A control method for a PCS energy storage system

Abstract

This invention discloses a control method for a PCS energy storage system. In the field of energy storage control technology, it solves the technical problem of the inability to predict power fluctuation trends in advance, which easily leads to a surge in transient losses. This invention collects historical operating data and performs cleaning, normalization, and feature engineering processing. Combined with the characteristics of the scenario, a time series model is selected, enabling prediction of power change trends at the minute to hour level in advance. Based on the prediction results, the adjustment amount and timing are quantified, reducing the adjustment delay from the current second level to the millisecond level, thus proactively addressing the surge in losses caused by power fluctuations. Based on the core information extracted from the prediction results, precise matching of adjustment methods such as PCS energy storage, active power filters, gas turbines, and on-load tap changers of transformers is achieved, avoiding resource waste or insufficient capacity. Simultaneously, closed-loop feedback corrects prediction errors, improving the accuracy of the adjustment amount and balancing loss reduction, system stability, and operational economy.

Description

Technical Field

[0001] This invention relates to the field of energy storage control technology, specifically a control method for a PCS energy storage system. Background Technology

[0002] With the large-scale grid connection of new energy power generation and the rapid growth of user-side energy storage demand, PCS, as the core energy conversion device between energy storage system and grid / load, directly affects the economy and stability of energy storage system through its operating efficiency and loss control.

[0003] Existing technologies lack a comprehensive calculation model for resistive and reactive losses, relying solely on a single parameter to determine the need for adjustment. This can easily overlook the combined effects of the two types of losses, leading to premature or delayed adjustment. Furthermore, current adjustments depend on real-time data feedback and lack a power prediction mechanism, making it impossible to anticipate power fluctuation trends and resulting in adjustment delays. This is particularly problematic in ultra-short-term power surge scenarios, which can trigger a surge in transient losses. Additionally, the absence of a prediction error correction mechanism means that excessive deviations between predicted and actual values ​​can lead to inaccurate adjustment, further exacerbating losses or affecting system stability. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a control method for a PCS energy storage system, which solves the problem that the inability to predict power fluctuation trends in advance can easily lead to a surge in transient losses.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a control method for a PCS energy storage system, which specifically includes the following steps:

[0006] Step S1: Obtain the operating voltage, operating current and line resistance of the PCS energy storage system, and calculate the total line loss. If it is greater than the preset loss threshold, generate a loss adjustment signal.

[0007] Step S2: Respond to the loss adjustment signal and adjust the line loss and reactance loss respectively. For the line loss adjustment, the voltage is increased or decreased based on the power change trend of the historical power data of the same period. For the reactance loss, it is achieved by optimizing the system impedance characteristics and reducing the harmonic current amplitude.

[0008] Step S3: Collect historical operating data of the PCS energy storage system, select a standard model based on the prediction duration and application scenario characteristics, obtain the current scenario features and substitute them into the standard model to output the prediction results;

[0009] Step S4: Extract power change characteristics and related parameter trends from the power prediction results, match the corresponding adjustment measures, and formulate the corresponding adjustment amount and timing based on the matched adjustment measures.

[0010] As a further aspect of the present invention, the method for calculating the total line loss is as follows:

[0011] According to formula P 损，阻 =I 2 R is used to calculate the resistive loss of the PCS energy storage system, where I is the current and R is the resistance; the corresponding inductive reactance X is then obtained. L Harmony Anti-X C And according to the formula X=X L -X C Calculate the corresponding reactance X, and then use the formula P 损，抗 =Q 2 X / U 2 The reactance loss is calculated.

[0012] The total line loss is obtained by summing the two.

[0013] As a further aspect of the present invention, the adjustment method where the power change trend is increasing is as follows:

[0014] The system obtains the power and reactive power before and after the power increase, and calculates the voltage increase value by combining the target voltage at the receiving end, the real-time line resistance R, and the line reactance X. The system then adjusts the current line voltage based on the voltage increase value.

[0015] As a further aspect of the present invention, the adjustment method in which the power change trend decreases is as follows:

[0016] The voltage at the beginning of the line is obtained, and the voltage drop is calculated by combining the line output power and the line reactive power. The current line voltage is then adjusted based on the voltage drop value. Comprehensive line loss optimization and adjustment information is generated.

[0017] As a further aspect of the present invention, the method for optimizing the system impedance characteristics is as follows:

[0018] Obtain the system's equivalent inductance and reactive power compensation capacitance, and calculate the system's potential resonant frequency; process the deviation between this frequency and the system's main harmonic frequencies to generate impedance optimization information.

[0019] As a further aspect of the present invention, the method for reducing the amplitude of harmonic current is as follows:

[0020] A passive filter is used, and an LC filter circuit is designed for a specific harmonic order. The LC filter circuit has low impedance at the target harmonic frequency, which shunts the harmonic current and generates harmonic current optimization information.

[0021] As a further aspect of the present invention, the method for outputting the prediction result is as follows:

[0022] Historical working data is preprocessed to obtain preprocessed data. A prediction model is selected as the standard model based on the prediction duration and application scenario characteristics. The standard model is trained based on the preprocessed data. During the training process, time-series slicing is used to divide the data, and rolling window cross-validation is used instead of ordinary cross-validation. At the same time, grid search combined with Bayesian optimization is used to fine-tune the model hyperparameters, obtain the current scene features, and substitute them into the trained standard model to output the power prediction results.

[0023] As a further aspect of the present invention, the preprocessing includes:

[0024] Data cleaning: Remove outlier values ​​caused by sensor malfunctions and fill in missing values ​​using interpolation or the mean of adjacent time points;

[0025] Data normalization: standardizing parameters of different magnitudes to the same range;

[0026] Feature engineering: Extracting key features, including power change rate, periodic fluctuation features, and weather change precursor features.

[0027] As a further aspect of the present invention, the method for matching the corresponding adjustment means is as follows:

[0028] Power change characteristics and related parameter trends are extracted from the power prediction results. Power change characteristics include the magnitude, rate, and duration of predicted power changes; related parameter trends include changes in energy storage SOC and grid voltage; and adjustment methods are matched based on this core information.

[0029] If the core information corresponds to a short-term power surge, match the PCS energy storage battery capacity adjustment; if the core information corresponds to a short-term reactive power fluctuation, match the active power filter capacity adjustment; if the core information corresponds to an hourly power deficit, match the gas turbine or diesel generator power adjustment; if the core information corresponds to a short-term voltage deviation, match the transformer on-load tap changer.

[0030] As a further aspect of the present invention, the method for determining the corresponding adjustment amount and adjustment timing based on the matching adjustment means is as follows:

[0031] Quantitative adjustment amount: Obtain the predicted change amount after error correction. Error correction includes: comparing the predicted power with the actual power and calculating the error value = actual power - predicted power; if the error value exceeds the preset error threshold, the error correction calculation adjustment amount is triggered. The error coefficient is set based on the prediction reliability.

[0032] Quantitative adjustment timing: Obtain the predicted power change time and calculate the adjustment timing according to the formula: Adjustment timing = Predicted power change time - Equipment response delay time.

[0033] This invention provides a control method for a PCS energy storage system. Compared with the prior art, it has the following advantages:

[0034] This invention establishes a comprehensive calculation model for resistive and reactive losses, enabling real-time quantification of total line losses. It adapts loss thresholds based on system rated power and operating scenarios, avoiding errors from human experience. Adjustment is triggered only when thresholds are exceeded, reducing ineffective adjustments and lowering auxiliary energy consumption. For scenarios with increasing or decreasing power, precise adjustment amounts are calculated to ensure voltage adjustment perfectly matches power fluctuations and line parameters. This avoids losses exceeding limits and prevents voltage deviations from exceeding load tolerance, thus improving power supply stability.

[0035] This invention employs a combined impedance optimization and harmonic current shunting strategy. First, it calculates the potential resonant frequency of the system and adjusts the capacitor / reactor parameters to deviate from the main harmonic frequency, thus avoiding resonance at its source. Then, it designs an LC passive filter circuit for specific harmonics to achieve harmonic current shunting, reduce reactance loss, avoid the risk of harmonic amplification, and improve the system's anti-interference capability.

[0036] This invention collects historical operating data and performs cleaning, normalization, and feature engineering processing. By combining the data with the characteristics of the scenario and selecting a time series model, it can predict power change trends in advance at the minute to hour level. Based on the prediction results, it quantifies the adjustment amount and timing, reducing the adjustment delay from the current second level to the millisecond level, thus proactively addressing the surge in losses caused by power fluctuations. Based on the core information extracted from the prediction results, it accurately matches adjustment methods such as PCS energy storage, active power filters, gas turbines, and on-load tap changers of transformers, avoiding resource waste or insufficient capacity. At the same time, by correcting prediction errors through closed-loop feedback, it can improve the accuracy of adjustment, balancing loss reduction, system stability, and operational economy. Attached Figure Description

[0037] Figure 1 This is a diagram illustrating the steps and methods of the present invention. Detailed Implementation

[0038] 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 embodiments of the present invention, and not all embodiments. 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.

[0039] Please see Figure 1 This application provides a control method for a PCS energy storage system, which specifically includes the following steps:

[0040] Step S1: Obtain the operating parameters of the PCS energy storage system, including operating voltage, current, and resistance. Calculate the line loss of the energy storage system based on the resistance value, specifically according to formula P. 损，阻 =I 2 R is used to calculate the resistive loss of the PCS energy storage system, where I is the current and R is the resistance. Simultaneously, the corresponding inductive reactance X is obtained. L Harmony Anti-X C And according to the formula X=X L -X C Calculate the corresponding reactance X, and then use the formula P 损，抗 =Q 2 X / U 2 The reactance loss is calculated, and the sum of the resistive loss and reactance loss is calculated to obtain the line loss P. 总 ;

[0041] The obtained line loss P 总 Compared with the corresponding loss threshold, and the specific value of the loss threshold is set by the operator, if the line loss P 总 If the line loss P is greater than the loss threshold, it indicates that the PCS energy storage system needs to perform loss reduction processing and generate a loss adjustment signal; conversely, if the line loss P is less than the loss threshold, the PCS energy storage system needs to perform loss reduction processing and generate a loss adjustment signal. 总 If the loss is less than the loss threshold, it means that the loss of the PCS energy storage coefficient is within an acceptable range, and a continuous monitoring signal will be generated.

[0042] Step S2: Analyze the generated loss adjustment signal, and adjust the line loss and reactance loss separately based on the loss adjustment signal. The specific method for analyzing the line loss is as follows:

[0043] Acquire historical operating data of the PCS energy storage system, and simultaneously acquire corresponding historical data from the same period. Analyze the power changes corresponding to the historical data from the same period to determine whether the power increased or decreased. For cases of increased power, the line voltage needs to be increased and adjusted. Record the power before and after the increase as P1 and P2, respectively. Simultaneously, record the corresponding reactive power before and after the power increase as Q1 and Q2, respectively. Then, apply the formula... The voltage increase value was calculated. V r R is the target voltage at the receiving end, R is the real-time resistance, and X is the line reactance. The phase coefficient of the three-phase AC system is given by the calculated voltage increase value. Adjust the current voltage to the standard.

[0044] If the power decreases, the line voltage needs to be reduced and adjusted. The line's starting voltage U1 is then obtained, and the formula is applied... The voltage drop value was calculated. Where P is the output power of the line, Q is the reactive power of the line, and the calculated voltage drop is used. Adjustments are made to the standard to generate voltage regulation information;

[0045] Based on the above analysis, information for optimizing and adjusting line losses is generated.

[0046] To analyze reactance loss, the approach involves optimizing system impedance characteristics and reducing harmonic current amplitude. For optimizing system impedance characteristics, the resonant frequency is calculated using the formula... The potential resonant frequency f of the system is calculated. res Where L is the system equivalent inductance, C is the reactive power compensation capacitor, and the calculated potential resonant frequency f is used. res The impedance optimization information is generated by deviating from the main harmonic frequency. Common harmonics in the system are the 3rd, 5th, 7th, and 11th harmonics. If the system has a 5th harmonic, the corresponding main harmonic frequency f is... h If the resonant frequency is 250Hz, then the compensation capacitor capacity needs to be adjusted or a small-capacity reactor needs to be connected in series to achieve the desired resonant frequency f. res Deviation from the main harmonic frequency f h ;

[0047] To reduce the amplitude of harmonic currents, existing harmonic currents are removed from the system using filtering devices. Specifically, passive filter technology is employed, with LC filter circuits designed for specific harmonic orders to provide low-impedance paths, shunt harmonic currents, and generate optimized harmonic current information. For example, for the 5th harmonic, f... h =250Hz, design an LC parallel circuit X C5 =X L5 This makes the circuit have low impedance at 250Hz, diverting the 5th harmonic current to the filter instead of flowing through the system impedance.

[0048] The two factors are combined to generate reactive loss adjustment information.

[0049] Step S3: Collect historical operating data of the PCS energy storage system, such as the charging and discharging power of the energy storage PCS, load power, and new energy power generation. Simultaneously, perform data cleaning, data normalization, and feature engineering on the obtained historical operating data. Data cleaning involves removing outliers caused by sensor malfunctions and filling missing values ​​using interpolation or the average of adjacent time points. Data normalization involves standardizing parameters of different magnitudes to the same range. Feature engineering involves extracting key features, such as power change rate, periodic fluctuation characteristics, and precursory features before sudden weather changes, such as cloud movement speed. Preprocessed data is obtained. Based on the prediction duration and scenario characteristics, a suitable prediction model is selected and designated as the standard model. Specific prediction models include time series models and machine learning or deep learning models. The former is suitable for periodic power fluctuations, while the latter is suitable for nonlinear and complex fluctuations. The standard model is trained based on the obtained preprocessed data, and cross-validation is performed using time series data. The specific processing methods are as follows:

[0050] The time series data is divided into segments using time series slicing, such as using data from January to September 2024 for training, data from October for validation, and data from November for testing. Rolling window cross-validation is used instead of ordinary cross-validation. In addition to the usual mean absolute error and root mean square error, peak error and orientation accuracy are added. Grid search + Bayesian optimization is used to fine-tune the hyperparameters.

[0051] Next, the current scene features are obtained and substituted into the model to output the prediction results.

[0052] Step S4: Based on the obtained prediction results, and simultaneously obtaining core information from the prediction results, specifically including power change characteristics and related parameter trends, the power change characteristics include the magnitude change, rate of change, and duration of the predicted power. The related parameter trends specifically represent the changes in parameters strongly correlated with power, such as the energy storage SOC decreasing from 80% to 60% and the grid voltage decreasing from 10.5kV to 10.2kV. Adjustment measures are then matched based on the obtained core information, with the specific matching method as follows:

[0053] If the core information corresponds to a short-term power surge, then the corresponding PCS energy storage battery capacity is adjusted. If the core information corresponds to a short-term reactive power fluctuation, then the active power filter capacity is adjusted. If the core information corresponds to an hourly power deficit, such as a continuous 100kW power deficit in photovoltaic power within 2 hours, then the gas turbine or diesel generator power is adjusted. If the core information corresponds to a short-term voltage deviation, then the transformer on-load tap changer is adjusted. Based on the above analysis, the corresponding adjustment measures are generated.

[0054] Next, corresponding regulation strategies are formulated based on the regulation methods. Specifically, the PCS energy storage system is regulated by quantifying the regulation amount and timing. The specific regulation methods are as follows:

[0055] For the quantitative adjustment amount, the predicted change amount is obtained, and the predicted change amount here is the value after error correction. The specific error correction method is to compare the predicted power with the actual power and calculate the error value = actual power - predicted power. At the same time, the obtained error value is compared with the threshold. If the error value is greater than the threshold, the error correction is triggered. The calculated error value is used as the standard for corresponding adjustment. For example, if the error = -10kW, the actual power is 10kW lower than the prediction, then the energy storage will discharge an additional 10kW, increasing from the original 92kW to 102kW.

[0056] The corresponding adjustment amount is calculated according to the formula: adjustment amount = predicted change amount × (1 - error coefficient). The error coefficient is set based on the prediction reliability. For example, if the photovoltaic power is predicted to drop by 100kW after 1 hour (error rate 8%), then the energy storage pre-adjustment amount = 100kW × (1 - 0.08) = 92kW, with 8kW reserved for real-time adjustment and replenishment.

[0057] To determine the timing for quantitative adjustment, the predicted power change time is obtained, and the corresponding adjustment timing is calculated according to the formula: Adjustment Timing = Predicted Power Change Time - Equipment Response Delay Time. For example, if the load reactive power is predicted to increase by 100 kvar in 5 minutes and the SVG response delay is 50 ms, then the adjustment timing = 5 × 60 s - 0.05 s ≈ 299.95 s, which means that the SVG adjustment command is issued approximately 5 minutes in advance.

[0058] The data in the above formulas are all calculated using numerical values, without substituting the units of the parameters. In addition, the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0059] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A control method of a PCS energy storage system, characterized by, The method specifically comprises the following steps: Step S1, the working voltage, working current and line resistance of the PCS energy storage system are obtained, and the total line loss is calculated, and if it is greater than the preset loss threshold, a loss adjustment signal is generated; Step S2, in response to the loss adjustment signal, the line loss and the reactive power loss are adjusted respectively, for line loss adjustment, the voltage is increased or decreased based on the power change trend of the historical power data, and for the reactive power loss, the system impedance characteristics are optimized and the harmonic current amplitude is reduced; Step S3, historical working data of the PCS energy storage system are collected, a standard model is selected according to the prediction length and the application scene characteristics, the current scene characteristics are substituted into the standard model, and a prediction result is output; Step S4, the power change characteristics and the associated parameter trend are extracted from the power prediction result, and the corresponding adjustment means are matched, and the corresponding adjustment amount and adjustment time are formulated based on the matched adjustment means.

2. The control method of a PCS energy storage system according to claim 1, characterized by, The way to calculate the total line loss is: According to the formula P 损，阻 = I 2 R, where R is the resistance of the PCS energy storage system and I is the current. X L X C and the capacitive reactance X L X C X, then according to the formula P 损，抗 = Q 2 X / U 2 , the reactance loss is calculated; The sum of the two is the total line loss.

3. The control method of a PCS energy storage system according to claim 1, characterized by, The adjustment mode when the power change trend is increased is: The power, reactive power before the power increases, and the power, reactive power after the power increases are obtained, the target voltage of the receiving end, the real-time line resistance R and the line reactance X are combined, the voltage increase value is calculated, and the current line voltage is adjusted based on the voltage increase value.

4. The control method of a PCS energy storage system according to claim 1, characterized by, The adjustment mode when the power change trend is decreased is: The line head voltage is obtained, the line output power and the line reactive power are combined, the voltage reduction value is calculated, and the current line voltage is adjusted based on the voltage reduction value; comprehensive line loss optimization adjustment information is generated.

5. The control method of a PCS energy storage system according to claim 1, wherein, The way to optimize the system impedance characteristics is: The system equivalent inductance and the reactive power compensation capacitor are obtained, the system potential resonance frequency is calculated, the system main harmonic frequency is deviated, and impedance optimization information is generated.

6. The control method of a PCS energy storage system according to claim 1, characterized by, The way to reduce the harmonic current amplitude is: A passive filter is used, an LC filter circuit is designed for a specific harmonic number, the LC filter circuit has low impedance at the target harmonic frequency, the harmonic current is shunted, and harmonic current optimization information is generated.

7. The control method of a PCS energy storage system according to claim 1, characterized by, The way to output the prediction result is: The historical working data are preprocessed to obtain preprocessed data, a prediction model is selected as a standard model according to the prediction length and the application scene characteristics, the standard model is trained based on the preprocessed data, time series slicing is used to divide the data during the training process, and a rolling window cross-validation is used instead of ordinary cross-validation; at the same time, the model hyperparameters are optimized by using grid search combined with Bayesian optimization, the current scene characteristics are obtained, which are substituted into the trained standard model, and the power prediction result is output.

8. The control method of a PCS energy storage system according to claim 7, characterized by, The preprocessing includes: Data cleaning: removing abnormal values caused by sensor failure, and filling missing values by interpolation or adjacent time mean; Data normalization: normalizing parameters of different magnitudes to the same interval; Feature engineering: extracting key features, including power change rate, periodic fluctuation characteristics and weather mutation precursor characteristics.

9. The control method of a PCS energy storage system according to claim 1, characterized by, The way to match the corresponding adjustment means is: The power change characteristics and the associated parameter trend are extracted from the power prediction result, the power change characteristics include the amplitude change amount, change rate and duration of the predicted power; The associated parameter trend includes SOC change of energy storage and voltage change of power grid; the core information includes power change characteristics and associated parameter trend, and the matching adjustment means is based on the core information; If the core information corresponds to short-term power mutation, the matching adjustment means is the capacity adjustment of energy storage battery of PCS; if the core information corresponds to short-term reactive power fluctuation, the matching adjustment means is the capacity adjustment of active filter; if the core information corresponds to power shortage in hours, the matching adjustment means is the power adjustment of gas turbine or diesel generator; if the core information corresponds to short-term voltage deviation, the matching adjustment means is the on-load voltage regulation of transformer.

10. The control method of a PCS energy storage system according to claim 1, wherein, The matching adjustment means formulates the corresponding adjustment amount and adjustment time in the following manner: Quantitative adjustment amount: the predicted change amount is obtained after error correction, and the error correction includes: comparing the predicted power with the actual power, calculating the error value = actual power - predicted power; if the error value exceeds the preset error threshold, the error correction is triggered to calculate the adjustment amount; Quantitative adjustment time: the time when the predicted power change occurs is obtained, and the adjustment time is calculated according to the formula adjustment time = time when predicted power change occurs - equipment response delay time.

Images