Virtual power plant cooperative scheduling test method

Abstract

The invention discloses a virtual power plant collaborative scheduling test method, which comprises the following steps: step 1, constructing a simulation test platform which comprises a layered fusion test environment of a physical simulation layer, a digital twinborn layer and a market simulation layer; 2, designing a multi-dimensional test scene based on a hierarchical fusion test environment, and covering extreme and typical working condition scenes; step 3, acquiring resource parameters of the physical simulation layer, market environment parameters of the market simulation layer and power grid constraint parameters, deploying a collaborative scheduling algorithm in the digital twin layer, driving the physical simulation layer and the market simulation layer to cooperatively operate by the simulation test platform according to a scene sequence of the multi-dimensional test scene, and acquiring operation data; and step 4, carrying out quantitative evaluation on the operation data according to the technical advancement dimension and the engineering grounding dimension. The comprehensive quantitative evaluation is realized through the technology advancement dimension and the engineering feasibility dimension, and the core problems that the technology and the market direction of the existing test method are separated and the evaluation dimension is single are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of smart energy and power system simulation testing technology, and in particular to a virtual power plant collaborative scheduling testing method. Background Technology

[0002] Virtual power plants are a key carrier for building new power systems. Through advanced information and communication technologies and software systems, they can aggregate and coordinate the optimization of diversified resources such as distributed power sources, energy storage systems, and controllable loads, and participate in grid operation and electricity market transactions. This is of great significance for improving grid flexibility and promoting the consumption of new energy sources.

[0003] Currently, the industry has formed two core technological directions for the development of virtual power plants: one is the technology-driven "integrated source-grid-load-storage" direction, which focuses on achieving real-time, safe, and efficient coordination of various resources through technological means to ensure the stable operation of the power grid; the other is the market-driven "market-oriented operation" direction, which focuses on incentivizing resource entities to participate through market mechanisms to achieve overall profitability and sustainable commercial development of virtual power plants.

[0004] However, existing virtual power plant testing methods have significant limitations: Most testing platforms either focus solely on the technical coordination simulation of power generation, grid, load, and storage, neglecting the verification of market trading strategies; or they focus only on market return simulation, lacking a detailed characterization of the interaction between the underlying equipment's physical constraints and grid security. This failure to effectively integrate the two leads to one-sided test results, unable to meet the actual needs of a dual-engine approach to virtual power plant technology and business.

[0005] Some testing methods pursue the advancement of theoretical algorithms, but the testing environment is idealized and does not take into account issues such as device response delay, communication network jitter, hardware cost constraints, and compatibility of heterogeneous devices in actual engineering. As a result, the test conclusions have poor feasibility.

[0006] Existing evaluation systems mostly focus on technical performance indicators or single economic indicators, lacking a comprehensive quantitative evaluation system that can simultaneously measure technological advancement and engineering feasibility, making it difficult to comprehensively guide the optimization of virtual power plant dispatch strategies and engineering selection. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of existing virtual power plant testing methods, such as fragmented technical directions, incomplete coverage of test scenarios, and lack of quantitative evaluation in both technical advancement and engineering feasibility. These shortcomings lead to one-sided test results that are difficult to guide practical applications. This invention provides a virtual power plant collaborative scheduling testing method. By constructing a layered integrated test environment that includes a physical simulation layer, a digital twin layer, and a market simulation layer, and designing multi-dimensional test scenarios covering extreme and typical operating conditions, the invention deploys collaborative scheduling algorithms and collects operational data. It conducts quantitative evaluation from both technical advancement and engineering feasibility dimensions, achieving the goal of integrated testing of source-grid-load-storage and market-oriented operation scheduling, and comprehensively and objectively verifying the collaborative scheduling strategy of virtual power plants.

[0008] The objective of this invention is achieved through the following technical solution: The virtual power plant collaborative scheduling test method includes the following steps: Step 1: Construct a simulation testing platform, which includes a layered and integrated testing environment comprising a physical simulation layer, a digital twin layer, and a market simulation layer. Step 2: Based on the layered fusion test environment, design multi-dimensional test scenarios to cover extreme and typical working conditions; Step 3: Obtain the resource parameters of the physical simulation layer, the market environment parameters of the market simulation layer, and the power grid constraint parameters. Deploy the collaborative scheduling algorithm in the digital twin layer. The simulation test platform drives the physical simulation layer and the market simulation layer to run collaboratively according to the scenario sequence of multi-dimensional test scenarios and collects the running data. Step 4: Quantitatively evaluate the operational data according to the dimensions of technological advancement and engineering feasibility.

[0009] Preferably, the physical simulation layer includes typical virtual power plant equipment and distribution network topology. The typical virtual power plant equipment includes a photovoltaic / wind power inverter simulator, an energy storage converter, a programmable load simulation device, and a smart gateway. The simulation test platform is equipped with sensors and a fast control unit, all of which are connected to the typical virtual power plant equipment.

[0010] Preferably, the digital twin layer integrates a multi-timescale resource prediction model, a power grid flow calculation model, and an equipment control model. The digital twin layer is based on an edge-cloud collaborative computing architecture and interacts with the physical simulation layer in real time through a communication network.

[0011] Preferably, the market simulation layer is equipped with an electricity market simulation engine to simulate the day-ahead market, intraday market, real-time balancing market, as well as frequency regulation and standby ancillary service markets.

[0012] Preferably, the multi-dimensional testing scenarios include integrated source-grid-load-storage collaborative scenarios and market-oriented operation and scheduling scenarios. The integrated source-grid-load-storage collaborative scenarios include high volatility scenarios, fault ride-through and recovery scenarios, power quality management scenarios, and multi-regional collaborative scenarios. The market-oriented operation and scheduling scenarios include multi-market joint clearing scenarios, price-driven demand response scenarios, incentive-driven demand response scenarios, and extreme market volatility scenarios.

[0013] Preferably, in step 3, the process of collecting operational data specifically involves recording the entire process data during continuous testing over a set time period. The entire process data includes dispatch instructions, actual equipment output, grid node voltage / frequency, market transaction list, and revenue settlement.

[0014] As a preferred approach, the virtual power plant collaborative scheduling test method also preprocesses the collected operating data, uses wavelet transform algorithm to filter out noise, automatically identifies and removes distorted data caused by instantaneous communication interruptions or equipment malfunctions based on the isolated forest algorithm, and combines data completion algorithm to repair missing data.

[0015] Preferably, in step 4, a dynamic weight allocation mechanism is used to allocate the weights of the technological advancement dimension and the engineering feasibility dimension.

[0016] The beneficial effects of this invention are as follows: This invention achieves comprehensive quantitative evaluation through both technological advancement and engineering feasibility dimensions, effectively solving the core problems of existing testing methods being disconnected from technology and market orientation, and having a single evaluation dimension. By constructing a three-layer integrated testing environment encompassing physical, digital, and market aspects, and combining the multi-dimensional scenarios of integrated source-grid-load-storage and market-oriented operation, quantitative evaluation is conducted from both technological advancement and engineering feasibility dimensions. This allows the test results to simultaneously cover technological synergy effectiveness and commercialization potential, providing a more comprehensive decision-making basis for the research, verification, and selection of scheduling strategies.

[0017] Relying on a hardware-in-the-loop simulation platform, sensors and rapid control units, as well as a data preprocessing process for wavelet transform noise reduction and isolated forest outlier removal, the system accurately simulates the physical characteristics of equipment, fault states and market fluctuations. Data acquisition ensures that the testing process fits the actual engineering scenario, greatly improving the credibility and feasibility of the test conclusions.

[0018] This invention offers comprehensive and forward-looking scenario coverage, adapting to the diverse application needs of virtual power plants. The test scenarios encompass extreme conditions such as high volatility and extreme market fluctuations, as well as typical technical collaboration and market-oriented operation scenarios. This not only verifies the adaptability of existing dispatching strategies, but its scenario design and evaluation system can also guide the development of high-quality dispatching solutions that balance technical performance, economic efficiency, and compatibility, accelerating the large-scale application of virtual power plant technology. Attached Figure Description

[0019] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0022] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0023] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0024] Example: Virtual power plant collaborative scheduling testing methods, such as Figure 1 As shown, it includes the following steps: Step 1: Construct a simulation testing platform, which includes a layered and integrated testing environment comprising a physical simulation layer, a digital twin layer, and a market simulation layer. Step 2: Based on the layered fusion test environment, design multi-dimensional test scenarios to cover extreme and typical working conditions; Step 3: Obtain the resource parameters of the physical simulation layer, the market environment parameters of the market simulation layer, and the power grid constraint parameters. Deploy the collaborative scheduling algorithm in the digital twin layer. The simulation test platform drives the physical simulation layer and the market simulation layer to run collaboratively according to the scenario sequence of multi-dimensional test scenarios and collects the running data. Step 4: Quantitatively evaluate the operational data according to the dimensions of technological advancement and engineering feasibility.

[0025] The physical simulation layer includes typical virtual power plant equipment and distribution network topology. The typical virtual power plant equipment includes photovoltaic / wind power inverter simulators, energy storage converters, programmable load simulation devices, and smart gateways. The simulation test platform is equipped with sensors and fast control units, all of which are connected to the typical virtual power plant equipment.

[0026] The digital twin layer integrates multi-timescale resource prediction models, power grid flow calculation models, and equipment control models. Based on an edge-cloud collaborative computing architecture, it interacts with the physical simulation layer in real time via a communication network. This integrated digital twin layer possesses 3D visualization capabilities for operational status and historical operating condition backtracking, enabling transparent monitoring, precise traceability, and in-depth analysis of the testing process.

[0027] The market simulation layer includes a power market simulation engine that simulates day-ahead, intraday, real-time balancing, frequency regulation, and ancillary service markets. It supports custom market clearing algorithms, price fluctuation models, and demand response incentive mechanisms, and includes a new market risk simulation module, providing a highly realistic game environment for virtual power plant market-based dispatch strategies.

[0028] The multi-dimensional testing scenarios include integrated source-grid-load-storage collaborative scenarios and market-oriented operation and dispatch scenarios. The integrated source-grid-load-storage collaborative scenarios include high volatility scenarios (PV power drop of ±50% / h, wind power ramp-up rate of ±30% / h), fault ride-through and recovery scenarios (voltage dips, line short circuits, equipment offline faults), power quality management scenarios (voltage harmonic exceedances, three-phase imbalance adjustment), and multi-regional collaborative scenarios (joint dispatch of virtual power plants across distribution networks). The market-oriented operation and dispatch scenarios include multi-market joint clearing scenarios (coordinated bidding between the energy market and multiple ancillary service markets), price-driven demand response scenarios (dynamic response to time-of-use / real-time / peak electricity prices), incentive-driven demand response scenarios (coordinated response and revenue distribution among multiple stakeholders), and extreme market volatility scenarios (sudden price changes, temporary adjustments to market rules).

[0029] Before collecting operational data, the simulation test platform is initialized and three types of core parameters are entered: resource parameters (equipment capacity, efficiency, ramp rate, cost), grid constraint parameters (topology, line capacity, safe operation upper and lower limits), and market environment parameters (electricity price model, rule parameters, competitor behavior model). Batch import, parameter template storage, and one-click switching are supported. The process of collecting operational data is as follows: during no less than 168 hours of continuous testing, the entire process data is recorded at a millisecond frequency. The entire process data includes dispatch instructions, actual equipment output, grid node voltage / frequency, market transaction list, and revenue settlement.

[0030] The virtual power plant collaborative scheduling test method also preprocesses the collected operation data, uses wavelet transform algorithm to filter out noise, automatically identifies and removes distorted data caused by instantaneous communication interruptions or equipment abnormalities based on the isolated forest algorithm, and combines data completion algorithm to repair missing data.

[0031] Specifically, in this embodiment, the data denoising process employs a wavelet transform algorithm to process random noise in the test data. The noise in the test data mainly originates from electromagnetic interference from sensors in the physical simulation layer, signal fluctuations in the communication link, and minor disturbances during equipment operation. This type of noise exhibits irregular distribution characteristics, and if not filtered out, it can lead to false fluctuations in key data such as equipment output and grid voltage. In practice, the collected raw data is first decomposed using multi-scale wavelet decomposition, separating it into low-frequency approximate components (containing core effective information) and high-frequency detail components (mainly noise). Then, based on the different data types, an appropriate threshold function is selected to process the high-frequency detail components, suppressing the amplitude of the noise-corresponding components. Finally, through inverse wavelet transform, the processed low-frequency components and the optimized high-frequency components are reconstructed to obtain smoothed data after denoising, preserving the original true trend of change in the data while eliminating meaningless random disturbances.

[0032] Outlier removal employs the Isolation Forest algorithm, focusing on processing distorted data caused by momentary communication interruptions or equipment malfunctions. In long-term continuous testing, situations may arise such as data jumps caused by temporary communication link interruptions, or abnormal values ​​triggered by equipment failures (e.g., sudden output drops in photovoltaic inverters due to failure, or abnormal transaction prices caused by data transmission delays in the market simulation layer). These distorted data differ significantly from normal data, and retaining them would severely impact evaluation accuracy. In implementation, the pre-processed, denoised data is first divided into several data subsets according to time series, each subset corresponding to continuous operating data within a fixed time period. Multiple isolation trees are constructed based on these data subsets. By randomly selecting features and segmentation thresholds, data points are gradually isolated. Distorted data, due to its features deviating from the normal range, is isolated to leaf nodes more quickly. Reasonable anomaly judgment criteria are set, and the isolation results of each isolation tree are comprehensively voted on to identify discrete distorted data caused by momentary communication interruptions and continuous distorted data caused by equipment malfunctions. These are automatically removed from the dataset to ensure the authenticity of the remaining data.

[0033] The data completion phase combines the time-series characteristics and correlation features of the test data, employing an appropriate completion algorithm to repair missing data. Data loss mainly stems from scenarios such as brief device offline times and communication link switching delays, with durations ranging from milliseconds to minutes, and usually occurring alongside consecutive normal data. During implementation, a comprehensive detection of data loss is first conducted to identify the missing location, duration, and corresponding data type. For short-duration missing data, interpolation of adjacent data is used to complete the data based on the continuity of the time series, and the reasonable range of missing values ​​is calculated using the trend of normal data before and after the missing data. For long-duration missing data or data lacking key parameters, a data association model is constructed by combining the correlation of multi-dimensional data and integrating historical data characteristics from similar scenarios to generate complete data consistent with the overall data trend. After completion, consistency verification is performed to ensure that the completed data does not logically conflict with the preceding and following normal data, guaranteeing the integrity and coherence of the dataset.

[0034] Through the above preprocessing process, the test run data is purified and repaired, effectively solving problems such as noise interference, distortion and anomalies, and partial missing data. This forms a complete and reliable evaluation dataset, providing high-quality data support for the subsequent dual-dimensional quantitative evaluation of technological advancement and engineering feasibility, and ensuring that the evaluation results can objectively reflect the actual performance of the collaborative scheduling strategy.

[0035] In step 4, a dynamic weight allocation mechanism is used to allocate the weights of the technological advancement dimension and the engineering feasibility dimension.

[0036] The aforementioned dimensions of technological advancement include rapid response capability (average delay from command issuance to effective action of the first device ≤ 80ms), precise control capability (average absolute error percentage between planned output and actual output ≤ 2%), efficient absorption capability (actual utilization rate of new energy ≥ 98% during testing), system stability support capability (improvement of grid frequency / voltage deviation ≥ 30%), and multi-resource synergy efficiency (increase in comprehensive resource utilization efficiency ≥ 20%).

[0037] The feasibility of the project includes economic efficiency (comprehensive investment cost per kilowatt of dispatch capacity ≤ 95% of the industry average), reliability (successful issuance rate of control commands and successful return rate of status data within the test cycle ≥ 99.95%), ease of use and maintainability (system deployment time reduced by ≥ 40%, average fault repair time ≤ 30 min), compatibility (support for access of ≥ 10 types of heterogeneous devices and adaptation to mainstream communication protocols), and policy and standard compliance (compliance with current power grid dispatching regulations and communication protocols such as IEC61850 ≥ 90%).

[0038] Finally, the system automatically generates a structured comprehensive evaluation report, which includes indicator scores, radar comparison charts, typical scenario operation curves, abnormal operating condition analysis, and targeted optimization suggestions, such as optimizing energy storage charging and discharging strategies to improve arbitrage profits in the spot market and enhancing communication redundancy design to improve reliability under extreme operating conditions.

[0039] The following are supplementary and optimized technical features of the technical solution in this embodiment: (I) Supplementary details on test environment setup Physical simulation layer: The main controller is built using RT-LAB and dSPACE real-time simulation systems and is connected in the loop with actual low-power photovoltaic inverters, energy storage BMS, load boxes and other hardware. It supports full-process testing of equipment controllers and adds a new equipment fault simulation module.

[0040] Digital Twin Layer: Developed using Python + Unity3D, it supports the integration of different vendor models according to the FMI standard, enhancing model interoperability and reusability; the data protocol supports industry standards such as MQTT, IEC104, and 61850, and a new local computing redundancy design for edge nodes is added to avoid test interruptions caused by cloud failures.

[0041] Market Simulation Layer: Employs agent-based modeling to simulate the behavior of diverse market participants. Adds a market rule configuration module to support rapid adaptation to different regional market operation models. Includes 5 built-in typical market scenario templates for direct use.

[0042] (II) Supplementary Specific Parameters for Test Scenarios Integrated source-grid-load-storage scenario: In scenarios with a high proportion of new energy access, the energy storage charging and discharging efficiency is ≥95%, and the load peak-valley difference is ≥4:1; In fault scenarios, the fault duration of the photovoltaic inverter can be customized (1-10min), and the short-circuit fault clearing time of the distribution network line is ≤2s.

[0043] Market-based dispatch scenario: In the spot market bidding scenario, the daily electricity price fluctuation range is 0.2-1.5 yuan / kWh, and the virtual power plant's declared capacity can be dynamically adjusted (0.5-5MW); in the demand response scenario, the grid triggers response during peak hours, and the incentive price is 0.3-1.0 yuan / kWh, which can be set in stages.

[0044] (III) Supplementary Calculation of Evaluation Indicators All indicators are calculated based on preprocessed clean data, and the weighted summation method is used to calculate the overall score.

[0045] The weight allocation employs the Analytic Hierarchy Process (AHP) combined with entropy weighting, balancing subjective needs with objective data characteristics and supporting user-defined weight coefficients. First, a three-tiered structure of "comprehensive evaluation, criteria layer, and indicator layer" is constructed. A review panel composed of technical experts, engineering and maintenance personnel, and marketing operations staff compares the importance of the criteria layer and each indicator pairwise, constructing a judgment matrix. After consistency verification, subjective weights are calculated to adapt to the needs of different application scenarios. Second, based on preprocessed test data, the information entropy of each evaluation indicator is calculated. Objective weights are determined according to the entropy value; a smaller entropy value indicates higher data dispersion and stronger discrimination, corresponding to a larger indicator weight, reflecting the objective characteristics of the data itself. Finally, a weighted summation method is used to merge subjective and objective weights to generate an initial comprehensive weight. A custom interface is also provided, allowing users to directly adjust the weight coefficients of the criteria layer or indicator layer according to actual testing objectives, flexibly adapting to different testing scenario requirements.

[0046] Operational complexity is broken down into three directly quantifiable metrics: Mean Time To Repair (MTTR), system availability, and deployment time.

[0047] (iv) Supplementing the comprehensive evaluation report The report now includes sections on typical fault condition analysis and comparison of the effects before and after strategy optimization. It supports exporting to multiple file formats such as PDF and Excel, and features a built-in report template customization function to adapt to the needs of different application scenarios.

[0048] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0049] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A virtual power plant collaborative scheduling test method, characterized by: Includes the following steps: Step 1: Construct a simulation testing platform, which includes a layered and integrated testing environment comprising a physical simulation layer, a digital twin layer, and a market simulation layer. Step 2: Based on the layered fusion test environment, design multi-dimensional test scenarios to cover extreme and typical working conditions; Step 3: Obtain the resource parameters of the physical simulation layer, the market environment parameters of the market simulation layer, and the power grid constraint parameters. Deploy the collaborative scheduling algorithm in the digital twin layer. The simulation test platform drives the physical simulation layer and the market simulation layer to run collaboratively according to the scenario sequence of multi-dimensional test scenarios and collects the running data. Step 4: Quantitatively evaluate the operational data according to the dimensions of technological advancement and engineering feasibility.

2. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, The physical simulation layer includes typical virtual power plant equipment and distribution network topology. The typical virtual power plant equipment includes photovoltaic / wind power inverter simulators, energy storage converters, programmable load simulation devices, and smart gateways. The simulation test platform is equipped with sensors and fast control units, all of which are connected to the typical virtual power plant equipment.

3. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, The digital twin layer integrates a multi-timescale resource prediction model, a power grid flow calculation model, and an equipment control model. Based on an edge-cloud collaborative computing architecture, the digital twin layer interacts with the physical simulation layer in real time through a communication network.

4. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, The market simulation layer is equipped with an electricity market simulation engine to simulate the day-ahead market, intraday market, real-time balancing market, as well as frequency regulation and standby ancillary service markets.

5. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, The multi-dimensional testing scenarios include integrated source-grid-load-storage collaborative scenarios and market-oriented operation and scheduling scenarios. The integrated source-grid-load-storage collaborative scenarios include high volatility scenarios, fault ride-through and recovery scenarios, power quality management scenarios, and multi-regional collaborative scenarios. The market-oriented operation and scheduling scenarios include multi-market joint clearing scenarios, price-driven demand response scenarios, incentive-driven demand response scenarios, and extreme market volatility scenarios.

6. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, In step 3, the process of collecting operational data specifically involves recording all data during continuous testing over a set time period. This data includes dispatch instructions, actual equipment output, grid node voltage / frequency, market transaction list, and revenue settlement.

7. The virtual power plant collaborative scheduling test method according to claim 1 or 6, characterized in that, The collected operational data is also preprocessed, with wavelet transform algorithm used to filter out noise, isolated forest algorithm used to automatically identify and remove distorted data caused by instantaneous communication interruptions or equipment malfunctions, and data completion algorithm used to repair missing data.

8. The virtual power plant collaborative scheduling test method according to claim 1, characterized in that, In step 4, a dynamic weight allocation mechanism is used to allocate the weights of the technological advancement dimension and the engineering feasibility dimension.

Images