A multi-source data fusion processing method and system for a new power system

Abstract

This invention relates to the field of multi-source data fusion, and more particularly to a multi-source data fusion processing method and system for novel power systems. The method includes: acquiring electrical parameters and numerical weather forecast data of the power grid at renewable energy power plants; calculating grid instability based on the electrical parameters; calculating a grid sensitivity factor; calculating a power fluctuation potential index, where the power fluctuation potential index is positively correlated with the grid sensitivity factor and the current remaining adjustable power; the current remaining adjustable power is the difference between the rated installed capacity of the renewable energy power plant and its current actual active power; training a prediction model using historical electrical parameters and numerical weather forecast data; and using the trained prediction model to predict the power of the renewable energy power plant at future times. This invention improves the accuracy of power prediction for renewable energy power plants.

Description

Technical Field

[0001] This invention relates to the field of multi-source data fusion, and in particular to a multi-source data fusion processing method and system for new power systems. Background Technology

[0002] In the process of building a new power system with new energy sources as the mainstay, the grid connection ratio of renewable energy sources such as photovoltaic and wind power continues to rise. Their power generation exhibits significant intermittency and volatility, directly affected by meteorological factors, posing a severe challenge to the safe and stable operation of the power grid. Therefore, accurate ultra-short-term power forecasting for new energy power plants is crucial for real-time grid dispatching, frequency control, and stability maintenance. Existing technologies mostly employ gradient boosting decision tree models (such as XGBoost), which integrate phasor measurement unit data from new energy power plants with numerical weather prediction data, simply concatenating meteorological data and grid status data as input features for modeling.

[0003] Chinese patent document CN112069673B discloses a method for estimating surface PM2.5 concentration based on gradient boosting decision tree. The method includes: 1. Preprocessing ground-observed PM2.5 concentration data to obtain average PM2.5 concentration data for each station. 2. Processing remote sensing AOD product data to obtain AOD data with better control quality and expand the spatial coverage of AOD data. 3. Preprocessing meteorological and auxiliary data for unified processing. 4. Integrating the data to ensure consistency in both space and time. 5. Conducting exploratory analysis of the data to eliminate multivariate collinearity. 6. Constructing a PM2.5 concentration estimation model using the gradient boosting decision tree method.

[0004] However, when using the gradient boosting decision tree model for prediction, the feature construction simply involves splicing and statistically aggregating multi-source data, failing to reflect the physical coupling relationship between meteorological disturbances and the dynamic response of the power grid. It treats meteorological changes (cause) and power grid state (effect) as independent variables, ignoring the differences in power fluctuations caused by the same meteorological disturbance under different power grid states. This static decoupling process is ill-suited to reflecting the nonlinear dynamic characteristics of the system's transient processes, resulting in low prediction accuracy for the gradient boosting decision tree model. Summary of the Invention

[0005] To improve the prediction accuracy of gradient boosting decision tree models, this invention provides a multi-source data fusion processing method and system for new power systems.

[0006] In a first aspect, the present invention provides a multi-source data fusion processing method for novel power systems, employing the following technical solution:

[0007] A multi-source data fusion processing method for new power systems includes the following steps:

[0008] Acquire electrical parameters and numerical weather forecast data of the power grid for new energy power plants. The numerical weather forecast data includes the total solar irradiance at the current and future times. Calculate the grid instability based on the electrical parameters. The grid instability is positively correlated with the rate of change of frequency, the rate of deviation of voltage, and the deviation of power factor. Calculate the grid sensitivity factor. The grid sensitivity factor is positively correlated with the grid instability and with the rate of change of total solar irradiance at future times.

[0009] The power fluctuation potential index is calculated, which is positively correlated with the grid sensitivity factor and the current remaining adjustable power. The current remaining adjustable power is the difference between the rated installed capacity of the renewable energy power station and the current actual active power. The prediction model is trained using electrical parameters from historical data and numerical weather prediction data, and the trained prediction model is used to predict the power of the renewable energy power station at future times.

[0010] By calculating grid instability, grid sensitivity factor, and power fluctuation potential index, multi-source data were deeply integrated to quantify the impact of meteorological disturbances on power fluctuations under different grid conditions. By constructing key features that can reflect physical coupling relationships and using these features to train prediction models, the accuracy of future power prediction for new energy power plants was improved.

[0011] Preferably, the method for calculating grid instability includes: calculating the frequency difference between time t and the previous sampling time, and normalizing the frequency difference to obtain the relative rate of change of frequency; calculating the absolute value of the difference between the voltage phasor amplitude at time t and the rated voltage amplitude, and obtaining the voltage deviation rate by the ratio of the absolute value of the difference to the rated voltage amplitude; calculating the difference between the voltage phase angle and the current phase angle at time t, and obtaining the power factor deviation by the difference between the absolute value of the cosine of the difference and 1; and taking the sum of the relative rate of change of frequency, the voltage deviation rate, and the power factor deviation as the grid instability.

[0012] By combining the relative rate of change of frequency, which reflects the dynamic characteristics of the power grid, the voltage deviation rate, which reflects the static characteristics, and the power factor deviation, which reflects the reactive power support capability, the real-time operating status of the power grid can be comprehensively and quantitatively evaluated in multiple dimensions, making the calculated power grid instability indicators more accurate and reliable.

[0013] Preferably, the expression for power grid instability is:

[0014]

[0015] In the formula, It refers to the instability of the power grid; and They are The power grid frequency at the current time and the previous sampling time; It is a phasor measurement unit with a fixed sampling interval; It is the rated frequency of the power grid; It is a time constant; yes Voltage phasor amplitude at time; It is the rated voltage amplitude; and These are the voltage phase angle and the current phase angle, respectively. It is a cosine function.

[0016] The above formula can be used to calculate power grid instability, enhancing the practicality and operability of the technical solution.

[0017] Preferably, the expression for the power grid sensitivity factor is:

[0018]

[0019] In the formula, It is the power grid sensitivity factor; and They are the future Time and Total solar irradiance at any given time; It is the total solar irradiance under standard test conditions; It refers to the instability of the power grid.

[0020] The study couples the intensity of external meteorological disturbances, i.e., the rate of change of total solar irradiance, with the internal vulnerability of the power grid, i.e., the grid instability. This dynamically reflects how a meteorological disturbance of the same magnitude will trigger more severe power fluctuations when the grid itself is more unstable, thus accurately quantifying the grid's sensitivity to meteorological changes.

[0021] Preferably, the calculation method of the power fluctuation potential index includes: multiplying the grid sensitivity factor by the current remaining adjustable power to obtain the power fluctuation potential index.

[0022] Preferably, the method for obtaining the electrical parameters of the power grid of the new energy power station is as follows: the power grid electrical parameters, including voltage phasor amplitude, current phasor amplitude, voltage phase angle, current phase angle and power grid frequency, are obtained by using a phasor measurement unit installed at the grid inlet of the new energy power station.

[0023] By installing a phasor measurement unit at the grid connection port, accurate phasor data such as voltage, current, and frequency can be acquired at high frequency and synchronously, providing an accurate data source for calculating key indicators such as grid instability.

[0024] Preferably, the current actual active power is calculated by multiplying the voltage phasor amplitude, current phasor amplitude, and the cosine of the difference between the voltage phase angle and the current phase angle at time t to obtain the current actual active power.

[0025] By utilizing the voltage, current amplitude, and phase angle data acquired in real time by the phasor measurement unit, the accuracy and real-time performance of the current power value, which serves as the basis for calculating the remaining adjustable power, are ensured.

[0026] Preferably, the numerical weather prediction data includes the total solar irradiance at the current time and the total solar irradiance at the future predicted time.

[0027] This ensures the amount of meteorological variation required to calculate the power grid sensitivity factor, thus providing the necessary data input for assessing the intensity of future meteorological disturbances.

[0028] Preferably, the prediction model is a gradient boosting decision tree model.

[0029] Gradient boosting decision tree models can better learn and fit the nonlinear mapping relationship between the constructed enhancement features and future power generation, thereby making full use of the advantages of these deeply fused features and further improving the accuracy and generalization ability of the prediction model.

[0030] Secondly, this invention provides a multi-source data fusion processing system for a new type of power system, employing the following technical solution:

[0031] A multi-source data fusion processing system for a new type of power system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the multi-source data fusion processing method for a new type of power system as described above.

[0032] The aforementioned method for multi-source data fusion processing for new power systems is used to generate a computer program, which is then stored in a memory for loading and execution by a processor. This allows for the creation of a system based on the memory and processor, making it convenient to use.

[0033] The present invention has the following technical effects:

[0034] By constructing three physically meaningful intermediate features—grid instability, grid sensitivity factor, and power fluctuation potential index—the coupling relationship between meteorological disturbances and grid dynamic response was quantified. These deeply fused enhanced features were used as key inputs to train the prediction model, thereby improving the accuracy of power prediction for renewable energy power plants. Attached Figure Description

[0035] Figure 1This is a flowchart of a multi-source data fusion processing method for a new type of power system according to the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] This invention discloses a multi-source data fusion processing method for novel power systems, referring to... Figure 1 This includes the following steps:

[0038] S1: Acquire multi-source data and perform preprocessing.

[0039] Multi-source data is acquired, including total solar irradiance and future forecasts from numerical weather prediction (NMR) to characterize meteorological disturbances; and voltage phasor amplitudes, current phasor amplitudes, and grid frequency collected at fixed intervals by the phasor measurement unit (PMU) to assess the real-time dynamic response capability of the power grid. Since the frequency of NMR data is much lower than that of PMU data, direct fusion would lead to time mismatch. Therefore, cubic spline interpolation is used to up-frequency the NMR data, achieving accurate alignment of multi-source data under a unified time reference, thus completing data preprocessing.

[0040] Specifically, it acquires high-frequency phasor measurement unit (PMU) data and minute-level numerical weather prediction (NWP) data from the grid connection interface of new energy power plants.

[0041] Phasor measurement unit data: obtained through phasor measurement unit equipment installed at the grid connection interface of the ground-mounted photovoltaic power station, with a fixed sampling interval. (e.g., every 20 milliseconds) Continuously and synchronously collect power grid electrical parameters, including: Voltage phasor amplitude at time 1 , Current phasor amplitude at time 1 and Power grid frequency at time .

[0042] Numerical weather forecast data: Numerical weather forecast data is obtained through numerical weather forecast products. This data provides real-time meteorological conditions and weather forecasts for a future period of time for ground-mounted photovoltaic power stations. The numerical weather forecast data used includes: Total solar irradiance at time Relative to the current time The Future Predicted total solar irradiance at any time For example, the total solar irradiance in the next 10 or 15 minutes.

[0043] For low-frequency numerical weather forecast data, cubic spline interpolation is used for up-frequency processing to ensure accurate alignment with phasor measurement unit data in terms of timestamps, forming a multi-dimensional input time series under a unified time reference, thus completing data preprocessing.

[0044] S2: Calculate the power fluctuation potential index.

[0045] a. Calculate the instability of the power grid.

[0046] Calculate the power grid frequency acquired by the phasor measurement unit. The frequency change rate between the current sampling time and the previous sampling time is normalized to obtain the relative frequency change rate. Voltage phasor amplitude, voltage phase angle, and current phase angle are collected by the phasor measurement unit to calculate the voltage deviation rate and power factor deviation. Finally, these three factors are combined to calculate the grid instability. The calculation expression is as follows:

[0047]

[0048] In the formula, It refers to the instability of the power grid; and They are The power grid frequency at the current time and the previous sampling time; It is a phasor measurement unit with a fixed sampling interval; It is the rated frequency of the power grid; It is a time constant, used to eliminate dimensions, and its value is... ; yes Voltage phasor amplitude at time; It is the rated voltage amplitude; and These are the voltage phase angle and the current phase angle, respectively, which are directly and synchronously acquired by the phasor measurement unit, and represent... The instantaneous phase angle of the voltage phasor and current phasor relative to the system reference phase at any given moment; It is a cosine function; It is the absolute value sign.

[0049] This is the relative rate of change of frequency per unit time. The larger this value, the more severe the system frequency fluctuations, the higher the dynamic strain pressure, the weaker the power grid's immunity to disturbances, and the higher the power grid instability. The higher the value, the lower the value; conversely, when the frequency is stable, this term approaches 0, indicating that the system is in a steady state and the grid instability is low. The lower.

[0050] This is the voltage deviation rate. The larger the value, the worse the static voltage stability, and it may be approaching the voltage instability boundary, indicating grid instability. The higher the value, the lower the voltage phasor amplitude; conversely, the lower the value, the smaller the value when the voltage phasor amplitude is close to the rated voltage, indicating that the system is operating well and the grid instability is low. The lower.

[0051] It is the power factor deviation. The smaller the absolute value of the cosine of the difference between the voltage phase angle and the current phase angle, the larger this value is, indicating that the proportion of reactive current used to maintain voltage in the power grid is higher and the system's anti-interference ability is weaker; conversely, the smaller this value is, the higher the system efficiency and the stronger the stability.

[0052] For example, the rated frequency of the power grid is obtained through experimental calibration. Therefore Sampling interval ; , Rated voltage Voltage phasor amplitude Voltage phase angle Current phase angle The instability of the power grid is then:

[0053]

[0054] b. Calculate the power grid sensitivity factor.

[0055] Using numerical weather prediction data, we can obtain future... Time and The change in total solar irradiance at time t, normalized to standard test conditions. The basic meteorological disturbance intensity is obtained; combined with the power grid instability. Calculate the power grid sensitivity factor The expression is as follows:

[0056]

[0057] In the formula, It is the power grid sensitivity factor; and They are the future Time and Total solar irradiance at any given time; This is the total solar irradiance under standard test conditions. According to international standards in the photovoltaic industry, its value is... ; It refers to the instability of the power grid.

[0058] This indicates the intensity of basic meteorological disturbances. The larger the value, the more drastic the changes in solar radiation during the forecast period, and the stronger the external meteorological disturbances. (Grid sensitivity factor) The larger the value, the weaker the meteorological disturbance; conversely, the closer the value is to 0, the weaker the power grid sensitivity factor. The smaller.

[0059] It is a dynamic amplification factor used to modulate the intensity of basic meteorological disturbances, when the power grid instability... The higher the value, the more fragile the system; the same meteorological disturbance will trigger more severe power fluctuations, therefore the disturbance intensity needs to be amplified. As the value approaches 0, the dynamic amplification factor approaches 1, and the grid sensitivity factor... It is approaching the intensity of the basic meteorological disturbance.

[0060] For example, it is obtained through experimental calibration. Total solar irradiance at time Numerical weather forecast of total solar irradiance 10 minutes later ; Then the power grid sensitivity factor .

[0061] c. Calculate the power fluctuation potential index based on the grid sensitivity factor.

[0062] Since the actual impact of meteorological disturbances on photovoltaic power generation is also limited by the current remaining regulation capacity of the power plant, using only the grid sensitivity factor cannot reflect the key constraint of whether there is regulation space. Because the lower the actual active power, the larger the remaining adjustable power, and the higher the potential for power change under the same disturbance, the remaining adjustable power is obtained by calculating the current actual active power output based on voltage, current amplitude, and phase angle, and subtracting it from the rated installed capacity, thus characterizing the remaining regulation space of the photovoltaic power plant. Furthermore, by combining the grid sensitivity factor, a power fluctuation potential index is constructed, achieving an accurate characterization of future power fluctuation potential.

[0063] Voltage amplitude acquired by phasor measurement unit Current amplitude And the voltage phase angle difference and the current phase angle difference, calculate The actual active power output of the ground-mounted photovoltaic power station at all times, combined with the rated installed capacity of the ground-mounted photovoltaic power station. The remaining adjustable power is obtained. This is further combined with the grid sensitivity factor. Calculate the power fluctuation potential index The calculation expression is as follows:

[0064]

[0065] In the formula, It is a power fluctuation potential index; It is the power grid sensitivity factor; This is the rated installed capacity of the ground-mounted photovoltaic power station; , They are The magnitudes of the voltage phasor and current phasor at any given moment; and It is the phase angle between voltage and current.

[0066] yes The actual active power output of the ground-mounted photovoltaic power station at any given time is directly calculated from the real-time measurement data of the phasor measurement unit.

[0067] This refers to the remaining adjustable power. The larger this value, the lower the current actual power generation level of the ground-mounted photovoltaic power station, and the greater the adjustment space available to respond to weather changes. This indicates a potential power fluctuation index. The larger the value, the more limited the remaining adjustment space; conversely, the smaller the value, the greater the power fluctuation potential index. The smaller.

[0068] Power grid sensitivity factor The larger the value, the more amplified the same meteorological disturbance is under the current vulnerable state of the power grid, and the stronger its driving effect on power fluctuations. This indicates a higher power fluctuation potential index. The larger; conversely, The smaller the value, the weaker the meteorological disturbance and the weaker its impact on power fluctuations; the lower the power fluctuation potential index. The smaller.

[0069] For example, the current amplitude is obtained through experimental calibration. Preset rated installed capacity of photovoltaic power station The power fluctuation potential index is:

[0070] .

[0071] This result indicates that the current power grid is slightly unstable. When the forecast period is 10 minutes, the potential fluctuation in photovoltaic power generation could reach approximately [missing information]. Kw.

[0072] S3: Use an enhanced training set to train the gradient boosting decision tree model to predict photovoltaic power generation.

[0073] Traditional prediction models rely solely on historical, conventional feature data, making it difficult to capture the dynamic modulation effect of grid conditions on photovoltaic (PV) responses. The power fluctuation potential index, which characterizes the variable potential of future power, is introduced as a key enhancement feature into the input features, while conventional feature data is retained. Furthermore, a gradient boosting decision tree model is used to learn the nonlinear mapping relationship between this feature set and actual PV power generation, achieving adaptive, high-precision prediction of meteorological disturbance responses under different grid operating conditions.

[0074] Construct training samples for each moment in the historical data. Obtain input features With target value Among them, input features The multi-source data to be acquired includes: Voltage phasor amplitude at time 1 , Current phasor amplitude at time 1 , Power grid frequency at time , Total solar irradiance at time Relative to the current time The Future Predicted total solar irradiance at any time and power fluctuation potential index Target value For the future The actual photovoltaic power generation at any given time.

[0075] Furthermore, historical phasor measurement unit and numerical weather prediction data are collected for each time point. Generate paired training samples The augmented training set composed of these samples is then fed into a gradient boosting decision tree model (such as XGBoost) for training. The gradient boosting decision tree model will learn from the input features. To the target value The nonlinear mapping relationship.

[0076] At the present moment Obtain input features and calculate the power fluctuation potential index. Together they constitute the input features Input features By inputting the data into a trained gradient boosting decision tree model, the future can be predicted. Photovoltaic power generation at any given time.

[0077] This invention also discloses a multi-source data fusion processing system for a new type of power system, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a multi-source data fusion processing method for a new type of power system according to the present invention is implemented.

[0078] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0079] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-source data fusion processing method for a new power system, characterized in that, Including the following steps: Acquire electrical parameters and numerical weather forecast data of the power grid for new energy power plants. The numerical weather forecast data includes the total solar irradiance at the current and future times. Calculate the grid instability based on the electrical parameters. The grid instability is positively correlated with the rate of frequency change, voltage deviation rate, and power factor deviation. The power grid sensitivity factor is calculated. It is positively correlated with the power grid instability and with the rate of change of total solar irradiance at future times. The power fluctuation potential index is calculated, and it is positively correlated with the grid sensitivity factor and the current remaining adjustable power. The current remaining adjustable power is the difference between the rated installed capacity of the new energy power station and the current actual active power. The prediction model is trained using electrical parameters from historical data and numerical weather prediction data, and the trained prediction model is used to predict the power of the new energy power station at future times. The power fluctuation potential index can characterize the variable potential of future power. It is introduced as a key enhancement feature into the input features, while retaining electrical parameters and numerical weather forecast data to form the input features of the prediction model.

2. The multi-source data fusion processing method for new power systems according to claim 1, characterized in that, The calculation method for grid instability includes: calculating the frequency difference between time t and the previous sampling time, and normalizing the frequency difference to obtain the relative rate of change of frequency; calculating the absolute value of the difference between the voltage phasor amplitude at time t and the rated voltage amplitude, and obtaining the voltage deviation rate by the ratio of the absolute value of the difference to the rated voltage amplitude; calculating the difference between the voltage phase angle and the current phase angle at time t, and obtaining the power factor deviation by the difference between the absolute value of the cosine of the difference and 1; and taking the sum of the relative rate of change of frequency, the voltage deviation rate, and the power factor deviation as the grid instability.

3. The multi-source data fusion processing method for new power systems according to claim 1, characterized in that, The expression for power grid instability is: In the formula, It refers to the instability of the power grid; and They are The power grid frequency at the current time and the previous sampling time; It is a phasor measurement unit with a fixed sampling interval; It is the rated frequency of the power grid; It is a time constant; yes Voltage phasor amplitude at time; It is the rated voltage amplitude; and These are the voltage phase angle and the current phase angle, respectively. It is a cosine function.

4. The multi-source data fusion processing method for a new type of power system according to claim 1, characterized in that, The expression for the power grid sensitivity factor is: In the formula, It is the power grid sensitivity factor; and They are the future Time and Total solar irradiance at any given time; It is the total solar irradiance under standard test conditions; It refers to the instability of the power grid.

5. The multi-source data fusion processing method for a new type of power system according to claim 1, characterized in that, The calculation method of the power fluctuation potential index includes multiplying the grid sensitivity factor by the current remaining adjustable power to obtain the power fluctuation potential index.

6. The multi-source data fusion processing method for a new type of power system according to claim 1, characterized in that, The method for obtaining the electrical parameters of the power grid of new energy power plants is as follows: the electrical parameters of the power grid, including voltage phasor amplitude, current phasor amplitude, voltage phase angle, current phase angle and power grid frequency, are obtained by using the phasor measurement unit installed at the grid inlet of the new energy power plant.

7. The multi-source data fusion processing method for a new type of power system according to claim 5, characterized in that, The current method for calculating actual active power includes multiplying the voltage phasor amplitude, current phasor amplitude, and the cosine of the difference between the voltage phase angle and the current phase angle at time t to obtain the current actual active power.

8. The multi-source data fusion processing method for a new type of power system according to claim 1, characterized in that, Numerical weather forecast data includes the total solar irradiance at the current time and the total solar irradiance at the predicted future time.

9. The multi-source data fusion processing method for a new type of power system according to claim 1, characterized in that, The prediction model is a gradient boosting decision tree model.

10. A multi-source data fusion processing system for a new type of power system, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a multi-source data fusion processing method for a novel power system according to any one of claims 1-9.

Images