Intelligent scheduling and parallel cooperative control system for peak clipping and valley filling of lithium battery pack

Abstract

The invention discloses an intelligent scheduling and parallel cooperative control system for peak load shifting of a lithium battery pack, particularly relates to the field of energy storage management systems, and is used for solving the problem of a breaking point between plan generation and execution response. The method comprises the following steps: receiving and aligning a station-level charging and discharging curve, the state and temperature trend of each battery cluster and the working domain of each converter through a time mark model-in module to generate parallel feasible domain mapping, and a projection consensus module transcribes a peak clipping and valley filling target and combines a flow distribution stable boundary and a protection boundary to generate a plan candidate set; the shaping synchronization module executes step shaping and phase synchronization to output a station level plan according to a switching position, and the flow distribution alignment module calculates a prior flow distribution track and balance time slot issuing execution according to the station level plan and a parallel feasible region. And a rolling correction module collects execution deviation and temperature lag information to iteratively reestimate parallel feasible region mapping and a prior flow distribution track to generate an update plan.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy storage management systems, and more particularly, to a lithium battery pack peak shaving and valley filling intelligent scheduling and parallel collaborative control system. BACKGROUND

[0002] In the current energy management system, the energy storage station at the park and substation has become the core component of power peak shaving and valley filling. These stations are usually equipped with large-scale lithium battery packs to cope with power load fluctuations and time-of-use price changes. In the prior art, the scheduling platform mainly generates a station-level charge-discharge curve based on the real-time load curve and the time-of-use price mechanism. This curve is used as a power instruction sequence to be sent to the converter and parallel battery cluster to realize the daily operation of charge filling and discharge peak shaving. At the execution level, a group of parallel battery clusters and multiple converters respond to the instructions in coordination. The state of charge, equivalent internal resistance and temperature trajectory of each battery cluster often differ significantly, and the balancing capability and protection boundaries also differ due to manufacturing processes, aging levels and environmental factors. This parallel configuration aims to improve the overall capacity and response speed of the system, but the current situation of the prior art is that scheduling mainly relies on macro planning at the station level, ignoring the dynamic coupling of bottom-level physical constraints, resulting in actual power trajectories being influenced by current distribution, heat dissipation paths and protection interlocks. In terms of defects, the current system usually uses static flow allocation rules or simple feedback control, which cannot real-time align the planned timing with physical boundaries. For example, when the load peak and valley switch, the battery cluster with higher voltage or lower impedance preferentially absorbs current, forming inter-loop current between branches, and accompanied by heat hot spot accumulation; in addition, the balancing module is often activated only after the event, causing parameter lag and frequent intervention of compensation actions. These defects are particularly prominent in the power grid environment with high penetration of renewable energy sources, such as photovoltaic or wind power fluctuations leading to sudden load changes, further amplifying system response delays and energy losses, limiting the economic benefits and reliability of energy storage stations.

[0003] The technical problem mainly lies in the disconnection between the plan generation and the response execution, which is caused by the bias of the station-level scheduling towards curve smoothing and benefit optimization, while ignoring the real constraints at the parallel level. Specifically, in the actual application scenario of the energy storage station in the park or substation, when the scheduling platform generates the charge-discharge curve according to the load trend and the electricity price rhythm and issues instructions, the parallel battery clusters and the converter on the execution side need to cooperatively handle the power demand. However, due to the inconsistency of the state of charge, the equivalent internal resistance and the temperature trajectory of each battery cluster, as well as the difference in open-circuit voltage and the thermal accumulation effect, the problem occurs when the planned trajectory changes rapidly or switches, for example, during the peak-valley transition. The battery cluster with higher voltage or lower impedance will preferentially carry the current, leading to the formation of circulating current among branches, accompanied by the superposition of thermal hotspots. In order to catch up with the planned curve, the system constantly modifies the flow distribution rules, causing the alternating occurrence of compensation actions and protection actions. This disconnection is particularly pronounced in high-frequency load fluctuation scenarios, such as the sharp increase in peak-time electricity consumption in industrial parks or the response of substations to sudden power grid instability. The misalignment of the planned timing and physical constraints will amplify the response deviation and parameter lag, and the balancing action will often be delayed, failing to timely alleviate the coupling effect. The consequences include the continuous deviation of the station-level curve and the execution result, the compression of the apparent available capacity, the decrease in the stability of the benefit leading to the accumulation of economic losses, the exacerbation of the health status differentiation causing uneven battery cluster aging, ultimately shortening the system life and increasing the maintenance cost, and restricting the large-scale promotion of the energy storage station in the smart grid.

[0004] To solve the above problems, a technical solution is provided. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a lithium battery pack peak clipping and valley filling intelligent scheduling and parallel cooperative control system. The time scale into the mold module receives and aligns the station-level charge-discharge curve, the state and temperature trend of each battery cluster, and the working domain of each converter, generates a parallel feasible region mapping on a unified time axis, the projection consensus module converts the peak clipping and valley filling target into an executable target sequence within the parallel feasible region mapping, and combines the flow distribution stability boundary and the protection boundary to generate a plan candidate set. The shaping and synchronization module performs step shaping and phase synchronization processing on the switching position of the plan candidate set to output a station-level plan that meets the constraints of the feasible region and the stability boundary. The flow distribution alignment module calculates the prior flow distribution trajectory and the balancing time slot for each battery cluster according to the station-level plan and the parallel feasible region, and issues them to the converter and the balancing unit in time sequence for execution. The rolling correction module collects the execution deviation and temperature lag information to iteratively optimize the parallel feasible region mapping and the prior flow distribution trajectory to generate an updated next time period plan and flow distribution trajectory. The unified modeling domain of the plan timing and the physical constraints is jointly constrained and decided, thereby solving the problems raised in the background art.

[0006] To achieve the above-mentioned purposes, the present application provides the following technical solutions: The lithium battery pack peak shaving intelligent scheduling and parallel collaborative control system comprises: The time scale input module receives and aligns the station-level charge-discharge curve, the state of each battery cluster and the temperature trend, and the working domain of each converter, generates a parallel feasible region mapping on a unified time axis, and projects the consensus module. The projection consensus module: in the parallel feasible region mapping, the peak shaving target is converted into an executable target sequence, and the flow distribution stability boundary and protection boundary are combined to generate a plan candidate set. The shaping synchronization module: for the switching position of the plan candidate set, step shaping and phase synchronization processing are performed, and the station-level plan satisfying the feasible region and stability boundary constraints is output. The flow distribution alignment module: according to the station-level plan and the parallel feasible region, the prior flow distribution trajectory and the equalization time slot are calculated for each battery cluster, and are sequentially issued to the converter and the equalization unit for execution. The rolling correction module: collect the execution deviation and temperature lag information, and re-estimate the parallel feasible region mapping and the prior flow distribution trajectory in an iterative optimization manner to generate an updated next period plan and flow distribution trajectory.

[0007] Further, the time scale input module receives the station-level charge-discharge curve represented as a power and time sequence, the state of each battery cluster including a state of charge sequence and an equivalent internal resistance sequence, the temperature trend as a temperature sequence, and the working domain of each converter represented as a power upper limit sequence and a power lower limit sequence. The sequences are mapped to a unified time axis based on the least common multiple sampling interval through a dedicated interface into the module buffer, and the alignment process uses a linear interpolation method.

[0008] Further, the time scale input module generates a parallel feasible region mapping on a unified time axis. The power range set that can be supported by all battery clusters and converters is calculated through constraint aggregation. First, the feasible power range of each battery cluster including the lower limit and the upper limit is calculated, then the intersection of all battery cluster ranges and the working domain range of each converter is calculated, and further intersection with the station-level charge-discharge curve constraint is calculated.

[0009] Further, the projection consensus module converts the peak shaving target into an executable target sequence within the parallel feasible region mapping. The peak shaving target is extracted as a power target sequence through boundary projection calculation. For each time point, if it exceeds the lower bound or upper bound of the parallel feasible region mapping, it is projected to the nearest boundary. The minimum maximum operation is used to limit the power target sequence value between the lower bound and the upper bound of the parallel feasible region mapping.

[0010] Further, the projection consensus module merges the flow allocation stable boundary and the protection boundary into the executable target sequence, defines the flow allocation stable boundary as a current allocation margin sequence based on the allocation limit of the equivalent internal resistance sequence of each battery cluster by constraint superposition update, defines the protection boundary as a safety threshold sequence based on the joint limit of the temperature sequence and the state of charge sequence, and then updates the executable target sequence to the minimum value of the original sequence multiplied by the current allocation margin sequence and the safety threshold sequence.

[0011] Further, the projection consensus module generates a plan candidate set based on the merged executable target sequence, generates a plurality of candidate sequences by introducing a margin disturbance to the merged executable target sequence at each time point to calculate a set of alternative power sequence sets, and forms a final plan candidate set by verifying that all candidate sequences do not exceed the parallel feasible region mapping.

[0012] Further, the shaping synchronization module performs step shaping processing on the switching position of the plan candidate set, determines the switching position by differential scanning to calculate a power difference sequence at adjacent time points and compare it with a preset jump threshold to mark the time points, defines a shaping window for the switching position, applies linear gradient to update the candidate sequence, and verifies that the updated sequence is still located within the parallel feasible region mapping.

[0013] Further, the shaping synchronization module performs phase synchronization processing on the shaped plan candidate set, realizes phase consistency by time shift adjustment to calculate the time shift value of the phase offset sequence of each sequence relative to the reference sequence, applies time shift correction and merges the flow allocation stable boundary and the protection boundary, and ensures that the synchronized sequence satisfies the stable boundary constraint by product constraint.

[0014] Further, the flow allocation alignment module calculates the prior flow allocation trajectory of each battery cluster according to the station-level plan and the parallel feasible region mapping, generates a distribution proportion equal to the reciprocal of the equivalent internal resistance sequence divided by the total admittance by impedance proportional distribution, calculates the prior flow allocation trajectory as the station-level plan power value multiplied by the distribution proportion, adjusts the trajectory to incorporate the state of charge sequence, and verifies that the trajectory does not exceed the range of a single battery cluster in the parallel feasible region mapping.

[0015] Further, the flow allocation alignment module calculates the equalization time slot of each battery cluster according to the prior flow allocation trajectory and the parallel feasible region mapping, compares the maximum deviation between the prior flow allocation trajectory and the state of charge sequence by deviation scanning to mark the starting time point when the maximum deviation exceeds the equalization threshold, scans a continuous time period to form a list of equalization time slot windows, incorporates the temperature sequence to define the end point of the window, and ensures that the time slot does not overlap the fast-changing segment of the station-level plan and is located in the low-power area of the parallel feasible region mapping.

[0016] The technical effects and advantages of the lithium battery pack peak clipping and valley filling intelligent scheduling and parallel collaborative control system are as follows: The application generates parallel feasible region mapping through the time scale mold module, transcribes the target and merges the boundary to generate the plan candidate set through the projection consensus module, smoothes the switching position output station level plan through the shaping synchronization module, calculates the prior flow trajectory and the balance time slot through the flow alignment module, and updates the plan through the rolling correction module, forms a closed-loop cooperative control, places the planning time sequence and physical constraints in the unified modeling domain, and realizes common decision-making; effectively solves the response deviation caused by internal resistance difference, parameter deviation and circulating current formation problem caused by thermal lag, reduces compensation action and protection alternation, improves station level curve tracking accuracy, compresses apparent available capacity loss, stabilizes income target, balances battery cluster health status, prolongs system life, and through iterative optimization, adapts to dynamic load and price changes, and ensures efficient and safe operation. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The structure diagram of the lithium battery pack peak clipping and valley filling intelligent scheduling and parallel cooperative control system is shown. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0019] Embodiment 1: Figure 1 The lithium battery pack peak clipping and valley filling intelligent scheduling and parallel cooperative control system is given, which comprises: The time scale mold module: receiving and aligning the station level charging and discharging curve, the state and temperature trend of each battery cluster, and the working domain of each converter, generating parallel feasible region mapping on the unified time axis.

[0020] The projection consensus module: transcribing the peak clipping and valley filling target into an executable target sequence in the parallel feasible region mapping, merging the flow stability boundary and the protection boundary, and generating a plan candidate set.

[0021] The shaping synchronization module: performing step shaping and phase synchronization processing on the switching position of the plan candidate set, and outputting a station level plan meeting the feasible region and stability boundary constraints.

[0022] The flow alignment module: calculating the prior flow trajectory and the balance time slot for each battery cluster according to the station level plan and the parallel feasible region, and issuing to the converter and the balancing unit for execution in time sequence.

[0023] Rolling Correction Module: Collects execution deviation and temperature lag information, re-evaluates the parallel feasible region mapping and prior flow trajectory in an iterative optimization manner, and generates an updated plan and flow trajectory for the next time period.

[0024] Energy storage stations in the park and substation need to generate charge and discharge curves through the dispatch platform to respond to load trends and electricity price rhythms. However, the execution level involves the coordinated response of parallel battery clusters and converters. The differences in the state of charge, equivalent internal resistance and temperature trajectory of the battery clusters, as well as the inconsistency between balancing capabilities and protection boundaries, highlight the breakpoint between the planned trajectory and physical constraints.

[0025] The above emphasizes that the planning generation process focuses on the smoothness of the station-level curve and the profit target. However, the real constraints of the parallel layer come from the coupling effects inside and outside the battery cluster, such as the response offset caused by the difference in internal resistance and open circuit voltage, and the parameter lag caused by heat accumulation. Therefore, the time-scaled input module needs to first receive and align multi-source data to construct a parallel feasible domain mapping on a unified time axis, thereby providing a basic constraint framework for subsequent modules.

[0026] The specific processing logic of the time stamp input module: The receiving station-level charge / discharge curves, the status and temperature trends of each battery cluster, and the operating domain of each converter.

[0027] The charge / discharge curves generated by the scheduling platform need to be integrated with the battery cluster and inverter data at the execution level. These inputs are collected to ensure the integrity of subsequent alignment processing.

[0028] The receiving process collects the following data through a dedicated interface: the station-level charge-discharge curve is represented as a power-time series. ,in Station-level power requirements For each time point; the state of each battery cluster includes a sequence of states of charge. ( For battery cluster indexing, Refers to the first (Percentage of state of charge of each battery cluster) and equivalent internal resistance sequence ( Refers to the first (Equivalent internal resistance of each battery cluster); temperature trend is a temperature sequence. ( Refers to the first (Temperature values ​​of individual battery clusters); the operating domain of each converter is represented as a power upper limit sequence. and power lower limit sequence ( For converter indexing, Refers to the first The maximum allowable power of each converter Refers to the first The minimum allowed power of the converter). These data are transmitted into the module buffer at fixed sampling intervals.

[0029] After the data reception is completed, a multi-source sequence set is formed to provide the original input for alignment and ensure the preliminary integration of planning and execution constraints.

[0030] The station-level charge-discharge curve, the state of each battery cluster and the temperature trend, and the operating domain of each converter are aligned to a unified time axis.

[0031] The received station-level charge-discharge curve, the state of each battery cluster and the temperature trend, and the operating domain of each converter may have sampling time differences or inconsistent intervals, and need to be aligned based on the timing dialogue requirements to eliminate the beat misalignment between the planned trajectory and the physical constraints.

[0032] The alignment process uses linear interpolation to map all sequences to a unified time axis The unified time axis defined for the module is based on the least common multiple sampling interval): for the station-level charge-discharge curve, the formula is used and are adjacent time points of the original sequence; similarly, for the state of charge sequence of each battery cluster and are adjacent time points of the state of charge sequence, the equivalent resistance sequence and are adjacent time points of the equivalent resistance sequence; the temperature trend sequence and are adjacent time points of the temperature sequence; the upper limit and the lower limit , , and of the operating domain of each converter are the adjacent time points of the power upper and lower limit sequences. After interpolation, all sequences are synchronized on the unified time axis .

[0033] After alignment, a sequence set on the unified time axis is generated to provide a timing-consistent basis for the modeling of parallel constraints and avoid deviations caused by response bias and thermal hysteresis.

[0034] The parallel feasible region mapping is generated on the unified time axis.

[0035] The aligned station-level charge-discharge curve, the state of each battery cluster and the temperature trend, and the operating domain of each converter need to be integrated into a constraint framework, and a mapping is constructed based on the coupling effect to reflect the actual feasible power trajectory. ​​​​​

[0036] The generating process constrains the parallel feasible region mapping by aggregating the computation , defined as the union of all battery cluster and converter supportable power ranges on a unified time axis : compute the feasible power range of each battery cluster , where the lower bound ( refers to the open-circuit voltage of the th battery cluster, refers to the temperature-dependent thermal resistance value determined by the temperature sequence ); the upper bound ( refers to the equalization capability margin derived from the difference between the state-of-charge sequence and the equivalent internal resistance sequence ); then aggregate all battery cluster ranges and converter operating domains to obtain the parallel feasible region mapping , ensuring that the mapping covers the station-level charge-discharge curve constraints.

[0037] The parallel feasible region mapping reflects the common boundaries of current distribution, heat dissipation paths, and protection interlocks in the invention scenario, providing a basis for executable target projection consensus modules.

[0038] The time axis embedding module achieves the temporal integration of station-level charge-discharge curves and battery cluster, converter constraints by receiving, aligning, and generating parallel feasible region mappings on a unified time axis.

[0039] The time axis embedding module has generated parallel feasible region mappings on a unified time axis to integrate station-level charge-discharge curves and battery cluster, converter constraints, but the projection consensus module needs to convert peak shaving and valley filling targets into executable sequences on this basis, while merging flow distribution and protection boundaries to bridge the dialogue between plan benefits and physical safety, avoiding the amplification of health state differentiation caused by execution deviation.

[0040] The specific processing technology logic of the projection consensus module: Convert the peak shaving and valley filling target into an executable target sequence within the parallel feasible region mapping.

[0041] The parallel feasible region mapping has defined a power range set on a unified time axis. According to the benefit tracking requirements, the peak shaving and valley filling target is mapped to this range to form an executable sequence to avoid the tension between the plan and constraints.

[0042] The conversion process calculates the executable target sequence , where refers to the executable power sequence on a unified time axis : extract the peak shaving and valley filling target as a power target sequence , where denotes the ideal power demand based on the load trend and the price rhythm; for each time point , if exceeds the lower or upper limit of the parallel feasible region mapping , it is projected to the nearest boundary using the formula , where denotes the lower limit of the parallel feasible region mapping at , denotes the upper limit of the parallel feasible region mapping at , and the minimum maximum operation logically limits the target within the feasible region to maintain the safety boundary.

[0043] The executable target sequence reflects the preliminary consensus of the peak shaving target and the parallel constraint in the invention scenario, avoiding the current preferential allocation and the circulation formation caused by the fast-changing trajectory.

[0044] The stable boundary and the protection boundary of the distribution are merged into the executable target sequence.

[0045] The executable target sequence has been limited within the parallel feasible region mapping, and now needs to be further integrated into the distribution and protection constraints based on the coupling effect to strengthen the stability and safety of the sequence.

[0046] The merging process updates the executable target sequence through constraint superposition: first, define the distribution stable boundary as the current allocation margin sequence , where denotes the allocation limit based on the equivalent internal resistance sequence of each battery cluster, calculated by the formula , where the product operation aggregates the relative impedance deviation of all battery clusters to quantify the stable margin, logically ensuring that the margin is less than 1, indicating the risk of uneven distribution; the protection boundary is the safety threshold sequence , where denotes the joint limit of the temperature sequence and the state of charge sequence , calculated by the formula , where denotes the reference temperature threshold (determined by the battery cluster specification), denotes the temperature-sensitive scale (which can be determined by the difference between the temperature upper limit and the reference temperature divided by the natural logarithm of the target decay rate, for example =(60°C-25°C) / ln(2), so that the exponential function reaches the preset decay level at the upper limit), and the exponential operation logically simulates the exponential decay effect of heat accumulation on protection; then update the executable target sequence to , and the product logic ensures that the sequence is contracted within the merged boundary.

[0047] After the merging is completed, the executable target sequence integrates the boundaries of the distribution law and the protection interlocking in the invention scenario, reducing the deviation caused by compensation actions and thermal hysteresis.

[0048] A candidate set of plans is generated based on the merged executable target sequence.

[0049] The merged executable target sequence already contains the flow stability boundary and protection boundary. The projection consensus module needs to extend the sequence into multiple candidates to cover potential trajectory variations based on the decision consensus of the invention scenario.

[0050] The generation process calculates a candidate set of plans through sequence variation wherein refers to a set of alternative power sequence sets: to the merged executable target sequence At each time point a margin perturbation is introduced to generate a candidate sequence wherein refers to the perturbation term, calculated by the formula The alternating sign and the exponential denominator logic ensure that the perturbation gradually decreases, covering the variation path within the feasible region; all candidate sequences need to be verified not to exceed the parallel feasible region mapping to form the final candidate set of plans.

[0051] The candidate set of plans provides multiple trajectory options in the invention scenario that meet the consensus constraints, supporting shaping to improve the stability of benefits and capacity utilization.

[0052] The projection consensus module realizes the peak clipping and valley filling goal and the boundary consensus of the parallel feasible region mapping by transcribing, merging, and generating the candidate set of plans.

[0053] The projection consensus module has generated a candidate set of plans, which contains multiple alternative power sequences that incorporate the flow stability boundary and protection boundary. The shaping synchronization module needs to perform step shaping and phase synchronization processing for the switching positions of these candidate sets to smooth fast-changing trajectories and align timing, thereby eliminating the risk of circulating current formation and hot spot superposition, ensuring that the output station-level plan meets the parallel feasible region mapping and stability boundary constraints.

[0054] The specific processing technology logic of the shaping synchronization module: Identify the switching positions in the candidate set of plans. The candidate set of plans provides multiple sets of alternative power sequences, based on the response offset caused by fast-changing or switching of the planned trajectory, scan these sequences to locate the power jump points, and thereby specifically address potential current prioritization problems.

[0055] The identification process determines the switching position through differential scanning: for each candidate sequence in the candidate set of plans calculate the power difference sequence wherein refers to the candidate sequence in the unified time axis to ; then compare the difference with preset transition threshold, if the absolute difference exceeds the transition threshold, mark the time point as switching position, where the transition threshold is determined by the maximum difference of battery cluster equivalent internal resistance sequence to reflect the offset sensitivity caused by internal resistance difference; all marked points form switching position set , where is the switching time point list of the th candidate sequence.

[0056] Switching position set accurately captures fast-changing and switching points of planned trajectory, which tend to cause higher voltage or lower impedance battery cluster to preferentially carry current, thereby providing targeted position guidance for subsequent step shaping, avoiding the formation and superposition of circulating current between parallel branches, and reducing the frequency of compensation actions and maintaining the balanced distribution of health state.

[0057] Step shaping is performed for switching positions. The switching position set has marked the transition points in the planned candidate set, and step shaping needs to be applied to these positions to smooth the power transition in a gradual manner to suppress the intervention of protection actions based on heat accumulation and parameter hysteresis.

[0058] The shaping process updates the candidate sequence by interpolation smoothing: for each switching position of the th candidate sequence , define the shaping window as to , where is the shaping time width, determined by the average rate of change of temperature sequence to match the hysteresis characteristics of thermal timing; then apply linear ramping within the window, using the formula , where is the sequence value after shaping at window time point , the formula ensures smooth transition without introducing abrupt changes through linear slope; the updated sequence needs to be verified to still be within the parallel feasible region mapping .

[0059] The shaped candidate sequence achieves step smoothing at switching positions, which effectively disperses the current distribution pressure caused by parameter hysteresis and the delayed intervention of balancing actions, preventing lower impedance battery clusters from being overburdened and causing circulating current and heat hotspot superposition, thereby reducing the need for flow distribution rules and improving the stability of apparent available capacity, ensuring that the execution result is more consistent with the station-level curve and reducing the degree of differentiation of health state.

[0060] Phase synchronization processing is performed on the shaped candidate set. The shaped candidate sequences have been smoothed out, and phase alignment needs to be performed on these sequences based on parallel physical constraints and timing dialogue to synchronize the battery cluster response and incorporate them into the distribution stability boundary.

[0061] The synchronization process achieves phase consistency through time shift adjustment: for all shaped candidate sequences Calculate the phase shift sequence of each sequence. ,in Refers to the first The time shift of each sequence relative to the reference sequence (selected as the first candidate sequence) is given by the formula... ,in The time-shift variable is used; the argmin operation logic finds the time shift with the smallest absolute difference to align the waveform; then, time-shift correction is applied. and merge the distribution stability boundary With protection boundary Through product constraints Ensure that the synchronization sequence satisfies the stability boundary constraints.

[0062] After synchronization, the candidate sequences achieve phase alignment. This process enhances the coordination of the parallel system and reduces the alternation of protection threshold intervention and compensation actions caused by the difference in internal resistance and open circuit voltage. This stabilizes the profit target and prevents the amplification of health status differentiation by execution deviation. It ensures that all sequences comply with the feasible domain and boundary constraints on a unified time axis to support the final planned output.

[0063] The output station-level plan satisfies the feasible region and stability boundary constraints. The synchronized candidate sequences have been phase-aligned and boundary-constrained. Based on a common decision-making framework, the optimal station-level plan is selected from these sequences to bridge the gap between plan smoothing and execution safety.

[0064] The output process generates station-level plans through optimized aggregation. ,in The final station-level power sequence refers to all synchronized candidate sequences. The aggregate value is calculated using the formula. ,in The total number of candidate sequences; final verification Mapping in parallel feasible region Internal distribution stability boundary With protection boundary limit.

[0065] The station-level plan fully satisfies the feasible region and stable boundary constraints in the invention scenario, smooths the overall trajectory by aggregating the synchronous sequence, effectively compresses the deviation degree and improves the capacity utilization for the mutual pulling of the plan and physical constraints, thereby maintaining the long-term stability of the revenue and reducing the differentiation amplitude of the battery cluster health status, ensuring efficient and safe collaborative control of the energy storage station in peak shaving tasks.

[0066] The shaping and synchronization module realizes the conversion of the plan candidate set to the constraint satisfaction trajectory by identifying the switching position, performing step shaping and phase synchronization processing, and outputting the station-level plan.

[0067] The shaping and synchronization module has output the station-level plan, which satisfies the parallel feasible region mapping and flow distribution stable boundary, and the flow distribution alignment module needs to calculate the prior flow distribution trajectory and equalization time slot for each battery cluster according to the station-level plan and the parallel feasible region mapping, and issue it in time sequence to the converter and equalization unit for execution, thereby bridging the current distribution between the plan decision and execution layers, avoiding circulating current and compensation actions caused by internal resistance difference and thermal hysteresis, and ensuring accurate response of parallel collaboration.

[0068] The specific processing technology logic of the flow distribution alignment module: According to the station-level plan and the parallel feasible region mapping, the prior flow distribution trajectory of each battery cluster is calculated. The station-level plan has defined the final power sequence on the unified time axis and is constrained by the parallel feasible region mapping. The flow distribution alignment module first calculates the power share allocated to each battery cluster based on the current distribution and heat dissipation path requirements of the parallel system, combined with the equivalent internal resistance sequence and state of charge sequence of the battery cluster, to pre-avoid the risk of higher voltage or lower impedance battery cluster preferentially carrying current.

[0069] The prior flow distribution trajectory is generated by impedance proportional distribution: for each time point on the unified time axis, the power value of the station-level plan is extracted , and the relative impedance contribution of each battery cluster is calculated. First, the reciprocal sum of all battery cluster equivalent internal resistance sequences is taken as the total admittance , then the allocation proportion of each battery cluster is equal to its equivalent internal resistance sequence reciprocal divided by the total admittance, which reflects that the battery cluster with lower impedance should bear a larger share to balance the current; then the prior flow distribution trajectory is the station-level plan power value multiplied by this allocation proportion, where is the pre-allocated power trajectory of the battery cluster on the unified time axis ; the trajectory is further adjusted to incorporate the state of charge sequence , using the formula , where the subtraction logic of the relative state of charge deviation ensures that the low state of charge battery cluster reduces the burden to prevent over-discharge; the final trajectory needs to be verified not to exceed the single battery cluster range of the parallel feasible region mapping .

[0070] The prior flow distribution trajectory provides each battery cluster with a pre-allocated power sequence, optimizes the current distribution path through dynamic proportional adjustment of impedance and state of charge to reduce the formation of circulating current between parallel branches and the superposition effect of thermal hot spots caused by response shifts due to internal resistance differences and open circuit voltage differences, thereby reducing the frequency of flow distribution rule modifications triggered to catch up with the plan, maintaining the effective play of balancing capacity, and ensuring that subsequent balancing time slot calculations are based on reliable distribution to improve the accuracy of overall execution and the uniformity of the health state.

[0071] The balancing time slots for each battery cluster are calculated based on the prior flow distribution trajectory and the parallel feasible region mapping. The prior flow distribution trajectory has allocated power shares to each battery cluster and adjusted the charge deviation. Based on the balancing capacity and the difference in protection boundaries, further identification of the imbalance interval in the trajectory is needed to calculate the time slot window to allow the balancing unit to intervene during off-peak periods, avoiding the amplification of heat accumulation and parameter hysteresis.

[0072] Generate balancing time slots through deviation scanning: for each time point on the unified time axis, compare the prior flow distribution trajectories of all battery clusters and the state of charge sequence , find the maximum deviation between the state of charge sequences, if the deviation exceeds the balancing threshold, mark the starting time point, where the balancing threshold is determined by the minimum value of the protection boundary to match the safety margin; then scan the continuous time period until the deviation falls, forming the balancing time slot , where is the balancing intervention time window list of the th battery cluster, each window is defined as the starting to ending time point; incorporate the temperature sequence , use the formula , where the window end point is equal to the starting point plus the temperature deviation divided by the temperature sensitive scale multiplied by the basic time step, the temperature deviation subtraction logic prolongs the balancing time slot of the high-heat cluster to enhance heat dissipation, is the basic interval of the unified time axis; the final time slot needs to ensure that it does not overlap with the fast-changing segment of the station-level plan and is located in the low-power region of the parallel feasible region mapping .

[0073] The balancing time slot provides each battery cluster with a list of intervention windows, addressing the parameter hysteresis caused by heat accumulation and the subsequent intervention of balancing actions. Through dynamic scanning of deviations and temperatures, the time slot reserves adjustment space to prevent the alternating occurrence of compensation actions and protection actions, thereby compressing the loss of apparent available capacity and improving the long-term performance of revenue stability, ensuring that the power converter and balancing unit can respond in time to reduce the deviation between the station-level curve and the execution result, and balance the health state differentiation of the battery clusters.

[0074] The prior flow distribution trajectory and the balancing time slot are sequentially issued to the converter and the balancing unit for execution. The balancing time slot has defined an intervention window based on the prior flow distribution trajectory and the protection boundary. The flow distribution alignment module needs to package these trajectories and time slots into instruction sequences based on the timing dialogue and interlocking mechanism of the invention scenario, and issue them in sequence according to a unified time axis, in order to achieve real-time response of parallel coordination and avoid additional actions caused by beat misalignment.

[0075] The issuing process transmits instructions by timing packaging: the prior flow distribution trajectory and the balancing time slot are integrated into a time-ordered package. For each unified time axis time point, a power instruction sequence for the first converter is generated , where the power instruction sequence is equal to the sum of the prior flow distribution trajectories of the associated battery cluster, and is limited to the power upper limit sequence and the power lower limit sequence of each converter operating domain; and a time slot activation sequence for the balancing unit is generated , where the activation sequence is set to an enable value within the balancing time slot and to a disable value otherwise; these sequences are issued to the corresponding converter and balancing unit in ascending order according to the unified time axis using a dedicated communication protocol, ensuring that the transmission delay is less than the basic time step to maintain synchronization; and the instructions are verified before issuance to ensure that they do not violate the flow distribution stability boundary and the protection boundary .

[0076] After the issuance is completed, the prior flow distribution trajectory and the balancing time slot are converted into executable instruction sequences, and the issuance is coordinated by timing-ordered packaging to respond to the converter and the balancing unit, reducing additional flow distribution and compensation actions caused by slight beat misalignment, thereby stabilizing the tracking of the station-level curve and the profit target, and preventing the amplification of health state differentiation, ensuring efficient parallel coordination control and capacity utilization optimization of the energy storage station in peak shaving and valley filling tasks.

[0077] The flow distribution alignment module realizes the alignment conversion of the station-level plan to the battery cluster execution layer by calculating the prior flow distribution trajectory, the balancing time slot, and sequentially issuing them.

[0078] The flow distribution alignment module has issued the prior flow distribution trajectory and the balancing time slot to the converter and the balancing unit for execution, thereby realizing the preliminary distribution of the station-level plan. The rolling correction module needs to collect execution deviation and temperature lag information to iteratively optimize the parallel feasible region mapping and the prior flow distribution trajectory, generate an updated next time period plan and flow distribution trajectory, thereby dynamically bridging the break point between the plan and the execution, and avoiding the amplification of long-term deviation, health state differentiation, and decline in profit stability.

[0079] The specific processing technology logic of the rolling correction module: Collecting execution deviation and temperature lag information. Prior to the dispatch trajectory and the equilibrium time slot have been issued in sequence, based on the additional dispatch and compensation actions that occur on the execution side in the invention scenario, and the parameter lag caused by heat accumulation, the rolling correction module first collects data from the real-time feedback of the converter and the battery cluster to quantify the difference between the planned trajectory and the actual response, so as to provide accurate deviation basis for subsequent iterative optimization, and avoid the alternation of response deviation and protection action to amplify the compression of apparent available capacity.

[0080] The following information is obtained by real-time monitoring of the interface: execution deviation is defined as the difference between the station-level planned power value and the actual output power sequence, where the actual output power sequence is composed of the power measurement values fed back by the converter, and the difference is calculated at each time point on the unified time axis and accumulated as a deviation sequence , where is the execution power deviation on the unified time axis ; the temperature lag information is calculated by comparing the predicted value (thermal model derived based on the prior dispatch trajectory) of the temperature sequence with the actual sensor temperature value, to obtain a lag sequence , where is the average of the temperature difference of all battery clusters on the unified time axis ; at the same time, the actual updated values of the state of charge sequence and the equivalent internal resistance sequence are collected to supplement the deviation analysis; all collected data are buffered and stored at fixed intervals to ensure coverage of the recent execution period to reflect the immediate impact of circulating current formation and thermal hot spot superposition.

[0081] The execution deviation and temperature lag information form a set of quantitative sequences, which capture the dynamic performance of compensation actions and protection actions through real-time difference and lag calculation, prevent the continuous deviation of the station-level curve and the execution result, and provide data support for re-estimating the parallel feasible region mapping, so as to ensure that the iterative optimization can accurately correct the deviation caused by the difference in internal resistance and the difference in open circuit voltage, and maintain the intervention time of the equilibrium capacity to reduce the long-term accumulation effect of the health state differentiation, and improve the stability and capacity utilization efficiency of the overall benefit of the energy storage station.

[0082] Re-estimating the parallel feasible region mapping in an iterative optimization manner. The execution deviation and temperature lag information have quantified the execution difference in the recent period, and the rolling correction module needs to use these information to iteratively adjust the power range of the battery cluster and the converter based on the common constraint decision framework of the invention scenario, to update the parallel feasible region mapping to reflect the actual coupling effect and avoid the compression of the feasible region caused by the thermal time lag and the delay of the equilibrium action.

[0083] The re-estimation process updates the parallel feasible region mapping using a cyclic iteration method: initialize the current parallel feasible region mapping is the previous value, then incorporates the execution bias sequence in each iteration round , using the formula , where denotes the updated parallel feasible region mapping on the unified time axis , the lower and upper bounds are shrunk by this formula, the subtraction logic deducts the bias and simulates the hysteresis decay by the exponential function, where denotes the hysteresis sensitive scale (determined by the historical average of the temperature sequence), the exponential operation ensures the dimensionless proportion and logically amplifies the range adjustment at high hysteresis, keeping the power unit consistent; meanwhile, the feasible power range of each battery cluster is updated, using the actual state of charge sequence and the equivalent internal resistance sequence to recalculate the lower and upper bounds, and the intersection of all ranges is aggregated; the iteration continues until the bias sequence converges to the preset standard, typically no more than a specified upper limit to control the computational overhead.

[0084] The updated parallel feasible region mapping reflects the immediate impact of execution bias and temperature hysteresis, this re-estimation dynamically defines the safety boundary through iterative shrinkage and exponential decay, preventing frequent modifications of compensation actions and the superimposed circulating current heat hotspots, thereby improving the feasibility of the planned trajectory and reducing the downward trend of yield stability, ensuring that the subsequent a priori allocation trajectory is re-estimated based on a more accurate constraint framework to optimize parallel collaborative response and maintain the balanced distribution of battery cluster health status and the recovery of apparent available capacity.

[0085] The a priori allocation trajectory is re-estimated in an iterative optimization manner. The updated parallel feasible region mapping has incorporated bias and hysteresis adjustments, and the rolling correction module needs to continue to iterate and optimize the allocation share of each battery cluster based on the allocation rules and thermal timing constraints of the invention scenario, to re-estimate the a priori allocation trajectory to adapt to the physical boundaries of actual execution, avoiding excessive intervention of after-the-fact balancing actions.

[0086] The re-estimation process continues the aforementioned iterative framework to adjust the a priori allocation trajectory: for each time point on the unified time axis, starting from the station-level planned power value , the allocation proportion is recalculated, using the inverse sum of the updated equivalent internal resistance sequence as the total admittance, and the proportion of each battery cluster is the inverse of its own divided by the total admittance; then a temporary trajectory is generated, and the temperature hysteresis sequence is incorporated, using the formula , where denotes the updated a priori allocation trajectory of the th battery cluster, the exponential function logically simulates decay through the product of hysteresis and internal resistance, where The internal resistance sensitivity scale (determined by the historical maximum of the equivalent internal resistance sequence) logically reduces the allocation of high-lag clusters to protect the boundary; iteratively adjust until the trajectory deviation is less than the preset requirement, and verify that the trajectory is located in the updated parallel feasible region mapping The internal resistance sensitivity scale (determined by the historical maximum of the equivalent internal resistance sequence) logically reduces the allocation of high-lag clusters to protect the boundary; iteratively adjust until the trajectory deviation is less than the preset requirement, and verify that the trajectory is located in the updated parallel feasible region mapping

[0087] The updated prior flow distribution trajectory optimizes the power distribution of each battery cluster, adapts to deviation feedback through iterative proportional and exponential decay, prevents higher voltage or lower impedance battery clusters from preferentially consuming current and triggering circulating current and thermal hotspots, thereby reducing the need for continuous correction of flow distribution rules and improving the pre-exertion of balancing capability, ensuring that the next period plan is generated based on reliable trajectories to reduce the deviation of station-level curves and execution results, and maintain the minimization of health state differentiation and long-term stability of the income target.

[0088] The updated next period plan and flow distribution trajectory are generated. The updated prior flow distribution trajectory has been iteratively integrated with deviation and lag, and the rolling correction module needs to be based on the joint decision and timing of the invention scenario to aggregate these trajectories to generate the station-level plan and flow distribution trajectory for the next period, to achieve continuous correction to bridge the gap between plan generation and execution.

[0089] The generation process outputs an updated sequence through trajectory aggregation: for the next period (fixed window after uniform time axis shift), calculate the updated station-level plan Update the prior flow distribution trajectory for all battery clusters The sum of which is The next period uniform time axis; then verify that the sum is located in the updated parallel feasible region mapping Inside, and merge the flow distribution stability boundary And the protection boundary Ensure plan shrinkage through minimum constraint; finally output the updated station-level plan and the updated prior flow distribution trajectory of each battery cluster as the next cycle input.

[0090] The updated next period plan and flow distribution trajectory provide a continuously optimized trajectory set, dynamically adjust the response path through aggregation and boundary constraints to prevent amplification of additional actions and health state differentiation caused by slight timing misalignment, thereby improving the tracking accuracy of peak shaving and valley filling targets and compressing the risk of income stability decline, ensuring that the energy storage station achieves efficient capacity utilization and safe boundary maintenance in parallel cooperative control.

[0091] The rolling correction module realizes the rolling conversion of execution feedback to the next period plan by collecting deviation and lag information, iteratively re-estimating the parallel feasible region mapping and prior flow distribution trajectory, and generating an updated sequence.

[0092] The above formulas are all de-dimensioned to calculate the numerical values, the formulas are obtained by collecting a large amount of data to simulate a formula of the most recent real situation, and the preset parameters in the formulas are set by the person skilled in the art according to the actual situation.

[0093] It should be noted that the system of the present application can be deployed in the device itself to realize embedded application, or can be run on PC or other terminal with user interface, so as to meet various hardware environments and use requirements.

[0094] The above only describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that the described embodiments can be modified in various ways without departing from the spirit and scope of the present application for those skilled in the art. Therefore, the above drawings and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

[0095] It should be noted that in this document, relational terms such as first and second and the like can merely be used to distinguish one entity or action from another, without necessarily requiring or implying any actual such relationship or order between or among the entities or actions. Also, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0096] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A lithium battery pack peak shaving and valley filling intelligent scheduling and parallel collaborative control system, characterized in that, include: Time-scaled model module: Receives and aligns station-level charge and discharge curves, the status and temperature trend of each battery cluster, and the operating domain of each converter, and generates a parallel feasible domain mapping on a unified time axis; Projection consensus module: Within the parallel feasible region mapping, the peak shaving and valley filling objectives are rewritten into an executable sequence of objectives, and the distribution stability boundary and protection boundary are merged to generate a set of plan candidates; Shaping and Synchronization Module: Performs step shaping and phase synchronization processing on the switching positions of the candidate plan set, and outputs a station-level plan that satisfies the feasible region and stability boundary constraints; Distribution alignment module: Calculates the prior distribution trajectory and balancing time slot for each battery cluster based on the station-level plan and parallel feasible domain, and sends them to the converter and balancing unit for execution according to the timing sequence; Rolling Correction Module: Collects execution deviation and temperature lag information, re-evaluates the parallel feasible region mapping and prior flow trajectory in an iterative optimization manner, and generates an updated plan and flow trajectory for the next time period.

2. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 1, characterized in that: The time-stamped input module receives station-level charge and discharge curves as power and time series, the state of each battery cluster including the state of charge series and equivalent internal resistance series, the temperature trend as a temperature series, and the operating domain of each converter as the upper power limit series and the lower power limit series. These are transmitted to the module buffer through a dedicated interface at a fixed sampling interval. The alignment process uses a linear interpolation method to map all sequences to a unified time axis based on the least common multiple sampling interval.

3. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 2, characterized in that: The time-scaled model module generates a parallel feasible domain mapping on a unified time axis. Through constraint aggregation, it calculates the set of power ranges that all battery clusters and converters can support. First, it calculates the feasible power range of each battery cluster, including the lower limit and the upper limit. Then, it aggregates the ranges of all battery clusters and finds the intersection with the operating domain range of each converter, and finds a further intersection with the station-level charge and discharge curve constraints.

4. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 1, characterized in that: The projection consensus module rewrites the peak shaving and valley filling objectives into an executable objective sequence within the parallel feasible region mapping. It extracts the peak shaving and valley filling objectives as a power objective sequence through boundary projection calculation. For each time point, if it exceeds the lower or upper bound of the parallel feasible region mapping, it projects to the nearest boundary and uses the minimum-maximum operation to limit the power objective sequence value to between the lower and upper bounds of the parallel feasible region mapping.

5. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 4, characterized in that: The projection consensus module merges the current distribution stability boundary and the protection boundary into the executable target sequence. It updates and defines the current distribution stability boundary as the current distribution margin sequence based on the distribution limit of the equivalent internal resistance sequence of each battery cluster through constraint superposition. The protection boundary is the safety threshold sequence based on the joint limit of the temperature sequence and the state of charge sequence. Then, the executable target sequence is updated to the original sequence multiplied by the minimum value of the current distribution margin sequence and the safety threshold sequence.

6. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 5, characterized in that: The projection consensus module generates a set of candidate plans based on the merged executable target sequence. It calculates a set of alternative power sequences through sequence mutation. It introduces margin perturbation into the merged executable target sequence at each time point to generate multiple candidate sequences. All candidate sequences are verified to not exceed the parallel feasible region and are mapped to form the final set of candidate plans.

7. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 1, characterized in that: The shaping and synchronization module performs step shaping processing on the switching position of the candidate set of plans. It determines the switching position by differential scanning, calculates the power difference sequence between adjacent time points, compares it with the preset jump threshold, marks the time point, defines a shaping window for the switching position, applies linear gradual update to the candidate sequence, and the updated sequence verification is still within the parallel feasible region mapping.

8. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 7, characterized in that: The shaping and synchronization module performs phase synchronization processing on the shaped candidate set of plans. It achieves phase consistency by adjusting the time shift, calculates the phase shift value of each sequence relative to the reference sequence, applies time shift correction, and merges the distribution stability boundary and protection boundary. It ensures that the synchronization sequence meets the stability boundary constraints through product constraints.

9. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 1, characterized in that: The current distribution alignment module calculates the prior current distribution trajectory of each battery cluster based on the station-level plan and the parallel feasible region mapping. It generates a distribution ratio equal to the reciprocal of the equivalent internal resistance sequence divided by the total admittance through impedance ratio allocation. The prior current distribution trajectory is the station-level planned power value multiplied by this distribution ratio. The trajectory is adjusted to integrate into the state of charge sequence. The trajectory verification does not exceed the range of a single battery cluster mapped by the parallel feasible region.

10. The intelligent scheduling and parallel collaborative control system for peak shaving and valley filling of lithium battery packs according to claim 9, characterized in that: The distribution alignment module calculates the equalization time slots for each battery cluster based on the prior distribution trajectory and the parallel feasible region mapping. By comparing the prior distribution trajectory with the state of charge sequence through deviation scanning, the starting time point when the maximum deviation exceeds the equalization threshold is marked. The module scans continuous time periods to form an equalization time slot window list, incorporates the temperature sequence to define the end point of the window, and ensures that the time slots do not overlap with the fast-changing sections of the station-level plan and are located in the low-power region of the parallel feasible region mapping.

Images