A source network load storage multi-dimensional coupling dynamic supply and demand linkage optimization algorithm

Abstract

The present application relates to the technical field of smart grid, and discloses a source-grid-load-storage multi-dimensional coupling dynamic supply-demand linkage optimization algorithm, comprising: obtaining real-time active power output data of the power supply side, node voltage data and frequency deviation data of the power grid side, baseline power consumption data of the load side and real-time state of charge data of the energy storage side; calculating dynamic inertia margin according to inherent physical response delay time of each end; when the dynamic inertia margin is lower than the safety benchmark threshold, generating a dynamic supply-demand linkage instruction sequence, increasing the load side disconnection priority baseline, and constructing an exponential damping term according to the real-time voltage change rate and the dynamic inertia margin, attenuating the power supply side output ramp rate to limit the energy storage side response slope, so that the energy storage response action is delayed to the physical node where the power supply side output ramp curve coincides, the present application eliminates the multi-end regulation phase misalignment, suppresses the node voltage harmonic oscillation and active power overshoot, and guarantees the power grid transient stability.

Description

Technical Field

[0001] This invention relates to a multi-dimensional coupled dynamic supply and demand linkage optimization algorithm for source, grid, load and storage, belonging to the field of smart grid technology. Background Technology

[0002] Grid coordinated dispatch is typically implemented in layers according to different time scales. Traditional control methods generally assume that baseload dispatch and fast frequency regulation are independent and issue control commands separately within a preset fixed time window. In power supply systems dominated by traditional thermal power units, generator units have large rotational inertia, and the physical response times of each unit are relatively similar. Therefore, this layered control method can usually maintain the transient stability of the grid. However, with the large-scale integration of distributed renewable energy and energy storage devices using converters into the grid, the physical response times of the power generation side, grid side, load side, and energy storage side are gradually showing orders of magnitude differences. The output power of renewable energy generation has a variable... Due to the volatility of load-side electricity demand, which typically changes in a minute-by-minute cycle, energy storage converters can respond within milliseconds. When a grid node experiences a sudden increase in load and a sharp drop in renewable energy output simultaneously, the existing hierarchical dispatching method will initiate peak shaving and frequency regulation control separately. Since the inherent physical response delay time of each side is not fully considered during the control process, the actual actions corresponding to different control commands are prone to time asynchrony. At this time, the timing of the active power injected by the energy storage converter may not be synchronized with the change in the grid voltage phase, which will cause harmonic oscillations of node voltage and active power overshoot, and exacerbate voltage and power angle instability under weak grid conditions.

[0003] Simply increasing energy storage capacity or raising control frequency is insufficient to eliminate errors caused by asynchronous responses on different sides. This can actually increase equipment investment costs and potentially further deteriorate power quality at local nodes. At the control method level, to improve the transient support capability of low-inertia power systems, existing technologies attempt to achieve multi-resource coordination by adjusting unit control parameters. For example, Chinese invention patent application CN121602359A discloses a source-grid-load-storage inertia coordinated control method and device. This application calculates the rotational inertia adjustment coefficient based on the inertia power of each inertia resource and combines it with the system frequency change rate. The scheme dynamically adjusts the virtual rotational inertia parameters. However, this scheme implicitly relies on the physical premise that the converter operates without delay and is highly synchronized with the system phase. It only relies on local frequency parameter changes to adaptively adjust the control parameters. It fails to fundamentally solve the inherent physical response delay problem caused by the order-of-magnitude difference between the power supply side, grid side, load side and energy storage side. When faced with sudden changes in multi-terminal linkage, this adaptive adjustment that is out of sync with the actual response timing is very likely to cause phase misalignment between the millisecond-level action of the energy storage side and the second-level ramp-up of the power supply side. This can lead to harmonic oscillations of node voltage and active power overshoot, making it difficult to guarantee transient stability under weak grid conditions.

[0004] Therefore, how to reduce the phase deviation caused by the different physical response delays on each side, suppress node voltage harmonic oscillations and active power overshoot, and maintain dispatch stability under communication fluctuations and critical energy storage conditions are the technical problems that need to be solved. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of this invention is as follows: A multi-dimensional coupled dynamic supply and demand linkage optimization algorithm for source-grid-load-storage systems, comprising the following steps: Step S1: Obtain real-time active power output data from the power source side, node voltage data and frequency deviation data from the grid side, baseline power consumption data from the load side, and real-time state of charge data from the energy storage side. Step S2: Calculate the dynamic inertia margin for multi-terminal response hysteresis alignment based on the inherent physical response delay times of the power supply side, grid side, load side, and energy storage side. Step S3: When the dynamic inertia margin is lower than the preset safety benchmark threshold, a dynamic supply and demand linkage command sequence is generated to limit the command response slope on the energy storage side and raise the priority benchmark line for cutting off interruptible loads on the load side. Step S4: When limiting the command response slope of the energy storage side, the real-time voltage change rate is extracted based on the grid-side node voltage data, and the power output ramp rate is obtained based on the real-time active power output data of the power source side. The ratio of the absolute value of the real-time voltage change rate to the dynamic inertia margin is calculated to obtain the voltage-inertia coupling factor. The negative value of the voltage-inertia coupling factor is used as the exponent of the natural logarithm base to construct an exponential damping term. The power output ramp rate of the power source side is multiplied by the exponential damping term to obtain the exponential damping attenuation value. The exponential damping attenuation value is used as the limit boundary of the command response slope of the energy storage side, and the high-frequency response action of the energy storage side is delayed to the physical node that coincides with the power output ramp curve of the power source side.

[0006] Preferably, in step S4, when the grid-side node voltage data fluctuates and the dynamic inertia margin decreases, the exponential damping term decays exponentially with the increase of the voltage-inertia coupling factor, thus limiting the command response slope of the energy storage side to below the power output ramp rate of the power source side.

[0007] Preferably, before calculating the dynamic inertia margin in step S2, a pre-sampling fault-tolerant hedging step is included: step S301, monitoring the time interval variance between two adjacent physical parameter samplings; step S302, when the time interval variance exceeds a preset communication delay safety threshold, automatically freezing the current dynamic inertia margin update, and reusing the dynamic inertia margin calculation result of the previous effective sampling period as the current output.

[0008] Preferably, in step S301, the zero-order hold module within a historical 500ms sliding time window is used to monitor the time interval variance; the upper limit of the network jitter time span corresponding to the communication delay safety threshold is set to 20ms.

[0009] Preferably, it also includes a boundary condition nonlinear dimensionality reduction scheduling step: Step S501, continuously monitor the real-time state of charge data of the energy storage side; Step S502, when the real-time state of charge data of the energy storage side is lower than the preset first charge threshold or higher than the preset second charge threshold, cut off the calculation and modulation of the dynamic supply and demand linkage command sequence by the baseline power consumption data of the load side; Step S503, control the energy storage side to exit the bidirectional peak shaving and valley filling mode and switch to the unidirectional frequency support mode, prioritize the remaining active power throughput space of the energy storage side to smooth the frequency deviation in the voltage data and frequency deviation data of the grid side node, and stop responding to the power output fluctuation of the power supply side and the change of power demand of the load side.

[0010] Preferably, the first charge threshold is set to 15%, and the second charge threshold is set to 85%.

[0011] Preferably, step S3, which involves raising the cutoff priority baseline for interruptible loads on the load side, includes the following steps: Step S31, calculating the difference between the dynamic inertia margin and the safety baseline threshold to obtain the inertia deficit; Step S32, based on the inertia deficit, searching for the corresponding cutoff response ladder in the preset interruptible load priority matrix; Step S33, increasing the response weight of each load node within the cutoff response ladder, and when issuing the dynamic supply and demand linkage command sequence, prioritizing the cutoff of interruptible loads with response weights higher than the cutoff priority baseline, using the ladder response of load switching to offset the remaining fluctuation power of the real-time active power output on the power supply side.

[0012] Preferably, after step S4, an emergency physical isolation and safety clamping step is also included: Step S801, when the real-time voltage change rate exceeds the preset extreme voltage safety limit and the dynamic inertia margin is lower than the preset extreme inertia boundary, the power output limit of the energy storage side is locked; Step S802, a discrete cut-off control signal is sent to the load side to forcibly cut off the preset first-level interruptible load that is within the preset threshold of the physical distance from the smart grid node.

[0013] Preferably, steps S1 to S4 are run in a preset control cycle, and within each control cycle: the dynamic inertia margin and exponential damping decay value are dynamically updated based on the real-time active power output data of the power source side, the node voltage data and frequency deviation data of the grid side, the baseline power consumption data of the load side, and the real-time state of charge data of the energy storage side.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In the multi-dimensional coupling dynamic supply and demand linkage optimization of power generation, grid, load and energy storage, by comprehensively utilizing the operating data of the power source side, grid side, load side and energy storage side, and considering the inherent physical response delay time of each side, the adjustment actions of different devices are more coordinated in time. By combining the voltage changes of the grid side nodes to limit the command response slope of the energy storage side, it is possible to avoid the energy storage side acting too fast and being out of sync with the power output changes of the power source side, reduce node voltage harmonic oscillations and active power overshoot, and improve the transient stability of the power grid.

[0015] 2. By monitoring the sampling time interval before calculating the dynamic inertia margin and pausing updates when there is significant jitter in the communication network, and using the calculation results from the previous effective sampling period, the impact of communication delay, data loss, and sampling fluctuations on the calculation results can be reduced. This avoids the control system from making incorrect adjustments due to abnormal data, and ensures that the supply and demand linkage control can remain continuous and reliable even when communication is unstable.

[0016] 3. When the real-time state of charge of the energy storage side is close to the upper and lower boundaries of the allowable range, the impact of load-side power consumption changes on the linkage control is reduced, and the energy storage side is given priority to undertake frequency support tasks. This can reduce the mutual constraints between multiple objectives under critical operating conditions, reduce the complexity of control calculations, and avoid control stagnation of energy storage equipment due to responding to multiple adjustment demands at the same time. This is conducive to maintaining the stability of key operating parameters such as grid frequency. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the dynamic supply and demand linkage process of multi-terminal response lag alignment in this invention. Figure 2 This is a diagram of the multi-terminal connection structure of the communication bus of the present invention, which includes source, network, load, and storage.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A multi-dimensional coupled dynamic supply and demand linkage optimization algorithm for source-grid-load-storage includes the following steps: Step S1: Obtain real-time active power output data from the power source side, node voltage data and frequency deviation data from the grid side, baseline power consumption data from the load side, and real-time state of charge data from the energy storage side. Step S2: Calculate the dynamic inertia margin for multi-terminal response hysteresis alignment based on the inherent physical response delay times of the power supply side, grid side, load side, and energy storage side. Step S3: When the dynamic inertia margin is lower than the preset safety benchmark threshold, a dynamic supply and demand linkage command sequence is generated to limit the command response slope on the energy storage side and raise the priority benchmark line for cutting off interruptible loads on the load side. Step S4: When limiting the command response slope of the energy storage side, the real-time voltage change rate is extracted based on the grid-side node voltage data, and the power output ramp rate is obtained based on the real-time active power output data of the power source side. The ratio of the absolute value of the real-time voltage change rate to the dynamic inertia margin is calculated to obtain the voltage-inertia coupling factor. The negative value of the voltage-inertia coupling factor is used as the exponent of the natural logarithm base to construct an exponential damping term. The power output ramp rate of the power source side is multiplied by the exponential damping term to obtain the exponential damping attenuation value. The exponential damping attenuation value is used as the limit boundary of the command response slope of the energy storage side, and the high-frequency response action of the energy storage side is delayed to the physical node that coincides with the power output ramp curve of the power source side.

[0021] Preferably, in step S4, when the grid-side node voltage data fluctuates and the dynamic inertia margin decreases, the exponential damping term decays exponentially with the increase of the voltage-inertia coupling factor, thus limiting the command response slope of the energy storage side to below the power output ramp rate of the power source side.

[0022] Preferably, before calculating the dynamic inertia margin in step S2, a pre-sampling fault-tolerant hedging step is included: step S301, monitoring the time interval variance between two adjacent physical parameter samplings; step S302, when the time interval variance exceeds a preset communication delay safety threshold, automatically freezing the current dynamic inertia margin update, and reusing the dynamic inertia margin calculation result of the previous effective sampling period as the current output.

[0023] Preferably, in step S301, the zero-order hold module within a historical 500ms sliding time window is used to monitor the time interval variance; the upper limit of the network jitter time span corresponding to the communication delay safety threshold is set to 20ms.

[0024] Preferably, it also includes a boundary condition nonlinear dimensionality reduction scheduling step: Step S501, continuously monitor the real-time state of charge data of the energy storage side; Step S502, when the real-time state of charge data of the energy storage side is lower than the preset first charge threshold or higher than the preset second charge threshold, cut off the calculation and modulation of the dynamic supply and demand linkage command sequence by the baseline power consumption data of the load side; Step S503, control the energy storage side to exit the bidirectional peak shaving and valley filling mode and switch to the unidirectional frequency support mode, prioritize the remaining active power throughput space of the energy storage side to smooth the frequency deviation in the voltage data and frequency deviation data of the grid side node, and stop responding to the power output fluctuation of the power supply side and the change of power demand of the load side.

[0025] Preferably, the first charge threshold is set to 15%, and the second charge threshold is set to 85%.

[0026] Preferably, step S3, which involves raising the cutoff priority baseline for interruptible loads on the load side, includes the following steps: Step S31, calculating the difference between the dynamic inertia margin and the safety baseline threshold to obtain the inertia deficit; Step S32, based on the inertia deficit, searching for the corresponding cutoff response ladder in the preset interruptible load priority matrix; Step S33, increasing the response weight of each load node within the cutoff response ladder, and when issuing the dynamic supply and demand linkage command sequence, prioritizing the cutoff of interruptible loads with response weights higher than the cutoff priority baseline, using the ladder response of load switching to offset the remaining fluctuation power of the real-time active power output on the power supply side.

[0027] Preferably, after step S4, an emergency physical isolation and safety clamping step is also included: Step S801, when the real-time voltage change rate exceeds the preset extreme voltage safety limit and the dynamic inertia margin is lower than the preset extreme inertia boundary, the power output limit of the energy storage side is locked; Step S802, a discrete cut-off control signal is sent to the load side to forcibly cut off the preset first-level interruptible load that is within the preset threshold of the physical distance from the smart grid node.

[0028] Preferably, steps S1 to S4 are run in a preset control cycle, and within each control cycle: the dynamic inertia margin and exponential damping decay value are dynamically updated based on the real-time active power output data of the power source side, the node voltage data and frequency deviation data of the grid side, the baseline power consumption data of the load side, and the real-time state of charge data of the energy storage side.

[0029] Example 1: This example employs a multi-dimensional coupled dynamic supply-demand linkage optimization algorithm for source-grid-load-storage systems to control a smart grid in a specific area, including distributed photovoltaic systems, energy storage systems, and controllable loads. When external sensors collect grid operation data, communication delays or transient data loss can create implicit negative event records that are not directly observed. The control system executing this algorithm first executes the initial state definition procedure and retrieves data from the data acquisition terminal within a preset time window. The system collects raw voltage signals, current signals, and distributed photovoltaic output power time-series data. The data acquisition terminal has a sampling frequency of 10Hz and a measurement accuracy of 0.01%. Based on the collected data, the control system generates real-time active power output data on the power source side, node voltage data and frequency deviation data on the grid side, and simultaneously acquires baseline power consumption data on the load side and real-time state of charge data on the energy storage side. Subsequently, the above data is input into the memory buffer of the central processing unit for data preprocessing. The system identifies the injection position of implicit negative event records by scanning the data breakpoints along the time axis, and initializes the initial state parameters to a vector containing the power deviation of each node.

[0030] Before calculating the dynamic inertia margin for multi-terminal response hysteresis alignment, the CPU performs pre-sampling fault-tolerant hedging. The zero-order hold module records the time interval between two adjacent physical parameter samplings within a historical 500-millisecond sliding time window. Before the next sampled data arrives, it holds the previous valid time interval and calculates the time interval variance based on the time interval sequence within the window. The communication delay safety threshold is calibrated according to the upper limit of the 20-millisecond network jitter time span. When the time interval variance exceeds the communication delay safety threshold, the CPU freezes the update of the current dynamic inertia margin and reuses the dynamic inertia margin calculation result of the previous valid sampling period as the current output. If the threshold is not exceeded, the calculation continues using the data of the current sampling period.

[0031] In this embodiment, the inherent physical response delay time on the power supply side is 2 seconds, the inherent physical response delay time on the grid side is 0.5 seconds, the inherent physical response delay time on the load side is 3 seconds, and the inherent physical response delay time on the energy storage side is 5 milliseconds. The central processing unit calculates the ratio of the inherent physical response delay time of each end to the inherent physical response delay time of the energy storage side to obtain the corresponding dimensionless time delay coefficient. Then, the change in the real-time active power output data on the power supply side, the power amount converted from the frequency deviation data on the grid side, the fluctuation of the baseline power consumption data on the load side, and the available power limit on the energy storage side are multiplied by the corresponding time delay coefficient and summed. The weighted sum is divided by the rated total power of the entire network to complete the normalization and obtain a scalar value with a value of 0 to 1, which serves as the dynamic inertia margin for multi-end response lag alignment. The safety benchmark threshold is calibrated according to the lower limit of the dynamic inertia margin in the stable operation sample of the power grid in this region.

[0032] When the dynamic inertia margin is lower than the safety baseline threshold, the central processing unit generates a dynamic supply and demand linkage instruction sequence. This sequence includes at least the energy storage side instruction response slope limiting instruction and the load side interruptible load shedding priority baseline line adjustment instruction. Each instruction uses the same control cycle identifier and determines the execution time based on the inherent physical response delay time of the power supply side, grid side, load side and energy storage side, so that the actual responses of each end are aligned in time.

[0033] When limiting the command response slope of the energy storage side, the central processing unit (CPU) extracts the real-time voltage change rate based on the grid-side node voltage data and sampling intervals at adjacent sampling times, and obtains the power supply output ramp rate based on the power supply real-time active power output data and sampling intervals at adjacent sampling times. The absolute value of the real-time voltage change rate is divided by the dynamic inertia margin to obtain the voltage-inertia coupling factor. The negative value of the voltage-inertia coupling factor is taken as the exponent of the natural logarithm base to construct an exponential damping term. The CPU multiplies the power supply output ramp rate by the exponential damping term to obtain the exponential damping attenuation value, and uses this exponential damping attenuation value as the limit boundary of the command response slope of the energy storage side. When the absolute value of the original command response slope of the energy storage side exceeds this limit boundary, its absolute value is limited to the exponential damping attenuation value, while keeping the original power adjustment direction unchanged. When the fluctuation of the grid-side node voltage data increases or the dynamic inertia margin decreases, the voltage-inertia coupling factor increases, and the exponential damping term decreases accordingly, so that the command response slope of the energy storage side remains below the power supply output ramp rate.

[0034] Based on the de-identified data, the central processing unit calls the implicit negative event record injection correction algorithm. According to the observable active power change trend and historical load characteristic matrix within adjacent sampling periods, it calculates the power compensation amount injected from the implicit negative event records. The specific calculation formula is as follows: ,in, The amount of power compensation injected for implicit negative event recording. This is a dimensionless smoothing adjustment coefficient calibrated based on historical operating data; The current sampling time The active power measurement value, The previous sampling time The active power measurement value, This represents the total number of sampling points; As a time decay factor, the central processing unit will The power adjustment command is superimposed on the corresponding power adjustment command in the dynamic supply and demand linkage command sequence to compensate for the power deficit caused by implicit negative event records.

[0035] Steps S1 to S4 are executed cyclically according to a preset control cycle. In each control cycle, the central processing unit reacquires real-time active power output data from the power source side, node voltage data and frequency deviation data from the grid side, baseline power consumption data from the load side, and real-time state of charge data from the energy storage side, and updates the dynamic inertia margin and exponential damping decay value of the multi-terminal response lag alignment accordingly. When an abnormal sampling time interval occurs, the dynamic inertia margin calculation result of the previous effective sampling cycle is maintained according to the pre-sampling fault tolerance offset rule.

[0036] To verify the effect of the implicit negative event record injection correction algorithm, a control group without executing the correction algorithm was set up under the same operating conditions. After running continuously for 72 hours, the test group using the algorithm of this embodiment limited the bus voltage fluctuation frequency to within 0.1Hz. Because the control group did not perform dynamic compensation for the implicit negative event records, the number of energy storage system actions increased by 45%, and the bus voltage oscillated above the preset threshold. The active power supply and demand of the test group maintained dynamic balance, and each control terminal was in a stable adjustment state.

[0037] Example 2: When voltage fluctuations caused by distributed photovoltaic grid connection and external random thermal convection disturbances work together, the input power of the electric heating system in the solid electrolyte composite film drying channel fluctuates accordingly. As a controllable industrial terminal on the load side, the electric heaters in each zone are connected to the dispatch controller of the smart grid node through a communication bus. After the smart grid node activates the source-grid-load-storage multi-dimensional coupling dynamic supply and demand linkage optimization algorithm due to photovoltaic power output fluctuations or frequency deviations, the dispatch controller sends the load shedding command or power adjustment command in the dynamic supply and demand linkage command sequence to the adjustable analog power supply, and participates in the active power balance of the grid by changing the input power of the electric heater.

[0038] The experiment employed a physical experimental platform comprising a multi-channel temperature acquisition matrix and an adjustable analog power supply. The multi-channel temperature acquisition matrix had a measurement range of 0℃ to 150℃, a resolution of 0.01℃, a measurement accuracy of ±0.05℃, and a sampling frequency of 20Hz. The adjustable analog power supply had a temperature control response time of less than 10ms and a power adjustment resolution of 0.1W. The temperature sampling period was determined based on the signal response lag time of the thermocouples in the drying tunnel and the data processing capability of the processor. Due to the random thermal convection disturbance caused by the high-speed fan in the drying tunnel during the experiment, the temperature sampling period was set to 0.50s to obtain temperature changes in a timely manner.

[0039] The experiment was conducted with an experimental group and a control group. The control group used a conventional proportional-integral-derivative control algorithm without heating power increment limiting and over-temperature alarm shutdown functions. The experimental group used discrete control logic, based on the target set temperature and the current temperature sample value. The difference between them and the proportional adjustment coefficient Calculate heating power increment And compare the calculation results with the maximum adjustment threshold. and minimum adjustment threshold In comparison, the calculation result is higher than At that time, the actual output is taken The calculation result is lower than At that time, the actual output is taken If the calculation result falls between the two, output the result that is calculated.

[0040] During operation in Zone 2 of the drying tunnel, the processor superimposed Gaussian white noise with a signal-to-noise ratio of 20dB onto the test signal. Simultaneously, the adjustable analog power supply introduced 50Hz power frequency interference harmonics with an amplitude of ±10% of the rated voltage. Random thermal convection disturbances were applied through damper adjustment. Three sampling points under different operating conditions were continuously selected. Both groups used the same initial input at each sampling point. Sampling point 1... Both are 71.5℃, and the processor is adjusted according to the target set temperature of 75℃ and the proportional adjustment coefficient. After calculation, the two groups Both were 17.5W, which is within the limit range. The test group was output at 17.5W.

[0041] Sampling point 2 simulates the extremely cold low-temperature start-up condition, with two groups of samples. Both are -35℃, calculated The experimental group output 550W, while the control group directly outputs 550W; the experimental group's calculated results were compared with... After comparison, the actual output was limited to 500W, and sampling point 3 simulated an external high-temperature heat flow overload condition. The two sets of... Both are 180℃, calculated The output was -525W, while the control group directly output -525W; the experimental group's calculated results were compared with... After comparison, the actual output is limited to -500W.

[0042] The highest safety alarm threshold temperature during the over-temperature shutdown test. Set to 90℃, under continuous external high heat source interference, the multi-channel temperature acquisition matrix measured three consecutive temperatures in the second zone of the drying tunnel. The temperatures were 91.2℃, 93.5℃, and 95.8℃, respectively, all higher than [the specified values]. The test group processor immediately shut off the heating power output, making The total heating power dropped to 0W and an abnormal termination signal was sent to the external controller. The control group did not perform continuous sampling over-temperature shutdown and continued to output regulating power, causing the temperature in the drying channel zone 1 and zone 2 to continue to rise.

[0043] In the proportional adjustment coefficient In the boundary test, When the setting is 1W / ℃, which is lower than the preset range, the electric heating system failed to compensate for sudden changes in distributed photovoltaic output and random thermal convection disturbances in a timely manner. Fluctuations occurred in the acetonitrile solvent evaporation process, and the pinhole defect rate on the composite film surface reached 12.5%. When set to 25W / ℃, which is higher than the preset range, the actual output is... and The repeated switching between these modes caused periodic fluctuations in the drying tunnel temperature field, ranging from ±6.8℃, resulting in wrinkles on the film surface. When set to 5W / ℃, the temperature fluctuation amplitude remains within ±0.3℃.

[0044] In the external thermal convection disturbance gradient test, the instantaneous change amplitudes of low-intensity, medium-intensity, and high-intensity disturbances were ±2℃, ±5℃, and ±8℃, respectively. After the same discrete control processing, the maximum temperature deviations in the drying tunnel were 0.12℃, 0.28℃, and 0.45℃, respectively, and the corresponding film product qualification rates were 99.2%, 98.5%, and 96.8%, respectively. As the disturbance amplitude increased, the maximum temperature deviation gradually increased, and the film product qualification rate decreased accordingly.

[0045] During the continuous operation of the multi-dimensional coupling dynamic supply and demand linkage optimization algorithm of source-grid-load-storage, the scheduling controller converts the load-side power adjustment command into a pulse width modulation signal, and adjusts the input current of the electric heater through a solid-state relay. After the slurry on the surface of the stainless steel foil is dried, the thickness of the composite film is maintained at 25 μm. The characteristic diffraction peak of the lithium lanthanum zirconium oxide ceramic phase is measured by X-ray diffractometer at a diffraction angle of 30.80°. The coating speed is maintained at 0.50 m / min, and no pinhole defects are observed on the surface of the solid electrolyte composite film output from the production line.

[0046] Example 3: The current power grid dispatch center performs dynamic coordinated control of multiple energy storage converters in the region. Each energy storage converter is in charging operation. The controller collects active power, reactive power, grid-side node voltage data, frequency deviation data, and real-time state of charge data of each smart grid node in real time through the communication bus.

[0047] The controller discretizes the collected voltage and current signals according to a preset sampling period to form a discrete raw data array containing multiple sampling points. Then, it performs a fast Fourier transform on the array to obtain the spectral distribution data and identifies the peak values ​​whose amplitudes are greater than a preset energy threshold. The preset energy threshold is set to 1.5% of the fundamental component amplitude, and the preset number threshold is set to 3. When there are 3 or more harmonic peak values ​​in the spectral distribution data whose amplitudes exceed 1.5% of the fundamental component amplitude, the controller determines that the total harmonic distortion of the current power grid operation exceeds the set range and initiates subsequent control.

[0048] After confirming that the harmonics exceed the standard, the controller calculates the dynamic adjustment amount based on the real-time state of charge data of the energy storage side and the available capacity of the energy storage converter. The dynamic adjustment amount is sent to the drive circuit of the inverter bridge through the control port to adjust the switching frequency and output voltage phase of the inverter bridge in order to suppress harmonics and smooth out active power fluctuations.

[0049] When the energy storage converter participates in active power ramping, the controller obtains the exponential damping attenuation value based on the real-time voltage change rate, dynamic inertia margin, and power supply ramping rate. This value is used as the limit boundary of the command response slope of the energy storage side. When the original command response slope of the energy storage side exceeds this limit boundary, the controller restricts its power increase slope and correspondingly postpones the effective time of the secondary dispatch command. This delays the high-frequency response action of the energy storage side until it coincides with the power supply ramping curve. This physical node is the moment when the power increase curve of the energy storage side coincides with the actual power supply ramping curve on the time axis after the command response slope of the energy storage side has been running according to the limited slope.

[0050] During the aforementioned secondary scheduling process, the energy storage converter retains its instantaneous primary frequency regulation and voltage support capabilities within 5 milliseconds. The delay limitation on the high-frequency response action of the energy storage side only applies to the secondary scheduling stage in which the energy storage side coordinates with the power supply side to perform large-capacity active power ramping. It does not limit the primary frequency regulation and voltage support actions. The actual physical ramping time of the generator set on the power supply side is 2 to 5 seconds. By limiting the command response slope of the energy storage side and adjusting the effective time of the secondary scheduling command, the controller ensures that the power increase curve of the energy storage side coincides with the power output ramping curve of the power supply side at the corresponding physical node, thus avoiding the premature injection of excessive active power by the energy storage side.

[0051] When the real-time voltage change rate exceeds the extreme voltage safety limit and the dynamic inertia margin is lower than the extreme inertia boundary, the controller performs emergency physical isolation and safety clamping. The controller first locks the upper limit of the power output on the energy storage side and stops increasing the active power output of the energy storage converter. Then, according to the physical distance between the smart grid node and each load node, it selects the first-level interruptible loads whose physical distance is within the preset threshold and sends discrete disconnection control signals to the load side to forcibly disconnect the selected first-level interruptible loads.

[0052] The processor cyclically performs data acquisition, spectrum analysis, dynamic adjustment calculation, energy storage side command response slope limitation, and emergency physical isolation and safety clamping according to the preset control cycle. Within each control cycle, the controller updates the dynamic inertia margin and exponential damping decay value based on the real-time acquired data, so that the energy storage converter continuously adjusts the power output according to the current grid operating status.

[0053] Example 4: This example combines Figures 1 to 2 This paper describes a multi-dimensional coupled dynamic supply and demand linkage optimization algorithm for source-grid-load-storage systems, such as... Figure 1As shown, step S1 involves acquiring real-time active power output data from the power source side, node voltage data and frequency deviation data from the grid side, baseline power consumption data from the load side, and real-time state of charge data from the energy storage side; step S2 involves calculating the dynamic inertia margin for multi-terminal response lag alignment based on the inherent physical response delay times of the power source side, grid side, load side, and energy storage side; step S3 involves generating a dynamic supply and demand linkage command sequence when the dynamic inertia margin is lower than a preset safety benchmark threshold, limiting the command response slope of the energy storage side, and raising the priority benchmark for disconnecting interruptible loads on the load side; step S4 involves... The real-time voltage change rate is extracted from the grid-side node voltage data, and the power output ramp rate is obtained based on the real-time active power output data of the power source side. The ratio of the absolute value of the real-time voltage change rate to the dynamic inertia margin is calculated to obtain the voltage-inertia coupling factor. The negative value of the voltage-inertia coupling factor is used as the exponent of the natural logarithm base to construct an exponential damping term. The power output ramp rate of the power source side is multiplied by the exponential damping term to obtain the exponential damping attenuation value. The exponential damping attenuation value is used as the limit boundary of the command response slope of the energy storage side, and the high-frequency response action of the energy storage side is delayed to the physical node that coincides with the power output ramp curve of the power source side.

[0054] like Figure 2 As shown, the smart grid node, data acquisition module, power supply side, energy storage system, and load-side controllable industrial terminal are all connected to the communication bus. The smart grid node includes a central processing unit, a zero-order hold module, and a memory buffer. The data acquisition module includes external sensors and a data acquisition terminal. The power supply side includes distributed power sources. The energy storage system includes an energy storage converter, an inverter bridge drive circuit, and a lithium battery pack. The load-side controllable industrial terminal includes a multi-channel temperature acquisition matrix, an adjustable analog power supply, an electric heater, and a solid-state relay.

[0055] Example 5: In this example, when the control system is first connected to a specific regional power grid, a pre-deployment calibration is performed on the multi-dimensional coupling dynamic supply and demand linkage optimization algorithm of source-grid-load-storage. The data acquisition module collects active power data of the tie line and real-time active power output data of the power source side with a sampling period of 100ms. The collected results are converted into state vectors and input into the control chip. The control chip calculates the active power deviation value of the tie line. Simultaneously read real-time state of charge data from the energy storage side. Then, based on the two calculations, the active power is dynamically adjusted to adjust the step size. : ,in, For dynamic adjustment of active power step size, The active power deviation value of the tie line is indicated by the subscript. Indicates a connecting line; This is real-time state of charge data for the energy storage side, with values ​​ranging from 0 to 1, indicated by subscripts. Indicates the state of charge; The rated power of the energy storage system is indicated by the subscript. Indicates the rated value; The first adjustment coefficient is set to 0.12; The second adjustment coefficient is set to 0.05.

[0056] When the real-time state of charge on the energy storage side is below 15%, the lithium battery pack enters the deep discharge region, increasing the risk of plate overvoltage polarization degradation; when it is above 85%, the lithium battery pack enters the overcharge high-voltage region, intensifying electrochemical side reactions. Therefore, in this embodiment, the first charge threshold is set to 15%, and the second charge threshold is set to 85%. The first and second adjustment coefficients are determined based on the sensitivity analysis of the rated power of the energy storage system and the maximum overload capacity of the tie line. In the experiment, power overshoot oscillation occurs when the first adjustment coefficient is greater than 0.18, and adjustment lag occurs when it is less than 0.08. After setting the first adjustment coefficient to 0.12 and the second adjustment coefficient to 0.05, the system converges to a steady state within 3 control cycles.

[0057] When the real-time state of charge data of the energy storage side is lower than the first charge threshold or higher than the second charge threshold, the control chip executes the boundary condition nonlinear dimensionality reduction scheduling step. In the subsequent calculation of the dynamic supply and demand linkage instruction sequence, the calculation and modulation of the baseline power consumption data of the load side is cut off. At the same time, the control chip controls the energy storage side to exit the bidirectional peak shaving and valley filling mode and switch to the unidirectional frequency support mode. Based on the rated power of the energy storage system and the currently occupied active power throughput, the remaining active power throughput space is determined. This space is preferentially allocated to smoothing frequency deviation and stops responding to power supply side output fluctuations and load side power demand changes.

[0058] When the real-time state of charge data on the energy storage side is between the first and second charge thresholds, the control chip will dynamically adjust the active power step size. The command is converted to an action level and sent to the control terminal of the energy storage inverter. By adjusting the power throughput of the energy storage inverter, the active power deviation of the tie line is mitigated. When the load exceeds 50kW for more than 200ms, the control chip activates the load-side discrete load reduction rule and reuses the data collected from the existing measurement and control interface to adjust the switching status of each load node.

[0059] In this embodiment, the interruptible load priority matrix is ​​a multi-row, four-column data matrix. Each row records the load node number, power response level, unit cut-off capacity, and recovery delay time. After the control chip obtains the inertia loss, it divides the inertia loss by the baseline single-level cut-off power and rounds the result up to determine the number of the cut-off response ladder. Then, it searches for the load node corresponding to the number of the cut-off response ladder in the interruptible load priority matrix, increases the response weight of each load node in the cut-off response ladder, and raises the cut-off priority baseline of the interruptible load on the load side.

[0060] When issuing dynamic supply and demand linkage command sequences, the control chip selects interruptible loads with response weights higher than the cutoff priority baseline, generates load cutoff commands in descending order of response weights, and executes them sequentially. The stepped response formed by load cutoff offsets the remaining fluctuating power of the real-time active power output on the power supply side. When the inertia loss is eliminated and the active power deviation of the tie line returns to normal, each load node resumes operation sequentially when the recovery delay time recorded in the interruptible load priority matrix is ​​reached, forming a stepped response for load switching.

[0061] Example 6: This example illustrates the operation of the source-grid-load-storage multi-dimensional coupling dynamic supply and demand linkage optimization algorithm within a continuous control cycle. The smart grid in a specific area includes distributed power sources, energy storage systems, and interruptible loads. The devices on each side are connected to the controller of the smart grid node through a communication bus.

[0062] Within each control cycle, the controller acquires real-time active power output data from the power source, node voltage and frequency deviation data from the grid, baseline power consumption data from the load, and real-time state of charge data from the energy storage. Before updating the dynamic inertia margin, the zero-order hold module uses a historical 500ms sliding time window to record the time interval between two adjacent physical parameter samplings and calculates the time interval variance. When the time interval variance exceeds the communication delay safety threshold, the controller freezes the update of the current dynamic inertia margin and reuses the dynamic inertia margin calculation result from the previous valid sampling cycle as the current output. The upper limit of the network jitter time span corresponding to the communication delay safety threshold is 20ms.

[0063] When the sampling status is normal, the controller calculates the dynamic inertia margin of multi-terminal response lag alignment based on the inherent physical response delay time of the power supply side, grid side, load side and energy storage side. When the dynamic inertia margin is lower than the safety benchmark threshold, the controller generates a dynamic supply and demand linkage command sequence, limits the command response slope of the energy storage side, and raises the priority benchmark line for disconnecting interruptible loads on the load side.

[0064] When limiting the command response slope of the energy storage side, the controller extracts the real-time voltage change rate based on the grid-side node voltage data and obtains the power supply side output ramp rate based on the real-time active power output data of the power supply side. The absolute value of the real-time voltage change rate is divided by the dynamic inertia margin to obtain the voltage-inertia coupling factor. The negative value of this factor is then taken as the exponent of the natural logarithm base to construct an exponential damping term. The controller multiplies the power supply side output ramp rate by the exponential damping term to obtain the exponential damping attenuation value, which is used as the limit boundary of the command response slope of the energy storage side. When the original command response slope of the energy storage side exceeds this boundary, the controller limits its change amplitude according to the exponential damping attenuation value, so that the high-frequency response action of the energy storage side is delayed until the physical node coincides with the power supply side output ramp curve.

[0065] The controller continuously monitors the real-time state of charge data of the energy storage side within the same control cycle. When the data is lower than the first charge threshold by 15% or higher than the second charge threshold by 85%, the controller cuts off the calculation and modulation of the dynamic supply and demand linkage command sequence by the baseline power consumption data of the load side, controls the energy storage side to exit the bidirectional peak shaving and valley filling mode and switch to the unidirectional frequency support mode, uses the remaining active power throughput space of the energy storage side to smooth out frequency deviation, and stops responding to power supply side output fluctuations and load side power demand changes.

[0066] When the real-time state of charge data on the energy storage side is between the first charge threshold and the second charge threshold, the controller obtains the inertia loss based on the difference between the dynamic inertia margin and the safety benchmark threshold, and searches for the corresponding cut-off response ladder in the interruptible load priority matrix. The controller increases the response weight of each load node in the cut-off response ladder. When issuing the dynamic supply and demand linkage command sequence, it prioritizes the cut-off of interruptible loads with response weights higher than the cut-off priority benchmark, and uses the ladder response of load switching to offset the remaining fluctuation power of the real-time active power output on the power supply side.

[0067] When the real-time voltage change rate exceeds the extreme voltage safety limit and the dynamic inertia margin is lower than the extreme inertia boundary, the controller locks the upper limit of power output on the energy storage side and sends a discrete cut-off control signal to the load side to forcibly cut off the first-level interruptible load whose physical distance from the smart grid node is within the preset threshold.

[0068] After the current control cycle ends, the controller retains the effective dynamic inertia margin and exponential damping decay value for this cycle, and reacquires the operating data of each side in the next control cycle. When the sampling status is normal, the controller updates the dynamic inertia margin and exponential damping decay value based on the newly acquired data; when the sampling status is abnormal, the pre-sampling fault tolerance offset continues to be executed to keep the dynamic supply and demand linkage process between each control cycle continuous.

[0069] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-dimensional coupled dynamic supply and demand linkage optimization algorithm for source-grid-load-storage systems, characterized in that, Includes the following steps: Step S1: Obtain real-time active power output data from the power source side, node voltage data and frequency deviation data from the grid side, baseline power consumption data from the load side, and real-time state of charge data from the energy storage side. Step S2: Calculate the dynamic inertia margin for multi-terminal response hysteresis alignment based on the inherent physical response delay times of the power supply side, grid side, load side, and energy storage side. Step S3: When the dynamic inertia margin is lower than the preset safety benchmark threshold, a dynamic supply and demand linkage command sequence is generated to limit the command response slope on the energy storage side and raise the priority benchmark line for cutting off interruptible loads on the load side. Step S4: When limiting the command response slope of the energy storage side, the real-time voltage change rate is extracted based on the grid-side node voltage data, and the power output ramp rate is obtained based on the real-time active power output data of the power source side. The ratio of the absolute value of the real-time voltage change rate to the dynamic inertia margin is calculated to obtain the voltage-inertia coupling factor. The negative value of the voltage-inertia coupling factor is used as the exponent of the natural logarithm base to construct an exponential damping term. The power output ramp rate of the power source side is multiplied by the exponential damping term to obtain the exponential damping attenuation value. The exponential damping attenuation value is used as the limit boundary of the command response slope of the energy storage side, and the high-frequency response action of the energy storage side is delayed to the physical node that coincides with the power output ramp curve of the power source side.

2. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, In step S4, when the grid-side node voltage data fluctuates and the dynamic inertia margin decreases, the exponential damping term decays exponentially with the increase of the voltage-inertia coupling factor, limiting the command response slope of the energy storage side to below the power output ramp rate of the power source side.

3. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, Before calculating the dynamic inertia margin in step S2, a pre-sampling fault-tolerant hedging step is also included: step S301, monitoring the time interval variance between two adjacent physical parameter samplings; step S302, when the time interval variance exceeds the preset communication delay safety threshold, automatically freezing the current dynamic inertia margin update, and reusing the dynamic inertia margin calculation result of the previous effective sampling period as the current output.

4. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 3, characterized in that, In step S301, the zero-order hold module within a historical 500ms sliding time window is used to monitor the time interval variance; the upper limit of the network jitter time span corresponding to the communication delay safety threshold is set to 20ms.

5. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, It also includes a boundary condition nonlinear dimensionality reduction scheduling step: Step S501, continuously monitor the real-time state of charge data of the energy storage side; Step S502, when the real-time state of charge data of the energy storage side is lower than the preset first charge threshold or higher than the preset second charge threshold, cut off the calculation and modulation of the dynamic supply and demand linkage command sequence by the baseline power consumption data of the load side; Step S503, control the energy storage side to exit the bidirectional peak shaving and valley filling mode and switch to the unidirectional frequency support mode, prioritize the remaining active power throughput space of the energy storage side to smooth the frequency deviation in the voltage data and frequency deviation data of the grid side node, and stop responding to the power output fluctuation of the power supply side and the change of power demand of the load side.

6. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 5, characterized in that, The first charge threshold is set to 15%, and the second charge threshold is set to 85%.

7. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, Step S3, which involves raising the priority baseline for interruptible load shedding on the load side, includes the following steps: Step S31, calculating the difference between the dynamic inertia margin and the safety baseline threshold to obtain the inertia deficit; Step S32, based on the inertia deficit, searching for the corresponding shedding response ladder in the preset interruptible load priority matrix; Step S33, increasing the response weight of each load node within the shedding response ladder, and prioritizing the shedding of interruptible loads with response weights higher than the shedding priority baseline when issuing dynamic supply and demand linkage command sequences, using the ladder response of load shedding to offset the remaining fluctuation power of the real-time active power output on the power supply side.

8. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, After step S4, there are also emergency physical isolation and safety clamping steps: Step S801, when the real-time voltage change rate exceeds the preset extreme voltage safety limit and the dynamic inertia margin is lower than the preset extreme inertia boundary, the power output limit of the energy storage side is locked; Step S802, a discrete cut-off control signal is sent to the load side to forcibly cut off the preset first-level interruptible loads that are within the preset threshold of the physical distance from the smart grid node.

9. The source-grid-load-storage multi-dimensional coupled dynamic supply-demand linkage optimization algorithm according to claim 1, characterized in that, Steps S1 to S4 are run in a preset control cycle. Within each control cycle, the dynamic inertia margin and exponential damping decay value are dynamically updated based on the real-time active power output data of the power source side, the node voltage data and frequency deviation data of the grid side, the baseline power consumption data of the load side, and the real-time state of charge data of the energy storage side.

Images