Power grid load and equipment reactive voltage dispatching system

Abstract

The invention discloses a power grid load and equipment reactive voltage dispatching system, relates to the technical field of power system operation and control, and aims to solve the technical problem that current power grid reactive voltage dispatching has hysteresis quality and is difficult to quickly stabilize power grid charges. Comprising a data acquisition module, a data processing module, a scheduling decision module, an execution control module and a communication module used for data interaction among the modules, and the data processing module is in communication connection with the data acquisition module and used for preprocessing, analyzing, predicting and storing acquired data. According to the method, through accurate prediction (based on the LSTM / SVM model) of the load trend in the future 15 minutes to 2 hours, the system can predict the voltage fluctuation risk, a scheduling strategy is generated in advance, passive response is changed into active defense, voltage collapse or out-of-limit accidents are effectively prevented, and the method has the advantages that repeated oscillation and conflicts of different devices are reduced, and the operation stability of a power grid is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system operation and control, in particular to a power grid load and equipment reactive voltage scheduling system. BACKGROUND

[0002] With the continuous development of social economy and large-scale access of new energy generation (such as wind power and photovoltaic power), the load characteristics of the power system are becoming increasingly complex, and the volatility and uncertainty are significantly enhanced. The dynamic change of the load directly impacts the reactive power balance of the system, causing voltage deviation or even instability. If not properly regulated, it will not only affect the power quality, but also may cause equipment damage or power grid oscillation and other safety accidents.

[0003] Currently, the traditional power grid reactive voltage scheduling method mainly relies on artificial experience and basic automation devices, which has the following main defects: the conventional monitoring equipment usually uses fixed cycle sampling, which is difficult to capture the instantaneous and rapid changes of the load and voltage, resulting in a lag in the data on which the scheduling decision is based; the load monitoring, reactive power regulation and voltage control are usually independent of each other, and there is a lack of unified and coordinated optimization scheduling strategy, making it difficult to achieve optimal allocation of the whole network resources. In view of this, we propose a power grid load and equipment reactive voltage scheduling system. SUMMARY

[0004] The present application aims to overcome the shortcomings of the prior art, adapt to the actual needs, and provide a power grid load and equipment reactive voltage scheduling system to solve the technical problems of current power grid reactive voltage scheduling lag and difficulty in quickly and smoothly regulating the power grid load.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a power grid load and equipment reactive voltage scheduling system, comprising a data acquisition module, a data processing module, a scheduling decision module, an execution control module and a communication module for data interaction between the modules; The data processing module is in communication connection with the data acquisition module, and is used for pre-processing, analyzing and predicting the collected data and storing them, wherein the analysis and prediction includes predicting the future load trend based on a load prediction model; The scheduling decision module is in communication connection with the data processing module, and is used for generating a scheduling strategy based on the analysis and prediction results by using a multi-objective optimization algorithm, and verifying and correcting the scheduling strategy by using a power grid virtual simulation model constructed by using digital twinning technology; The execution control module is in communication connection with the scheduling decision module, and is used for executing the verified scheduling strategy and collecting the state data after execution and feeding back to the data acquisition module.

[0006] The application can predict voltage fluctuation risks and generate scheduling strategies in advance through accurate prediction of future 15-minute to 2-hour load trends based on an LSTM / SVM model, changing passive response to active defense, effectively preventing voltage collapse or out-of-limit accidents. The virtual simulation model constructed by the digital twin technology verifies and corrects the scheduling strategy, which can discover potential risks and unreasonable instructions before actual execution, greatly reducing the possibility of power grid failure caused by mis-scheduling. The scheduling decision module considers various devices such as SVG, capacitors, transformer taps, etc., to generate coordinated control instructions, avoiding repeated oscillation and conflict between devices, and ensuring the power grid to operate near the optimal stability point.

[0007] Preferably, the data acquisition module includes smart meters and load monitoring terminals deployed at substations, distribution lines and user sides, configured to collect instantaneous values, average values and trend data of active and reactive loads at a sampling frequency of 1-5Hz, voltage transformers and smart voltage monitors deployed at key nodes of the power grid for collecting voltage amplitude and phase data, reactive power compensation device state monitoring units for collecting operating states, output values and fault information of capacitor banks, reactors, static var generators SVG and static var compensators SVC, interference factor monitoring units including harmonic monitors and new energy generation output monitoring terminals for collecting power grid harmonic content and wind power / photovoltaic output fluctuation data.

[0008] Preferably, the data preprocessing performed by the data processing module includes removing data noise and correcting errors using Kalman filtering or wavelet filtering algorithms, and the load prediction model is a model based on an LSTM neural network or a support vector machine, used to predict load change trends in the next 15 minutes to 2 hours.

[0009] Preferably, the scheduling decision module is specifically configured to establish a mathematical model between load changes and reactive power demand, voltage fluctuation, to determine the priority of reactive power regulation under different load scenarios, and based on the mathematical model and a multi-objective optimization algorithm, to generate at least one of the output instructions for the static var generator SVG, the switching instructions for the capacitor bank, the adjustment instructions for the transformer tap and the load control instructions.

[0010] Preferably, the execution control module includes a reactive device control unit that controls the power module output of the static var generator SVG, the switching of the capacitor bank / reactor and the electric adjustment of the transformer tap through a programmable logic controller PLC or an intelligent control unit, and a load control unit that sends peak-shaving or peak-avoiding instructions to the intelligent control terminal of interruptible load users, or remotely controls the switching on and off of user loads.

[0011] Preferably, the communication module adopts a hybrid communication mode of wired and wireless, the modules inside the substation adopt optical fiber communication with a transmission rate not less than 1000 Mbps; the power distribution line and the user side monitoring terminal and the data acquisition module adopt 5G / 4G wireless communication or power wireless private network, and the communication delay is not more than 100 milliseconds.

[0012] Preferably, the total delay of the system from data acquisition to device execution adjustment is not more than 200 milliseconds.

[0013] Preferably, the data processing module is further configured with a data quality evaluation unit for calculating the signal-to-noise ratio, the integrity rate and the proportion of abnormal data of the preprocessed data, and if the data quality does not meet the preset threshold, triggering the data acquisition module to reacquire or correct the data.

[0014] Preferably, the power grid virtual simulation model constructed in the scheduling decision module is a dynamic model containing topological structure, device parameters and real-time running state, which is updated synchronously by receiving real-time data stream of the data processing module, and is used to simulate the power grid voltage distribution and reactive power flow within 15 minutes after the execution of the scheduling strategy.

[0015] Preferably, the execution control module is further provided with a safety check unit for checking whether the received control instruction conforms to the safe operation range of the device and the safe operation constraint of the power grid before executing the scheduling strategy, and if the check fails, discarding the instruction and sending alarm information to the scheduling decision module.

[0016] Compared with the prior art, the present application has the following advantages: 1、The present application can predict the voltage fluctuation risk and generate scheduling strategy in advance by accurately predicting the load trend in the next 15 minutes to 2 hours (based on LSTM / SVM model), from passive response to active defense, effectively preventing voltage collapse or out-of-limit accidents. The virtual simulation model constructed by digital twin technology can verify and correct the scheduling strategy, which can find potential risks and unreasonable instructions before actual execution, greatly reducing the possibility of power grid failure caused by mis-scheduling. The scheduling decision module considers various devices such as SVG, capacitor and transformer tap, generates collaborative control instructions, avoids repeated oscillation and conflict between devices, and ensures the power grid to run near the optimal stable point.

[0017] 2、The system can quickly capture and suppress voltage sag, flicker and other problems caused by new energy fluctuation, load mutation, etc., and ensure the voltage quality at the user side, by the total delay of the system from data acquisition to device execution being not more than 200 milliseconds, combined with high-frequency data acquisition of 1-5 Hz. Through harmonic monitoring and new energy output monitoring, the system can take into account the harmonic influence when formulating the reactive voltage strategy, realize the cooperation of reactive power compensation and harmonic control, and further improve the power quality. Through accurate reactive voltage control, the voltage level of the power grid is maintained, and the device downtime or damage caused by unqualified voltage is reduced. At the same time, through interruptible load management, peak shaving and valley filling are realized with the minimum impact range when necessary, and the reliable power supply of most users is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0018] Fig. 1 The module schematic diagram of the application is shown in the figure; Fig. 2 The fast control flow and delay index schematic diagram is shown in the figure; Fig. 3 The data flow and closed-loop feedback schematic diagram is shown in the figure; Fig. 4 The core decision and digital twin verification schematic diagram is shown in the figure. DETAILED DESCRIPTION

[0019] As shown in the figure, the power grid load and device reactive voltage scheduling system of the application comprises a data acquisition module, a data processing module, a scheduling decision module, an execution control module and a communication module for data interaction between the modules; the system builds a complete "perception-analysis-decision-execution-feedback" automatic closed loop. Data acquisition is the "senses" of the system, data processing is information filtering and understanding, scheduling decision is judgment, execution control is action, and the communication module is signal transmission. The final feedback link ensures that the system can perceive the effect of its own action and realize self-adaptive optimization. Figs. 1 to 4

[0020] Communication module and system performance: Optical fiber inside the substation: The substation is the core of data aggregation and control instruction sending, and the internal data flow is large and the real-time requirement is extremely high. Gigabit optical fiber provides a high-speed, reliable and anti-interference backbone channel.

[0021] Remote wireless communication: For distribution lines and user side terminals that are widely distributed and difficult to wire, 5G / 4G or power wireless private network provides a flexible and economical access scheme. The low delay and large connection characteristics of 5G technology are particularly suitable for the needs of distributed measurement and control of power grids.

[0022] Engineering significance of delay index: ​Communication delay ≤ 100ms: This index ensures the timeliness of signal transmission from the remote monitoring point to the master station and from the master station to the remote execution unit. This is crucial for suppressing rapidly developing dynamic processes such as voltage collapse and power angle instability.

[0023] Total delay ≤ 200ms: This is a stringent performance index from the overall system level. It covers data acquisition, communication transmission, data processing, optimization decision, strategy simulation, instruction issuance, and execution mechanism action. Achieving this means that the system can effectively respond to dynamic changes in the power grid at the second level or even sub-second level, and has the ability to actively support grid stability, not just minute-level optimization scheduling The data processing module is in communication connection with the data acquisition module, used for pre-processing, analyzing and predicting and storing the collected data, the analysis and prediction including predicting future load trend based on load prediction model; The dispatching decision module is in communication connection with the data processing module, used for generating dispatching strategy through multi-objective optimization algorithm based on the analysis and prediction result, and verifying and correcting the dispatching strategy by using power grid virtual simulation model constructed by digital twinning technology; The integration of digital twinning technology into the dispatching decision module is the core innovation point of the system. It is not just an offline simulation tool, but a virtual mirror that is real-time synchronized and bidirectionally interactive with the physical power grid. Before the strategy is issued to the real device, it is "preplayed" in the virtual model, which can: Risk avoidance: Discover strategies that may cause voltage out-of-limit, device overload, etc. in advance to avoid misoperation.

[0024] Strategy optimization: Quickly perform multiple iterations in the virtual environment to find better solutions under multiple constraint conditions without worrying about impacting the actual power grid.

[0025] Training and deduction: Can be used for operator training and extreme fault scenario deduction.

[0026] The execution control module is in communication connection with the dispatching decision module, used for executing the verified dispatching strategy, and collecting the state data after execution and feeding back to the data acquisition module.

[0027] In the embodiment of the present application, the data acquisition module includes smart meters and load monitoring terminals deployed at substations, distribution lines and user sides, configured to collect instantaneous values, average values and trend data of active and reactive loads at a sampling frequency of 1-5 Hz, voltage transformers and smart voltage monitors deployed at key nodes of the power grid for collecting voltage amplitude and phase data, reactive power compensation device state monitoring units for collecting operating state, output value and fault information of capacitor banks, reactors, static var generators (SVG) and static var compensators (SVC), interference factor monitoring units including harmonic monitors and new energy generation output monitoring terminals for collecting power grid harmonic content and wind power / photovoltaic output fluctuation data; the significance of high-frequency sampling: the sampling frequency of 1-5 Hz (i.e. 1-5 times per second) is much higher than that of the traditional SCADA system (usually once per second or several seconds), which enables the system to capture the instantaneous fluctuations and transient processes of loads and voltages, such as disturbances caused by large motor starting, impact loads or rapid changes in new energy output. This is the basis for precise and rapid control.

[0028] Comprehensiveness of data types: Instantaneous values and average values: instantaneous values are used to analyze rapid changes and transient problems, and average values are used to evaluate overall trends and steady-state performance.

[0029] Trend data: through historical data sequences, it can be preliminarily judged whether the load is in an upward, downward or stable period, providing input for the prediction model.

[0030] Phase data: voltage phase is a key parameter for judging the stability of the power grid and the direction of reactive power flow, and is crucial for accurate calculation of power factor and stable control.

[0031] Interference factor monitoring: taking harmonics and new energy output fluctuations as special monitoring content, the positioning of the system towards modern power grids (high proportion of power electronic devices, high penetration rate of new energy) is clear. Harmonic data can be used to evaluate the filtering effect of reactive power compensation devices (such as SVG), and new energy output data is an important input for load prediction disturbances.

[0032] In the embodiment of the present application, the data preprocessing performed by the data processing module includes removing data noise and correcting errors using Kalman filtering or wavelet filtering algorithms, and the load prediction model is a model built based on LSTM neural network or support vector machine, which is used to predict the load trend in the next 15 minutes to 2 hours; the necessity of data preprocessing: the original data collected in the field usually contains noise (such as electromagnetic interference, sensor errors) and outliers (such as temporary communication interruptions).

[0033] Kalman Filter: Suitable for real-time and recursive optimal estimation of linear systems, it can effectively filter out random noise and smooth interpolate missing data.

[0034] Wavelet Filter: Particularly good at processing non-stationary signals, it can analyze data in both time and frequency domains, making it very suitable for capturing mutation points and different time scale features in load data.

[0035] Selection of Load Forecasting Model: LSTM Neural Network: As a type of recurrent neural network, LSTM is naturally good at handling time series data. It can "remember" long-term dependencies (such as the influence of "yesterday's same period load" on "current load"), and has strong advantages in capturing periodicity, trendiness and randomness of load characteristics.

[0036] Support Vector Machine: It performs stably in small sample cases and can solve non-linear regression problems through kernel functions, making it one of the classic load forecasting algorithms.

[0037] Prediction Length (15 minutes to 2 hours): This belongs to the category of ultra-short-term load forecasting, and its prediction results are directly used by the dispatching decision module for real-time and near-real-time optimization. The prediction accuracy directly determines the effectiveness of the dispatching strategy.

[0038] In the embodiments of the present application, the dispatching decision module is specifically configured to establish a mathematical model between load changes and reactive power demand, voltage fluctuations, to determine the priority of reactive power regulation under different load scenarios, and based on the mathematical model and a multi-objective optimization algorithm, to generate at least one of the output instructions of the static var generator SVG, the switching instructions of the capacitor bank, the adjustment instructions of the transformer tap, and the load control instructions; Mathematical model establishment: This model is essentially a quantitative relationship formula describing the relationship between load (P, Q), reactive power compensation device output (Qc), network parameters (such as transformer ratio), and node voltage (U). Through this model, the system can quantitatively analyze "in a certain load scenario, which device to adjust and how much to adjust, the effect of voltage improvement is most significant", so as to determine the priority of reactive power regulation. For example, preferentially adjust the SVG with fast response speed, and cooperate with the switching of the capacitor bank.

[0039] Multi-objective optimization algorithm: Its optimization objectives usually include: Objective 1: Maintain all node voltages within the qualified range specified by national standards (such as 0.95~1.05 p.u.).

[0040] Objective 2: Minimize grid network loss and improve economic operation efficiency.

[0041] Objective 3: Balance reactive power locally as much as possible to reduce long-distance transmission of reactive power.

[0042] Objective 4: Minimize the number of operations performed by control equipment (such as capacitor switching and transformer tap adjustment) to extend equipment life.

[0043] These objectives may be contradictory, and multi-objective optimization algorithms seek an optimal balance point (Pareto optimal solution) among these objectives.

[0044] The diversity of commands: The generated commands cover the control methods of the entire process from "source-grid-load", reflecting the idea of ​​comprehensive resource optimization. SVG commands are continuous, fast and fine-tuning, capacitor switching is discrete and large-capacity coarse-tuning, transformer tap change is changing the background voltage of the power grid, and load control is demand-side response, maintaining balance in emergency situations.

[0045] In an embodiment of the present invention, the execution control module includes a reactive power equipment control unit, which controls the power module output of the static var generator (SVG), the switching action of the capacitor bank / reactor, and the electric adjustment of the transformer taps through a programmable logic controller (PLC) or an intelligent control unit; and a load control unit, which is used to send peak-shaving or peak-avoidance commands to the intelligent control terminals of interruptible load users, or to remotely control the disconnection and connection of user loads.

[0046] In embodiments of the present invention, the communication module employs a hybrid wired and wireless communication method. Modules within the substation communicate via fiber optic communication with a transmission rate of no less than 1000 Mbps. The power distribution lines and user-side monitoring terminals communicate with the data acquisition module via 5G / 4G wireless communication or a dedicated power wireless network, with a communication delay of no more than 100 milliseconds. Regarding the reliability of the control unit: industrial-grade PLCs or dedicated intelligent control units are used, emphasizing their applicability in substations and power distribution sites with complex electromagnetic environments and extremely high reliability requirements. They possess high anti-interference capabilities and stable I / O interfaces, enabling direct drive of actuators such as switches and relays.

[0047] The deepening of load control: The implementation of load control units marks the evolution of the system from the traditional "grid-side one-way control" to "source-grid-load interaction". Sending peak-shaving / peak-avoidance commands to user-side smart terminals is a demand-side response based on the market or agreement; while remote direct control is usually applicable to important users who have signed interruptible load contracts, and is a backup measure when the system experiences an emergency power shortage.

[0048] In an embodiment of the present invention, the total delay of the system from data acquisition to device adjustment is no more than 200 milliseconds.

[0049] In an embodiment of the present invention, the data processing module is further configured with a data quality assessment unit, which is used to calculate the signal-to-noise ratio, integrity rate and abnormal data ratio of the preprocessed data. If the data quality does not meet the preset threshold, the data acquisition module is triggered to re-acquire or correct the data. By objectively assessing data health using quantitative indicators (signal-to-noise ratio, integrity rate, and percentage of outlier data), the system ensures the reliability of data used for subsequent load forecasting and optimization decisions, reducing the risk of misjudgment at the source. When data quality fails to meet standards, the system automatically triggers a "re-acquisition or data correction" action, forming a closed-loop control system for data quality from assessment to correction. This reduces manual intervention and enhances the system's autonomy. In scenarios where temporary equipment failures or communication interference cause data anomalies, this mechanism effectively identifies and attempts to repair them, ensuring stable system operation even under suboptimal conditions, rather than making dangerous scheduling decisions based on erroneous data.

[0050] In embodiments of the present invention, the power grid virtual simulation model constructed in the scheduling decision module is a dynamic model including topology, equipment parameters, and real-time operating status. It is synchronously updated by receiving real-time data streams from the data processing module, and is used to simulate the power grid voltage distribution and reactive power flow within the next 15 minutes after the scheduling strategy is executed. Before issuing scheduling instructions to the real power grid, simulation execution is performed in the virtual model. This can expose potential risks in advance, such as simulating overvoltage on a line or voltage exceeding limits at a node, thereby correcting the strategy before it causes actual impact. Transforming the output of the optimization algorithm from "theoretically optimal" to "practically feasible and safe" greatly reduces the safety risks that may arise from directly controlling real equipment. This is a safety redundancy design. Limiting the simulation to "within the next 15 minutes" closely matches the load forecast period (15 minutes to 2 hours mentioned in claim 3), ensuring the timeliness and relevance of the simulation, and balancing computational complexity with scheduling requirements.

[0051] In embodiments of the present invention, the execution control module further includes a security verification unit, used to verify whether the received control commands comply with the safe operating range of the equipment and the safe operation constraints of the power grid before executing the scheduling strategy. If the verification fails, the command is discarded and an alarm message is sent to the scheduling decision module; commands that may damage the equipment are directly intercepted at the verification stage, preventing hardware damage caused by incorrect control commands. This complements the "pre-verification" of the digital twin model, forming a defense-in-depth system of "software simulation pre-verification + hardware pre-execution verification," greatly enhancing the overall security and robustness of the system. When a command is discarded, an alarm is immediately sent to the scheduling decision module. This allows the decision center to know about control failures immediately and trigger a new round of analysis and decision-making, avoiding system "blindness" or "deadlock," and maintaining the system's situational awareness and self-healing capabilities.

[0052] Working principle: This embodiment provides a power grid load and equipment reactive power and voltage dispatching system. When in use... S1: Data Acquisition and Upload The data acquisition module collects real-time panoramic operation data of the power grid and uploads it to the data processing module via the communication module. The specific data collected includes: instantaneous, average, and trend data of active and reactive loads in the power grid, collected at a sampling frequency of 1-5Hz through smart meters and load monitoring terminals; voltage amplitude and phase data of substation busbars, line ends, and key nodes at important user entrances, collected through voltage transformers and smart voltage monitors; operating status, real-time output value, and fault information of reactive power compensation equipment, and transformer tap position data; and harmonic monitoring instruments and new energy output monitoring terminals, collected data on the harmonic content of the power grid and fluctuations in wind and solar power output.

[0053] S2: Data Processing and Analysis The data processing module processes and analyzes the received raw data: Data preprocessing: Kalman filtering or wavelet filtering algorithms are used to filter the raw data to remove noise interference and correct data deviations caused by transformer errors and transmission delays.

[0054] Load forecasting and state analysis: Based on the load forecasting model built by LSTM neural network or support vector machine, the load change trend is predicted for the next 15 minutes to 2 hours; at the same time, the voltage deviation and reactive power deficit of each node of the current power grid are analyzed to identify abnormal states such as voltage over-limit and reactive power imbalance.

[0055] Data storage: The preprocessed real-time data and analysis results are stored in a distributed database for querying and retrieval.

[0056] S3: Intelligent Scheduling Decision Generation Based on the analysis results from the data processing module, the scheduling decision module generates the optimal scheduling strategy: Collaborative analysis: Establish a mathematical model between load changes, reactive power demand, and voltage fluctuations to clarify the reactive power adjustment priority of each node under the current operating scenario.

[0057] Multi-objective optimization: With the comprehensive optimization objectives of minimizing voltage deviation, network loss, and equipment operation frequency, particle swarm optimization or genetic algorithm is used to calculate and generate adjustment instructions and load control suggestions for reactive power compensation equipment; the adjustment instructions include the output value of static var generator (SVG), the number of capacitor banks / reactors switched on and off, and the adjustment range of transformer taps.

[0058] Strategy simulation verification: The generated scheduling strategy is verified by simulation in the power grid virtual simulation model built based on digital twin technology. If the simulation results do not meet the power grid operation constraints, the process returns to step S3 for re-optimization until a feasible and optimal scheduling strategy is generated.

[0059] S4: Scheduling instruction execution and closed-loop feedback The execution control module receives and executes scheduling instructions, and completes the system closed loop: Equipment control: Through a programmable logic controller (PLC) or intelligent control unit, drive the SVG to adjust its reactive power output, control the switching action of capacitor banks / reactors, and adjust the tap position of transformers; at the same time, send load regulation commands to the control terminal on the interruptible load user side.

[0060] Status feedback: Real-time acquisition of status data and grid parameters after the equipment performs adjustment actions, including the actual output value of the SVG, the actual switching status of the capacitor bank, and the actual voltage value of key nodes, and feeding this status data back to the data acquisition module, thereby forming a closed-loop control system to provide a basis for the next round of scheduling decisions.

[0061] S5: Communication guarantee Throughout the entire process from S1 to S4, the communication module ensures the real-time performance and reliability of data interaction at each stage. Fiber optic communication is used within the substation, while 5G / 4G wireless communication or a dedicated power wireless network is used between the substation and the distribution lines and user-side distributed terminals to ensure that the communication delay is no more than 100 milliseconds and the total end-to-end scheduling delay is no more than 200 milliseconds.

[0062] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A power grid load and device reactive voltage dispatch system, characterized by, The system comprises a data acquisition module, a data processing module, a scheduling decision module, an execution control module, and a communication module for data interaction between the modules. The data processing module is in communication connection with the data acquisition module, and is configured to pre-process, analyze, predict, and store the collected data, wherein the analysis and prediction include predicting future load trends based on a load prediction model. The scheduling decision module is in communication connection with the data processing module, and is configured to generate a scheduling strategy based on the analysis and prediction results by using a multi-objective optimization algorithm, and verify and correct the scheduling strategy by using a power grid virtual simulation model constructed by using digital twinning technology. The execution control module is in communication connection with the scheduling decision module, and is configured to execute the verified scheduling strategy and collect state data after execution and feed back to the data acquisition module. The data acquisition module comprises intelligent electric meters and load monitoring terminals deployed at substations, power distribution lines, and user sides, and is configured to collect instantaneous values, average values, and change trend data of active loads and reactive loads at a sampling frequency of 1-5 Hz, voltage transformers and intelligent voltage monitors deployed at key nodes of the power grid, and is configured to collect voltage amplitude and phase data, reactive power compensation device state monitoring units configured to collect operating states, output values, and fault information of capacitor banks, reactors, static var generators (SVGs), and static var compensators (SVCs), and interference factor monitoring units comprising harmonic monitors and new energy power generation output monitoring terminals configured to collect power grid harmonic content and wind power / photovoltaic output fluctuation data.

2. The grid load and device reactive voltage dispatch system of claim 1, wherein, The data pre-processing performed by the data processing module comprises removing data noise and correcting errors by using Kalman filtering or wavelet filtering algorithms, and the load prediction model is a model constructed based on an LSTM neural network or a support vector machine, and is configured to predict load change trends in the next 15 minutes to 2 hours.

3. The grid load and device reactive voltage dispatch system of claim 1, wherein, The scheduling decision module is specifically configured to establish a mathematical model between load changes and reactive power demand and voltage fluctuations to determine the priority of reactive power regulation under different load scenarios, and generate at least one of an output instruction for the static var generator (SVG), a switching instruction for the capacitor bank, an adjustment instruction for the transformer tap, and a load control instruction based on the mathematical model and the multi-objective optimization algorithm.

4. The grid load and device reactive voltage dispatch system of claim 1, wherein, The execution control module comprises a reactive device control unit configured to control the power module output of the static var generator (SVG), the switching action of the capacitor bank / reactor, and the electric adjustment of the transformer tap by using a programmable logic controller (PLC) or an intelligent control unit, and a load control unit configured to send peak-shaving or peak-avoiding instructions to intelligent control terminals of interruptible load users, or remotely control the switching off and switching on of user loads.

5. The grid load and device reactive voltage dispatch system of claim 1, wherein, The communication module adopts a hybrid communication mode of wire and wireless, and the modules within the substation adopt optical fiber communication with a transmission rate not less than 1000 Mbps; the monitoring terminals at the power distribution lines and user sides and the data acquisition module adopt 5G / 4G wireless communication or a power wireless private network, and the communication delay is not greater than 100 milliseconds.

6. The grid load and device reactive voltage dispatch system of claim 1, wherein, The total delay of the system from data acquisition to device execution adjustment is not greater than 200 milliseconds.

7. The grid load and device reactive voltage dispatch system of claim 1, wherein, ​ 8. The grid load and device reactive voltage dispatch system of claim 1, wherein, The data processing module is further configured with a data quality evaluation unit for calculating signal-to-noise ratio, completeness rate and proportion of abnormal data of the preprocessed data, and triggering the data acquisition module to reacquire or correct the data if the data quality does not satisfy a preset threshold.

9. The grid load and device reactive voltage dispatch system of claim 1, wherein, The power grid virtual simulation model constructed in the scheduling decision module is a dynamic model containing a topological structure, device parameters and a real-time running state, which is synchronously updated by receiving real-time data streams of the data processing module, and is used for simulating power grid voltage distribution and reactive power flow within 15 minutes in the future after execution of the scheduling strategy.

10. The grid load and device reactive voltage dispatch system of claim 1, wherein, The execution control module is further provided with a safety check unit for checking whether the received control instruction conforms to a device safety operation range and a power grid safety operation constraint before execution of the scheduling strategy, discarding the instruction and sending alarm information to the scheduling decision module if the checking fails.

Images