Equivalent inertia identification, characterization and evaluation method of new energy station under all operating conditions

Abstract

The application discloses a new energy station equivalent inertia identification, characterization and evaluation method considering all working conditions, relates to the new energy power generation technical field, and aims to solve the problems of insufficient working condition adaptability, single inertia cognition dimension and poor application of evaluation results in the prior art. The application defines typical operation working conditions of the new energy station and constructs a working condition perceiver; inertia parameter identification is performed in a dual mode of environmental excitation and event driving, and the identification results are fused and calibrated; a multi-dimensional inertia characterization vector is constructed; a multi-level evaluation system is built based on the characterization vector, adaptive graded early warning and risk root diagnosis are realized through fuzzy comprehensive judgment; and finally, decision support and visualization are completed. The application improves the accuracy and robustness of working condition adaptability and inertia identification, and provides core support for power grid dispatching operation and safety evaluation.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation technology, and in particular to a method for identifying, characterizing and evaluating the equivalent inertia of new energy power plants that takes into account all operating conditions. Background Technology

[0002] The global energy structure is shifting towards a green and low-carbon model, driving a continuous increase in the proportion of new energy sources such as wind and solar power in the power system. Synchronous generator units, connected to the grid via power electronic converters, rely on their rotor inertia to release kinetic energy to resist power disturbances and maintain frequency stability. However, new energy units are decoupled from the grid frequency and lack inertial response capabilities. As large-capacity synchronous generator units are replaced by new energy sources, the overall equivalent rotational inertia of the power system decreases, increasing the system's sensitivity to disturbances, raising the rate of frequency change, and shrinking the stability boundary, making it prone to large-scale power outages caused by frequency instability.

[0003] To address this issue, the power industry has improved control strategies to enable renewable energy power plants to simulate the inertial response of synchronous machines, providing virtual inertia. The concept of power system inertia has also been expanded to include a generalized composite inertia encompassing both rotational inertia and virtual inertia. Accurately determining the composite inertia level at the grid connection point of renewable energy power plants has become a crucial prerequisite for grid dispatching and operation, safety assessment, and ancillary service market design.

[0004] Current mainstream technologies have the following limitations when dealing with the complex operating characteristics of new energy power plants: 1. The equivalent inertia of new energy power plants is strongly coupled with operating conditions and exhibits time-varying characteristics. Most methods do not fully consider this characteristic, leading to a decrease in the accuracy and robustness of identification results during operating condition switching. 2. Only outputting the static scalar equivalent inertia time constant cannot comprehensively characterize the response speed, duration, and other qualities supported by inertia, making it difficult to support accurate assessment and risk prediction. 3. The lack of a complete technical closed loop encompassing data perception, state representation, risk warning, and control decision-making limits its practical value in real-time scheduling decisions. Summary of the Invention

[0005] The main objective of this invention is to provide a method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations that takes into account all operating conditions, thereby solving the problems of insufficient adaptability to operating conditions, single cognitive dimension, weak applicability of evaluation results, and low applicability.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions, comprising the following steps:

[0007] S1: Define various typical operating conditions of new energy power stations, and build an operating condition sensor to collect power station operating data in real time, identify and output the operating condition label at the current moment; S2: Inertia parameter identification is performed using a parallel-running environmental excitation mode and an event-driven mode. The environmental excitation mode is based on an improved forgetting factor recursive least squares algorithm, utilizing power system normal disturbance data for continuous recursive identification. The event-driven mode is triggered upon detecting a disturbance event, using the overall least squares method to perform high-precision batch processing identification based on data within the event time window. The identification results of the environmental excitation mode are then fused and calibrated based on the identification results of the event-driven mode. S3: Based on the identified equivalent inertia, construct a multi-dimensional inertia representation vector that includes equivalent inertia time constant, inertia response delay, inertia energy contribution potential and inertia support credibility, so as to comprehensively describe the inertia support capability of new energy power plants. S4: Based on the multi-dimensional inertia representation vector, construct a multi-level evaluation index system, calculate the comprehensive evaluation index, and use the fuzzy comprehensive evaluation method to perform adaptive hierarchical early warning and risk root cause diagnosis on the evaluation results; Step S5: Display the evaluation results of Step S4 through a visualization platform, and generate and recommend control strategies for new energy power stations based on the diagnostic conclusions and the associated strategy knowledge base.

[0008] In the preferred embodiment, step S1 includes at least the following typical operating conditions: a first operating condition using maximum power point tracking (MPPT) mode. G 1. Second operating condition using constant power scheduling mode G 2. Third operating condition using power-limited operation mode G 3. Fourth operating condition using droop control mode G 4. Fifth operating condition using virtual synchronous machine mode G 5; The operating condition sensor outputs the current operating condition label based on real-time control commands, grid connection point power, grid frequency and its rate of change, combined with a preset operating condition transition model that includes active switching, automatic transition and hysteresis return mechanisms.

[0009] In the preferred scheme, the modeling process of the improved forgetting factor recursive least squares algorithm includes: Establish a discretized difference equation describing the relationship between the change in active power, the rate of change of frequency, and the frequency deviation: ; in, , representing the change in active power at the grid connection point of the new energy power station at the k-th sampling time. These represent the active power of the new energy power plant at the grid connection point at the k-th and (k-1)-th sampling times, respectively. This is a discrete approximation of the rate of change of frequency; This refers to the power grid frequency deviation. For the firstk The model error term at each sampling time. 'a' represents the dynamic parameter to be identified in the model; 'a' is directly related to the equivalent inertia. Derivation of the equivalent inertial time constant Calculation formula: ; in, The rated frequency; Set an adaptive forgetting factor ,in, This serves as the baseline value for the forgetting factor. Within the sliding time window variance The adjustment coefficients enable adaptive adjustment of the algorithm's tracking speed and smoothness. A covariance reset mechanism triggered by operating conditions is added. When the operating conditions change, the recursive covariance matrix is ​​reset to [value]. It converges to the new operating condition; among them, To reset the scalar, It is a unit array.

[0010] In the preferred embodiment, step S2, the identification of inertia parameters in the event-driven mode, includes: When a grid disturbance or a power surge at a power station is detected, high-density data within a preset time window before and after the event is locked. The overall least squares method is used to solve the problem within the time window to obtain a high-precision estimate of the inertia. The high-precision inertia estimate is used as the baseline true value, and the continuous inertia estimate output by the environmental excitation mode is weighted and calibrated by a fusion filter.

[0011] In the preferred embodiment, in step S3, the multi-dimensional inertia representation vector is expressed as: ; in, These represent the equivalent inertial time constant, inertial response delay, inertial energy contribution potential, and inertial support reliability of the new energy power station at the k-th sampling time, respectively. The peak shift of the cross-correlation function between the rate of frequency change and the inertial response power is calculated to determine the... ; Calculations based on equivalent inertial time constant and power station active power margin And is subject to active margin constraints; It is a composite index that comprehensively estimates uncertainty, time-varying stationarity, and physical realizability.

[0012] In the preferred scheme, step S4 includes a multi-level evaluation index system comprising a basic index layer, a comprehensive evaluation layer, and a trend prediction layer. The basic indicators include the equivalent inertia time constant, inertia response delay, inertia energy contribution potential, and inertia support credibility. The comprehensive evaluation layer is used to calculate the static adequacy index, which characterizes the adequacy of inertia, and the dynamic quality index, which characterizes the quality of inertia. The trend prediction layer is used to predict future trends of indicators based on historical sequences.

[0013] In the preferred embodiment, step S4, the adaptive hierarchical early warning and risk root cause diagnosis specifically includes: Membership functions with multiple early warning levels are established for the static sufficiency index, the dynamic quality index, and their short-term forecast trends. A fuzzy evaluation matrix is ​​constructed based on the membership degree of each indicator, and combined with a preset weight vector, the comprehensive membership degree of each warning level is calculated through fuzzy comprehensive evaluation. The final warning level is determined based on the principle of maximum membership. When an early warning is triggered, a hierarchical diagnostic mechanism based on a decision tree is activated to sequentially determine the main category of the problem, accurately locate the weak dimension, and output the risk root cause diagnosis conclusion.

[0014] In the preferred scheme, step S4, the hierarchical diagnostic mechanism based on decision trees, specifically includes: First-level diagnosis: Compare the deviations of the static adequacy index and dynamic quality index from the safety threshold to determine the main category of risk root cause, which is static inertia shortage, dynamic performance defect or compound problem; The second layer of diagnosis: For static inertia shortage, the operating condition label is used to determine whether the control mode is not activated or the parameters / spare are insufficient; for dynamic performance defects, the inertia response delay, inertia energy contribution potential, and the contribution of inertia support reliability to the dynamic quality index are analyzed to locate the problem as response lag, insufficient support endurance, or unreliable estimation; for compound problems, the static inertia shortage and dynamic performance defects are analyzed in parallel to output a list of multiple root causes. The third layer of diagnosis maps the indicator problem to a specific, actionable description and outputs the diagnostic confidence level.

[0015] In the preferred embodiment, in step S5, the strategy knowledge base is stored in the form of rules or graphs. The conditions of the rules or graphs are linked to the early warning level, the risk root cause diagnosis results, and the power grid operation mode. The actions of the rules or graphs include information monitoring, strategy suggestions, resource requests, and equipment control. Candidate strategies are intelligently matched according to the diagnosis conclusions, and simulation verification is performed to evaluate the expected effects. The visualization platform adopts a multi-screen collaborative layout, including at least one or more combinations of the following partitions: The overall network inertia security overview area is used to display system-level macro-level summary information; Multi-site geographic topology early warning zone, used to mark the early warning level of each site on the geographic connection map; The core integrated dashboard is used to display the static adequacy index and dynamic quality index of the key stations in real time; Multi-dimensional representation radar chart is used to dynamically display the multi-dimensional inertia representation vector of the focal station. The inertia time series trend curve area is used to display the historical trajectory and predicted trend of key indicators; The strategy recommendation window and execution console are used to display recommended strategies and operation buttons; Deep linkage is achieved based on event-driven mechanisms to respond to the scheduler's selected actions.

[0016] This invention provides a method for identifying, characterizing, and evaluating the equivalent inertia of new energy power plants considering all operating conditions. By defining multiple typical operating conditions of new energy power plants and constructing an operating condition sensor, it employs parallel environmental excitation and event-driven modes to identify inertia parameters. Based on the identification results of the event-driven mode, it fuses and calibrates the identification results of the environmental excitation mode. A multi-dimensional inertia representation vector and a multi-level evaluation index system are constructed, a comprehensive evaluation index is calculated, and a fuzzy comprehensive evaluation method is used to adaptively grade and warn of risks and diagnose root causes based on the evaluation results. The evaluation and warning results are displayed through a visualization platform, and based on the diagnostic conclusions and a strategy knowledge base, control strategies are generated and recommended. This improves the power grid's ability to manage the inertia of new energy power plants, enhances the adaptability of operating conditions, and improves the accuracy and robustness of inertia identification. It solves the problems of insufficient adaptability to operating conditions, a single dimension of inertia cognition, and poor applicability of evaluation results in existing technologies. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a general technical framework diagram of an embodiment of the present invention; Figure 2 This is a diagram of the operating mode and conversion model of a new energy power station according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the dual-mode cooperative inertia online identification process according to an embodiment of the present invention; Figure 4 This is a structural diagram of the online inertia assessment, early warning, and decision recommendation system according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the layout of the inertia situation visualization platform according to an embodiment of the present invention. Detailed Implementation

[0018] like Figure 1-5As shown, a method for identifying, characterizing, and evaluating the equivalent inertia of a new energy power station considering all operating conditions includes the following steps: S1: Define various typical operating conditions of new energy power stations, and build an operating condition sensor to collect power station operating data in real time, identify and output the operating condition label at the current moment.

[0019] S2: Inertia parameter identification is performed using a parallel operation of an environmental excitation mode and an event-driven mode. The environmental excitation mode is based on an improved forgetting factor recursive least squares algorithm and uses power system normal disturbance data for continuous recursive identification. The event-driven mode is triggered when a disturbance event is detected, and uses the overall least squares method to perform batch identification using data within the event time window. The identification results of the environmental excitation mode are then fused and calibrated based on the identification results of the event-driven mode.

[0020] S3: Based on the identified equivalent inertia, construct a multi-dimensional inertia representation vector that includes the equivalent inertia time constant, inertia response delay, inertia energy contribution potential, and inertia support credibility.

[0021] S4: Based on the multi-dimensional inertia representation vector, a multi-level evaluation index system is constructed, a comprehensive evaluation index is calculated, and the fuzzy comprehensive evaluation method is used to adaptively classify and warn of evaluation results and diagnose the root causes of risks.

[0022] S5: Display the evaluation results of step S4 through a visualization platform, and generate and recommend control strategies for new energy power stations based on the diagnostic conclusions and the associated strategy knowledge base.

[0023] In this embodiment, an operating condition sensor is constructed for various typical operating conditions of new energy power plants; inertia parameters are identified using a parallel environmental excitation mode and an event-driven mode, and the identification results are fused and calibrated; then, a multi-dimensional inertia representation vector is constructed based on the identified equivalent inertia; a multi-level evaluation index system is constructed to perform adaptive hierarchical early warning and risk root cause diagnosis on the evaluation results; finally, the evaluation and early warning results are displayed through a visualization platform, realizing panoramic real-time perception and intelligent evaluation of inertia support capabilities, improving the power grid's control over the inertia of new energy power plants, enhancing operating condition adaptability, and improving the accuracy and robustness of inertia identification.

[0024] like Figure 1 As shown, its core evolution is reflected in the following aspects: the evaluation layer has evolved from judging by a single indicator to a three-layer evaluation system that includes basic indicators, comprehensive indices, and trend predictions; the early warning mechanism has been upgraded from triggering with a fixed threshold to adaptive grading based on fuzzy membership, and an automatic risk root cause tracing function has been added; the decision output has been enhanced from status prompts to a linked executable strategy library, providing a complete solution from diagnosis to suggestions, and presented through an interactive visual cockpit.

[0025] This embodiment illustrates the technical solution through each step.

[0026] Step S1: Definition and perception of all operating conditions of new energy power stations; First, we define five core operating conditions related to inertia for new energy power plants. G i : G 1 (Maximum Power Point Tracking (MPPT) mode): The power station aims to maximize power generation efficiency and does not respond to frequency changes, with an equivalent virtual inertia of zero.

[0027] G 2 (Constant Power Dispatch Mode): The station operates according to a given planned curve, with a constant power setpoint, and usually does not provide inertia support.

[0028] G 3 (Power-limited operation mode): Due to grid constraints, the system operates at reduced capacity, which provides an upward adjustment margin and may have the potential to provide downward inertia support (power absorption).

[0029] G 4 (Sag Control Mode): The power station adjusts the active power output according to the frequency deviation by a fixed ratio to provide primary frequency regulation. Its dynamic response is faster than that of traditional inertia, providing high-quality virtual inertia.

[0030] G 5 (Virtual Synchronous Machine VSG Mode): The station fully simulates the rotor motion equation of the synchronous machine, while providing inertial and damping support, and providing high-quality virtual inertia.

[0031] In this embodiment, by dividing the operating states of new energy power stations with different inertia support capabilities, the coupling relationship between the operating conditions of the power station and inertia is clearly defined, which improves the matching between the algorithms of each link and the actual operating state of the power station and enhances the operating condition adaptability of the entire technical system.

[0032] like Figure 2 As shown, a new energy power station operation mode and conversion model with control strategy as the core is constructed. Five steady-state conditions and their dynamic conversion paths and triggering conditions are clearly defined, forming a complete perception logic closed loop. Two basic conversion forms between operating conditions are also clarified: active switching based on dispatching instructions and automatic transition in response to grid status.

[0033] In this embodiment, the various operating condition switching mechanisms include: 1) Power plants typically begin with the pursuit of power generation efficiency. G 1 (Maximum Power Eye Tracking) or Execution Plan G 2 (Constant Power Scheduling): The two can be directly switched between each other through scheduling commands, or enter restricted operation.G 3. The critical automatic switching link occurs in frequency support scenarios: when the grid frequency deviation continuously exceeds the threshold ε1, the operating condition will change from... G 3 Automatic jump to G 4 (droop control), begin providing one frequency modulation.

[0034] 2) If higher support quality is required for scheduling, it can be further switched to G 5 (Virtual Synchronous Machine Mode) fully simulates the inertia and damping characteristics of a synchronous generator set. To ensure control stability and prevent frequent oscillations near the critical point, a hysteresis return mechanism is employed: when the frequency deviation recovers to a more stringent, smaller threshold value... hour, G 4. Relocation to G 3.

[0035] 3) In emergency situations, it can be obtained from G 1 or G 2 Direct Forced Cut-in G 5 modes.

[0036] In this embodiment, the dynamic transition path and triggering conditions between operating conditions are clearly defined. Combined with a multi-source data sensor, real-time operating conditions are accurately identified, and stable and reliable operating condition status labels are output. This avoids frequent oscillations of the operating conditions near the critical point and improves the accuracy of input parameters.

[0037] In this embodiment, the operating condition sensor analyzes the station-level control commands in real time. P cmd AGC status, grid connection power P cmd ,frequency f and its derivative d f / d t Combined with predefined rules (e.g., if |d f / d t |> Threshold and P cmd If it is in automatic frequency adjustment mode, then it is determined to be G 5) Real-time output of current operating condition labels G i ( k ), used for all subsequent algorithm modules.

[0038] It should be noted that, G 1- G The five core operating conditions are the steady-state modes of sustainable operation of new energy power plants during periods when the power grid is normal or under minor disturbances, determined by the autonomous control strategy of their grid-connected converters, based on the continuous power-frequency dynamic characteristics presented by the power plants to the outside world.

[0039] In practical applications, specific external events (such as power step surges under dispatch instructions) or grid fault conditions (such as high / low voltage ride-through) may trigger a power station to enter or pass through one or more of the aforementioned core operating conditions. For example, a power station in... G In a 5-mode (VSG) power station encountering a grid fault, it will first enter a transient process of low voltage ride-through, which is subject to mandatory standards, and its virtual inertia response may be temporarily suppressed; after the fault is cleared and the voltage is restored, it will return to normal operation. G 5. Steady-state operating conditions required by other scheduling instructions.

[0040] Therefore, the operating condition sensor in this embodiment is used to identify the steady-state core operating conditions in which the station operates for most of the time, thereby enabling a routine online assessment of its inertia support capability.

[0041] Step S2: Online identification of dual-mode collaborative inertia; 1. Environmental stimulus mode (resident, continuous): Recursive identification is performed using small disturbance data inherent in the power system. The core is the adaptive forgetting factor recursive least squares algorithm (FFRLS) with operating condition awareness.

[0042] In the preferred scheme, the adaptive forgetting factor recursive least squares algorithm for the environmental incentive mode specifically includes: Use applicable G 4 / G Discretized difference equations for 5 operating conditions: ; in, This indicates that the new energy power station's grid connection point is on the [number]th [year]. k The change in active power at each sampling time. These represent the grid connection points of the new energy power stations on the [number]th [day / year]. k The sampling time and the first k Active power at -1 sampling time; , representing a discrete approximation of the system's rate of change of frequency (ROCOF). The respective k The sampling time and the first k The power grid frequency at -1 sampling time. Indicates the sampling frequency; - , indicating the first k The power grid frequency deviation at each sampling time, The system's rated frequency; For the first k The model error term at each sampling time includes measurement noise, high-frequency dynamic or nonlinear characteristics not covered by the model, etc. These are the dynamic parameters to be identified in the model, which collectively characterize the frequency response characteristics of new energy power plants under specific operating conditions. a It is directly related to the system's equivalent inertia; It is a parameter vector composed of parameters to be identified, and the goal of the identification algorithm is to estimate the parameter vector online.

[0043] The equivalent inertial time constant of the new energy power station is: ; in, This is the system's rated frequency.

[0044] Set an adaptive forgetting factor ,in, This serves as the baseline value for the forgetting factor. Within the sliding time window variance The adjustment coefficient enables adaptive adjustment of the algorithm's tracking speed and smoothness.

[0045] A covariance reset mechanism triggered by operating conditions is added. When the operating conditions change, the recursive covariance matrix is ​​reset to [value]. It converges to the new operating condition; among them, To reset the scalar, It is a unit array.

[0046] Through adaptive forgetting factor Reset triggered by operating conditions: ; in, This serves as the baseline (lower limit) for the forgetting factor. When power fluctuations are infinitely large, the forgetting factor approaches this value, and is typically set to a positive number less than 1 (such as 0.94) to ensure that the algorithm has a minimum tracking speed. Within the sliding time window The variance, when the output is stable ( Hour, The algorithm has a long memory and produces smooth results; however, when the output fluctuates drastically... When (large), (e.g., 0.94), the algorithm quickly forgets old data, enhancing tracking speed; This is an adjustment coefficient used to control the power variance. Sensitivity to the effects of forgetting factors The larger the value, the greater the impact of power variance change on The more gradual the impact, the better.

[0047] When the operating condition sensor determines Switching occurs (e.g., from) G 2 jump to G5) indicates that the old model of the algorithm is no longer applicable, and it needs to start learning again. The algorithm applies the recursive covariance matrix. Perform a reset: ; in, To reset a scalar, it is a large positive number (such as 1000). It is a unit array. When the operating condition changes, it is through... Resetting the covariance matrix is ​​equivalent to significantly amplifying the gain of the new observation data at the current moment, enabling the algorithm to quickly forget the old model and accelerate convergence to the true parameters under the new operating conditions.

[0048] In this embodiment, in the environmental excitation mode, by adaptively adjusting the forgetting factor (dynamically adjusting the tracking speed and smoothness according to the power fluctuation variance) and combining the working condition triggering covariance reset mechanism, the algorithm achieves rapid convergence after the working condition switch, improves the real-time performance of inertia identification, and enhances the dynamic tracking capability and robustness of inertia identification.

[0049] The parameters involved in this embodiment are shown in Table 1.

[0050] Table 1. Parameter Values ​​for the Embodiment

[0051] 2. Event-driven mode (trigger, high precision): Triggered when one of the following conditions is met: (a) a large disturbance in the power grid that can be clearly identified (such as a remote short-circuit fault trip); (b) the substation actively performs a high-power step test.

[0052] After the event is triggered, the events before and after it are automatically locked. High-density data within a time window. The duration of data prior to the event trigger point. This refers to the duration of data following the event trigger point. Together, they define the total data window used for high-precision batch processing calculations. Its value fully covers the dynamic process of frequency and power in a single disturbance event.

[0053] On the current data window, the parameter vector is directly solved using the total least squares (TLS) method. θ TLS simultaneously considers the observation vector and output The error is more robust to noise interference than that of ordinary least squares method, thus yielding a high-precision inertia estimate of the event-driven mode output after a specific disturbance event. H eq-event .

[0054] High-precision inertia estimate output by event-driven mode Heq-event A fusion filter is input as the reference truth value, and the resident environmental excitation modes (using an improved FFRLS algorithm) are processed at the [number]th [time]. k The continuous inertia estimate output online at each sampling time point H eq-rls ( k Weighted calibration is performed to correct any potential cumulative biases, thereby improving the accuracy of long-term online identification.

[0055] In this embodiment, the event-driven model is used to reduce noise interference and improve the accuracy of the reference inertia value.

[0056] 3. Obtain the identification results of the two and perform fusion calibration.

[0057] In this embodiment, the high-precision identification results of the event-driven mode are weighted and calibrated to the continuous identification results of the environmental excitation mode by using a fusion filter. This corrects the cumulative deviation that may exist in the continuous identification, improves the real-time performance and accuracy of inertia identification, and enhances the accuracy and reliability of the inertia identification results.

[0058] Step S3: Multi-dimensional dynamic inertia characterization system; In the preferred scheme, the multi-dimensional inertia representation vector (No. k The sampling time (each sampling moment) is represented as: ; in, They represent the first k The equivalent inertial time constant, inertial response delay, inertial energy contribution potential, and inertial support reliability of the new energy power station at each sampling time.

[0059] Inertial response delay τ Characterizes the lag time from the occurrence of frequency change to the generation of inertial support power. This is achieved by calculating the rate of frequency change. With inertial response power Estimate by the peak shift of the cross-correlation function: ; in, The variable that maximizes the objective function. d The possible values ​​of ; d The number of sampling points for the response delay (the number of sampling points of delay in the power response relative to the frequency change); This indicates the cross-correlation calculation, which calculates the cross-correlation when... P inert Slide the sequence forward d At point 1, the degree of matching (correlation) between the two sequences within the time window. P inertThe sampling point offset of the sequence relative to the ROCOF sequence. Traversing different d Find the offset that maximizes the correlation; i Indicates the sampling time of the sequence. w The length of the sliding time window (number of sampling points) defines how much historical data (ROCOF sequence and...) is truncated when calculating the cross-correlation function. P inert The sequence will be analyzed, and the calculations here will use the sequence number from the current number. k Backtracking from each sampling point w Total points, w +1 consecutive sampled data; Indicates the first The theoretical value of the inertial response power at each sampling time is expressed as follows: .

[0060] Inertia energy contribution potential It represents the maximum theoretical energy that the station can release or absorb by inertia at the current operating point, reflecting the endurance of inertia support. Compared with the current active power margin of renewable energy power stations (Refers to the range of active power output that a renewable energy power station can instantly increase or decrease under its current operating conditions, without violating equipment safety limits.) Strong correlation: ; in, For the rated capacity of new energy power stations, For the maximum permissible frequency deviation (e.g., ±0.5Hz), this formula is essentially a change in kinetic energy.

[0061] At the same time, utilizing The following constraints are imposed on it: ; in, The expected duration of inertia support represents the shortest time during which a renewable energy power station is expected to continuously provide inertia support in the event of a frequency disturbance.

[0062] Inertia support credibility Characterize the current The reliability of an estimate is a comprehensive indicator: ; in, This indicates that the recursive algorithm (such as FFRLS) is in the first... k The estimation error covariance matrix at each sampling time point, whose trace (the sum of the diagonal elements) quantitatively reflects the algorithm's estimation of the current parameter values. The degree of uncertainty. The smaller the trace, the more certain and reliable the estimate; Represents the Frobenius norm, used for quantizing matrices. The overall size; The norm normalization benchmark is used to... Normalizing to the [0, 1] interval allows us to take the typical maximum value of the covariance matrix norm after the algorithm stabilizes; Represents the variance of the equivalent inertia estimate, based on H eq The variance calculated from recent time window historical data reflects its volatility and time-varying stationarity; The variance assessment threshold is a preset criterion value used to measure recent equivalent inertia. Estimate whether the volatility of the sequence is within an acceptable range of stationarity, when At that time, it was considered that the estimated value was sufficiently stable; This indicates the maximum continuous active power output that is set or allowed at a new energy power station.

[0063] The credibility of inertia support is comprehensively evaluated from three dimensions, the first of which is... norm The covariance matrix reflects the uncertainty of the estimate; the larger the covariance matrix (the more uncertain the estimate), the lower its contribution. The second term... Reflecting time-varying smoothness, the greater the historical fluctuations in the inertia estimate, the lower its contribution; the third term Reflecting physical feasibility, the higher the current active power regulation margin, the more sufficient the physical conditions for achieving current inertia support, and the greater its contribution. These represent the weights of the three dimensions: estimation uncertainty, time-varying stationarity, and physical realizability, respectively, and can be configured according to actual needs. The higher the value, the higher the credibility.

[0064] In this embodiment, the inertia response delay improves the accuracy of the lag time between quantization frequency changes and inertia response; the inertia energy contribution potential reflects the endurance of inertia support and improves the stability of inertia characterization; and the inertia support credibility improves the reliability of quantization identification results.

[0065] Step S4: Online inertia assessment, early warning, and decision recommendation; like Figure 4 As shown, the core objective is to build a multi-layered, adaptive, and actionable intelligent assessment and early warning engine.

[0066] 1. Multi-level evaluation indicator system; In the preferred embodiment, in step S4, the inertia representation vector from the representation layer is received: V inert ( k )=[ H eq ( k ), τ ( k ), E pot ( k ), C ( k Based on this, a three-tiered evaluation index system is constructed.

[0067] 1) Basic indicator layer: The four indicators of the characterization layer are directly adopted as the health check report of the inertia health status of the new energy power station.

[0068] 2) Comprehensive evaluation layer, used to calculate the static adequacy index characterizing the adequacy of inertia and the dynamic quality index characterizing the quality of inertia.

[0069] ①Static Satisfaction Index (SARI): ; in, Dynamic inertia requirements can be obtained through online rolling calculations or by querying pre-generated operating condition-demand mapping tables, based on real-time grid topology, start-up methods, and frequency stability requirements. Therefore, SARI directly reflects the adequacy of the quantity.

[0070] ② Dynamic Quality Index (DQI): ; in, , , These correspond to the three quality dimensions of new energy power station inertia: response delay, energy potential, and reliability. The weights in the algorithm reflect the emphasis placed on different quality characteristics during scheduling operations. This serves as a reference value for the inertial response delay, used to adjust the inertial response delay. Dimensionless, which can usually be taken as the control and communication delay under ideal conditions; The inertial energy demand threshold represents the expected value of the inertial energy that the station should contribute in a single frequency event, and is used to... Normalize to a reasonable range.

[0071] In summary, DQI comprehensively reflects the response speed, support potential, and support reliability of inertia.

[0072] 3) Trend prediction layer, based on H eqUsing short-term historical series of SARI and DQI, lightweight time series prediction algorithms (such as linear extrapolation or exponential smoothing) are employed to predict future trends. t The possible values ​​and trends of the indicator after a certain time (e.g., 5-15 minutes) enable early warning of inertia status.

[0073] In this embodiment, the multi-level evaluation index system enables a comprehensive evaluation of inertia support capability, improving the accuracy and real-time performance of predictions.

[0074] 2. Adaptive hierarchical early warning and source tracing mechanism; In this embodiment, adaptive graded early warning and risk root cause diagnosis specifically include: Dynamic classification of warning levels: (1) Fuzzification: Establish a four-level membership function for each indicator. The warning level is no longer simply defined by a fixed threshold. Instead, a fuzzy comprehensive evaluation method is adopted to form a smooth transition between four warning states: safety, caution, risk, and emergency. First, four-level membership functions are established for the current values ​​and short-term predicted trends of the Static Inertia Index (SARI) and the Dynamic Inertia Quality Index (DQI). Then, a fuzzy evaluation matrix composed of the above four indicators is constructed, and corresponding weight vectors are configured. The comprehensive membership vector for each level is obtained through weighted calculation. Finally, the immediate warning level is determined based on the principle of maximum membership (supplemented by a conservative strategy of choosing the highest membership).

[0075] This embodiment achieves adaptive fusion of multi-dimensional indicators, improving the accuracy of state assessment and the reliability of decision support, thereby enhancing the accuracy of early warning and the safety of power grid operation. For SARI, the membership functions at its four warning levels can be set as follows: ① Degree to which it belongs to the security level: .

[0076] ② The degree to which it belongs to the attention hierarchy: .

[0077] ③ The degree to which it belongs to the risk level: .

[0078] ④ The degree to which it belongs to the emergency level: .

[0079] Similarly, the membership functions of the Dynamic Quality Index (DQI) for the four warning levels are as follows.

[0080] Considering that DQI comprehensively reflects the speed, potential, and reliability of inertial response, its threshold setting focuses on the quality of dynamic performance: ① Degree to which it belongs to the security level: .

[0081] When dynamic quality is excellent ( When ), it is in a completely safe state.

[0082] ② The degree to which it belongs to the attention hierarchy: .

[0083] When the quality is good but not optimal ( It is in the range that requires careful observation.

[0084] ③ The degree to which it belongs to the risk level: .

[0085] When there are obvious defects in dynamic quality ( If the response is too slow or the potential is insufficient, it is clearly a risky situation.

[0086] ④ The degree to which it belongs to the emergency level: .

[0087] When the dynamic quality is extremely poor ( When an effective inertia support cannot be formed, an emergency warning is triggered.

[0088] (2) Fusion: Fuzzy comprehensive evaluation method to determine the final warning level A comprehensive decision is made by simultaneously considering the current values ​​of SARI and DQI and their short-term predicted trends ΔSARI and ΔDQI (e.g., the rate of change over the next 5 minutes).

[0089] ① Construct the evaluation matrix R For the k At any given time, a 4×4 fuzzy relation matrix is ​​formed. R ( k ), its elements r ij Indicates the first i The first indicator is relative to the first j Membership degree of each warning level: ; The membership functions of trend indicators ΔSARI and ΔDQI can be set as follows: when the trend is negative (deteriorating), the membership increases to lower security levels (such as risk, emergency).

[0090] ② Determine the weight vector W Define the weight vector for each indicator. W =[ w a , w b , w c , w d ],satisfy w a + w b + w c + w d =1.

[0091] In this embodiment This indicates that the current SARI and DQI have the highest weights (with a weight value of 0.35), followed by ΔSARI and ΔDQI (with a weight value of 0.15), based on the principle of prioritizing the current state and supplementing it with trend warnings.

[0092] ③ Calculate the comprehensive membership vector B By combining the weight vector and the evaluation matrix (such as using a weighted average operator), the overall membership degree of the four warning levels is obtained: ; in, b j Indicates that it ultimately belongs to the first j The severity of each warning level.

[0093] (3) Judgment: Determine the final warning level The final warning level is determined by employing the maximum membership degree principle supplemented by threshold judgment. First, based on the comprehensive membership degree vector... Calculate the preliminary warning level Subsequently, a judgment threshold was introduced. (Usually taken as 0.05~0.15), let and These are the highest and second-highest membership degrees, respectively. Final warning level. The judgment rules are as follows: ; in, This indicates the second-highest warning level.

[0094] The above rules ensure that when the state is critical or ambiguous, a more conservative result (i.e., a lower safety level and a higher warning level) will be output, thereby improving the safety of power grid operation.

[0095] In this embodiment, a fuzzy comprehensive evaluation method is used to achieve adaptive hierarchical early warning. Through multi-level membership functions, a smooth transition between four levels of early warning states—safety, caution, risk, and emergency—is achieved. By combining weight configuration and comprehensive membership calculation, both the current state and trend changes are taken into account. A conservative decision rule is adopted, which improves the accuracy of early warning and the safety of power grid operation.

[0096] This embodiment constructs an automatic tracing mechanism based on a hierarchical diagnostic decision tree: When an early warning is triggered, the deviations of SARI and DQI from their safety thresholds are first compared to determine the main problem category as static inertia shortage, dynamic performance defect, or a combined problem. Then, drill-down analysis is performed: for static inertia shortage, real-time operating conditions are considered... G i The defects are categorized as either mode inactive or insufficient parameters / spare; for dynamic performance defects, analysis is used to address them. τ , E pot , C The contribution of a decline in DQI is precisely pinpointed as a lag in response, insufficient resilience, or unreliable estimation. Finally, the output includes a conclusion with a detailed description of the root cause and diagnostic confidence levels.

[0097] When a warning triggers an alert at the level of "attention" or higher, the risk root cause diagnosis engine is automatically activated. Based on a preset diagnostic decision tree, this engine deconstructs the comprehensive assessment index layer by layer to quickly locate the most likely root cause of the risk.

[0098] In this embodiment, the core logic of the diagnostic process is as follows: First-level diagnosis: Determining the main category of the problem First, determine whether the main category (primary issue) causing the IHSI decline or warning trigger is insufficient static quantity or dynamic quality degradation.

[0099] ; in, This is the static adequacy safety threshold. This is a dynamic quality and safety threshold.

[0100] Second-level diagnosis: Precisely identifying weak points Based on the results of the first layer, drill down to obtain specific characterization indicators: ① If the main category is static inertia shortage The root cause points directly to H eqInsufficient. Further integration with operating condition information is needed. G i Determine the cause: If G i = G 1 / G 2. The root cause is that the control mode is not activated (the station has not entered inertia support mode). If G i = G 5. The root cause is either the virtual inertia parameter setting is too low or the active power reserve is exhausted (through...). P margin judge).

[0101] ② If the main category is dynamic performance defect Analyzing the sub-components of DQI, the following specific defects were identified: ; in, This represents the critical threshold for inertial response delay. This represents the critical threshold for the reliability of inertia support.

[0102] ③ If the main category is a composite problem The above rules are used for parallel analysis to output a list of multiple risk root causes.

[0103] Third-level diagnosis: Associating physical objects with confidence levels The diagnostic engine maps abstract metric issues to concrete, actionable descriptions, along with diagnostic confidence scores, including: ① Output format Diagnostic conclusion: Confidence level [High / Medium / Low] - [Site Name] is currently facing [Specific Root Cause].

[0104] ②Example Output ( a Diagnostic conclusion: High confidence level - Photovoltaic power station A is currently facing a problem where the virtual inertia parameter setting is too low, resulting in an equivalent inertia of only 4.2s, which is lower than the required value of 5.8s.

[0105] ( b Diagnostic conclusion: Medium confidence level - Wind farm B is currently facing a severe response lag. τ =0.28s) and insufficient support endurance ( E pot The composite problem of (120MJ).

[0106] 3. Executable strategy library and decision recommendation; The generated control strategy was further verified through rapid simulation.

[0107] This embodiment improves the speed of early warning response and the accuracy of locating the root cause of risk through the above steps.

[0108] In this embodiment, the strategy knowledge base is stored in the form of rules or graphs. The conditions of the rules or graphs are linked to the early warning level, the risk root cause diagnosis results, and the power grid operation mode. The actions of the rules or graphs include information monitoring, strategy suggestions, resource requests, and equipment control. Candidate strategies are intelligently matched according to the diagnosis conclusions, and simulation verification is performed to evaluate the expected effects.

[0109] The executable strategy knowledge base for inertia problems is stored in the form of IF (condition), THEN (action) rules or a more flexible graph format. Conditions are linked to early warning levels, source tracing results, and current grid operation modes (such as reserve capacity and tie-line power). Actions are divided into the following categories: (1) Information category: Enhance the monitoring of relevant data.

[0110] (2) Recommendations: It is recommended to increase the rotation of the hydropower unit for standby and to check the VSG control parameters of station A.

[0111] (3) Request type: Request AGC to adjust the output plan of station B and reserve inertia support capacity.

[0112] (4) Control type: Automatic start-up of grid-type energy storage in station C.

[0113] For risk levels of 25 and above, the effectiveness of recommended strategies can be quickly evaluated based on simple power flow and frequency response simulation models. The expected results (such as the projected SARI improvement value) can then be presented to the scheduler to assist in decision-making. Strategy recommendations are based on intelligent matching of conclusions drawn from the aforementioned rapid tracing of risk root causes.

[0114] This embodiment establishes a mapping rule base from diagnostic conclusions to recommendation strategies, as shown in Table 2.

[0115] Table 2. Recommended Example Table of Mapping Rules

[0116] Based on the warning level and diagnostic confidence, the matched candidate strategies are ranked, and the top 1-3 ranked strategies are sent to the strategy simulation and recommendation stage for pre-evaluation of their effects.

[0117] This embodiment realizes closed-loop management from risk diagnosis to strategy execution, improving the continuity and reliability of method application.

[0118] Step S5: Visual interaction and decision support; In this embodiment, the visualization platform adopts a multi-screen collaborative layout, including at least: Area A: Overall Network Inertia Security Overview It displays macro-level summary information such as the equivalent inertia level of the entire system, the proportion of low-inertia stations, and the current highest warning level in the form of a statistical panel. It is updated infrequently and is used for a quick overview of the overall situation.

[0119] Area B: Multi-site geographic topology early warning On the power grid geographical wiring map, different colors (green / yellow / red) and pulse effects are used to mark the early warning level of each new energy power station in real time, so as to realize the spatial positioning of risks.

[0120] Area C: Core Integrated Dashboard It adopts a concentric circle dual pointer instrument to display the precise values ​​of SARI (outer ring) and DQI (inner ring) of the focal station and its safe zone in real time.

[0121] Area D: Multi-dimensional characterization radar chart Dynamically display the multi-dimensional representation vector of the focal station in the form of a radar chart. H eq ( k ), τ ( k ), E pot ( k ), C ( k [ ], which visually presents the healthy shape of its inertia support capability.

[0122] Area E: Inertia Time Series Trend Curve The historical trajectory, real-time value, and short-term forecast range of key indicators such as Heq are displayed side by side, revealing their evolution trend.

[0123] Section F: Strategy Recommendation Window and Execution Console The strategy recommendation window displays a list of 1-3 priority strategies for the current risk. Each strategy includes the expected effect after simulation verification (e.g., it is expected to improve SARI by 10%) and operation buttons such as confirm execution and details.

[0124] In this embodiment, the six zones are deeply interconnected through event-driven mechanisms: when a dispatcher clicks on any alarm site in Zone B (geographic topology), the content in Zones C, D, and E will automatically switch to focus on displaying the site's real-time assessment, multi-dimensional representation, and historical trends; simultaneously, Zone F will immediately refresh, pushing customized strategies targeting the current root causes of risk at that site. The dispatcher can review the strategies in Zone F and confirm their execution with one click. Subsequently, the actual effects of the strategy execution will be automatically tracked, and the feedback data will be used to optimize the assessment model and strategy knowledge base, thereby forming a complete intelligent business closed loop.

[0125] The various areas of the above inertial situation visualization platform are not displayed in isolation, but are a highly interconnected organic whole: when a site icon flashes an alarm on the geographic topology map, the panoramic dashboard automatically switches to the real-time SARI / DQI data of that site; clicking the icon synchronously updates the multi-dimensional radar chart and spatiotemporal trend curve area, displaying the site's detailed representation vector and historical situation; at the same time, the strategy recommendation window automatically lists the recommended strategies and simulation results for the current risks of that site, realizing seamless navigation from network-wide situation awareness to single-site in-depth diagnosis and then to measure generation.

[0126] Dispatchers can click on any alarm station to drill down and view its detailed characterization data, historical events, and policy execution records. For suggested policies, dispatchers can confirm and issue them with one click, track the effects of policy execution, form a business closed loop, continuously optimize the policy library, and build a complete industrial-grade solution from algorithm core to intelligent decision-making to interactive implementation.

[0127] This embodiment achieves closed-loop optimization of scheduling operations and effect tracking, improving the adaptability and practicality of the method.

[0128] In application, this embodiment realizes panoramic real-time perception and intelligent assessment of the inertia support capability of new energy power stations, and improves the power grid's ability to control the inertia of new energy power stations. Through the organic combination of full-condition perception, dual-mode collaborative identification, multi-dimensional characterization, adaptive early warning and tracing, and visual decision support, the identification accuracy and real-time performance, early warning accuracy, inertia identification accuracy, robustness and operating condition adaptability are improved, and the stability of power grid operation is enhanced.

[0129] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for identifying, characterizing, and evaluating the equivalent inertia of a new energy power station considering all operating conditions, characterized in that, Includes the following steps: S1: Define various typical operating conditions of new energy power stations, and build an operating condition sensor to collect power station operating data in real time, identify and output the operating condition label at the current moment; S2: Inertia parameter identification is performed using a parallel-running environmental excitation mode and an event-driven mode. The environmental excitation mode is based on an improved forgetting factor recursive least squares algorithm, utilizing power system normal disturbance data for continuous recursive identification. The event-driven mode is triggered upon detecting a disturbance event, using the overall least squares method to perform high-precision batch processing identification based on data within the event time window. The identification results of the environmental excitation mode are then fused and calibrated based on the identification results of the event-driven mode. S3: Based on the identified equivalent inertia, construct a multi-dimensional inertia representation vector that includes equivalent inertia time constant, inertia response delay, inertia energy contribution potential and inertia support credibility, so as to comprehensively describe the inertia support capability of new energy power plants. S4: Based on the multi-dimensional inertia representation vector, construct a multi-level evaluation index system, calculate the comprehensive evaluation index, and use the fuzzy comprehensive evaluation method to perform adaptive hierarchical early warning and risk root cause diagnosis on the evaluation results.

2. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, It also includes step S5: displaying the evaluation results of step S4 through a visualization platform, and generating and recommending new energy power station control strategies based on the diagnostic conclusions and the associated strategy knowledge base.

3. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, In step S1, the various typical operating conditions include at least the first operating condition using maximum power point tracking (MPPT) mode. G 1. Second operating condition using constant power scheduling mode G 2. Third operating condition using power-limited operation mode G 3. Fourth operating condition using droop control mode G 4. Fifth operating condition using virtual synchronous machine mode G 5; The operating condition sensor outputs the current operating condition label based on real-time control commands, grid connection point power, grid frequency and its rate of change, combined with a preset operating condition transition model that includes active switching, automatic transition and hysteresis return mechanisms.

4. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, In step S2, the modeling process of the improved forgetting factor recursive least squares algorithm includes: Establish a discretized difference equation describing the relationship between the change in active power, the rate of change of frequency, and the frequency deviation: ； in, , representing the change in active power at the grid connection point of the new energy power station at the k-th sampling time. These represent the active power of the new energy power plant at the grid connection point at the k-th and (k-1)-th sampling times, respectively. This is a discrete approximation of the rate of change of frequency; This refers to the power grid frequency deviation. For the first k The model error term at each sampling time. 'a' represents the dynamic parameter to be identified in the model; 'a' is directly related to the equivalent inertia. Derivation of the equivalent inertial time constant Calculation formula: ； in, The rated frequency; Set an adaptive forgetting factor ,in, This serves as the baseline value for the forgetting factor. Within the sliding time window variance This is the adjustment coefficient; A covariance reset mechanism triggered by operating conditions is added. When the operating conditions change, the recursive covariance matrix is ​​reset to [value]. It converges to the new operating condition; among them, To reset the scalar, It is a unit array.

5. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, In step S2, the inertia parameter identification for event-driven mode includes: When a grid disturbance or a power surge at a power station is detected, high-density data within a preset time window before and after the event is locked. The overall least squares method is used to solve the problem within the time window to obtain a high-precision estimate of the inertia. The high-precision inertia estimate is used as the baseline true value, and the continuous inertia estimate output by the environmental excitation mode is weighted and calibrated by a fusion filter.

6. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, In step S3, the multi-dimensional inertia representation vector is expressed as: ； in, These represent the equivalent inertial time constant, inertial response delay, inertial energy contribution potential, and inertial support reliability of the new energy power station at the k-th sampling time, respectively. The peak shift of the cross-correlation function between the rate of frequency change and the inertial response power is calculated to determine the... ; Calculations based on equivalent inertial time constant and power station active power margin And is subject to active margin constraints; It is a composite index that comprehensively estimates uncertainty, time-varying stationarity, and physical realizability.

7. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 1, characterized in that, In step S4, the multi-level evaluation index system includes a basic index layer, a comprehensive evaluation layer, and a trend prediction layer; The basic indicators include the equivalent inertia time constant, inertia response delay, inertia energy contribution potential, and inertia support credibility. The comprehensive evaluation layer is used to calculate the static adequacy index, which characterizes the adequacy of inertia, and the dynamic quality index, which characterizes the quality of inertia. The trend prediction layer is used to predict future trends of indicators based on historical sequences.

8. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 7, is characterized in that... In step S4, the adaptive hierarchical early warning and risk root cause diagnosis specifically includes: Membership functions with multiple early warning levels are established for the static sufficiency index, the dynamic quality index, and their short-term forecast trends. A fuzzy evaluation matrix is ​​constructed based on the membership degree of each indicator, and combined with a preset weight vector, the comprehensive membership degree of each warning level is calculated through fuzzy comprehensive evaluation. The final warning level is determined based on the principle of maximum membership. When an early warning is triggered, a hierarchical diagnostic mechanism based on a decision tree is activated to sequentially determine the main category of the problem, accurately locate the weak dimension, and output the risk root cause diagnosis conclusion.

9. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 8, is characterized in that... In step S4, the hierarchical diagnostic mechanism based on decision trees is as follows: First-level diagnosis: Compare the deviations of the static adequacy index and dynamic quality index from the safety threshold to determine the main category of risk root cause, which is static inertia shortage, dynamic performance defect or compound problem; The second layer of diagnosis: For static inertia shortage, the operating condition label is used to determine whether the control mode is not activated or the parameters / spare are insufficient; for dynamic performance defects, the inertia response delay, inertia energy contribution potential, and the contribution of inertia support reliability to the dynamic quality index are analyzed to locate the problem as response lag, insufficient support endurance, or unreliable estimation; for compound problems, the static inertia shortage and dynamic performance defects are analyzed in parallel to output a list of multiple root causes. The third layer of diagnosis maps the indicator problem to a specific, actionable description and outputs the diagnostic confidence level.

10. The method for identifying, characterizing, and evaluating the equivalent inertia of new energy power stations considering all operating conditions as described in claim 2, characterized in that, In step S5, the strategy knowledge base is stored in the form of rules or graphs. The conditions of the rules or graphs are linked to the early warning level, the risk root cause diagnosis results, and the power grid operation mode. The actions of the rules or graphs include information monitoring, strategy suggestions, resource requests, and equipment control. Candidate strategies are intelligently matched according to the diagnosis conclusions, and simulation verification is performed to evaluate the expected effects. The visualization platform adopts a multi-screen collaborative layout, including at least one or more combinations of the following partitions: The overall network inertia security overview area is used to display system-level macro-level summary information; Multi-site geographic topology early warning zone, used to mark the early warning level of each site on the geographic connection map; The core integrated dashboard is used to display the static adequacy index and dynamic quality index of the key stations in real time; Multi-dimensional representation radar chart is used to dynamically display the multi-dimensional inertia representation vector of the focal station. The inertia time series trend curve area is used to display the historical trajectory and predicted trend of key indicators; The strategy recommendation window and execution console are used to display recommended strategies and operation buttons; Deep linkage is achieved based on event-driven mechanisms to respond to the scheduler's selected actions.

Images