Multi-source information fusion control method for improving stability of power system

Abstract

The invention relates to the technical field of power system decentralized control, and discloses a multi-source information fusion control method for improving the stability of a power system, which comprises the following steps: acquiring a state error signal, injecting an active detection signal, extracting a self-response component and an orthogonal demodulation intermediate component of the active detection signal, calculating a power grid stiffness context parameter based on the self-response component, and calculating the stability of the power system. And generating a detection channel quality index based on the time stability of the intermediate component, and determining whether to fuse context parameters to carry out adaptive control or adopt preset default control parameters to carry out safe backspacing according to whether the quality index meets a preset threshold. The problem that oscillation is easily induced by context blindness in fixed parameter control is solved, a channel quality self-arbitration mechanism established by the method can actively back to a safety default parameter when sensing information is polluted, and the engineering safety of self-adaptive control is ensured.

Description

Technical Field

[0001] This invention relates to a multi-source information fusion control method for improving the stability of power systems, belonging to the field of distributed control technology for power systems. Background Technology

[0002] In current power systems, distributed control systems (DCS) are a commonly used technology to ensure the coordinated operation of numerous distributed control units and maintain system frequency and voltage stability. In power grid environments dominated by large synchronous generators, the total system inertia is large, and the grid characteristics are relatively stable and predictable. Under these conditions, distributed control units generally adopt control laws based on fixed parameters of local measurements for stability control, which has proven effective in specific applications. However, with a high proportion of new energy sources being connected to the grid via inverters, the dynamic characteristics of the power system change, and the total system inertia shows a trend of real-time change and gradual decrease. This makes the assumption of constant grid characteristics, which is the basis of traditional distributed control, no longer fully applicable, leading to a fundamental constraint problem: the control parameters of distributed control nodes, such as control gain, are usually calibrated under high inertia and high stiffness grid conditions. When the system switches to a low inertia and low stiffness state due to factors such as fluctuations in new energy output without the node being aware of it in time, if the node continues to use the fixed control gain designed for a high-stiffness grid to respond to system disturbances, the control action is prone to over-response.

[0003] When a large number of nodes in a distributed control system simultaneously execute excessive responses based on this delayed judgment of the power grid state, the interaction between nodes can easily induce systemic low-frequency oscillations, affecting the expected effect of distributed control. To address this issue, attempts have been made to build high-speed, wide-area communication systems to obtain global state information, but this has brought engineering application obstacles such as high deployment costs, communication latency, and network security. Another approach, which uses passive observation algorithms based purely on local measurements to estimate power grid characteristics, is susceptible to interference from power grid background noise and its disturbance signals, making it difficult to meet the real-time and reliability requirements of some dynamic control requirements. In addition to the aforementioned shortcomings in direct sensing of power grid stiffness characteristics and communication architecture, the current industry is also exploring how to efficiently and reliably utilize multi-source data for auxiliary control decision-making. There are also fundamental flaws at the software level. For example, Chinese invention patent CN120449091A discloses a multi-source data fusion method and real-time monitoring device for power systems. This method aims to time-align multi-source data such as photovoltaic power and load demand through adaptive DTW, extract features by combining transformer models, and finally use the entropy weight method to dynamically allocate weights to achieve data fusion, thereby improving the stability of real-time monitoring. However, the core idea of ​​this scheme is to quantify and assign weights to the entropy value of the static information contribution of different data sources. The focus is on the inherent information value of the data sources, rather than the real-time assessment of the current detection channel quality or environmental interference. This means that once strong adjacent frequency interference or background noise occurs in the actual operation of the power system, the feature data used as the basis for control will be contaminated in real time.

[0004] Therefore, the technical problem to be solved by this invention is how to enable each local node in a distributed control system to autonomously perceive the key characteristics of its local power grid in a low-cost and high-real-time manner without relying on high-speed global communication. Summary of the Invention

[0005] This invention provides a multi-source information fusion control method to improve the stability of power systems. Its main purpose is to solve the problem mentioned in the background art, which is that the distributed control nodes lack real-time perception of the stiffness characteristics of the power grid, resulting in control parameter mismatch and easy induction of system oscillation.

[0006] To achieve the above objectives, this invention provides a multi-source information fusion control method for improving power system stability. This method is executed on at least one distributed control node of the distributed control system of the power system and includes the following steps: Acquire local routine state measurements of distributed control nodes and determine state error signals based on local routine state measurements; generate and inject preset active detection signals into the output of distributed control nodes; collect local response signals of distributed control nodes; Perform frequency-specific quadrature demodulation on the local response signal to extract the self-response component and obtain the in-phase and quadrature components for calculating the self-response component; calculate the context parameters characterizing the local power grid stiffness based on the active probe signal and the self-response component; monitor the time stability of the in-phase and quadrature components in real time to generate probe channel quality indicators. The sounding channel quality index is compared with a preset reliability threshold. When the sounding channel quality index meets the preset reliability threshold, the state error signal and context parameters are fused to adaptively generate control commands and execute the control commands. When the sounding channel quality index does not meet the preset reliability threshold, the fusion of context parameters is temporarily suspended, and the distributed control nodes use preset default control parameters to generate control commands and execute the control commands.

[0007] Preferably, the step of extracting the self-response component includes: performing a quadrature demodulation algorithm based on a specific frequency on the local response signal to filter out interference components of non-specific frequencies in the local response signal to obtain the self-response component; the step of calculating the context parameters includes: determining the context parameters based on the ratio between the amplitude of the active probe signal and the amplitude of the self-response component.

[0008] Preferably, the step of extracting the self-response component further includes: obtaining the phase difference between the active detection signal and the self-response component, and using the phase difference as a phase context parameter characterizing the local power grid impedance characteristics; the step of adaptively generating control commands further includes: fusing the phase context parameter to adaptively adjust the control strategy of the control commands.

[0009] Preferably, the method further includes: acquiring local frequency measurements of the distributed control node and determining a frequency error signal based on the local frequency measurements; generating and injecting a preset active active power detection signal with a frequency different from the active detection signal into the active power output terminal of the distributed control node; acquiring local frequency measurements and extracting the frequency self-response component caused by the active active power detection signal; calculating frequency context parameters characterizing local frequency stiffness in real time based on the active active power detection signal and the frequency self-response component; and fusing the frequency error signal and the frequency context parameters to adaptively generate active power control commands.

[0010] Preferably, the step of calculating the context parameters specifically involves: calculating the context parameters using the following formula. : ,in For context parameters, To actively detect the amplitude of the signal, This represents the amplitude of the self-response component.

[0011] Preferably, the step of real-time monitoring of the time stability of the in-phase component and the quadrature component includes: calculating the variance of the in-phase component and the quadrature component within a preset time window; and using the variance as a quality indicator of the probe channel.

[0012] Preferably, the active detection signal is a reactive power disturbance signal, the local routine state measurement is a local voltage measurement, the state error signal is a voltage error signal, the local response signal is a local voltage signal, and the self-response component is the voltage response component in the local voltage signal caused by the reactive power disturbance signal.

[0013] Preferably, the specific frequency is set to be higher than 2Hz; the amplitude of the active detection signal is in the range of 0.1% to 5% of the rated capacity of the distributed control node.

[0014] Preferably, the step of integrating phase context parameters to adaptively adjust the control command includes: when the phase context parameters indicate that the local power grid has a high reactance-resistance ratio characteristic, adopting a control strategy with reactive power regulating voltage as the main dimension; and when the phase context parameters indicate that the local power grid has a high resistance-reactance ratio characteristic, switching to a control strategy with active power regulating voltage as the main dimension.

[0015] Preferably, the step of integrating phase context parameters to adaptively adjust the control strategy of the control command includes: dynamically allocating the weights of the active power control component and the reactive power control component contained in the control command in voltage regulation according to the current value of the phase context parameters.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By actively generating a probe signal locally at the distributed control node and extracting the self-response component corresponding to the probe signal from the local response signal using an orthogonal demodulation mechanism, the control node can obtain context parameters that characterize the local power grid stiffness in real time. These context parameters are then used to dynamically adjust the node's own control gain, so that the controller's response strength matches the tolerance of its power grid environment, thus avoiding control oscillations caused by improper response when the power grid characteristics change in traditional fixed parameter control.

[0017] 2. The orthogonal extraction step calculates the amplitude of the self-response component to quantify the grid stiffness, while utilizing the phase difference information that is inevitably generated during the calculation process. This phase difference information is used to characterize the impedance characteristics of the local grid, enabling the distributed control nodes to not only adaptively adjust the control strength, i.e., the gain, but also to adaptively adjust the control strategy according to the characteristics of the local grid, such as the active or reactive power allocation method, thereby achieving dynamic optimization of the control dimension.

[0018] 3. The core mechanism can be applied in parallel to two key control dimensions of the system: sensing voltage response characteristics by injecting reactive power detection signals and sensing frequency response characteristics by injecting active power detection signals with frequencies orthogonal to them. This dual-axis parallel sensing and adaptive control approach enables distributed control nodes to utilize their local computing resources to simultaneously address dynamic changes in the power grid in both voltage and frequency dimensions, providing technical support for maintaining stable operation under different operating conditions. During the extraction of self-response components, the stability of intermediate computational components of the extraction algorithm is also monitored. By analyzing the time fluctuation of intermediate components, the quality of the current detection channel can be evaluated in real time. When strong adjacent frequency interference is detected, causing the extracted self-response components to be unreliable, the control logic will actively suspend adaptive adjustment and revert to preset conservative control parameters. This avoids executing erroneous control actions when the sensing information is contaminated, thereby improving the engineering applicability of the control method in complex disturbance environments. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the integrated control and security arbitration logic of the present invention. Figure 2 This is a comparison chart of the control effects of the present invention under sudden changes in power grid stiffness; Figure 3 This is a schematic diagram of the distributed control node implementation architecture of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0021] This invention provides a multi-source information fusion control method for improving power system stability. This method is executed on the local digital controller (using a digital signal processor (DSP) or FPGA) of a distributed control node, such as an inverter or energy storage unit. It mainly includes a local routine state error acquisition step, a grid context awareness step based on active probing, and an adaptive fusion control step based on channel quality arbitration. In a specific application scenario, taking a grid-connected inverter as a distributed control node as an example, this method performs a routine state error acquisition step. This node uses local sensors to sample local routine state measurements at its grid connection point at high speed. Taking voltage control as an example, this involves local voltage measurement. The local controller of this node will collect data in real time. With its internally set voltage reference value The comparison is performed to determine the state error signal, i.e., the voltage error signal, used for conventional feedback control. ,Should It is the basis for driving the distributed control nodes to respond; given that it is based solely on Unable to determine the current stiffness of the power grid, the distributed control nodes, while performing routine control, simultaneously initiate an active sensing mechanism. The node controller generates and injects a specific frequency active sensing signal into its output. For example, to detect QV stiffness, this signal could be a small-amplitude reactive power disturbance signal. This signal is digitally generated in the controller's pulse width modulation (PWM) logic and superimposed on the final control command. To ensure the effectiveness and safety of this detection signal, its key parameters are set as follows: First, a specific frequency. The selection of the frequency band must avoid the conventional low-frequency oscillation band of the power grid, such as the range of 0.1Hz to 2Hz, to avoid mutual interference. Simultaneously, it must be within the effective operating bandwidth of the controller and sensor; the preferred setting is to set it above 2Hz, such as 5Hz. Secondly, the amplitude of the actively detected signal... The choice of amplitude must be a trade-off between detectability and low interference. The amplitude must be large enough to be captured by local sensors, but small enough to avoid causing actual disturbance to the power grid. The preferred setting is to limit its amplitude to 0.1% to 5% of the rated capacity of the distributed control node.

[0022] Injection Simultaneously, the distributed control node rapidly acquires its local response signal, i.e., the local voltage signal. To extract from sources containing strong background noise Extracting only the components with medium to high signal-to-noise ratio The resulting weak self-response component This method uses a specific frequency-based approach. The orthogonal demodulation algorithm is executed on the local DSP of the distributed control node. The algorithmic description is as follows: The input is the locally acquired response signal. And two generated by the local controller and Synchronous quadrature reference signal and The processing steps include: Multiplying each signal by two orthogonal reference signals yields two high-frequency mixed signals. These mixed signals are then subjected to a digital low-pass filter (with a cutoff frequency much lower than...). This filters out interference components of non-specific frequencies from the local response signal; the output consists of two relatively stable DC components, namely the in-phase component and the quadrature component, which are the intermediate components used to calculate the self-response component; based on the acquired I and Q components, the self-response component... amplitude It is possible Calculations show that, subsequently, the node calculates context parameters characterizing the local power grid stiffness based on the active probe signal and the self-response component. This calculation step specifically involves calculating the amplitude of the active probe signal. With the amplitude of the self-response component The ratio between them is used to determine the context parameters. The specific calculation method is as follows: in, For context parameters, To actively detect the amplitude of the signal, This represents the amplitude of the self-response component; however, in the actual operation of a distributed control system, strong adjacent frequency interference may exist, causing fluctuations in the I / Q components output by the above quadrature demodulation algorithm, thus affecting the calculated amplitude. To avoid contamination, this method introduces a channel quality self-arbitration mechanism. This mechanism monitors the time stability of the in-phase and quadrature components in real time and generates a probe channel quality index. Specifically, it calculates the variance of the in-phase and quadrature components within a preset time window (e.g., 1 second) and then converts the variance or its sum into a single variance. As a quality indicator of the probe channel .

[0023] Distributed control nodes will probe channel quality indicators Compared with the preset reliability threshold The threshold is compared. The I / Q variance baseline value can be calibrated by multiplying it by a safety margin (e.g., 5 times) under confirmed interference-free operating conditions; based on the comparison results, a decision is made: when the probe channel quality index meets the preset reliability threshold, [the decision is made]. For example, it shows that when I / Q is stable and the channel is clean, the calculated context parameters... Reliable, at this point the node fuses the state error signal. With context parameters To adaptively generate control commands, for example, by adjusting the control gain. Adjusted to The function or feedforward control When the probe channel quality index does not meet the preset reliability threshold, For example, this shows that I / Q fluctuations indicate channel contamination. If unreliable, the node temporarily suspends the fusion of context parameters and causes the distributed control nodes to adopt preset default control parameters, i.e., fall back to the conservative fixed gain calibrated under fragile conditions. This generates control commands, thereby ensuring that distributed control nodes do not execute erroneous adaptive actions when sensing information is unreliable, thus guaranteeing system security.

[0024] Furthermore, the control method of the present invention can further deepen the fusion dimension. When performing the step of extracting the self-response component, the orthogonal demodulation algorithm can acquire the active detection signal while obtaining the I / Q components to calculate the amplitude. With self-response components phase difference between And the phase difference is used as a characterization of the local power grid impedance characteristics (i.e. The phase context parameters (ratio); accordingly, the step of adaptively generating control commands further includes: fusing the phase context parameters. The control strategy is adaptively adjusted to control commands; the specific strategy adjustment method is as follows: when the phase context parameter... When indicating that the local power grid has a high reactance-resistivity ratio characteristic, for example Approaching 90 degrees, a control strategy primarily based on reactive power regulation of voltage is adopted; when the phase context parameter When indicating that the local power grid has a high resistance-to-reactance ratio characteristic, for example If the temperature approaches 0 degrees, the control strategy switches to one that primarily regulates voltage based on active power; alternatively, it adjusts the control based on phase context parameters. The current value is used to dynamically allocate the weights of the active power control component and reactive power control component included in the control command in voltage regulation, so that the control strategy of the distributed control node can also adapt to changes in grid characteristics. Furthermore, the active detection self-arbitration adaptive fusion control method disclosed in this invention can be reused in parallel for other control axes of the system to solve the Pf axis inertia context blindness problem in low-inertia grids. This parallel method includes: acquiring local frequency measurements of the distributed control node. And determine the frequency error signal based on local frequency measurements. Generate and inject a second active active power detection signal into the active power output terminal of the distributed control node. To ensure alignment with the QV axis The detections do not interfere with each other. frequency (e.g., 6Hz) is set to be the same as frequency (For example, 5Hz) different; then, local frequency measurements are collected. and through work in The second orthogonal demodulation algorithm at the frequency point extracts the active active power detection signal from it. The resulting frequency self-response component Based on active power detection signals With frequency self-response components Real-time calculation of frequency context parameters characterizing local frequency stiffness Finally, the distributed control nodes fuse the frequency error signals. With frequency context parameters It adaptively adjusts its active power control gain to generate active power control commands. This parallel mechanism can also be configured with corresponding self-arbitration security logic, enabling distributed control nodes to achieve robust adaptive control on both the QV and Pf axes simultaneously, thus comprehensively improving the system's stability in low-inertia, high-uncertainty power grid environments.

[0025] Example 1: This example demonstrates the operation of the technical solution under specific conditions. In power systems with a high proportion of renewable energy integration, the total system inertia and short-circuit capacity fluctuate in real time with wind and solar power output. In this example, multiple distributed control nodes, i.e., grid-connected inverters, are all equipped with the aforementioned integrated control method. At a specific operating moment, the power grid encounters a complex disturbance: due to large-area photovoltaic cloud shading, the system's equivalent rotational inertia decreases rapidly in a short period, making the system vulnerable on the Pf active frequency axis; subsequently, a large industrial load suddenly enters the grid, causing an active power surge, resulting in all distributed control nodes measuring frequency error signals. For conventional distributed control nodes that have not deployed this invention, their Pf control gain... It is a fixed value calibrated under high inertia conditions, when it is based on this... When using this fixed high gain for response, the output active power will far exceed the current capacity of the fragile power grid, resulting in an over-response and inducing systemic low-frequency oscillations of the Pf axis. However, on the distributed control nodes deployed with this invention, the parallel active detection mechanism of the Pf axis (i.e., injection) And extract The system had been running continuously before the disturbance occurred. During the cloud cover that caused the system inertia to decrease, the local DSP of this node had already calculated the frequency context parameters. The decrease in frequency stiffness was quantified in real time, i.e. The amplitude increased, leading to The value decreases, this frequency context parameter The signal is sent to the Pf control loop in real time; therefore, when this large industrial load is applied, the trigger frequency error signal is activated. At that time, the control law of that node ( (Based on the current extremely low) The value is automatically adjusted to control the active power gain. Adjusted to a lower level to match the current fragile grid, the node outputs only moderate active power in response. This helps to avoid excessive response and keep the system frequency stable.

[0026] Simultaneously with this Pf-axis adjustment, this complex disturbance also evoked an active sensing signal in the power grid with a frequency close to that of the QV-axis. Strong background noise (e.g., 5Hz) causes drastic time fluctuations in the in-phase (I) and quadrature (Q) components used by this node to calculate the QV axis self-response components. At this point, the channel quality self-arbitration mechanism in this method is triggered, and the local DSP generates a probe channel quality index by calculating the time stability (e.g., variance) of the I / Q components. It quickly exceeded the preset reliability threshold. The arbitration logic determines the context parameters of the current QV axis. Contaminated and unreliable, the node controller immediately executes safety protection actions: on the one hand, the Pf axis probe channel operates in... For example, 6Hz is undisturbed, frequency context parameters Reliable, the Pf-axis adaptive control executed normally, ensuring frequency stability; on the other hand, the controller temporarily suspended the QV-axis context parameters. The fusion of these parameters, and the reversion of its QV control law to the preset default control parameters. This operation avoids nodes based on contaminated data. The information execution error QV adaptive adjustment ensures the safety of voltage control. This process demonstrates the coordinated operation of multi-axis parallel sensing and channel quality arbitration in the scheme of this invention, and the adaptive adjustment of the Pf axis (based on...) ) and QV axis safe back-off (based on These processes occur in parallel on the same node without interfering with each other, ensuring that the distributed control system can maintain both control effectiveness and robustness when facing complex power grid conditions and strong signal interference.

[0027] Example 2: To objectively verify the adaptive capability and security of the method of the present invention in distributed control nodes under the influence of changes in grid stiffness and signal interference, a hardware-in-the-loop test platform was built. This platform includes a real-time simulator running a detailed grid model with a simulation step size set to 10 microseconds, and a distributed control node hardware controller that implements the method of the present invention as the test object. The simulator is used to simulate the grid connection point, whose grid stiffness can be dynamically adjusted by the simulation model to simulate the switching between a robust grid state and a fragile grid state. The test setup includes two control groups and the sample group of the present invention: Control Group 1: The node uses traditional fixed parameter control, and the control gain is calibrated under robust grid conditions. Control group 2: The nodes only use the adaptive control part of this invention, that is, only the context parameters are fused. To adjust the gain, but disable the channel quality self-arbitration mechanism; Sample group 1 of this invention: The node adopts the complete method of this invention, including based on Adaptive control and based on probe channel quality indicators The security arbitration and rollback logic; all samples running active detection, their QV axis active detection signals Specific frequency All are set to 5.0Hz, amplitude Set to 1.0% of the node's rated capacity; this is the default control parameter for safe rollback. Set as This value is a conservative setting for detecting channel quality indicators. Reliability threshold After offline calibration, the value was set to 0.1; during the experiment, in At that time, the simulator switches the power grid state from strong to vulnerable. At that time, a step load disturbance lasting 2 seconds was applied at the grid connection point to simulate the generation of voltage error, and the maximum voltage oscillation amplitude during the node response process was recorded; in addition, in operating conditions 5 and 6, While switching to the vulnerable power grid, an additional background interference signal with a frequency of 5.1 Hz and a strong amplitude was injected to simulate the working condition of the detection channel being contaminated. The test data are shown in Table 1.

[0028] Table 1: Performance Comparison Data of Various Control Methods under Different Operating Conditions Table 1 shows that under operating conditions 1 and 2, both control group 1 and sample group 1 of this invention can remain stable under a robust power grid. Under operating condition 3, when the power grid switches to a vulnerable state, the fixed gain of control group 1... Mismatch with grid characteristics led to excessive node response, causing an 8.2% voltage oscillation. In operating condition 4, after grid switching, the active detection mechanism of sample group 1 calculated that the grid stiffness had decreased to 0.72, and the control logic automatically adjusted the control gain accordingly, so that the voltage oscillation was suppressed to 1.5% when encountering the same disturbance. Operating conditions 5 and 6 were used to verify the robustness when the detection channel was contaminated. In operating condition 5, 5.1Hz adjacent frequency interference caused fluctuations in the I / Q components of control group 2. The calculated... The contamination level was 9.5. Based on this error message, the controller performed excessive gain adjustment, causing a large oscillation of 10.5%. In operating condition 6, when sample group 1 of this invention faced the same interference, its calculated... Although both were contaminated to a score of 9.5, their channel quality self-arbitration mechanism simultaneously detected I / Q component fluctuations, leading to a decrease in the channel quality index. If the value increases to 0.85, which is higher than the reliability threshold of 0.1, the control logic determines that the perceived information is unreliable, immediately stops the fusion of context parameters, and causes the controller to revert to the preset default control parameters. During operation, the system oscillation was eventually suppressed to 2.5%. The data comparison in Table 1 shows that control group 1 became unstable under condition 3 (fragile power grid); control group 2 became unstable under condition 5 (fragile + interference); and sample group 1 of this invention maintained stable operation of the system under all conditions (including condition 4 (fragile power grid) and condition 6 (fragile + interference).

[0029] Example 3: This example combines Figures 1 to 3 This paper describes a multi-source information fusion control method for improving power system stability, such as... Figure 1 As shown, the process executes an active detection and response acquisition step, namely, injecting a specific frequency signal and acquiring the local response, and performing orthogonal demodulation and component extraction steps to extract the self-response component and intermediate component from the response signal. Based on this process, two branches are started in parallel. One branch calculates the grid stiffness context parameter, which is based on the amplitude of the self-response component and characterizes the grid stiffness. The other branch generates a detection channel quality index, which is achieved by monitoring the time stability of the intermediate component. Then, the channel quality arbitration step determines whether the index meets the preset reliability threshold. If the determination is yes, it indicates that the channel is reliable, and adaptive fusion control is executed, which fuses the context parameter with a separately acquired state error signal, such as a voltage error signal. If the determination is no, it indicates that the channel is contaminated, and safe backoff control is executed, using preset default control parameters. Finally, the parameters determined by both adaptive fusion control and safe backoff control are used to generate and execute control commands.

[0030] like Figure 2 As shown in the figure, the horizontal axis represents time in seconds, the left vertical axis represents the grid stiffness S(t), and the right vertical axis represents the voltage oscillation amplitude (%). The solid line curve of grid stiffness S(t) shows that the grid stiffness drops from 4.15 to 0.72 at 5.0 seconds. In contrast, the dashed line curve of voltage oscillation controlled by fixed parameters shows a large oscillation of 8.2% after the stiffness decreases, while the dotted line curve of voltage oscillation using the method of this invention suppresses the oscillation amplitude to 1.5% under the same operating conditions. Figure 3 As shown, the core of this architecture is a local digital controller (DSP / FPGA), on which a multi-source information fusion control method is deployed. This method logically includes a grid stiffness autonomous sensing module and a channel quality self-arbitration module. The controller interacts with the grid mainline connection point through a hardware interface layer. The power execution unit (PWM) is responsible for injecting active detection signals into the grid and outputting adaptive power adjustment, while the local measurement unit (sensor) is responsible for collecting local measurement signals from the grid and feeding these signals back to the local digital controller to close the control loop.

[0031] Example 4: This example describes a standardized offline calibration procedure for determining key operating parameters of distributed control nodes. This procedure is executed on a hardware-in-the-loop test platform, which has the same functional specifications as the platform in Example 2 and is capable of simulating different stiffnesses. The power grid operating conditions, and can inject frequency. With amplitude Controllable adjacent frequency interference signals; the local voltage measurement at this node has 16-bit sampling accuracy; the first step in the calibration process is to determine the active detection signal. The parameter, i.e., a specific frequency and amplitude To determine the frequency The lower limit is used to simulate normal low-frequency oscillations of the power grid in the range of 0.1Hz to 2.0Hz in the simulator. Starting from 1.0Hz and gradually increasing, it was found that below 2.0Hz, the I / Q components of the quadrature demodulation algorithm output were unstable due to interference from conventional oscillations. However, above 2.0Hz, taking 5.0Hz as an example, the I / Q components were stable. Therefore, it was determined that... The lower limit is 2.0Hz, and in this embodiment, it is preferably 5.0Hz. Based on a frequency setting of 5.0Hz, calibrate its amplitude. The scope; in Under the condition of a fragile power grid set to 0.5 pu, Starting with 0.01% of the node's rated capacity and increasing incrementally, it was found that below 0.1%, the extracted self-response component... The amplitude is submerged in local measurement noise, resulting in insufficient I / Q component signal-to-noise ratio. When the amplitude is between 0.1% and 5.0%, the I / Q component signal-to-noise ratio is good, and the voltage disturbance to the grid's point of common coupling is less than 0.1%, meeting the grid connection requirements. When the amplitude exceeds 5.0%, the voltage disturbance to the grid exceeds the allowable range, therefore, it is determined that... The preferred range is 0.1% to 5.0% of the node's rated capacity, and in this embodiment, it is preferably 1.0%.

[0032] The second step in the calibration process is to determine the cutoff frequency of the digital low-pass filter in the quadrature demodulation algorithm. and the time window used to calculate variance ; The setting needs to balance the ability to suppress interference from adjacent frequencies with the response speed of I / Q components, through... Under Hz conditions, injection A strong interference with a frequency difference of 0.5 Hz is used to test different... Value, found in Above 0.2Hz, the 0.5Hz difference frequency component leaks into the I / Q circuit, causing fluctuations. When set to 0.1Hz, the I / Q components remain stable, therefore Preferably 0.1Hz; time window The settings need to ensure that they can reliably reflect the fluctuations of the I / Q components. Under Hz conditions, injection Interference with a difference frequency of 0.2 Hz and a period of 5 seconds was found in... When the time interval is less than 2.5 seconds, or half a cycle, the variance calculation value is unstable, while... When the time interval is set to 5 seconds, the variance can stably reflect the fluctuation, therefore The preferred time is 5 seconds. The third step in the calibration process is to determine the sound channel quality index. Reliability threshold ;exist Hz and Under the condition of s, without any nearby interference, the active detection was operated for a long time, taking 300 seconds as an example, and the data was recorded. That is, the background baseline value of the sum of I / Q variances, measured to its maximum value. The value is 0.018; to ensure sufficient safety margin, the value will be... Set it to 5 times the maximum value of this baseline, that is To facilitate engineering adjustments, take The fourth step in the calibration process is to determine the default control parameters. In the simulator, the grid stiffness is set to the most vulnerable operating condition that the system may encounter during operation, in order to... Taking 0.5 PU as an example, under this operating condition, all adaptive logic is turned off, and only fixed-gain control is used. By gradually reducing the control gain and applying a step disturbance, a conservative gain value is found that ensures the system remains stable even under the most vulnerable operating conditions, such as oscillations below 3%. This value is determined to be... In this embodiment, this value is approximately the subscript value of the robust power grid. 0.2 times that of the above procedures, before the distributed control nodes are officially put into operation, they obtain a complete set of operating parameters with clear physical basis and calibration procedures, including Frequency and amplitude, filter cutoff frequency, variance calculation window, reliability threshold and default control parameters .

[0033] Example 5: In one node, the strategy adaptive logic of the QV axis and the stiffness adaptive logic of the Pf axis run in parallel in the local DSP. The strategy adaptive logic of the QV axis is used to adapt to the phase context parameters acquired in real time. The system dynamically allocates weights for the active power control component and reactive power control component used for voltage regulation; the Pf axis stiffness adaptive logic is used to adjust the values ​​based on real-time frequency context parameters. Adjust the active power control gain to match the current system inertia. Under specific operating conditions, when nodes simultaneously sense that the grid exhibits a high resistance-to-reactance ratio characteristic, i.e., the phase context parameter... Approaching 0 degrees, and low-frequency stiffness characteristics, i.e., frequency context parameters. When the value is very low, the two parallel logics mentioned above generate control conflicts. At this time, the QV strategy logic requests an increase in the weight of active power in voltage regulation, i.e. The Pf stiffness logic increases, while it requests limiting the response gain of all active power. To reduce frequency instability, an arbitration mechanism is employed to resolve this conflict, prioritizing the frequency stability of the Pf axis. This is achieved through adaptive logic for Pf stiffness, and is related to the frequency context parameters. Related active control gain This is used as the maximum allowable gain upper limit for the active power regulation channel of the entire node; correspondingly, the phase context parameters are calculated by the QV strategy adaptive logic. Related active power regulation component weights Before being sent to the actuator, with Comparison; active power control gain ultimately used for voltage regulation It was identified as and The smaller value in the equation; this arbitration step ensures that when a node utilizes active power to regulate voltage, its action range is limited to the safe boundary that the current system frequency stiffness can withstand, thereby achieving coordinated stability in both QV and Pf control dimensions.

[0034] Example 6: This example describes the use of phase context parameters in calibrating distributed control nodes. With local power grid impedance characteristics ( A standardized engineering procedure for mapping the relationship between (ratio) and (value). This procedure is executed during the offline testing phase or the field commissioning phase before the node controller is deployed. Its purpose is to provide a reproducible and quantifiable decision-making basis for the adaptive function of the control strategy. This calibration procedure can accurately set the equivalent impedance of the node grid connection point (ratio). and The test platform, such as a hardware-in-the-loop simulator or adjustable impedance source, is used for execution; settings are configured. The benchmark condition with a higher ratio, pu、 pu ( Taking a node as an example, the node is started and the QV axis active detection mechanism in the method of this invention is run, that is, a specific frequency is injected. Such as a 5.0Hz active detection signal After the system stabilizes, the nodes extract the self-response components using an orthogonal demodulation algorithm. The reference phase context parameters under this working condition are calculated. Record the measured value, such as .

[0035] Maintain the magnitude of the total impedance of the power grid Basically constant, systematically adjusted. and The value is used to simulate a series of values ​​with different The ratio of the power grid operating conditions; taking the gradient sequence as an example, the following settings are made sequentially. The ratios are 0.3, 0.5, 0.7, 1.0, 1.5, and 2.0; in each At each ratio setpoint, the active detection and quadrature demodulation steps are repeated, and the phase context parameters in the steady state are recorded. The measured values; through this gradient test, a set of... and Data points showing the correspondence between ratios, such as ( ), ( ), ( ), ( ), ( ), ( Finally, the node's local controller will use this set of discrete data points obtained through calibration ( Ratio and The measured values ​​are either fixed into an internal lookup table or generated as a continuous mapping function through curve fitting algorithms, such as piecewise linear interpolation or polynomial fitting. This lookup table or function constitutes the decision baseline for the control strategy in the specific implementation that adaptively adjusts the control commands, enabling the node to perform operations based on real-time measurements. By looking up tables or performing calculations, it can be converted into a value for the current power grid. The quantitative assessment of characteristics is used to determine whether a strategy of regulating voltage with reactive power should be adopted (e.g., Or switch to a strategy that regulates voltage based on active power (such as...) ) or in the transition zone ( Dynamic weight allocation is performed.

[0036] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-source information fusion control method for improving the stability of a power system, wherein the method is executed on at least one distributed control node of the distributed control system of the power system, characterized in that, Includes the following steps: Acquire local routine state measurements of distributed control nodes and determine state error signals based on local routine state measurements; Generate and inject preset active detection signals into the output of the distributed control nodes; Collect local response signals from distributed control nodes; Perform quadrature demodulation based on a specific frequency on the local response signal to extract the self-response component and obtain the in-phase and quadrature components used to calculate the self-response component; Based on the active detection signal and self-response component, the context parameters characterizing the stiffness of the local power grid are calculated; Real-time monitoring of the temporal stability of in-phase and quadrature components is used to generate sound channel quality indicators. The probe channel quality index is compared with a preset reliability threshold; When the probe channel quality index meets the preset reliability threshold, the state error signal and context parameters are fused to adaptively generate control commands and execute the control commands. When the probe channel quality index does not meet the preset reliability threshold, the fusion of context parameters is temporarily suspended, and the distributed control nodes use preset default control parameters to generate and execute control commands.

2. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, The step of extracting the self-response component includes: performing a quadrature demodulation algorithm based on a specific frequency on the local response signal to filter out interference components of non-specific frequencies in the local response signal to obtain the self-response component; the step of calculating the context parameters includes: determining the context parameters based on the ratio between the amplitude of the active probe signal and the amplitude of the self-response component.

3. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, The step of extracting the self-response component further includes: obtaining the phase difference between the active detection signal and the self-response component, and using the phase difference as a phase context parameter characterizing the local power grid impedance characteristics; the step of adaptively generating control commands further includes: fusing the phase context parameter to adaptively adjust the control strategy of the control commands.

4. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, Also includes: Acquire local frequency measurements of distributed control nodes and determine frequency error signals based on local frequency measurements; Generate and inject a preset active active power detection signal with a frequency different from that of the active detection signal into the active power output terminal of the distributed control node; Collect local frequency measurements and extract the frequency self-response component caused by the active active power detection signal; Based on the active active detection signal and frequency self-response component, frequency context parameters characterizing local frequency stiffness are calculated in real time. The frequency error signal and frequency context parameters are fused to adaptively generate active power control commands.

5. The multi-source information fusion control method for improving power system stability according to claim 2, characterized in that, The steps for calculating the context parameters are as follows: Calculate the context parameters using the following formula. : ,in For context parameters, To actively detect the amplitude of the signal, The amplitude of the self-response component.

6. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, The steps for real-time monitoring of the time stability of in-phase and quadrature components include: calculating the variance of in-phase and quadrature components within a preset time window; and using the variance as a quality indicator of the probe channel.

7. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, The active detection signal is the reactive power disturbance signal, and the local routine state measurement is the local voltage measurement; the state error signal is the voltage error signal, the local response signal is the local voltage signal, and the self-response component is the voltage response component in the local voltage signal caused by the reactive power disturbance signal.

8. The multi-source information fusion control method for improving power system stability according to claim 1, characterized in that, The specific frequency is set to be higher than 2Hz; the amplitude of the active detection signal is in the range of 0.1% to 5% of the rated capacity of the distributed control node.

9. The multi-source information fusion control method for improving power system stability according to claim 3, characterized in that, The steps of integrating phase context parameters to adaptively adjust control commands include: when the phase context parameters indicate that the local power grid has a high reactance-resistance ratio characteristic, adopting a control strategy with reactive power regulating voltage as the main dimension; and when the phase context parameters indicate that the local power grid has a high resistance-reactance ratio characteristic, switching to a control strategy with active power regulating voltage as the main dimension.

10. A multi-source information fusion control method for improving power system stability according to claim 3, characterized in that, The step of integrating phase context parameters to adaptively adjust the control strategy of control commands includes: dynamically assigning weights to the active power control component and reactive power control component contained in the control commands in voltage regulation based on the current value of the phase context parameters.

Images