Intelligent management method for parameters of solar energy storage battery and inverter

Abstract

The application discloses an intelligent management method for parameters of a solar energy storage battery and an inverter, and particularly relates to the fields of new energy micro-grid and power electronic control technology. Photovoltaic output power, load power, battery state of charge and temperature data are collected, and the photovoltaic power is decomposed into a trend component used for upper layer decision and an oscillation component used for triggering intervention. When the oscillation component simultaneously satisfies amplitude overrun and change rate overrun, the expected power is frozen, a non-integer order sliding mode surface is constructed by using a fractional order calculus operator with memory genetic characteristics, and the fractional order is dynamically adjusted according to the real-time battery temperature rise rate and state of charge, so as to constrain the change rate of the output current instruction. Meanwhile, the inverter is switched to a virtual synchronous generator mode, the virtual inertia coefficient is coupled with the current order, so as to absorb or release power impact energy. The method can reduce repeated flip of the charge and discharge instruction, and inhibit rapid reversal of the battery terminal voltage.

Description

Technical Field

[0001] This invention relates to the field of new energy microgrids and power electronic control technology, specifically to an intelligent management method for parameters of solar energy storage batteries and inverters. Background Technology

[0002] In solar photovoltaic energy storage systems, the coordinated control of batteries and inverters typically employs rule engines or machine learning methods to achieve intelligent parameter management. Existing technologies, based on preset optimization goals such as maximizing self-consumption rate or extending battery life, collect data on photovoltaic power, load power, and battery state of charge, make decisions within fixed time windows (e.g., seconds to minutes), generate charging and discharging commands, and send them to the inverter for execution.

[0003] However, actual photovoltaic output power is affected by cloud shadows, atmospheric turbulence, and other factors, resulting in drastic millisecond-level fluctuations (e.g., power drops sharply within one second and recovers rapidly due to broken cloud cover). Because the existing decision-making cycle is much slower than the power fluctuation cycle, the decision result is severely mismatched with the instantaneous actual power state: the upper-level logic continuously issues commands opposite to the instantaneous power, forcing the battery to rapidly switch between charging and discharging states at a frequency of several hertz, forming high-frequency oscillations. This oscillation not only triggers the risk of internal thermal runaway in the battery (temperature rise rate exceeding the protection threshold) but also causes the inverter power transistors to burn out due to repeated voltage spikes and reverse recovery losses, while conventional battery management systems cannot promptly cut off the circuit due to response delays. To address these issues, existing improvement schemes attempt to increase the decision frequency or add low-pass filters, but the former leads to wasted computational resources and a response lag paradox, while the latter cannot fundamentally eliminate the energy propagation path of the oscillations, and neither addresses the dynamic instability caused by cross-scale mismatch at the control mechanism level. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent management method for the parameters of solar energy storage batteries and inverters, so as to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent management method for parameters of solar energy storage batteries and inverters, comprising: Data on photovoltaic output power, load power, battery state of charge, and temperature are collected, and the photovoltaic power is decomposed into a trend component for upper-level decision-making and an oscillation component for triggering intervention. When the oscillation component simultaneously satisfies both amplitude and rate of change limits, a non-integer sliding surface is constructed using a fractional-order calculus operator with memory inheritance characteristics. The desired power given by the upper-level decision is used as input, so that the rate of change of the output current command obeys the constraint boundary determined by the fractional-order order of the real-time battery temperature rise rate and the state of charge dynamically adjusted. During the fractional sliding mode control period, the inverter is switched to virtual synchronous generator mode and its virtual inertia coefficient is coupled with the current order of the fractional sliding surface in real time, so that the inverter can automatically absorb or release power surge energy and further block the rapid reversal of the battery terminal voltage. The upper-level decision update is only triggered when the actual current change rate of the fractional sliding mode control output exceeds the preset delay for a continuous period of time and the deviation cannot be suppressed even after the fractional order has been adjusted to the limit value; otherwise, the original decision instruction is maintained.

[0006] Preferably, the data collected include photovoltaic output power, load power, battery state of charge, and temperature, including: Acquire photovoltaic output power, load power, battery state of charge, and temperature data, and record the acquisition time of each data; group data whose acquisition time deviation is within a preset allowable range under the same sampling time scale into valid data groups; when the acquisition time deviation exceeds the preset allowable range, discard the corresponding data group and wait for the next sampling time scale to re-merge; the highest value among the battery cell temperatures is used as the temperature data.

[0007] Preferably, the photovoltaic power is decomposed into trend components and oscillation components, including: During initial operation, the average photovoltaic output power from multiple consecutive valid data sets is taken as the initial hold value of the gradual change buffer zone; The current photovoltaic output power is compared with the previous photovoltaic output power at each sampling time scale, and the amplitude is judged when the change direction is consistent across multiple consecutive sampling time scales. The temperature limit is determined based on the temperature data, and the gradual change buffer is updated when the absolute difference between the current photovoltaic output power and the current hold value of the gradual change buffer does not exceed the temperature limit. The current hold value of the slow-change temporary storage region is determined as the trend component, and the difference between the current photovoltaic output power and the trend component is determined as the oscillation component.

[0008] Preferably, the determination of the temperature-corresponding limit includes: Divide the temperature data into multiple continuous temperature ranges; Configure a corresponding allowable range of photovoltaic power variation for each temperature range; When the temperature data enters the high temperature range, a lower allowable range of photovoltaic power change is selected as the corresponding temperature limit. The temperature-corresponding limit is used to control whether the gradual change temporary storage area is updated, so that the trend component of the battery temperature rises is less affected by the instantaneous photovoltaic power change.

[0009] Preferably, the oscillation component simultaneously satisfies both amplitude exceeding the limit and rate of change exceeding the limit, including: Calculate the absolute value of the current oscillation component and the rate of change between the current oscillation component and the oscillation component at the previous sampling time scale; Compare the absolute value of the current oscillation component with the preset amplitude limit; Compare the rate of change with a preset rate of change limit; When the absolute value of the current oscillation component exceeds the preset amplitude limit and the rate of change exceeds the preset rate of change limit, the expected power given by the upper-level decision under the current sampling time scale is frozen.

[0010] Preferably, the formation of the memory input for fractional-order calculus operators includes: After the expected power is determined, the power deviation is read from multiple consecutive sampling time scales before the freezing time. The power deviation is the result of subtracting the trend component from the photovoltaic output power at the same sampling time scale. A decreasing retention factor is assigned to each power deviation based on the time period from most recent to oldest. Multiply each power deviation by its corresponding retention factor and sum them to obtain the memory input used to construct the non-integer order sliding surface.

[0011] Preferably, the dynamic adjustment of the fractional order includes: The real-time battery temperature rise rate is determined based on the change between the current temperature data and the temperature data at the previous sampling time point. The temperature rise order deduction is determined based on the degree of deviation of the real-time battery temperature rise rate from the temperature rise rate safety limit. When the battery's state of charge is outside the safe charge range, the charge correction amount is determined based on the degree to which it deviates from the boundary of the safe charge range; The result after deducting the temperature rise order reduction and the charge correction from the baseline fractional order is used as the candidate fractional order, and the current fractional order is not higher than the fractional order under the previous sampling time scale and not lower than the lowest order limit value.

[0012] Preferably, the formation of the non-integer order sliding surface and constraint boundary includes: Determine the rate of change of the reference current based on the rated charge and discharge current; Based on the degree of reduction of the current fractional order relative to the reference fractional order, tighten the constraint boundary of the output current command change rate. The desired power is converted into the target current, and the current error is obtained by subtracting the actual output current from the target current. After converting the memory input into a memory current, it works together with the current fractional order to affect the current error, resulting in a non-integer order sliding surface value, which makes the output current command approach the target current within the constraint boundary.

[0013] Preferably, the virtual inertia coefficient is coupled in real time with the current order of the fractional-order sliding surface, including: When fractional sliding mode control is active, the inverter is switched from current follower operation to virtual synchronous generator operation, and the output current command of the previous sampling time scale and the inverter output power are used as the initial values ​​for switching. The inertia amplification ratio is determined based on the degree of reduction in the current fractional order relative to the baseline fractional order. The current virtual inertia coefficient is determined between the reference virtual inertia coefficient and the maximum virtual inertia coefficient according to the inertia amplification ratio; When the oscillation component is positive, the inverter is controlled to absorb the power surge energy first; when the oscillation component is negative, the inverter is controlled to release the power surge energy first.

[0014] Preferably, re-triggering the upper-level decision update includes: The actual output current change rate is calculated for each sampling time scale, and the over-limit duration is accumulated when the actual output current change rate is greater than the current constraint boundary. Clear the over-limit duration when the actual output current change rate is not greater than the current constraint boundary. When the over-limit duration exceeds the preset delay and the current fractional order reaches the lowest order limit, continue to determine whether the deviation between the target current and the actual output current is increasing in the same direction. The upper-level decision update is triggered only when the deviation increases in the same direction and the deviation increment exceeds the preset deviation increment limit; otherwise, the expected power and the original decision command are frozen.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention differs from existing methods that directly use instantaneous photovoltaic power in upper-level decision-making. Its core lies in first separating the photovoltaic output power into trend and oscillation components. Through methods such as sampling timescale merging, maintaining a gradual change buffer, and tightening temperature-related limits, the upper-level decision-making process only receives the temperature-constrained trend component. Therefore, millisecond-level disturbances such as cloud cover will not directly drive repeated changes in the charging and discharging strategy, solving the problem of frequent command direction reversals caused by the mismatch between the decision-making cycle and the power fluctuation cycle, and reducing the battery charging and discharging switching frequency.

[0016] This invention further differs from solutions that simply increase the decision frequency or use low-pass filtering. It freezes the desired power when the oscillation component exceeds the limit and adjusts the fractional order using memory input, real-time battery temperature rise rate, and state of charge, dynamically tightening the output current command change rate. Simultaneously, it switches the inverter to virtual synchronous generator operation and increases the virtual inertia coefficient with the current order. This method ensures that power surges are first absorbed or released by the inverter, preventing their rapid transmission to the battery terminal and suppressing rapid voltage reversal, increased temperature rise, and repeated surges to the inverter's power devices. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart of the intelligent management method for parameters of solar energy storage batteries and inverters according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0020] Example 1, please refer to Figure 1 As shown in this embodiment, the intelligent management method for the parameters of the solar energy storage battery and inverter includes: In this invention, the steps of collecting photovoltaic output power, load power, battery state of charge, and temperature data are used to form corresponding data sets before upper-level decision-making, and to separate the trend component for upper-level decision-making and the oscillation component for triggering intervention from the photovoltaic output power. The data acquisition period is set to 1 millisecond to 20 milliseconds, preferably 5 milliseconds. The acquisition time is recorded for each acquisition, and the data set processed in the same session includes photovoltaic output power, load power, battery state of charge, and temperature data. The highest temperature value among the individual battery cells is used as the temperature data for subsequent gradual change buffer zone updates.

[0021] When forming data sets, photovoltaic output power, load power, battery state of charge, and temperature data are synchronously merged using the same sampling timescale. The synchronization merging method is as follows: calculate the difference between the latest and earliest sampling times within the same data set. If the difference is less than 2 milliseconds, the data set proceeds to subsequent processing; if the difference exceeds 2 milliseconds, the data set does not participate in the determination of trend and oscillation components and waits for the next sampling timescale to be re-merged. This processing method avoids discrepancies between photovoltaic output power and battery temperature caused by communication delays or sampling order, thus preventing the gradual change buffer from being incorrectly updated.

[0022] After the photovoltaic (PV) output power is processed, candidate positions are simultaneously written to both the gradual variation buffer and the abrupt change buffer. The gradual variation buffer is used to maintain the low-frequency variation results of the PV output power, while the abrupt change buffer is used to store the instantaneous deviation of the PV output power relative to the trend component. During initial operation, the average PV output power of three consecutive valid data sets is taken as the initial hold value for the gradual variation buffer. This average value is calculated by adding the PV output power of the three valid data sets and then dividing by 3. After the initial hold value is formed, the trend component is equal to the current hold value of the gradual variation buffer.

[0023] The gradual change buffer is not updated with each sampling. The difference between the photovoltaic output power of the nth valid sample and the photovoltaic output power of the (n-1)th valid sample is calculated. A difference greater than 0 is recorded as an upward trend, a difference less than 0 as a downward trend, and a difference equal to 0 as a hold trend. If the direction of change in three consecutive valid samples is upward, downward, or hold, the photovoltaic output power is considered to have a continuous and consistent direction of change. If the direction of change in any valid sample is opposite to that of the previous valid sample, the gradual change buffer is not updated, and the original hold value continues to be used as the trend component.

[0024] After determining that the direction of change is continuous and consistent, the temperature-corresponding limit for photovoltaic output power is also determined based on temperature data. When the temperature data is below 35℃, the temperature-corresponding limit is 4% of the rated photovoltaic power; when the temperature data reaches 35℃ but is below 45℃, the temperature-corresponding limit is 2.5% of the rated photovoltaic power; when the temperature data reaches 45℃ but is below 55℃, the temperature-corresponding limit is 1% of the rated photovoltaic power; and when the temperature data reaches 55℃, the temperature-corresponding limit is 0.5% of the rated photovoltaic power. When the absolute difference between the photovoltaic output power of the nth valid sample and the current holding value of the gradual change buffer does not exceed the temperature-corresponding limit, the gradual change buffer is updated to the photovoltaic output power of the nth valid sample; when the absolute difference exceeds the temperature-corresponding limit, the gradual change buffer is not updated.

[0025] The trend component is determined as follows: after each effective sampling, the current hold value of the slowly varying temporary storage area is read, and this current hold value is used as the trend component for upper-level decision-making. Therefore, the photovoltaic power received by the upper-level decision-making is not directly equal to the instantaneous photovoltaic output power, but rather equal to the hold value that satisfies the requirements of synchronized sampling, continuous and consistent change direction, and temperature-constrained change amplitude. This avoids frequent changes in upper-level decision-making due to millisecond-level disturbances.

[0026] The oscillation component is determined as follows: Under the same valid sampling time scale, the difference between the photovoltaic output power and the trend component is calculated. This difference is written into the mutation temporary storage area as an instantaneous deviation and serves as the oscillation component used to trigger intervention. A positive oscillation component indicates that the instantaneous photovoltaic output power is higher than the trend component; a negative oscillation component indicates that the instantaneous photovoltaic output power is lower than the trend component; and a zero oscillation component indicates that the instantaneous photovoltaic output power is consistent with the trend component. The mutation temporary storage area stores the oscillation components under the most recent 5 valid sampling time scales and overwrites them in the order of sampling to facilitate subsequent judgment of oscillation component amplitude exceeding limits and rate of change exceeding limits.

[0027] Through the above processing, the continuous and slow changes in photovoltaic output power are preserved in the trend component, while the rapid deviations caused by cloud shading, enhanced edges of broken clouds, and transient disturbances on the inverter side are preserved in the oscillation component. The trend component is used for upper-level decision-making, which can reduce repeated changes in charging and discharging commands due to instantaneous fluctuations; the oscillation component is used to trigger intervention, which can identify millisecond-level power surges while keeping the upper-level decisions unchanged, providing input basis for subsequent fractional-mode sliding control to limit the rate of change of output current commands.

[0028] When the oscillation component simultaneously satisfies both amplitude and rate of change limits, a non-integer sliding surface is constructed using a fractional-order calculus operator with memory inheritance characteristics. The desired power given by the upper-level decision is used as input, so that the rate of change of the output current command follows the constraint boundary determined by the fractional-order order of the real-time battery temperature rise rate and the state of charge dynamically adjusted.

[0029] This implementation method is used to freeze the expected power given by the upper-level decision when the oscillation component simultaneously satisfies the limits of amplitude and rate of change, and during the freezing period, the rate of change of the output current command is constrained by a non-integer-order sliding mode surface.

[0030] The current sampling time is taken as the nth sampling time, and the sampling period is preferably 5 milliseconds. The absolute value of the oscillation component is calculated by taking the absolute value of the current oscillation component. The rate of change of the oscillation component is calculated by subtracting the oscillation component under the previous sampling time from the current oscillation component, taking the absolute value, and dividing the resulting value by the sampling period. The preset amplitude limit is preferably 5% of the rated photovoltaic power, with the setting rule being no less than 3% and no more than 8% of the rated photovoltaic power. The preset rate of change limit is preferably 10% per second of the rated photovoltaic power, with the setting rule being no less than 6% per second and no more than 15% per second of the rated photovoltaic power. When the absolute value of the oscillation component exceeds the preset amplitude limit and the rate of change of the oscillation component exceeds the preset rate of change limit, the expected power given by the upper-level decision under the current sampling time is recorded as the frozen expected power; during the freezing period, the expected power given by the upper-level decision later does not replace the frozen expected power until the exit condition is met.

[0031] After the desired power is frozen, the power deviation is read from the eight consecutive sampling time points prior to the freezing time. The power deviation is equal to the photovoltaic output power minus the trend component at the same sampling time point. Eight retention coefficients are set from most recent to furthest in time: the retention coefficient for the most recent sampling time point is 1, for the second-to-last sampling time point it is 0.78, for the third-to-last sampling time point it is 0.61, for the fourth-to-last sampling time point it is 0.48, for the fifth-to-last sampling time point it is 0.37, for the sixth-to-last sampling time point it is 0.29, for the seventh-to-last sampling time point it is 0.23, and for the eighth-to-last sampling time point it is 0.18. The memory input is obtained by multiplying each of the eight power deviations by its corresponding retention coefficient, and then summing the eight products. This memory input retains the attenuation effect of the power deviation before freezing, ensuring that subsequent current adjustments are not based solely on a single instantaneous sampling value.

[0032] The real-time battery temperature rise rate is calculated based on the temperature data at the continuously sampled time scale. The calculation process is as follows: subtract the temperature data at the previous sampling time scale from the current temperature data, and divide the difference by the sampling period; when the difference is less than 0, the real-time battery temperature rise rate is set to 0. The preferred safety limit for the temperature rise rate is 0.2℃ per second, with a rule of not less than 0.1℃ per second and not more than 0.4℃ per second. The temperature rise order deduction is calculated as follows: divide the real-time battery temperature rise rate by the safety limit for the temperature rise rate, and multiply the resulting ratio by 0.15; when the calculation result exceeds 0.25, the temperature rise order deduction is set to 0.25.

[0033] The safe charging range is preferably 20% to 90%. When the state of charge is below 20%, the charge deviation is equal to 20% minus the current state of charge; when the state of charge is above 90%, the charge deviation is equal to the current state of charge minus 90%; when the state of charge is between 20% and 90%, the charge deviation is 0. The charge correction is calculated by dividing the charge deviation by 10% and then multiplying by 0.1; when the calculation result exceeds 0.2, the charge correction is taken as 0.2.

[0034] The preferred baseline value for the fractional order is 0.85, and the preferred minimum value is 0.35. The candidate fractional order at the current sampling time scale is calculated by subtracting the temperature rise order deduction from 0.85, and then subtracting the charge correction. If the candidate fractional order is lower than 0.35, the current fractional order is set to 0.35; if the candidate fractional order is not lower than 0.35, it is compared with the fractional order at the previous sampling time scale, and the current fractional order is the one with the lower value. This method ensures that when the battery temperature rise rate increases or the state of charge deviates from the safe charge range, the fractional order can only be maintained or decreased, preventing the constraint boundary from repeatedly widening during oscillations.

[0035] The constraint boundary for the output current command rate of change is determined by the current fractional order. The preferred reference current rate of change is twice the rated charge / discharge current per second, with a setting rule of not less than one time per second and not more than three times the rated charge / discharge current per second. The constraint boundary is calculated by multiplying the reference current rate of change by the current fractional order and then dividing by 0.85. When the current fractional order is 0.85, the constraint boundary equals the reference current rate of change; when the current fractional order decreases to 0.35, the constraint boundary decreases to 0.4118 times the reference current rate of change, thus ensuring that the output current command rate of change decreases synchronously with battery thermal load and charge overrun.

[0036] The target current corresponding to the frozen expected power is calculated by dividing the frozen expected power by the current battery terminal voltage. The current error equals the target current minus the actual output current. The non-integer order sliding surface is composed of the current error and the memory input. The calculation process is as follows: first, divide the memory input by the current battery terminal voltage to obtain the memory current; then, multiply the memory current by the current fractional order; finally, add the current error to this product to obtain the non-integer order sliding surface value. The output current command approximates the target current according to the sign of the non-integer order sliding surface value, and the change in the output current command under adjacent sampling times must not exceed the constraint boundary multiplied by the sampling period. When the absolute value of the difference between the target current and the output current command under the previous sampling time is less than the allowable change, the current output current command is equal to the target current; when the absolute value of the difference is not less than the allowable change, the current output current command only increases or decreases the allowable change in the direction of the difference sign. As a result, the expected power is still continuously tracked, the output current command will not change abruptly due to millisecond-level oscillation components, and the rapid reversal of battery terminal voltage and repeated impacts on inverter power devices are suppressed.

[0037] During the fractional sliding mode control period, the inverter is switched to virtual synchronous generator mode, and its virtual inertia coefficient is coupled with the current order of the fractional sliding surface in real time, so that the inverter automatically absorbs or releases power surge energy, further blocking the rapid reversal of the battery terminal voltage.

[0038] During the fractional-mode sliding mode control period, the inverter switches from current-following operation to virtual synchronous generator operation. The switching time coincides with the freezing of the desired power. After the switch, the frozen desired power continues to be used as the power command, and the desired power re-given by the upper-level decision during the freezing period is not accepted. To avoid abrupt changes in operating state, the output current command of the sampling time before the switch is used as the initial current command after the switch, and the inverter output power of the sampling time before the switch is used as the initial electromagnetic power for virtual synchronous generator operation.

[0039] The virtual synchronous generator operates using virtual rotor angular velocity for power regulation. The rated angular velocity is taken as the grid's rated angular frequency; at 50 Hz, the rated angular velocity equals 2 multiplied by pi and then multiplied by 50. At each sampling timescale, the power difference is first calculated, which equals the frozen expected power minus the actual inverter output power. Then, the virtual angular acceleration is calculated, which equals the power difference minus the damping power, divided by the virtual inertia coefficient. The damping power equals the virtual damping coefficient multiplied by the virtual angular velocity deviation, which equals the current virtual angular velocity minus the rated angular velocity. The current virtual angular velocity equals the virtual angular velocity at the previous sampling timescale, plus the product of the virtual angular acceleration and the sampling period. The inverter output phase equals the output phase at the previous sampling timescale, plus the product of the current virtual angular velocity and the sampling period.

[0040] The virtual inertia coefficient is coupled in real-time with the current order of the non-integer order sliding surface. The preferred reference virtual inertia coefficient is 0.08 seconds, with a setting range of 0.04 to 0.15 seconds. The preferred maximum virtual inertia coefficient is 0.32 seconds, with a setting range of 3 to 5 times the reference virtual inertia coefficient. The current virtual inertia coefficient is obtained through the following calculation process: First, subtract the current fractional order from the reference fractional order (0.85) to obtain the order reduction; then divide the order reduction by 0.5 to obtain the inertia amplification ratio; then subtract the reference virtual inertia coefficient from the maximum virtual inertia coefficient to obtain the increaseable inertia; multiply the increaseable inertia by the inertia amplification ratio, and add it to the reference virtual inertia coefficient to obtain the candidate virtual inertia coefficient. When the candidate virtual inertia coefficient is lower than the reference virtual inertia coefficient, the current virtual inertia coefficient is the reference virtual inertia coefficient; when the candidate virtual inertia coefficient is higher than the maximum virtual inertia coefficient, the current virtual inertia coefficient is the maximum virtual inertia coefficient. Therefore, when the current fractional order decreases from 0.85 to 0.35, the virtual inertia coefficient continuously increases from the baseline virtual inertia coefficient to the maximum virtual inertia coefficient.

[0041] When the oscillation component is positive, it indicates that the instantaneous photovoltaic output power is higher than the trend component. Under the influence of the current virtual inertia coefficient, the inverter first absorbs the power surge energy. The absorbed amount is obtained by multiplying the absolute value of the oscillation component by the absorption ratio, which is preferably 0.6 and set within a range of 0.4 to 0.8; the absorbed amount must not exceed 20% of the rated inverter power. When the oscillation component is negative, it indicates that the instantaneous photovoltaic output power is lower than the trend component. Under the influence of the current virtual inertia coefficient, the inverter first releases the power surge energy. The released amount is obtained by multiplying the absolute value of the oscillation component by the release ratio, which is preferably 0.6 and set within a range of 0.4 to 0.8; the released amount must not exceed 20% of the rated inverter power. The power difference after absorption or release is entered into the virtual angular acceleration calculation, so that the power surge is primarily reflected as a limited change in virtual angular velocity, rather than being directly transmitted to the battery terminal.

[0042] To limit the reversal speed of the battery terminal voltage, the rate of change of the battery terminal voltage is calculated for each sampling time point. The calculation process is as follows: subtract the battery terminal voltage at the previous sampling time point from the current battery terminal voltage, take the absolute value, and divide by the sampling period. The preferred limit for the reversal speed of the battery terminal voltage is 8% per second of the rated battery voltage, with a setting range of 5% to 12% per second of the rated battery voltage. When the rate of change of the battery terminal voltage reaches this limit, the current virtual inertia coefficient is increased by 10% based on the original calculation result, and the increased value does not exceed the maximum virtual inertia coefficient; at the same time, the allowable change in the output current command at the next sampling time point is reduced to 80% of the original allowable change. Through the continuous coupling between the virtual inertia coefficient and the current fractional order, the inverter first assumes the role of absorption or release when a power surge occurs. The transmission time of the DC-side power surge to the battery terminal is lengthened, and the speed at which the battery terminal voltage changes from positive to negative is limited, thereby reducing the temperature rise and inverter power device surge caused by rapid switching of battery charging and discharging directions.

[0043] The upper-level decision update is only triggered when the actual current change rate of the fractional sliding mode control output exceeds the preset delay for a continuous period of time and the deviation cannot be suppressed even after the fractional order has been adjusted to the limit value; otherwise, the original decision instruction is maintained.

[0044] During the fractional-mode sliding control period, the actual output current is acquired at each sampling time scale, and the actual current change rate is determined based on the actual output current at the adjacent sampling time scale. The calculation process for the actual current change rate is as follows: the actual output current at the current sampling time scale is subtracted from the actual output current at the previous sampling time scale, the absolute value of the difference is taken, and then divided by the sampling period to obtain the actual current change rate at the current sampling time scale. The sampling period remains the aforementioned 5 milliseconds. The actual current change rate is used to determine whether the output current is still effectively limited by the constraint boundary.

[0045] At each sampling time point, the actual current change rate is compared with the current constraint boundary. If the current actual current change rate is greater than the current constraint boundary, the over-limit duration is increased by one sampling period based on the cumulative value of the previous sampling time point; if the current actual current change rate is not greater than the current constraint boundary, the over-limit duration is reset to zero. To avoid erroneous updates to upper-level decisions due to a single sampling spike, the preset delay is preferably 40 milliseconds, with the rule being no less than 4 consecutive sampling periods and no more than 12 consecutive sampling periods; when the sampling period is 5 milliseconds, the preset delay is between 20 and 60 milliseconds, preferably 40 milliseconds. Only when the over-limit duration is greater than the preset delay will subsequent judgments proceed.

[0046] The minimum order limit is preferably 0.35, with a setting rule of not lower than 0.25 and not higher than 0.45. When the current fractional order is greater than the minimum order limit, even if the over-limit duration is greater than the preset delay, the upper-level decision update will not be retried. Instead, the fractional order will continue to decrease based on the real-time battery temperature rise rate and charge correction amount, and the constraint boundary of the output current command change rate will continue to tighten. When the current fractional order is equal to the minimum order limit and the over-limit duration is greater than the preset delay, it is determined that the lower-level constraint has reached the most stringent state, and then the deviation change between the target current and the actual output current is compared.

[0047] The deviation is obtained by subtracting the actual output current from the target current. If the absolute value of the current deviation is greater than the absolute value of the deviation under the previous sampling time mark, and the current deviation has the same sign as the deviation under the previous sampling time mark, it is considered that the deviation is increasing in the same direction. To eliminate sampling quantization errors, the deviation amplification judgment also includes a minimum deviation increment limit. The minimum deviation increment limit is preferably 0.5% of the rated charge / discharge current, and the setting rule is not less than 0.2% and not more than 1% of the rated charge / discharge current. If the difference between the absolute value of the current deviation and the absolute value of the deviation under the previous sampling time mark is greater than the minimum deviation increment limit, and the deviation signs remain consistent, the deviation is considered to be increasing in the same direction; otherwise, the deviation is not considered to be increasing in the same direction.

[0048] Re-triggering the upper-level decision update requires three conditions to be met simultaneously: the actual current change rate is greater than the over-limit duration of the current constraint boundary is greater than the preset delay; the current fractional order is equal to the lowest order limit value; and the deviation between the target current and the actual output current increases in the same direction. When all three conditions are met, the frozen expected power is unfrozen, and the current trend component, load power, battery state of charge, and temperature data are re-sent to the upper-level decision, replacing the frozen expected power with the newly obtained expected power. After the replacement, the over-limit duration is reset to zero, and the deviation comparison restarts from the next sampling time scale.

[0049] If any of the three conditions is not met, the desired power and the original decision command are maintained at their current levels. During this period, the fractional-order sliding mode control continues to limit the rate of change of the output current command according to the current fractional-order limit, and the virtual synchronous generator continues to absorb or release power surge energy according to the current virtual inertia coefficient. Through this processing method, the upper-level decision will not be frequently updated due to short-term current spikes, transient sampling errors, or disturbances that can still be eliminated by the lower-level constraints. Only when the constraint boundary has been tightened to the lowest-order limit value, the actual rate of change of current continues to exceed the limit, and the deviation continues to worsen, is the upper-level decision allowed to be recalculated. This reduces the repeated flipping of charge and discharge commands and maintains the continuity of the battery terminal voltage change process.

[0050] 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 method for intelligent management of parameters of a solar energy storage battery and an inverter, characterized in that, include: Data on photovoltaic output power, load power, battery state of charge, and temperature are collected, and the photovoltaic power is decomposed into a trend component for upper-level decision-making and an oscillation component for triggering intervention. When the oscillation component simultaneously satisfies both amplitude and rate of change limits, a non-integer sliding surface is constructed using a fractional-order calculus operator with memory inheritance characteristics. The desired power given by the upper-level decision is used as input, so that the rate of change of the output current command obeys the constraint boundary determined by the fractional-order order of the real-time battery temperature rise rate and the state of charge dynamically adjusted. During the fractional sliding mode control period, the inverter is switched to virtual synchronous generator mode and its virtual inertia coefficient is coupled with the current order of the fractional sliding surface in real time, so that the inverter can automatically absorb or release power surge energy and further block the rapid reversal of the battery terminal voltage. The upper-level decision update is only triggered when the actual current change rate of the fractional sliding mode control output exceeds the preset delay for a continuous period of time and the deviation cannot be suppressed even after the fractional order has been adjusted to the limit value; otherwise, the original decision instruction is maintained.

2. The intelligent management of solar energy storage cell and inverter parameters method according to claim 1, wherein, Collect photovoltaic output power, load power, battery state of charge, and temperature data, including: Acquire photovoltaic output power, load power, battery state of charge, and temperature data, and record the acquisition time of each data; group data whose acquisition time deviation is within a preset allowable range under the same sampling time scale into valid data groups; when the acquisition time deviation exceeds the preset allowable range, discard the corresponding data group and wait for the next sampling time scale to re-merge; the highest value among the battery cell temperatures is used as the temperature data.

3. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 2, wherein, Photovoltaic power is decomposed into trend components and oscillation components, including: During initial operation, the average photovoltaic output power from multiple consecutive valid data sets is taken as the initial hold value of the gradual change buffer zone; The current photovoltaic output power is compared with the previous photovoltaic output power at each sampling time scale, and the amplitude is judged when the change direction is consistent across multiple consecutive sampling time scales. The temperature limit is determined based on the temperature data, and the gradual change buffer is updated when the absolute difference between the current photovoltaic output power and the current hold value of the gradual change buffer does not exceed the temperature limit. The current hold value of the slow-change temporary storage region is determined as the trend component, and the difference between the current photovoltaic output power and the trend component is determined as the oscillation component.

4. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 3, wherein, The determination of temperature-related limits includes: Divide the temperature data into multiple continuous temperature ranges; Configure a corresponding allowable range of photovoltaic power variation for each temperature range; When the temperature data enters the high temperature range, a lower allowable range of photovoltaic power change is selected as the corresponding temperature limit. The temperature-corresponding limit is used to control whether the gradual change temporary storage area is updated, so that the trend component of the battery temperature rises is less affected by the instantaneous photovoltaic power change.

5. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 4, wherein, The oscillating component simultaneously satisfies both amplitude exceeding the limit and rate of change exceeding the limit, including: Calculate the absolute value of the current oscillation component and the rate of change between the current oscillation component and the oscillation component at the previous sampling time scale; Compare the absolute value of the current oscillation component with the preset amplitude limit; Compare the rate of change with a preset rate of change limit; When the absolute value of the current oscillation component exceeds the preset amplitude limit and the rate of change exceeds the preset rate of change limit, the expected power given by the upper-level decision under the current sampling time scale is frozen.

6. The intelligent management method for parameters of solar energy storage batteries and inverters according to claim 1, characterized in that, The formation of the memory input for fractional calculus operators includes: After the expected power is determined, the power deviation is read from multiple consecutive sampling time scales before the freezing time. The power deviation is the result of subtracting the trend component from the photovoltaic output power at the same sampling time scale. A decreasing retention factor is assigned to each power deviation based on the time period from most recent to oldest. Multiply each power deviation by its corresponding retention factor and sum them to obtain the memory input used to construct the non-integer order sliding surface.

7. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 6 wherein, Dynamic adjustment of fractional order, including: The real-time battery temperature rise rate is determined based on the change between the current temperature data and the temperature data at the previous sampling time point. The temperature rise order deduction is determined based on the degree of deviation of the real-time battery temperature rise rate from the temperature rise rate safety limit. When the battery's state of charge is outside the safe charge range, the charge correction amount is determined based on the degree to which it deviates from the boundary of the safe charge range; The result after deducting the temperature rise order reduction and the charge correction from the baseline fractional order is used as the candidate fractional order, and the current fractional order is not higher than the fractional order under the previous sampling time scale and not lower than the lowest order limit value.

8. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 7, wherein, The formation of non-integer order sliding surfaces and constraint boundaries includes: Determine the rate of change of the reference current based on the rated charge and discharge current; Based on the degree of reduction of the current fractional order relative to the reference fractional order, tighten the constraint boundary of the output current command change rate. The desired power is converted into the target current, and the current error is obtained by subtracting the actual output current from the target current. After converting the memory input into a memory current, it works together with the current fractional order to affect the current error, resulting in a non-integer order sliding surface value, which makes the output current command approach the target current within the constraint boundary.

9. The method for intelligent management of solar energy storage cell and inverter parameters as claimed in claim 8, wherein, The virtual inertia coefficient is coupled in real time with the current order of the fractional-order sliding surface, including: When fractional sliding mode control is active, the inverter is switched from current follower operation to virtual synchronous generator operation, and the output current command of the previous sampling time scale and the inverter output power are used as the initial values ​​for switching. The inertia amplification ratio is determined based on the degree of reduction in the current fractional order relative to the baseline fractional order. The current virtual inertia coefficient is determined between the reference virtual inertia coefficient and the maximum virtual inertia coefficient according to the inertia amplification ratio; When the oscillation component is positive, the inverter is controlled to absorb the power surge energy first; when the oscillation component is negative, the inverter is controlled to release the power surge energy first.

10. The intelligent management method for parameters of solar energy storage battery and inverter according to claim 9, characterized in that, Re-triggering the upper-level decision update includes: The actual output current change rate is calculated for each sampling time scale, and the over-limit duration is accumulated when the actual output current change rate is greater than the current constraint boundary. Clear the over-limit duration when the actual output current change rate is not greater than the current constraint boundary. When the over-limit duration exceeds the preset delay and the current fractional order reaches the lowest order limit, continue to determine whether the deviation between the target current and the actual output current is increasing in the same direction. The upper-level decision update is triggered only when the deviation increases in the same direction and the deviation increment exceeds the preset deviation increment limit; otherwise, the expected power and the original decision command are frozen.

Images