing reactive power. The master station and substations coordinate according to predetermined voltage targets and bandwidth: rapid response devices provide initial support, and once the voltage stabilizes, the synchronous condenser assumes long-term support, aiming for a smooth relay and orderly division of labor. This approach has become standard practice in engineering projects. Strategy triggering relies heavily on criteria such as whether the grid connection point voltage enters the bandwidth or exceeds thresholds. The control link spans different time scales, and data comes from multiple sources of measurement and multiple control units. The overall operation is highly sensitive to the stability of state determination and switching timing.

[0003] The method of distinguishing between steady-state and transient states based on fixed bandwidth and thresholds, and switching accordingly, is prone to unstable state determination in actual operation. This leads to repeated takeover and concession between the fast-acting device and the synchronous condenser. This manifests as the target voltage and device output fluctuating near the bandwidth edge, with commands becoming inconsistent in strength. Some devices idle, while others passively cancel each other out, making it difficult to maintain consistent coordination. Frequent switching on and off introduces secondary disturbances, simultaneously damaging voltage quality and operational efficiency.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-timescale reactive power fusion resource coordination control method and system based on a synchronous condenser. This method generates time-series data frames by collecting bus voltage and phasors, device status and limits. Phase angle consistency is determined on these time-series data frames, and the phase angle dual-window stability and bottoming-out response residual ratio are calculated. Sequential bottoming-out dominance learning is used to generate handover reliability coefficients and output state label objects. When in a transient state, a bottoming-out command object is generated to specify the execution of rapid reactive power resources. When transitioning to a steady state, a relay sequence object is generated to issue the synchronous condenser takeover target and the rapid reactive power resource exit target. Deviations are verified during execution, and the objects are recalculated and updated in a single operation. This achieves smooth relay and orderly division of labor, improving voltage quality and operating efficiency, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A multi-timescale reactive power fusion resource coordination control method based on synchronous condensers includes the following steps: S1: Obtain bus voltage and phasor, device status and limit, complete unified clock alignment, and generate timing data frames covering synchronous condensers, static var compensators and grid-connected inverters; S2: Provide phase angle consistency judgment on the time series data frame, use phase angle consistency judgment to uniformly constrain the segment selection and sample selection of phase angle double window stability and bottoming response residual ratio, use sequential bottoming dominance learning to generate handover credibility coefficient, and output state label object based on handover credibility coefficient; S3: When the status label object is transient, generate a backup instruction object based on the device's feasible domain and current operating constraints; S4: When the state label object transitions to a steady state, a relay sequence object is generated based on the handover time and the feasible region of the device. S5: During the execution of the relay sequence object, check the target voltage deviation and the difference between the command execution. If the cancellation condition is not met, trigger a re-identification and recalculation, update the status label object, the bottoming command object and the relay sequence object, and output the final takeover result.

[0007] In a preferred embodiment, step S1 includes collecting bus voltage, phasor, device status, and device limit values ​​from the station and master station as raw inputs, correcting the timestamps using GPS timing signals, aggregating measurements taken at the same sampling time as records to form a preliminary time-series data frame, removing duplicate records and filling in missing records, aligning device fields using a unified naming system, and generating a multi-field structure containing sampling time, bus voltage, phasor, device status, device limit values, and real-time output.

[0008] In a preferred embodiment, step S2 includes calculating the difference trajectory of each device phasor relative to the bus phasor based on the time-series data frame and with the bus phasor as a reference to form a consistent segment; selecting short-window trajectory and long-window trajectory within the consistent segment to calculate the phase angle dual-window stability to quantify the convergence characteristics; retrieving steady-state segments from the operation record to construct a nearest-neighbor response mapping to calculate the bottom-line response residual ratio to evaluate the deviation; scanning parameter sequence cumulative trend evidence within the consistent segment to generate a handover confidence coefficient; and outputting a state label object based on the handover confidence coefficient.

[0009] In a preferred embodiment, the parameter sequence refers to the phase angle dual-window stability sequence and the bottoming response residual ratio sequence; the state labels include steady state and transient state.

[0010] In a preferred embodiment, step S3 includes extracting device limits, device status, current reactive power output, commissioning mode, and ramping capability from the time-series data frame, integrating the device feasible domain, sorting and allocating the bottom-line targets according to bus voltage trends and response capabilities, and writing them into the first part of the bottom-line instruction object. The second part is written with the handover confidence coefficient partitioning, the phase angle consistency determination persistence and the bottom-line response residual ratio convergence direction as the cancellation condition, and the earlier of the cancellation condition satisfaction time and the earliest time point in the available output range of the synchronous condenser as the handover time, and writing it into the third part. The bottom-line instruction object is marked to specify the execution of fast reactive power resources.

[0011] In a preferred embodiment, step S4 includes extracting the upper and lower limits of the feasible domain of the synchronous condenser from the handover time when the state label object transitions to a steady state, using the segmented linearly increasing output trajectory of the ramp capability as the takeover target, and using the output trajectory of rapidly decreasing reactive resources in the opposite direction as the exit target, ensuring that the output change signs are opposite at the same time point, marking the target value of each node, and generating a relay sequence object with the target execution device and the effective time.

[0012] In a preferred embodiment, step S5 includes extracting the real-time bus voltage and real-time device output from the timing data frame during the execution of the relay sequence object, calculating the target voltage deviation and command execution difference for continuous monitoring, and maintaining the trajectory until the takeover is completed if the cancellation condition is met.

[0013] In a preferred embodiment, if the condition is not met, the phase angle consistency determination is re-executed, the phase angle dual-window stability and the bottoming response residual ratio are calculated, the sequential bottoming dominance learning is entered to update the handover reliability coefficient, a new state label object and change time are generated, the bottoming instruction object and relay sequence object are reconstructed, and the final takeover result is output.

[0014] A multi-timescale reactive power fusion resource coordination control system based on synchronous condensers includes: The data acquisition module acquires bus voltage and phasors, device status and limits, completes unified clock alignment, generates timing data frames covering synchronous condensers, static var compensators and grid-connected inverters, and outputs timing data frames. The state determination module provides phase angle consistency determination on the time-series data frame. The phase angle consistency determination is used to uniformly constrain the segment selection and sample selection of phase angle dual-window stability and bottoming response residual ratio. Sequential bottoming dominance learning is used to generate handover confidence coefficients, and the state label object is output based on the handover confidence coefficients. When the status label object is transient, the bottom-line generation module generates a bottom-line instruction object based on the device's feasible domain and current operating constraints. When the state tag object transitions to a steady state, the relay construction module generates a relay sequence object based on the handover time and the feasible region of the device. The execution verification module checks the target voltage deviation and the instruction execution difference during the execution of the relay sequence object. If the cancellation condition is not met, a re-identification and recalculation is triggered to update the status label object, the bottoming instruction object, and the relay sequence object, and output the final takeover result.

[0015] The technical effects and advantages of this invention based on a multi-timescale reactive power fusion resource coordination control method and system using a synchronous condenser are as follows: This invention combines the unified generation of time-series data frames with the constraint of phase angle consistency determination to ensure the accuracy of multi-source data during metastable state switching. This ensures that the generation of the bottoming instruction object and the relay sequence object is based on reliable evidence, avoids repeated disturbances caused by traditional threshold criteria, and achieves a smooth connection between rapid reactive power resource bottoming and long-term synchronous condenser support, thereby improving the continuity of voltage stability.

[0016] The execution period deviation verification mechanism is further coordinated with the re-identification and recalculation process to dynamically correct the status label object and instruction sequence, prevent output overlap or deviation amplification, optimize the orderly division of labor among devices, reduce the impact of secondary disturbances on the power grid, and improve overall operating efficiency. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser according to the present invention.

[0018] Figure 2 This is a schematic diagram of the multi-timescale reactive power fusion resource coordination control system based on a synchronous condenser according to the present invention. Detailed Implementation

[0019] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: Figure 1 This invention presents a multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser, comprising: S1: Obtain bus voltage and phasors, device status and limits, complete unified clock alignment, and generate timing data frames covering synchronous condensers, static var compensators and grid-connected inverters.

[0021] S2: Provide phase angle consistency determination on the time-series data frame, use phase angle consistency determination to uniformly constrain the segment selection and sample selection of phase angle dual-window stability and bottoming response residual ratio, use sequential bottoming dominance learning to generate handover credibility coefficient, and output state label objects based on handover credibility coefficient.

[0022] S3: When the status label object is transient, a backup instruction object is generated based on the device's feasible domain and current operating constraints. The backup instruction object includes the cancellation condition and handover time, and specifies the execution of fast reactive resources.

[0023] S4: When the status label object transitions to a steady state, a relay sequence object is generated based on the handover time and the feasible domain of the device. This object is used to issue takeover targets to the synchronous condenser and exit targets to the rapid reactive resources, while maintaining a non-overlapping order.

[0024] S5: During the execution of the relay sequence object, check the target voltage deviation and the difference between the command execution. If the cancellation condition is not met, trigger a re-identification and recalculation, update the status label object, the bottoming command object and the relay sequence object, and output the final takeover result.

[0025] In power systems, reactive power resources, such as synchronous condensers, need to work in conjunction with static var compensators (SVCs) and grid-connected inverters to achieve voltage stability and reactive power sharing across multiple time scales. However, inconsistencies in the timing of multi-source measurement data often lead to unstable state determination and cause repeated fluctuations in device output. Step S1 generates a unified time-series data frame, ensuring data alignment in time and format, laying a solid foundation for subsequent phase angle consistency determination and handover reliability coefficient calculation. This step starts with the acquisition of raw inputs, gradually achieving correction, aggregation, cleaning, and alignment, ultimately outputting a standardized structure to support the smooth execution of control methods.

[0026] S1-1 Data Acquisition and Timestamp Correction.

[0027] In power systems, the coordinated control of multi-source reactive power resources faces the problem of inconsistent data timing, which may lead to deviations in state determination and fluctuations in device output. Achieving time alignment through a unified timing signal can ensure the accuracy and real-time performance of subsequent analysis, providing a reliable foundation for multi-timescale collaboration.

[0028] Bus voltage, phasors, device status, and device limits are collected from the station and from the master station, respectively, and these measurements are used as raw inputs. The timestamps of each channel are corrected using the GPS timing signal as a unified reference.

[0029] The specific process is as follows: First, the deviation between each timestamp and the unified time synchronization signal is calculated. The deviation is defined as the channel timestamp minus the corresponding time of the unified time synchronization signal. If the deviation exceeds a preset threshold, a linear interpolation algorithm is applied for adjustment. The linear interpolation formula is expressed as: ,in This indicates the adjusted timestamp. Indicates the timestamp of the preceding adjacent record. Indicates the timestamp of the next adjacent record. Indicates the current deviation. This indicates the deviation from the previous adjacent record. This represents the deviation of the next adjacent record. A linear interpolation algorithm is used to ensure that the adjusted value is based on a linear distribution of adjacent points, controlling the deviation within a preset threshold. After correction, the original value and adjusted value of each timestamp are recorded to form the corrected original dataset.

[0030] S1-2 Data aggregation and record formation.

[0031] Based on the original dataset after timestamp correction, the scattered measurement points need to be integrated into a continuous sequence to support subsequent calculations such as phase angle consistency determination. The aggregation process must strictly match the sampling time to avoid information loss or redundancy, thereby improving the robustness of the control method.

[0032] Based on the corrected original dataset, the data is sorted according to the sampling time. For bus voltage, phasor, device status, device limits, and real-time output at the same sampling time, the corresponding measurements are merged into a single record. This includes traversing all channels, matching entries with equal timestamp values, and integrating them into a record structure containing multiple fields, each corresponding to a specific measurement type, such as the bus voltage field storing voltage amplitude and phase angle. Multiple consecutive such records are arranged chronologically to form a preliminary time-series data frame. Each record in this preliminary time-series data frame represents comprehensive information for a complete sampling time, supporting subsequent cleaning operations.

[0033] S1-3 Data Cleaning and Missing Report Filling.

[0034] The initial time-series data frames may contain duplicate or missing entries, which can interfere with the sample selection for sequential bottom-line dominant learning. Targeted cleaning and imputation can maintain data continuity and prevent the amplification of biases during the generation of state label objects.

[0035] On the initial time-series data frame, all records are scanned to identify duplicate timestamp entries. Only the first record appearing in the time sequence is retained, and the remaining duplicate records are deleted. For missing records, a Kalman filter algorithm is used to estimate the value based on the bus voltage, phasor, device status, device limits, and real-time output of adjacent records. The Kalman filtering process includes a prediction step and an update step: the prediction step uses a state transition model to calculate the estimated value, using the formula: ,in This represents the predicted state vector at the current moment (including measurements such as bus voltage). Represents the state transition matrix. This represents the updated state vector from the previous time step. Represents the control input matrix. This represents the control input from the previous moment. The update step then integrates the measured values ​​to correct the prediction, using the following formula: ,in This represents the updated state vector at the current moment. Indicates Kalman gain, Represents the current measured vector. This represents the observation matrix. Estimated values ​​are labeled with the original displacement markers as the estimation type, while measured values ​​are labeled with the measured type. After processing, the record sequences of the time-series data frames are free of duplication and significant omissions, providing a basis for field alignment.

[0036] S1-4 Field Alignment and Structure Generation

[0037] The cleaned time-series data frames need to be uniformly named to cover different device types; otherwise, it will hinder the generation of backup instruction objects and relay sequence objects. Standardized field alignment enables seamless integration and supports reactive power fusion coordination among multiple devices.

[0038] By adopting a unified naming system, the device status, device limits, and real-time output of synchronous condensers, static var compensators (SVCs), and grid-connected inverters are aligned. For example, the status field of a synchronous condenser is named "Device Status Synchronous Condenser," the limit field of an SVC is named "Limited Value Static Var Compensator," and the output field of a grid-connected inverter is named "Real-Time Output Inverter," ensuring that the corresponding field names for all device types are standardized. The generated multi-field structure includes sampling time, bus voltage, phasor, device status, device limits, and real-time output, with each field labeled with the device type and aligned. This multi-field structure is the final timing data frame, which can be directly used for phase angle consistency determination in step S2.

[0039] After step S1 is completed, a time-series data frame containing a standardized multi-field structure is formed. This data frame covers the timing information of all devices, ensuring the stability and accuracy of state determination in multi-timescale coordination and avoiding secondary disturbances caused by data inconsistency.

[0040] In power systems, although time-series data frames provide a unified input, inconsistencies in phase angle trajectories and fluctuations in response residuals often lead to repeated transient-to-steady-state transitions, affecting the efficiency of reactive power resource allocation. Step S2 introduces phase angle consistency determination as a constraint, and combined with parameter calculation and sequential learning, accumulates convergence evidence to generate a handover reliability coefficient, thereby accurately outputting state label objects and change times, supporting the reliable generation of backup command objects and relay sequence objects. This step emphasizes time-series scanning of trend evidence, ensuring that disturbances caused by fixed thresholds are avoided in multi-timescale coordination, achieving smooth relay.

[0041] S2-1 Phase Angle Consistency Determination.

[0042] After the time-series data frame provides standardized input, a reference benchmark for the phase angle trajectory needs to be established first to avoid interference from inconsistent sections in subsequent parameter calculations. Based on the time-series data frame, and using the bus phasor as the reference object, the difference trajectory between each device phasor and the bus phasor is calculated.

[0043] For each record, the sign of the phase angle increase / decrease relationship is extracted. A positive sign corresponds to an increase in phase angle, and a negative sign corresponds to a decrease in phase angle. The sign switch is used as the inflection point criterion. Simultaneously, the reverse appearance of the phase angle change direction is checked in adjacent records. If the change direction of the current record is opposite to that of the next record, it is confirmed, forming a set of inflection points within the same observation window. The observation window is a fixed-length time series segment, sliding segment by segment from the beginning of the time series data frame. The absolute value of the inflection point time difference is calculated and compared with a preset alignment window. If the absolute value of the inflection point time difference falls into the preset alignment window, the alignment is considered successful. Sequential consistency is determined by comparing the matching degree between the order of inflection point appearance and the bus reference order. The matching degree is calculated as the ratio of the number of sequentially matched inflection points to the total number of inflection points. If the matching degree is higher than a preset standard, consistency is considered. The phase drift direction of the bus and the device is considered; if the drift is unidirectional, consistency is supported. The result of consistency or inconsistency is given, and the duration of the current consistency period is recorded as the duration from the start of the determination to the current record. A consistent segment is thus formed. Subsequent calculations only begin after a consistent segment is formed. If no consistent segment is formed, the previous state label object remains unchanged, and the observation window is extended by one record length to continue scanning the time-series data frames. The consistent segment and the determination result are output as constraints for parameter calculation.

[0044] S2-2 Phase Angle Double Window Stability Calculation.

[0045] Once the uniformity zone is established, the stabilization characteristics of short and long windows must be quantified within the constrained trajectory segment to capture evidence of voltage convergence from transient to steady state. The phase angle dual-window stability calculation consists of two steps: The first step involves selecting short-window and long-window trajectories within the consistent region. The short-window trajectory is the most recent fixed short-duration sub-segment, and the long-window trajectory is a fixed long-duration sub-segment encompassing the short window. The sum of the phase angle changes within the short window is compared with the sum within the long window. If both have the same sign and the absolute value of the short-window sum is less than that of the long window, then they converge in the same direction. The number of oscillations is obtained through sign switching counts. If the number of oscillations in the short window is not higher than that in the long window, the condition is met. A short-window stabilization marker is set to 1 if it is stable and 0 otherwise, and a long-window stabilization marker is set to 1 if it is stable and 0 otherwise. The two markers are added together to form the initial stability.

[0046] The second step involves matching the timestamps of the bus inflection point and the device inflection point within the alignment section. If the time difference falls within a preset alignment window, the alignment is successful. The proportion of successfully aligned inflection points is counted as the alignment ratio. The initial stability is multiplied by the alignment ratio to obtain the phase angle dual-window stability, which is dimensionless. This calculation and sample selection are limited to records within the alignment section.

[0047] S2-3 Calculation of the residual ratio of the bottoming response.

[0048] To assess the residual effect of rapid reactive power resource support, a nearest neighbor mapping must be constructed from historical steady-state segments to quantify the deviation between the current response and the expected response.

[0049] In the time-series data frame operation record, steady-state segments that match the current record of the device status and device limit, and whose power flow direction is consistent, are retrieved. The power flow direction is determined by the reactive power symbol. The selection criteria for steady-state segments are: unchanged device status, no device limit triggered, and monotonic bus voltage trajectory or no more than one change in direction. From the selected segments, pairs of small changes in reactive power output and small changes in bus voltage are extracted to form a nearest neighbor response mapping. A linear regression algorithm is used to fit the mapping, and the regression slope represents the voltage sensitivity to reactive power.

[0050] The actual reactive power change during the current support period is substituted into the model to obtain the predicted voltage change. This is compared with the actual bus voltage change, and the difference is calculated and divided by the predicted voltage change to obtain the support response residual ratio, which is dimensionless. The calculation segment and sample selection are constrained by the consistency segment. The output support response residual ratio sequence is input together with the phase angle dual-window stability sequence for sequential learning.

[0051] S2-4 Sequential bottom-line dominance learning and state output.

[0052] Once the input parameters are ready, sequential scanning is used to accumulate trend evidence to generate a handover confidence coefficient and trigger a state change.

[0053] Within the consistent segment, the latest values ​​of the phase angle dual-window stability index and the bottoming response residual ratio are scanned sequentially over time. The continuous duration of the phase angle dual-window stability index increase is recorded (accumulated when the current value is higher than the previous value, otherwise reset); the continuous duration of the bottoming response residual ratio decrease is recorded (accumulated when the current value is lower than the previous value, otherwise reset). If both continuous durations simultaneously exceed the preset evidence window without reverse interruption, the handover confidence coefficient is increased by one unit; if either shows a reverse change, it is decreased by one unit. The handover confidence coefficient is initialized to an intermediate value, and partitions are designated as lower, middle, and higher partitions. The lower partition corresponds to values ​​below the lower threshold, the middle partition corresponds to values ​​between the lower and upper thresholds, and the higher partition corresponds to values ​​above the upper threshold. When entering a new partition for the first time, the entry time is used as the change time.

[0054] The higher partition outputs the steady-state state label object and change time, the lower partition outputs the transient state label object and change time, and the intermediate partition retains the existing state label object and updates the timestamp. The handover reliability coefficient, state label object, and change time are output for use in steps S3 and S4.

[0055] After step S2 is completed, a handover reliability coefficient and status tag object are generated. This object marks the time of change based on accumulated evidence, ensuring orderly division of labor in reactive power fusion coordination and improving voltage quality and operating efficiency.

[0056] In power systems, although status tag objects mark transient phases, the backup of rapid reactive power resources is easily affected by device limits and operational constraints, leading to uneven output distribution or handover delays. By generating a backup instruction object in step S3, objectives and conditions can be comprehensively constrained to ensure accurate execution and orderly exit of transient backup, supporting smooth takeover by synchronous condensers. This step focuses on defining the feasible domain and triggering mechanisms, enabling priority support for rapid devices in multi-timescale coordination and avoiding amplification of disturbances under traditional criteria.

[0057] Determination of the feasible region of device S3-1.

[0058] Once the status label object is confirmed to be in a transient state, it is necessary to first define the operating boundaries of each device by comprehensively considering multiple factors in order to guide the allocation of the bottom-line target.

[0059] The device limits, device status, current reactive power output, commissioning mode, and ramp-up capability are extracted from time-series data frames. Device limits include upper and lower bounds for reactive power output; device status covers operating or fault modes; current reactive power output is taken from real-time records; commissioning mode distinguishes between automatic and manual control; and ramp-up capability represents the adjustment rate. Integrating these factors, a linear programming algorithm is used to solve for the feasible region of the device. The model formula is as follows: The main body is ,in Indicates the maximum feasible reactive power output. Represents the target coefficient vector. This represents a vector of decision variables, containing the current reactive power output and adjustment amounts. The constraint matrix is ​​represented by integrating device limits and climbing ability. The constraint right-hand vector is represented based on the device state and commissioning mode. Static upper and lower limits and dynamic boundaries are obtained by solving for fast reactive power resources and synchronous condensers respectively. The output device feasible region serves as the basis for generating the bottom-line command object, ensuring that the allocation conforms to the actual constraints.

[0060] The target coefficient vector is determined by the optimization objective of the linear programming problem. Specifically, it is obtained by setting it as a unit vector to maximize the total reactive power output. Each component corresponds to an element of the decision variable vector. The output priority based on the feasible region of the device is extracted from the current reactive power output field of the time-series data frame and standardized to 1 to emphasize balanced distribution.

[0061] The decision variable vector is constructed from the real-time output and device limit fields of the time-series data frame. It includes the current reactive power output as the initial value and the adjustment amount as the variable. The range of the adjustment amount is limited by the ramp-up capability and the commissioning mode. It is dynamically updated from the device status field through iterative solution to reflect transient constraints.

[0062] The constraint matrix is ​​constructed by integrating the linear relationship between the device limit and the climbing ability. Each row corresponds to an inequality constraint. For example, the upper limit constraint row is set to the coefficient 1 of the decision variable vector and corresponds to the limit value, the lower limit constraint row is set to -1, and the climbing constraint row is calculated from the climbing ability field based on the adjustment rate per unit time.

[0063] The constraint right-hand vector is obtained from the time-series data frame based on the device status and commissioning mode. For example, in the operating mode, the right-hand value is the device limit minus the current reactive power output; in the fault mode, it is set to zero; and in the automatic commissioning mode, it is adjusted to the ramping capacity multiplied by the time step to match the dynamic boundary.

[0064] S3-2 Generation and allocation of bottom-line targets.

[0065] Based on the feasible region of the device, a backup target is generated for rapidly increasing reactive power resources. The allocation order is sorted according to the bus voltage trend and response capability: the bus voltage trend is determined by the voltage difference between adjacent records in the time-series data frame, with a positive difference indicating an increase; the response capability is calculated as the ramp-up capability multiplied by the current reactive power output. After sorting, the total backup demand is calculated using the following formula: ,in This indicates the total demand for a safety net. This represents the proportionality coefficient. Indicates the voltage target. This represents the bus voltage. Resource allocation is performed on a per-resource basis. For example, the target for a static var compensator (SVC) is the upper limit of the device's feasible domain minus its current reactive power output, distributing the total backup demand proportionally to its response capacity. The backup target is written into the first part of the backup instruction object, specifying the executing device as a fast reactive resource. The backup target and executing device are output.

[0066] S3-3 The composition of the cancellation conditions.

[0067] Once the bottom-line target is generated, it constitutes the cancellation condition. The core elements are the partitioning of the handover confidence coefficient, the persistence of the phase angle consistency determination, and the convergence direction of the bottom-line response residual ratio: the partitioning is a lower, middle, or higher partition; the persistence is the duration of the consistency segment; and the convergence direction is the decreasing trend of the residual ratio. Through a logical AND operation, if the partition is higher than the lower partition, the persistence duration exceeds a preset window, and the residual ratio decreases, then it is triggered. The formula is: ,in This represents the truth value of the cancellation condition. Indicates the partition value. Indicates the upper bound of the lower partition. Indicates duration. Indicates the default window. This indicates the change in the residual ratio. The cancellation condition is written into the second part of the bottoming instruction object. The cancellation condition is output after processing is complete.

[0068] S3-4 Determination of handover time and marking of the bottom support instruction object.

[0069] After the cancellation condition is met, the handover time is determined. The earlier of the time the cancellation condition is met and the earliest time point within the available output range of the synchronous condenser is selected: the meeting time is the condition trigger timestamp, and the earliest point within the available range is calculated using the feasible region and climbing capacity of the synchronous condenser. The comparison formula is as follows: ,in Indicates the time of handover. Indicates the moment of satisfaction. Indicates the earliest available point. Writes the third part of the backup instruction object. Marks the object with the backup target, execution device, cancellation conditions, and handover time. Outputs the backup instruction object, specifying the execution of fast reactive resources.

[0070] The first part of the support instruction refers to the support target for fast reactive power resources. This target is assigned specific values ​​based on the feasible domain of each fast reactive power resource and sorted according to bus voltage trends and response capabilities to achieve orderly execution. The second part refers to the cancellation condition, which is based on the partitioning of the handover reliability coefficient and the continuity of phase angle consistency determination, supplemented by the convergence direction of the support response residual ratio, forming a trigger mechanism to stop the support. The third part refers to the handover time, which is selected as the earlier of the time when the cancellation condition is met and the earliest time when the synchronous condenser reaches the available output range, as the time anchor and execution boundary.

[0071] After step S3 is completed, a backup instruction object is formed. This object defines the transient execution boundary, ensuring coordination between rapid reactive power resource backup and synchronous condenser takeover, thereby improving voltage stability and operating efficiency.

[0072] In power systems, although the bottom-line command object defines transient boundaries, the relay during the steady-state phase can easily lead to output cancellation due to improper trajectory design, affecting reactive power sharing efficiency. By generating a relay sequence object in step S4, trajectories in opposite directions can be constructed based on the handover time, ensuring non-overlapping division of labor between synchronous condensers and fast reactive power resources, and achieving orderly transfer of long-term support across multiple time scales. This step emphasizes node definition and synchronous verification, supporting the stability of the control method and avoiding oscillations under fixed bandwidth.

[0073] The S4-1 camera takes over the formation of the target.

[0074] After the status label object transitions to a steady state, starting from the handover moment, the upper and lower limits of the feasible region and the ramp-up capability of the synchronous condenser are extracted. The output trajectory is divided into multiple segments, each segment linearly increasing from the current reactive power output to the next node value. Trajectory nodes are defined by equal time intervals or operating condition switching events. Equal time intervals refer to fixed time increments, while operating condition switching events refer to the point in time when the bus voltage trend reverses or the device state changes. Intra-segment values ​​are calculated using linear interpolation, with the following formula: ,in This represents the output value at time t. Indicates the initial output value of the segment. Indicates the output value at the end of the segment. Indicates the start time of the segment. Indicates the end time of the segment. Linear interpolation is used to ensure that the trajectory conforms to the climbing capacity rate and does not exceed the feasible region of the device. Outputs the takeover target of the synchronous condenser and sends it to the synchronous condenser.

[0075] S4-2 Generation of fast reactive resource exit target.

[0076] After the synchronous condenser takes over the target, it generates exit targets for fast reactive power resources from the handover moment. The output trajectory decreases in the opposite direction to the takeover target; if the output of the takeover target increases, the output of the exit target decreases. It ensures that at the same time point, one type of device strengthens while another weakens, achieved by comparing the sign of output changes at each timestamp. A positive sign indicates an increase, and a negative sign indicates a decrease. If the signs are the same, the exit trajectory rate is adjusted. Segment boundaries are determined by equal time intervals or operating condition switching events, chosen as one; for example, equal time intervals are used when there are no operating condition switching events. The exit targets for fast reactive power resources are output and distributed to the static var compensator and grid-connected inverter.

[0077] S4-3 Labeling and generation of relay sequence objects.

[0078] After identifying the two target trajectories, each trajectory node is labeled with its target value, target actuator, and effective time: the target value is the node's output value, the target actuator is a synchronous condenser or rapid reactive power resource, and the effective time is the accumulated timestamp starting from the handover moment. The trajectories are integrated to form a relay sequence object. Non-overlapping is verified by sorting by timestamps, using a bubble sort algorithm to compare timestamps and direction signs pair by pair. The relay sequence object is output for use in step S5.

[0079] After step S4 is completed, a relay sequence object is formed. This object integrates non-intersecting trajectories to ensure a smooth relay in steady-state coordination, thereby improving the efficiency of reactive power fusion resources and voltage quality.

[0080] In power systems, although the relay sequence object defines a steady-state trajectory, real-time deviations during the execution period often lead to takeover interruptions, affecting the continuity of reactive power sharing. Step S5 introduces deviation verification and single-time recalculation, which dynamically corrects the object, ensuring the ultimate realization of voltage stability in multi-timescale coordination. This step emphasizes monitoring mechanisms and update logic, supporting closed-loop optimization of the control method and avoiding efficiency losses under fixed thresholds.

[0081] S5-1 Definition and Continuous Monitoring of Deviation.

[0082] After the relay sequence object is executed, the real-time bus voltage, real-time unit output, and voltage target are extracted from the time-series data frame. The target voltage deviation is calculated using the following formula: ,in This indicates a deviation from the target voltage. This indicates the real-time bus voltage. Indicates the voltage target. The instruction execution difference is calculated using the formula: ,in This indicates poor instruction execution. Indicates the real-time output of the device. This indicates the command target, taken from the relay sequence object node. It continuously monitors the deviation sequence, scanning new records in the time-series data frame every sampling period, with the sampling period defined as a fixed time interval. The deviation sequence is output for judgment purposes.

[0083] S5-2 Determination of cancellation conditions.

[0084] After the deviation sequence, the cancellation condition is evaluated. The cancellation condition comes from the second part of the support command object. If the target voltage deviation is less than the preset deviation window and the command execution difference is less than the preset execution window, then the condition is met, and the relay sequence object trajectory is maintained until the takeover is completed. The judgment formula is: ,in Indicates the truth value of the condition. Indicates the preset deviation window. This indicates the preset execution window. If the conditions are not met, a re-identification and recalculation will be triggered. The judgment result will be output.

[0085] S5-3 Re-identification and Re-computation Execution.

[0086] If the condition is not met, the phase angle consistency determination is re-executed, forming a new consistent segment on the current time-series data frame. The phase angle dual-window stability and the bottoming response residual ratio are recalculated, and the new parameters are used to enter the sequential bottoming dominance learning, updating the handover confidence coefficient and producing a new state label object and change time. The new parameters refer to the segment samples under the influence of the current deviation sequence. The bottoming command object and relay sequence object are reconstructed based on the new object, the corrected output is completed, and execution is carried out according to the new trajectory. The updated object is output, triggered only once in the same process.

[0087] After step S5 is completed, the final takeover result is formed. This result integrates deviation correction to ensure the integrity of reactive power resource coordination and voltage quality.

[0088] Example 2: Figure 2 The present invention provides a multi-timescale reactive power fusion resource coordination control system based on a synchronous condenser, comprising: The data acquisition module obtains the bus voltage and phasor, device status and limit values, completes unified clock alignment, generates timing data frames covering the synchronous condenser, static var compensator and grid-connected inverter, and outputs timing data frames.

[0089] The state determination module provides phase angle consistency determination on the time-series data frame. The phase angle consistency determination is used to uniformly constrain the segment selection and sample selection of phase angle dual-window stability and bottoming response residual ratio. Sequential bottoming dominance learning is used to generate handover confidence coefficients, and state label objects are output based on the handover confidence coefficients.

[0090] When the status label object is transient, the bottom-line generation module generates a bottom-line instruction object based on the device's feasible domain and current operating constraints.

[0091] When the state tag object transitions to a steady state, the relay construction module generates a relay sequence object based on the handover time and the feasible domain of the device.

[0092] The execution verification module checks the target voltage deviation and the instruction execution difference during the execution of the relay sequence object. If the cancellation condition is not met, a re-identification and recalculation is triggered to update the status label object, the bottoming instruction object, and the relay sequence object, and output the final takeover result.

[0093] Specifically, the above description is only a preferred embodiment of this application and is not intended to limit this application.

[0094] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0095] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-time-scale reactive power fusion resource coordination control method based on a synchronous condenser, characterized in that, Including the following steps: S1: Obtain bus voltage and phasor, device status and limit values, complete unified clock alignment, and generate timing data frames; S2: Provide phase angle consistency judgment on the time series data frame, use phase angle consistency judgment to uniformly constrain the segment selection and sample selection of phase angle double window stability and bottoming response residual ratio, use sequential bottoming dominance learning to generate handover credibility coefficient, and output state label object based on handover credibility coefficient; S3: When the status label object is transient, generate a backup instruction object based on the device's feasible domain and current operating constraints; S4: When the state label object transitions to a steady state, a relay sequence object is generated based on the handover time and the feasible region of the device. S5: During the execution of the relay sequence object, check the target voltage deviation and the difference between the command execution. If the cancellation condition is not met, trigger a re-identification and recalculation, update the status label object, the bottoming command object and the relay sequence object, and output the final takeover result.

2. The multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 1, characterized in that: Step S1 includes collecting bus voltage, phasor, device status, and device limits from the station and master station as raw inputs, correcting the timestamps using GPS timing signals, aggregating measurements taken at the same sampling time to form preliminary time-series data frames, removing duplicate records and filling in missing records, aligning device fields with a unified naming system, and generating a multi-field structure containing sampling time, bus voltage, phasor, device status, device limits, and real-time output.

3. The multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 1, characterized in that: Step S2 includes calculating the difference trajectory of each device phasor relative to the bus phasor based on the time-series data frame and using the bus phasor as a reference to form a consistent segment; selecting short-window trajectory and long-window trajectory within the consistent segment to calculate the phase angle dual-window stability to quantify the convergence characteristics; retrieving steady-state segments from the operation record to construct a nearest-neighbor response mapping to calculate the bottom-line response residual ratio to evaluate the deviation; scanning parameter sequence cumulative trend evidence within the consistent segment to generate a handover confidence coefficient; and outputting a state label object based on the handover confidence coefficient.

4. The multi-time-scale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 3, characterized in that: in, The parameter sequence refers to the phase angle double-window steady-state sequence and the bottoming response residual ratio sequence; the state labels include steady state and transient state.

5. The multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 4, characterized in that: Step S3 includes extracting device limits, device status, current reactive power output, commissioning mode, and ramp-up capability from the time-series data frame to integrate the device feasible domain. The backup targets are sorted and allocated according to the bus voltage trend and response capability and written into the first part of the backup instruction object. The backup instruction object is partitioned by the handover confidence coefficient, and the phase angle consistency determination persistence and the backup response residual ratio convergence direction constitute the cancellation condition and written into the second part. The earlier of the cancellation condition satisfaction time and the earliest time point of the available output range of the synchronous condenser is taken as the handover time and written into the third part. The backup instruction object is marked to specify the execution of fast reactive power resources.

6. The multi-time-scale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 4, characterized in that: Step S4 includes extracting the upper and lower limits of the feasible domain of the synchronous condenser and the segmented linearly increasing output trajectory of the ramp capability from the handover time when the state label object transitions to a steady state, using the opposite direction of the rapidly decreasing reactive power output trajectory as the exit target, ensuring that the output change signs are opposite at the same time point, marking the target value of each node, and generating a relay sequence object with the target execution device and effective time.

7. The multi-timescale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 6, characterized in that: Step S5 includes extracting the real-time bus voltage and real-time device output from the timing data frame during the execution of the relay sequence object, calculating the target voltage deviation and command execution difference for continuous monitoring, and maintaining the trajectory until the takeover is completed if the cancellation condition is met.

8. The multi-time-scale reactive power fusion resource coordination control method based on a synchronous condenser according to claim 7, characterized in that: If the condition is not met, the phase angle consistency determination is re-executed, the phase angle dual-window stability and the bottoming response residual ratio are calculated, the sequential bottoming dominance learning is entered to update the handover reliability coefficient, a new state label object and change time are generated, the bottoming instruction object and relay sequence object are reconstructed, and the final takeover result is output.

9. A multi-time-scale reactive power fusion resource coordination control system based on a synchronous condenser, used to implement the multi-time-scale reactive power fusion resource coordination control method based on a synchronous condenser as described in any one of claims 1-8, characterized in that, include: The data acquisition module acquires bus voltage and phasors, device status and limits, completes unified clock alignment, generates timing data frames covering synchronous condensers, static var compensators and grid-connected inverters, and outputs timing data frames. The state determination module provides phase angle consistency determination on the time-series data frame. The phase angle consistency determination is used to uniformly constrain the segment selection and sample selection of phase angle dual-window stability and bottoming response residual ratio. Sequential bottoming dominance learning is used to generate handover confidence coefficients, and the state label object is output based on the handover confidence coefficients. When the status label object is transient, the bottom-line generation module generates a bottom-line instruction object based on the device's feasible domain and current operating constraints. When the state tag object transitions to a steady state, the relay construction module generates a relay sequence object based on the handover time and the feasible region of the device. The execution verification module checks the target voltage deviation and the instruction execution difference during the execution of the relay sequence object. If the cancellation condition is not met, a re-identification and recalculation is triggered to update the status label object, the bottoming instruction object, and the relay sequence object, and output the final takeover result.

Images