Wind-solar-hydrogen energy storage system multi-energy coordinated optimization method and device

Abstract

The present application provides a wind-solar-hydrogen energy storage system multi-energy coordinated optimization method and device, relates to the technical field of energy system optimization scheduling, and the present application constructs a wind-solar comprehensive output sequence, adopts a moving average method to construct a power smoothing value sequence, and carries out linear fitting to generate a power fitting curve, then adopts a moving T test method to identify an inflection point, the minimum continuous operation duration of an electrolytic cell is a constraint, the future time period is divided into a plurality of time windows, and according to the fitting accuracy of the power fitting curve, the time windows are divided into linear and nonlinear windows, in the linear window, the maximum allowable variable load rate is limited, the fitting accuracy is maximized, and the power reference value sequence is determined, in the nonlinear window, the power reference value is obtained according to the coefficient of variation, the mean and the standard deviation, finally, the power reference value of each time window is combined with the maximum allowable variable load rate of the equipment, and the smooth running power plan is used to suppress the power mutation of the electrolytic cell, so that the hydrogen production efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of energy system optimization and scheduling technology, specifically to a method and apparatus for multi-energy coordinated optimization of wind-solar-hydrogen energy storage systems. Background Technology

[0002] The technology of converting surplus wind and solar power into green hydrogen storage through electrolyzers has been applied to microgrids in large-scale wind and solar power bases and industrial parks. Its operating power must be dynamically adjusted according to the real-time changes in wind and solar power output. While the technology is being further advanced, avoiding energy waste caused by wind and solar curtailment has also become one of the goals.

[0003] Existing wind-solar-hydrogen energy storage coordinated optimization technologies mostly employ methods such as power smoothing filtering, energy storage-assisted regulation, and fixed threshold limiting to handle fluctuations in wind and solar power output. This involves smoothing the power curve through low-pass filtering and moving averages, or configuring electrochemical energy storage to quickly suppress short-term fluctuations, and then allocating the processed power command to the electrolyzer. Some technologies combine predictive control to adjust the electrolyzer load in advance to track wind and solar changes, achieving fluctuation smoothing and stable equipment operation to a certain extent. However, focusing solely on power curve smoothing ignores the minimum continuous operating time of the electrolyzer and the allowable load variation rate, easily leading to excessively large power command jumps or excessively short operating periods causing frequent start-ups and shutdowns of the electrolyzer, accelerating equipment aging and reducing hydrogen production efficiency. Using a uniform control strategy for periods with different fluctuation characteristics can result in insufficient margin leading to operational risks, or excessive margin causing energy waste.

[0004] Therefore, there is an urgent need for a multi-energy coordinated optimization technology that integrates equipment constraints and segments according to power characteristics to solve the problems of low matching degree and insufficient operational stability of existing equipment.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-energy coordinated optimization method and apparatus for wind-solar-hydrogen energy storage systems, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A multi-energy coordinated optimization method for wind-solar-hydrogen energy storage systems, comprising the following specific steps: Based on the maximum operating power, the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future time periods is adjusted by limiting, and then smoothed by the moving average method to obtain a smoothed power value sequence. The sliding T-test is used to identify the inflection points of the power smoothing value sequence. Based on the minimum continuous running time, a screening constraint is constructed, and the inflection points that meet the screening constraint are extracted as retained inflection points. In this way, the future period is divided into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve. The fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, and then each time window is divided into a linear window or a nonlinear window. The fitting slope is limited and adjusted based on the maximum allowable load rate, and the power smoothing value is limited and adjusted based on the minimum operating power. Combined with the fitting slope after the limit adjustment, the power reference value sequence for each linear window is determined to maximize the fitting accuracy. The mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window are calculated, and then the power smoothing value is corrected to determine the power reference value sequence for each nonlinear window. The power reference value sequence is limited and adjusted based on the maximum / minimum operating power to generate a boundary reference power sequence. The power jump value between adjacent time windows is calculated based on the boundary reference power sequence, and the boundary reference power sequence of the time window is adjusted in combination with the maximum allowable load change rate to complete the optimized setting of the electrolyzer operating power plan.

[0008] Furthermore, the method for limiting and adjusting the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future periods is as follows: The current time is preset as the start time, and the start time is pushed forward by 3 hours to determine the end time. The time interval between the start time and the end time is taken as the future time period. Within the future time period, the sampling time interval is set to be no less than 7 and no more than 13 sampling points. The predicted wind and solar power output values ​​are collected at each sampling point and arranged in chronological order to form a wind and solar power output sequence. Then, all wind and solar power output values ​​in the wind and solar power output sequence that are greater than the maximum operating power of the electrolyzer are adjusted to the maximum operating power of the electrolyzer to limit the wind and solar power output sequence.

[0009] Furthermore, the method for constructing filtering constraints based on the minimum continuous runtime is as follows: The selection constraint is that the time interval between adjacent retained inflection points is not less than the minimum continuous operating time of the electrolytic cell. The specific extraction process of retained inflection points is as follows: For all inflection points identified by the sliding T-test, they are arranged in order of sampling time to form an inflection point sequence. The first inflection point is taken as the retained inflection point, and other inflection points after it are traversed in order of sampling time until the time interval between an inflection point and this retained inflection point first meets the selection constraint. This inflection point is taken as the new retained inflection point, and other inflection points after this new retained inflection point are traversed in order of sampling time to find the next retained inflection point that meets the selection constraint. This process is repeated to extract several retained inflection points from the inflection point sequence.

[0010] Furthermore, the method for dividing each time window into a linear window or a nonlinear window is as follows: For any given time window, the power fitting values ​​of each sampling point within that time window are extracted from the power fitting curve, and the power smoothing values ​​of each sampling point within that time window are extracted from the power smoothing value sequence. The coefficient of determination is calculated, and this coefficient of determination is used as the goodness-of-fit value of that time window. A linear goodness-of-fit judgment threshold is preset, and the value of the linear goodness-of-fit judgment threshold is between 0.85 and 0.9. If the goodness-of-fit value of the time window is not less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a linear window. If the goodness-of-fit value of the time window is less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a nonlinear window.

[0011] Furthermore, the method for limiting and adjusting the fitted slope based on the maximum permissible load rate is as follows: Obtain the maximum allowable variable load rate of the electrolytic cell. For any linear window, extract its fitting slope. If the fitting slope is not greater than the maximum allowable variable load rate of the electrolytic cell, the adjusted fitting slope is itself. If the fitting slope is greater than the maximum allowable variable load rate of the electrolytic cell, the adjusted fitting slope is the maximum allowable variable load rate of the electrolytic cell.

[0012] Furthermore, by combining the fitting slope after amplitude limiting adjustment, the method for determining the power reference value sequence for each linear window to maximize fitting accuracy is as follows: For any linear window, a linear equation is constructed using the fitted slope after fixed limiting adjustment. Under this linear equation, the goal is to maximize the fitting accuracy of the smoothed power value after minimum operating power limiting adjustment. The unique intercept of this linear equation is then solved. Based on the linear equation with the fitted slope and unique intercept after fixed limiting adjustment as the benchmark, the power benchmark value sequence of this linear window is obtained.

[0013] Furthermore, the method for determining the power reference value sequence for each nonlinear window is as follows: For any nonlinear window, extract the smoothed power values ​​of all sampling points within that time window after adjustment by the minimum operating power limit, and calculate their mean, standard deviation, and coefficient of variation. Set the corresponding safety factor based on the coefficient of variation: when the coefficient of variation is less than the first boundary point, set the safety factor to 1; when the coefficient of variation is not less than the first boundary point and less than the second boundary point, set the safety factor to 1.5; when the coefficient of variation is not less than the second boundary point, set the safety factor to 2. Subtract the product of the safety factor and the standard deviation from the mean value and use it as the power reference value at all sampling points within that time window to construct the power reference value sequence for that time window.

[0014] Furthermore, the method for calculating the power jump value between adjacent time windows based on the boundary reference power sequence is as follows: For any retained inflection point, based on the boundary reference power sequence of each time window, the boundary reference power value of the retained inflection point in the previous time window and the boundary reference power value in the next time window are extracted, and the difference between the latter and the former is taken as the power jump value of the retained inflection point.

[0015] Furthermore, the method for adjusting the reference power of the time window is as follows: The maximum allowable load change rate includes the maximum allowable load increase rate and the maximum allowable load decrease rate. Based on the boundary reference power sequence of each time window, the power jump value of all reserved inflection points is obtained. For any reserved inflection point, if its power jump value is 0, the boundary reference power of the reserved inflection point is kept unchanged. Otherwise, if the power jump value is greater than 0, calculate the product of the maximum allowable load increase rate and the sampling time interval, and then round up the quotient of the power jump value and the product to obtain the minimum number of transition sampling points required for load increase; if the power jump value is less than 0, calculate the product of the maximum allowable load decrease rate and the sampling time interval, and then round up the quotient of the absolute value of the power jump value and the product to obtain the minimum number of transition sampling points required for load decrease; locate the two adjacent time windows to which the reserved inflection point belongs, and take the last N sampling points in the time window before the reserved inflection point as transition sampling points, and the value of N is the minimum number of transition sampling points required for load increase or load decrease; take the boundary reference power of the first transition sampling point as the starting point and the boundary reference power of the last transition sampling point as the ending point, and perform linear fitting to obtain the adjusted power value of each transition sampling point; Perform the above operation on all retained inflection points, replace the power values ​​of the corresponding sampling points in the boundary reference power sequence with the adjusted power values ​​of all transition sampling points, and finally obtain the replaced boundary reference power sequence. The power values ​​in the replaced boundary reference power sequence are the operating power plan of the electrolyzer.

[0016] Additionally, a multi-energy coordinated optimization device for wind-solar-hydrogen energy storage systems is provided, characterized in that: the system is used to execute the aforementioned multi-energy coordinated optimization method for wind-solar-hydrogen energy storage systems, including: The power smoothing module is used to smooth the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future time periods based on the maximum operating power, and then use the moving average method to obtain the smoothed power value sequence. The window classification module is used to identify the inflection points of the power smoothing value sequence using the sliding T-test method, and to construct screening constraints based on the minimum continuous running time. The inflection points that meet the screening constraints are extracted as retained inflection points, thereby dividing the future period into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve, and the fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, so that each time window is divided into a linear window or a nonlinear window. The benchmark value generation module is used to limit the fitting slope based on the maximum allowable load rate, limit the power smoothing value based on the minimum operating power, and combine the fitted slope after the limit adjustment to determine the power benchmark value sequence for each linear window with the maximum fitting accuracy. It also calculates the mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window, and then corrects the power smoothing value to determine the power benchmark value sequence for each nonlinear window. The baseline adjustment module is used to limit the power baseline value sequence according to the maximum / minimum operating power, generate the boundary baseline power sequence, calculate the power jump value between adjacent time windows based on the boundary baseline power sequence, and adjust the boundary baseline power sequence of the time window in combination with the maximum allowable load change rate, so as to complete the optimization setting of the electrolytic cell operating power plan.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention identifies and divides time windows by using a sliding T-test to identify inflection points, and divides linear and nonlinear windows by goodness-of-fit segmentation to determine the operating power of the electrolyzer that follows the trend of wind and solar power curves, or the operating power determined based on the mean and standard deviation characteristics. This achieves adaptive matching and control logic of the electrolyzer operating power based on the trend of the combined wind and solar power output within each divided time window, balancing the reduction of operating risks and the reduction of energy waste. This invention also significantly suppresses the problem of sudden power changes in electrolyzers by coupling the minimum continuous operating time and the maximum allowable load rate of the electrolyzer during the power adjustment process, thereby smoothing the power jump at the time window connection, improving hydrogen production efficiency and delaying the aging of the electrolyzer equipment. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a schematic diagram of the overall light output of the present invention; Figure 3 This is a future time period inflection point identification map for the present invention; Figure 4 This is a schematic diagram of the power reference value sequence of the present invention; Figure 5 This is a schematic diagram of the overall device structure of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: Please see Figures 1 to 4 The present invention provides a technical solution: A multi-energy coordinated optimization method for wind-solar-hydrogen energy storage systems, comprising the following specific steps: S1: Based on the maximum operating power, the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system is adjusted by limiting the amplitude in the future period, and then smoothed by the moving average method to obtain the smoothed power value sequence. When a wind-solar-hydrogen energy storage system generates surplus power, in order to realize the resource utilization of abandoned electricity, the surplus power can be input into an electrolyzer for electrolysis to produce hydrogen. At present, the machine learning prediction model built on meteorological forecast data and combined with the historical operation data of the station has been maturely applied to the intraday photovoltaic and wind turbine output prediction. This has made the wind and solar integrated output sequence generated by superimposing intraday photovoltaic and wind turbine output highly accurate, and the prediction error in the ultra-short term of 0-4 hours has reached a low level. Therefore, the current time is preset as the start time, and the end time is determined by pushing the start time forward by 3 hours. This avoids excessive deviation that could distort the operating power of the subsequent electrolyzers. The time interval between the start and end times is considered the future time period. Instantaneous predicted wind and solar combined output values ​​are obtained within this future time period, with a sampling interval set to be no less than 7 and no more than 13 sampling points. This is because the ultra-short-term fluctuations of wind and solar power, such as ramp / ramp events, typically complete significant trend shifts within 30 to 60 minutes. If the number of sampling points is less than 7, continuous variation patterns may not be captured. No more than 13 sampling points mean that the wind and solar combined output value at each sampling point is the arithmetic mean of the instantaneous predicted wind and solar combined output values ​​within at least 15 minutes prior to that sampling time, effectively filtering out high-frequency noise at the second to minute level, such as cloud shadow fluctuations in solar power and turbulent fluctuations in wind power. The predicted wind and solar combined output values ​​collected at each sampling point are arranged in chronological order to form a wind and solar combined output sequence. If the sampling interval is too small, the electrolyzer equipment is easily damaged due to frequent fluctuations. Figure 2 As shown, the combined wind and solar power output is sampled in 15-minute increments over a 3-hour future period. Meanwhile, the electrolyzer has a defined maximum operating power limit; exceeding this limit will cause excessive current density, resulting in severe ohmic and polarization heat, and permanent damage to the equipment. Therefore, all combined wind and solar power output values ​​in the sequence that exceed the electrolyzer's maximum operating power are adjusted to the electrolyzer's maximum operating power to limit the combined wind and solar power output sequence.

[0022] S2: The sliding T-test is used to identify the inflection points of the power smoothing value sequence. Based on the minimum continuous running time, a screening constraint is constructed, and the inflection points that meet the screening constraint are extracted as retained inflection points. In this way, the future period is divided into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve. The fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, and then each time window is divided into a linear window or a nonlinear window. The essence of the sliding T-test is to detect whether there is a significant difference between the means of two adjacent subsequences. It is sensitive to trend inflection points, and its statistical significance is such that the presence of a single noise fluctuation in the power sequence does not necessarily indicate a large fluctuation in the smoothed power value, thus preventing frequent load changes in the electrolyzer. Therefore, the sliding T-test is used to identify the inflection point of the smoothed power value sequence as a benchmark for adjusting the operating power trend of the electrolyzer. The logical method for identifying the inflection point is as follows: The sliding T-test is preset with a window half-width and a significance level. The significance level is set between 0.01 and 0.1, with 0.05 commonly used in engineering. Simultaneously, the minimum continuous operating time of the electrolytic cell is the shortest interval between two power adjustments. Within this time, a smooth, shock-free power adjustment can be achieved at a rate not exceeding the maximum allowable load change rate of the electrolytic cell. To avoid abrupt changes in power adjustment caused by the first inflection point being too close to the first sampling point or the last inflection point being too close to the last sampling point, the window half-width is set to be multiplied by the sampling time interval. The product must be no less than the minimum continuous operating time of the electrolytic cell. If the minimum continuous operating time of the electrolytic cell is 15 minutes, the half-width of the window can be set to 2 or 3, which can balance the detection capability and boundary loss. Obtain the minimum continuous operating time of the electrolytic cell, and use the difference between twice the half-width of the window and 2 as the degrees of freedom. Combine this with the significance level and look up the t-distribution table to obtain the significance threshold. Set the half-width of the window to m, and the total number of sampling points to n. Start from m+1 sampling points and end at nm sampling points. Take the corresponding sampling points as the sampling points to be tested. For any sampling point to be tested, perform the following operations: Using the sampling point to be tested as the benchmark, select consecutive sampling points forward with the same half-width as the window to form a left window sequence; using the sampling point to be tested as the benchmark, select consecutive sampling points backward with the same half-width as the window to form a right window sequence. Calculate the mean and variance of the left and right window sequences respectively, and then calculate the combined variance based on the variances of the two sequences. Based on the difference in mean between the left and right window sequences, the combined variance, and the half-width of the window, calculate the T-statistic using the T-test formula. When the absolute value of the T-statistic is greater than the significance threshold, the sampling point to be tested is determined to be an inflection point. The significance of the identified inflection point is that the previous trend of wind and solar power output, such as a steady increase, has ended, and the system is now in a stable operation or has begun a downward trend. At this point, the sliding T-test may identify multiple inflection points in a short period of time. To avoid frequent adjustments to the electrolytic cell's operating power as the inflection points change, a screening constraint is set: the time interval between adjacent retained inflection points is not less than the minimum continuous operating time of the electrolytic cell. The specific extraction process for retained inflection points is as follows: For all inflection points identified by the sliding T-test, they are arranged sequentially according to the sampling time to form an inflection point sequence. The first inflection point is taken as the retained inflection point, and other inflection points after it are traversed in the sampling time sequence until the time interval between an inflection point and this retained inflection point first meets the screening constraint. This inflection point is taken as the new retained inflection point, and other inflection points after this new retained inflection point are traversed in the sampling time sequence to find the next retained inflection point that meets the screening constraint. This process is repeated to extract several retained inflection points from the inflection point sequence, ensuring that each retained inflection point is executable. Figure 3 As shown, the T-test method was used to identify inflection points and select and retain them for the power smoothing value sequence.

[0023] The future time period is divided into several time windows based on the retention inflection points: all the extracted retention inflection points are arranged in ascending order of sampling time as the dividing benchmark. First, the time period between the first sampling point set in the future time period and the first retention inflection point is defined as the first time window. The time period between the last sampling point set in the future time period and the last retention inflection point is defined as the last time window. Then, the time periods between every two adjacent retention inflection points are sequentially defined as time windows, resulting in several continuous and non-overlapping time windows, thus completing the operation of dividing the future time period into several time windows. Within each defined time window, the least squares method is used to linearly fit the power smoothing value sequence contained in the window to obtain a power fitting curve. The slope of the power fitting curve corresponds to the theoretical load rate of the electrolytic cell. For any given time window, the power fitting value of each sampling point within that time window is extracted from the power fitting curve, and the power smoothing value of each sampling point within that time window is extracted from the power smoothing value sequence. The coefficient of determination is calculated, and this coefficient of determination is used as the goodness-of-fit value of that time window. A linear goodness-of-fit judgment threshold is preset, and the value of the linear goodness-of-fit judgment threshold is between 0.85 and 0.9. If the goodness-of-fit value of the time window is not less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a linear window, indicating that the power change trend within the time window is highly consistent with the linear law. If the goodness-of-fit value of the time window is less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a nonlinear window, indicating that the power fluctuation value within the time window is large.

[0024] S3: The fitting slope is limited and adjusted based on the maximum allowable load rate, and the power smoothing value is limited and adjusted based on the minimum operating power. Combined with the fitting slope after the limit adjustment, the power reference value sequence of each linear window is determined to maximize the fitting accuracy. The mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window are calculated, and then the power smoothing value is corrected to determine the power reference value sequence of each nonlinear window. The combined output of wind and solar power naturally fluctuates rapidly, and the original slope obtained from linear fitting will experience instantaneous and significant increases and decreases. If the electrolyzer power is adjusted directly according to this rate, it will exceed the dynamic response limit of the equipment. Therefore, the maximum allowable load rate of the electrolyzer is first obtained, which is the maximum amplitude of the allowable power change per unit time limited by the electrolyzer equipment's factory and operating procedures. For any linear window, its fitting slope is extracted. If the fitting slope is not greater than the maximum allowable load rate of the electrolyzer, it indicates that the wind and solar power changes naturally and gradually, within the tolerance range of the electrolyzer. In this case, the adjusted fitting slope is itself. If the fitting slope is greater than the maximum allowable load rate of the electrolyzer, it indicates that the wind and solar power output increases or decreases rapidly during this period, and the theoretical load rate of the electrolyzer exceeds the equipment's safe adjustment capability. Therefore, the adjusted slope is set to the maximum allowable load rate of the electrolyzer, forcibly constraining the actual load rate of the electrolyzer to always be within the safe range.

[0025] Since electrolyzers cannot be frequently started and stopped, in practical applications, they should be operated at their minimum operating power as much as possible. Otherwise, operating below this minimum operating power will not only significantly reduce hydrogen production efficiency and fail to meet hydrogen purity standards, but will also cause electrode passivation and accelerated component corrosion, seriously affecting equipment safety and service life. To ensure that the electrolyzer can maintain stable electrochemical reactions and long-term continuous operation, the power smoothing value is limited and adjusted based on the minimum operating power. All power smoothing values ​​lower than the minimum operating power are adjusted to the minimum operating power. For any linear window, a linear equation is constructed with a fixed fitting slope after the limit adjustment. Under this linear equation, the goal is to maximize the fitting accuracy after the minimum operating power limit adjustment. The unique intercept of this linear equation is then solved. Based on the linear equation with the fixed fitting slope and unique intercept as the benchmark, the power benchmark value sequence for this linear window is obtained. Specifically, the electrolyzer has continuous, linear, and gradually changing properties. The load is adjusted by a uniform ramp throughout the process, and it closely matches the actual combined wind and solar power output level at each sampling point.

[0026] For any nonlinear window, the power smoothing value after minimum operating power limiting adjustment within this type of time window does not exhibit a stable linear change pattern and fluctuates significantly. Therefore, linear following logic is unsuitable for adjustment, as it would cause repeated small load changes in the electrolytic cell. Instead, a logic that ensures constant power operation within this type of time window is adopted to correct the power smoothing value after minimum operating power limiting adjustment. This ensures stable operation of the electrolytic cell load while minimizing power wastage. Specifically, the logic for correcting the power smoothing value after minimum operating power limiting adjustment is as follows: Extract the power smoothing values ​​after minimum operating power limiting adjustment from all sampling points within this time window, and calculate their mean, standard deviation, and coefficient of variation. Based on the coefficient of variation, set the corresponding safety factor: when the coefficient of variation is less than the first boundary point, set the safety factor to 1; when the coefficient of variation is not less than the first boundary point and less than the second boundary point, set the safety factor to 1.5; when the coefficient of variation is not less than the second boundary point, set the safety factor to 2. Subtract the safety factor and standard deviation from the mean value as the power reference value for all sampling points within this time window. In the formula, Indicates the power reference value. This represents the mean. Indicates the safety factor. The standard deviation represents the overall average level of combined wind and solar power output within the time window, serving as the basis for determining the power benchmark value. The standard deviation quantifies the absolute fluctuation range of combined wind and solar power output within the window; a larger standard deviation indicates a higher degree of power dispersion and more severe instantaneous fluctuations. The coefficient of variation, the ratio of the standard deviation to the mean, quantifies the relative volatility of wind and solar power output within the time window; a smaller coefficient of variation indicates lower power dispersion and more stable output, while a larger coefficient of variation indicates higher power dispersion and stronger volatility. Therefore, the coefficient of variation is used as the grading basis to match differentiated safety factors. Based on the logic that the more severe the power fluctuation, the larger the safety factor value, and the inverse correlation between the power benchmark value and the power output, the power gap caused by a sudden drop in output is offset while minimizing power curtailment. The power smoothing values ​​of all sampling points within the time window, adjusted after minimum operating power limit, are corrected to the calculated power benchmark value, thereby constructing a power benchmark value sequence for the time window, ensuring the steady-state operation capability of the hydrogen production system under complex weather conditions.

[0027] S4: Adjust the power reference value sequence based on the maximum / minimum operating power to generate a boundary reference power sequence. Calculate the power jump value between adjacent time windows based on the boundary reference power sequence, and adjust the reference power of the time window in conjunction with the maximum allowable load rate. In determining the power reference value sequence for each linear window to maximize fitting accuracy, the power reference value of some sampling points may exceed the maximum / minimum operating power due to slope accumulation and intercept deviation. The power reference value of the nonlinear window is based on the mean minus the product of the safety factor and the standard deviation, which may also lead to the power reference value exceeding the maximum / minimum operating power. Therefore, the power reference value sequence is adjusted again based on the maximum / minimum operating power. That is, if the power reference value of a sampling point is greater than the maximum operating power of the electrolyzer, the power reference value of that sampling point is corrected to the maximum operating power; if the power reference value of a sampling point is less than the minimum operating power of the electrolyzer, the power reference value of that sampling point is corrected to the minimum operating power, and the power values ​​of the remaining sampling points remain unchanged. The power reference value sequence for each time window is calculated independently. The retained inflection point is the boundary point between two adjacent time windows. There may be cases where the boundary reference power of the same sampling point is different in two adjacent time windows. For any retained inflection point, based on the boundary reference power sequence of each time window, the boundary reference power value of the retained inflection point in the next time window and the boundary reference power value in the previous time window are extracted, and the difference between the two is used as the power jump value of the retained inflection point, reflecting the power change amplitude at the boundary of two adjacent windows. The maximum allowable load change rate includes the maximum allowable load increase rate and the maximum allowable load decrease rate, both of which are inherent hard constraints of the electrolytic cell. Based on the boundary reference power sequence of each time window, the power jump value of all retained inflection points is obtained. For any retained inflection point, if its power jump value is 0, it means that the boundary reference power of the overlapping samples in two adjacent time windows is consistent, so the boundary reference power of the retained inflection point remains unchanged. Otherwise, to address the power jump problem, the issue is transformed into a linear, gradual transition. This ensures that the load change rate of the electrolytic cell never exceeds its maximum allowable load change rate, thus avoiding damage to the electrolytic cell. If the power jump value is greater than 0, the product of the maximum allowable load increase rate and the sampling time interval is calculated, and the quotient of the power jump value and this product is rounded up to obtain the minimum number of transition sampling points required for load increase. If the power jump value is less than 0, the product of the maximum allowable load decrease rate and the sampling time interval is calculated, and the quotient of the absolute value of the power jump value and this product is rounded up to obtain the minimum number of transition sampling points required for load decrease. The two adjacent time windows to which the retained inflection point belongs are located, and the last N sampling points in the time window preceding the retained inflection point are taken as transition sampling points. The value of N is the minimum number of transition sampling points required for either load increase or load decrease. Linear fitting is performed with the boundary reference power of the first transition sampling point as the starting point and the boundary reference power of the last transition sampling point as the ending point to obtain the adjusted power value for each transition sampling point. Perform the above operation on all retained inflection points, replacing the power values ​​of the corresponding sampling points in the boundary reference power sequence with the adjusted power values ​​of all transition sampling points. The resulting boundary reference power sequence is the power plan for the electrolyzer. Figure 4 As shown, the replaced boundary reference power sequence constitutes the final operating power plan of the electrolyzer, transforming the discontinuous step power command into a smooth linear trajectory. While conforming to the actual wind and solar power output level at each sampling point, all operating powers are within the maximum / minimum operating power range of the electrolyzer, and the load rate compliance will not result in any jump shocks.

[0028] Please see Figure 5 The present invention also provides a multi-energy coordinated optimization device for a wind-solar-hydrogen energy storage system, used to execute the above-mentioned multi-energy coordinated optimization method for a wind-solar-hydrogen energy storage system, comprising: The power smoothing module is used to smooth the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future time periods based on the maximum operating power, and then use the moving average method to obtain the smoothed power value sequence. The window classification module is used to identify the inflection points of the power smoothing value sequence using the sliding T-test method, and to construct screening constraints based on the minimum continuous running time. The inflection points that meet the screening constraints are extracted as retained inflection points, thereby dividing the future period into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve, and the fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, so that each time window is divided into a linear window or a nonlinear window. The benchmark value generation module is used to limit the fitting slope based on the maximum allowable load rate, limit the power smoothing value based on the minimum operating power, and combine the fitted slope after the limit adjustment to determine the power benchmark value sequence for each linear window with the maximum fitting accuracy. It also calculates the mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window, and then corrects the power smoothing value to determine the power benchmark value sequence for each nonlinear window. The baseline adjustment module is used to limit the power baseline value sequence according to the maximum / minimum operating power, generate the boundary baseline power sequence, calculate the power jump value between adjacent time windows based on the boundary baseline power sequence, and adjust the boundary baseline power sequence of the time window in combination with the maximum allowable load change rate, so as to complete the optimization setting of the electrolytic cell operating power plan.

[0029] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0030] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0031] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0032] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A multi-energy coordinated optimization method for wind-solar-hydrogen energy storage systems, characterized in that, The specific steps include: Based on the maximum operating power, the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future time periods is adjusted by limiting, and then smoothed by the moving average method to obtain a smoothed power value sequence. The sliding T-test is used to identify the inflection points of the power smoothing value sequence. Based on the minimum continuous running time, a screening constraint is constructed, and the inflection points that meet the screening constraint are extracted as retained inflection points. In this way, the future period is divided into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve. The fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, and then each time window is divided into a linear window or a nonlinear window. The fitting slope is limited and adjusted based on the maximum allowable load rate, and the power smoothing value is limited and adjusted based on the minimum operating power. Combined with the fitting slope after the limit adjustment, the power reference value sequence for each linear window is determined to maximize the fitting accuracy. The mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window are calculated, and then the power smoothing value is corrected to determine the power reference value sequence for each nonlinear window. The power reference value sequence is limited and adjusted based on the maximum / minimum operating power to generate a boundary reference power sequence. The power jump value between adjacent time windows is calculated based on the boundary reference power sequence, and the boundary reference power sequence of the time window is adjusted in combination with the maximum allowable load change rate to complete the optimized setting of the electrolyzer operating power plan.

2. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 1, characterized in that: The method for limiting and adjusting the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future periods is as follows: The current time is preset as the start time, and the start time is pushed forward by 3 hours to determine the end time. The time interval between the start time and the end time is taken as the future time period. Within the future time period, the sampling time interval is set to be no less than 7 and no more than 13 sampling points. The predicted wind and solar power output values ​​are collected at each sampling point and arranged in chronological order to form a wind and solar power output sequence. Then, all wind and solar power output values ​​in the wind and solar power output sequence that are greater than the maximum operating power of the electrolyzer are adjusted to the maximum operating power of the electrolyzer to limit the wind and solar power output sequence.

3. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 2, characterized in that: The method for constructing filtering constraints based on minimum continuous runtime is as follows: The selection constraint is that the time interval between adjacent retained inflection points is not less than the minimum continuous operating time of the electrolytic cell. The specific extraction process of retained inflection points is as follows: For all inflection points identified by the sliding T-test, they are arranged in order of sampling time to form an inflection point sequence. The first inflection point is taken as the retained inflection point, and other inflection points after it are traversed in order of sampling time until the time interval between an inflection point and this retained inflection point first meets the selection constraint. This inflection point is taken as the new retained inflection point, and other inflection points after this new retained inflection point are traversed in order of sampling time to find the next retained inflection point that meets the selection constraint. This process is repeated to extract several retained inflection points from the inflection point sequence.

4. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 3, characterized in that: The method for dividing each time window into a linear window or a nonlinear window is as follows: For any given time window, the power fitting values ​​of each sampling point within that time window are extracted from the power fitting curve, and the power smoothing values ​​of each sampling point within that time window are extracted from the power smoothing value sequence. The coefficient of determination is calculated, and this coefficient of determination is used as the goodness-of-fit value of that time window. A linear goodness-of-fit judgment threshold is preset, and the value of the linear goodness-of-fit judgment threshold is between 0.85 and 0.

9. If the goodness-of-fit value of the time window is not less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a linear window. If the goodness-of-fit value of the time window is less than the linear goodness-of-fit judgment threshold, then the time window is determined to be a nonlinear window.

5. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 1, characterized in that: The method for limiting and adjusting the fitted slope based on the maximum permissible load rate is as follows: Obtain the maximum allowable variable load rate of the electrolytic cell. For any linear window, extract its fitting slope. If the fitting slope is not greater than the maximum allowable variable load rate of the electrolytic cell, the adjusted fitting slope is itself. If the fitting slope is greater than the maximum allowable variable load rate of the electrolytic cell, the adjusted fitting slope is the maximum allowable variable load rate of the electrolytic cell.

6. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 5, characterized in that: The method for determining the power reference value sequence for each linear window by combining the fitting slope after amplitude limiting adjustment to maximize fitting accuracy is as follows: For any linear window, a linear equation is constructed using the fitted slope after fixed limiting adjustment. Under this linear equation, the goal is to maximize the fitting accuracy of the smoothed power value after minimum operating power limiting adjustment. The unique intercept of this linear equation is then solved. Based on the linear equation with the fitted slope and unique intercept after fixed limiting adjustment as the benchmark, the power benchmark value sequence of this linear window is obtained.

7. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 4, characterized in that: The method for determining the power reference value sequence for each nonlinear window is as follows: For any nonlinear window, extract the smoothed power values ​​of all sampling points within that time window after adjustment by the minimum operating power limit, and calculate their mean, standard deviation, and coefficient of variation. Set the corresponding safety factor based on the coefficient of variation: when the coefficient of variation is less than the first boundary point, set the safety factor to 1; when the coefficient of variation is not less than the first boundary point and less than the second boundary point, set the safety factor to 1.5; when the coefficient of variation is not less than the second boundary point, set the safety factor to 2. Subtract the product of the safety factor and the standard deviation from the mean value and use it as the power reference value at all sampling points within that time window to construct the power reference value sequence for that time window.

8. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 1, characterized in that: The method for calculating the power jump value between adjacent time windows based on the boundary reference power sequence is as follows: For any retained inflection point, based on the boundary reference power sequence of each time window, the boundary reference power value of the retained inflection point in the previous time window and the boundary reference power value in the next time window are extracted, and the difference between the latter and the former is taken as the power jump value of the retained inflection point.

9. The multi-energy coordinated optimization method for wind-solar-hydrogen energy storage system according to claim 2, characterized in that: The method for adjusting the reference power of the time window is as follows: The maximum allowable load change rate includes the maximum allowable load increase rate and the maximum allowable load decrease rate. Based on the boundary reference power sequence of each time window, the power jump value of all reserved inflection points is obtained. For any reserved inflection point, if its power jump value is 0, the boundary reference power of the reserved inflection point is kept unchanged. Otherwise, if the power jump value is greater than 0, calculate the product of the maximum allowable load increase rate and the sampling time interval, and then round up the quotient of the power jump value and the product to obtain the minimum number of transition sampling points required for load increase; if the power jump value is less than 0, calculate the product of the maximum allowable load decrease rate and the sampling time interval, and then round up the quotient of the absolute value of the power jump value and the product to obtain the minimum number of transition sampling points required for load decrease; locate the two adjacent time windows to which the reserved inflection point belongs, and take the last N sampling points in the time window before the reserved inflection point as transition sampling points, and the value of N is the minimum number of transition sampling points required for load increase or load decrease; take the boundary reference power of the first transition sampling point as the starting point and the boundary reference power of the last transition sampling point as the ending point, and perform linear fitting to obtain the adjusted power value of each transition sampling point; Perform the above operation on all retained inflection points, replace the power values ​​of the corresponding sampling points in the boundary reference power sequence with the adjusted power values ​​of all transition sampling points, and finally obtain the replaced boundary reference power sequence. The power values ​​in the replaced boundary reference power sequence are the operating power plan of the electrolyzer.

10. A multi-energy coordinated optimization device for wind-solar-hydrogen energy storage systems, characterized in that: The device is used to execute the multi-energy coordinated optimization method for the wind-solar-hydrogen energy storage system as described in any one of claims 1-9: The power smoothing module is used to smooth the combined wind and solar power output sequence of the wind-solar-hydrogen energy storage system in future time periods based on the maximum operating power, and then use the moving average method to obtain the smoothed power value sequence. The window classification module is used to identify the inflection points of the power smoothing value sequence using the sliding T-test method, and to construct screening constraints based on the minimum continuous running time. The inflection points that meet the screening constraints are extracted as retained inflection points, thereby dividing the future period into several time windows. In each time window, the power smoothing value is linearly fitted to generate a power fitting curve, and the fitting slope and goodness of fit of each time window are determined by combining the power smoothing value sequence, so that each time window is divided into a linear window or a nonlinear window. The benchmark value generation module is used to limit the fitting slope based on the maximum allowable load rate, limit the power smoothing value based on the minimum operating power, and combine the fitted slope after the limit adjustment to determine the power benchmark value sequence for each linear window with the maximum fitting accuracy. It also calculates the mean, standard deviation and coefficient of variation of the power smoothing value in the nonlinear window, and then corrects the power smoothing value to determine the power benchmark value sequence for each nonlinear window. The baseline adjustment module is used to limit the power baseline value sequence according to the maximum / minimum operating power, generate the boundary baseline power sequence, calculate the power jump value between adjacent time windows based on the boundary baseline power sequence, and adjust the boundary baseline power sequence of the time window in combination with the maximum allowable load change rate, so as to complete the optimization setting of the electrolytic cell operating power plan.

Images