A method and system for intelligent power supply control of a forklift mobile robot based on dynamic analysis

Abstract

This invention discloses an intelligent power supply control method and system for a forklift mobile robot based on dynamic analysis, belonging to the field of intelligent power supply control technology. By further performing hierarchical screening and structured binding processing on the voltage and current change processes based on start-stop segments, the extraction process of battery internal resistance changes is transformed from the original overall statistical method to a refined identification method targeting the operating phase. This results in improved effects and technical advantages, and addresses the problem of accumulated deviations in SOC estimation under frequent start-stop conditions. Through the synchronous change window identification process, only continuous intervals where voltage and current change synchronously are found within the start-stop segments. This limits the data source for subsequent analysis to the actual energy change stage, excluding non-corresponding change intervals caused only by current disturbances or control fluctuations.

Description

Technical Field

[0001] This invention relates to the field of intelligent power supply control technology, specifically to an intelligent power supply control method and system for a fork-type mobile robot based on dynamic analysis. Background Technology

[0002] Forklift robots are crucial execution devices in warehousing and industrial material handling scenarios, widely used in automated warehousing systems, production line distribution, and regional logistics transportation. Within the operating system of a forklift robot, the battery power supply system not only provides the overall power output but also directly impacts driving control, lifting operations, and task scheduling stability.

[0003] In actual operation, forklift mobile robots typically operate in a state of frequent start-stop, short-distance movement, and intermittent lifting operations. This operating mode causes the power battery to be in a working environment of alternating low load and short-cycle load for a long time. Existing SOC estimation methods are mostly based on voltage or current change relationships established under stable operating conditions for state judgment. This method is applicable to continuous operation, but under frequent start-stop conditions, the internal state of the battery will undergo short-term, subtle changes. These changes do not form obvious voltage or current anomalies, and therefore are often not separately identified and modeled by existing power supply control systems.

[0004] The above situation arises because the operating rhythm of the forklift mobile robot exhibits a distinct repetitive start-stop characteristic. During the drive's start-up, stopping, and re-entry into operation, the battery's internal polarization state and energy release process undergo continuous short-cycle adjustments, causing the internal resistance to drift continuously and slightly within a local timeframe. When this type of drift is not identified separately, its impact accumulates during the SOC estimation process, causing the SOC change trend to gradually deviate from the true state of charge. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent power supply control method and system for forklift mobile robots based on dynamic analysis, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: an intelligent power supply control system for a fork-type mobile robot based on dynamic analysis, comprising a start / stop segment and reference segment generation module, an internal resistance candidate segment screening module, an internal resistance oscillation identification module, a SOC compensation command generation module, and a power supply control linkage module;

[0007] The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment.

[0008] The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence;

[0009] The internal resistance oscillation identification module is based on the binding sequence and the set of internal resistance candidate segments. It performs continuous discrimination on the changes in internal resistance between candidate segments to obtain oscillation indicators.

[0010] After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag.

[0011] The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

[0012] Preferably, the start / stop segment and reference segment generation module includes a drive operation state boundary identification unit and a structured segment index construction unit;

[0013] The drive operation state boundary recognition unit reads the existing state quantities of the fork mobile robot controller, including: drive enable state, drive speed command valid / invalid state, brake release state, lifting / lowering operation state, and steering command active state; and converts them into unified discrete state words to form a continuous state sequence.

[0014] When a change in the drive enable state from off to on or from on to off is detected, the boundary verification process is initiated, which includes two consecutive actions: triggering validity verification and stability solidification.

[0015] Trigger validity check: This determines whether the change in drive enable state is accompanied by speed command activation or brake release. If it is only a control signal jitter and no actual drive action is formed, then no start / stop boundary is generated.

[0016] Stability solidification: After triggering, continuously observe the changes in the driving state. If a reverse swing or multiple rapid switches occur in a short period of time, the interval is merged into a single start-stop event, and the boundary confidence level is recorded. If the swing is frequent, it is marked as a boundary interference interval.

[0017] When the drive enable state enters a continuous stable phase and the operation state is no longer in the lifting, lowering, or steering jogging switching range, the end boundary of the start-stop segment is confirmed, a start-stop segment with start and end times and a reliable marker is formed, and the start-stop segment is output as a set of start-stop segments in chronological order.

[0018] Preferably, for each start-stop segment, the structured segment index building unit first searches for candidate reference intervals in the adjacent intervals before and after; during the search process, it excludes areas of lifting switching, braking switching or frequent steering inching, and only retains continuous segments in areas where the drive command direction is stable and the speed command maintains a low load interval;

[0019] When a brief disturbance occurs within the candidate reference interval, such as a single turn adjustment or a short speed swing, the candidate interval is not discarded directly. Instead, interval pruning is performed, using the disturbance point as the dividing point, the longest sub-interval that satisfies the continuous stability condition on both sides is extracted, and the pruning behavior is recorded to retain the largest usable stable segment.

[0020] After obtaining the first and second candidate reference intervals, a unique matching decision process is executed: the candidate interval that is consistent with the running direction of the start and stop segments is selected first;

[0021] If the directions are consistent, select the interval that is closer to the start and stop segment time and has not been pruned;

[0022] If multiple intervals satisfy the conditions, the one with the longest continuity is selected; finally, only one reference segment is output, and a unique identifier is generated.

[0023] The unique identifier of the start / stop segment is bound to the identifier of the reference segment, encapsulated into a structured segment unit, and segment attribute information is attached, including whether the reference segment is before or after the start / stop segment, whether it has been pruned, and the boundary confidence level.

[0024] Write all structured fragment units into the structured fragment index sequence in the order of start and stop, and establish a bidirectional index relationship to obtain the structured fragment index sequence.

[0025] Preferably, the internal resistance candidate segment screening module includes a synchronous change window identification and boundary interference interval elimination unit and a direction consistency confirmation and reference segment binding sequence generation unit;

[0026] The synchronous change window identification and boundary interference interval elimination unit reads the structured fragment unit one by one according to the structured fragment index sequence and enters the start and stop fragment to perform the identification action;

[0027] Within the start-stop segment, find the continuous interval where voltage and current change synchronously:

[0028] If the current changes significantly but the voltage does not change accordingly, it is judged as an asynchronous interval and is not considered as a candidate window.

[0029] If the voltage and current show a consistent trend of change within the same time period, then the continuous segments are cut out to form a preliminary draft of the candidate window.

[0030] The incomparable boundaries that may exist within the start-stop segments are located, and a list of boundary interference intervals is generated, specifically including three categories:

[0031] Contactor switching interference zone: A short buffer zone extending forward and backward from the center of the contactor state change or enable swing is marked as unavailable.

[0032] Braking switching interference range: Centered on the braking release / re-braking switching point, the buffer zone is also extended; avoid treating the braking transient current change as an internal resistance change;

[0033] Command jump interference interval: When a speed command or torque command exhibits a sudden and short-duration spike, the area before and after the spike is merged into the interference interval to prevent single-point jumps from causing false synchronization.

[0034] Each of the initially selected candidate windows was compared with the boundary interference interval:

[0035] If a candidate window falls entirely within the interference range, it is discarded.

[0036] If the candidate window partially overlaps with the interference interval, then perform cut-off and retention trimming: remove the overlapping part, retain only the non-overlapping sub-intervals that still maintain continuity, and use the trimming result as a new candidate window;

[0037] If the pruned subintervals are too short or fragmented, they are marked as unusable windows.

[0038] Preferably, the direction consistency confirmation and reference segment binding sequence generation unit processes the candidate window. When the candidate window meets the synchronous change, but the voltage and current directions are opposite or misaligned, the direction consistency confirmation is performed on the candidate window.

[0039] The steps for consistency verification are: only retain windows where the voltage and current change in the same direction within the same window;

[0040] For windows with brief directional swings, if the swing occurs at the window edge, a secondary trim is performed to extract the main continuous segments with the same direction.

[0041] If the swing occurs in the middle of the window and causes the window to split, it is judged as an unmatchable window and is removed;

[0042] For each candidate window that has passed the directional consistency confirmation, it is not directly announced as an internal resistance candidate segment. Instead, it is returned to the structured segment index sequence to find the structured segment unit to which it belongs, read the attribute information of the corresponding reference segment, and confirm whether the window has a comparative significance.

[0043] If a reference fragment is marked as trimmed and has a low confidence level, then the conservative retention rule is applied, allowing only fragments with high window quality to be retained.

[0044] If the reference fragment has a high confidence level, the window will be retained according to normal rules;

[0045] For windows that meet the directional consistency and control conditions, they are solidified as internal resistance candidate segments and a unique number is generated; multiple internal resistance candidate segments are allowed within the same start-stop segment, and the sequence number is kept increasing, and it is recorded whether the source window of each candidate segment has been pruned; this is used to identify the residual influence of the boundary during subsequent oscillation identification.

[0046] For each candidate segment of internal resistance, establish a clear binding relationship:

[0047] Bind to the identifier of the start / stop segment; further bind to the identifier of the structured segment unit to which the start / stop segment belongs; and bind to the unique reference segment identifier within the structured segment unit; organize the above binding relationships in the order of the structured segment index sequence to form a binding sequence.

[0048] Preferably, the internal resistance oscillation identification module includes an intra-unit sequence comparison and fluctuation event extraction unit and a cross-unit continuity discrimination and oscillation identifier generation unit;

[0049] The unit reads the binding sequence of the sequential comparison and fluctuation event extraction unit, merges all internal resistance candidate segments according to the structured segment unit to which they belong, and sorts the internal resistance candidate segments according to the occurrence time within each structured segment unit to form a segment sequence;

[0050] Perform order comparison on the fragment sequence within the same structured fragment unit:

[0051] First, identify whether there is a main trend of continuous change in the same direction in the segment sequence, that is, whether the direction of change of internal resistance is consistent between adjacent segments;

[0052] If the directions are consistent, adjacent segments are merged into a continuous fluctuation event, and the number of segments covered by the event and the duration are recorded.

[0053] If the direction swings back and forth repeatedly, it is not directly regarded as an oscillation, but an event splitting action is triggered: the segment sequence is divided into multiple smaller fluctuation events based on the direction switching point, and marked as internal swing-type events.

[0054] For fluctuation events, read the source pruning mark and the position attributes of candidate segments within start and stop segments;

[0055] If an event contains a large number of source trimmed segments, or if the event is mainly distributed near the start / end of start / stop segments, then add a boundary proximity marker to the event;

[0056] If the events are mainly distributed in the middle of the start-stop segment and the segment sources are all unpruned, then add a middle segment stability tag;

[0057] Read the attributes of the reference segment corresponding to the structured segment unit in the bound sequence, such as whether it is pruned and the confidence level, and use them as the reference confidence marker for the fluctuation event to obtain the fluctuation event sequence;

[0058] The cross-unit continuity discrimination and oscillation identification generation unit reads oscillation events in the time sequence of structured segment units and establishes a comparison relationship between adjacent units: each structured segment unit is only compared with the adjacent units before and after it; to avoid background changes causing interference by crossing too many units; when comparing, the main oscillation event in each unit is selected first, and the one with the largest coverage and higher reference reliability is selected as the representative event of the unit; the remaining events are used as auxiliary evidence and only participate in the judgment when the representative event is unclear.

[0059] Perform continuity determination on representative events of multiple adjacent structured fragment units:

[0060] When multiple consecutive structured segment units all exhibit representative events with the same direction and shape, and none of them have boundary proximity markers, a rhythmic repetition chain is formed.

[0061] When the rhythmic repetition chain reaches the preset continuity requirement (which is manifested as "at least crossing multiple start and stop units and having a low number of interruptions"), then a rhythmic oscillation identifier is generated on all structured segment units covered by the rhythmic repetition chain.

[0062] When a structured segment unit has a representative event, but no event with the same direction and shape can be found in the adjacent units before and after it, the structured segment unit is classified as a single-segment oscillation.

[0063] When an event is marked with a boundary proximity marker and the event energy is mainly concentrated near the boundary of the start-stop segment, it is marked as a boundary oscillation marker, regardless of whether it repeats in adjacent units.

[0064] The obtained rhythmic oscillation identifiers, single-segment oscillation identifiers, and boundary oscillation identifiers are fitted to obtain oscillation identifiers.

[0065] Preferably, the SOC compensation instruction generation module includes a compensation instruction construction and intra-segment execution unit and a reference segment consistency verification unit;

[0066] The compensation instruction construction and intra-fragment execution unit are based on the structured fragment index sequence, reading the structured fragment unit and oscillation identifier one by one;

[0067] When the oscillation indicator is a rhythmic oscillation indicator, a standard compensation instruction is generated to locally correct the SOC estimation path inside the start and stop segments of the structured segment unit. This instruction is explicitly limited to: only locally correcting the SOC estimation path inside the start and stop segments of the structured segment unit, and requiring that the correction start and end boundaries be consistent with the start and stop segment boundaries to avoid compensation affecting the reference segment.

[0068] When the oscillation indicator is a single-segment oscillation indicator, no compensation instruction is generated; instead, an observation and recording instruction is generated, which only writes the observation status and the location of occurrence, without rewriting the SOC path; this prevents occasional disturbances from being treated as repeatable patterns.

[0069] When the oscillation indicator is a boundary oscillation indicator, a restrictive compensation instruction is generated. The restrictive compensation instruction has stricter action constraints: compensation is only allowed to take effect in the boundary area of ​​the start and stop segment, and it is only allowed to start if conditions such as "the proportion of boundary-adjacent events in this unit is not high and the confidence level of the reference segment is high" are met simultaneously; otherwise, the restrictive compensation instruction is only recorded as "to be enabled" and is not executed directly.

[0070] For standard compensation instructions and restrictive compensation instructions that have met the conditions, perform local correction of the SOC path within the segment, and simultaneously perform three types of boundary control:

[0071] Fragment boundary locking: Compensation is only allowed to take effect within the current start / stop fragment, and the start and end boundaries cannot be crossed;

[0072] Window continuity locking: If there are multiple internal resistance candidate segments within the start / stop segment, the compensation action is organized in the order of the candidate segments, and continuous correction is not allowed to cross the boundary interference interval.

[0073] State consistency locking: If a sudden change in the operation state occurs in the start-stop segment, such as a sudden switch to the lifting state, the compensation within that start-stop segment is split into multiple sub-segments for execution, and the reason for the split is recorded.

[0074] After the oscillation flag is executed, obtain the candidate compensation SOC status and compensation execution record.

[0075] Candidate compensated SOC state: contains the SOC output of the structured fragment unit, which may be compensated or uncompensated and maintained;

[0076] Compensation execution log: Records instruction type, standard / observation / restrictive, whether it was actually executed, execution coverage, whether it was split, whether the triggering condition was met, etc.; this execution log will be directly used by the next unit for consistency verification and rollback determination;

[0077] Preferably, the reference fragment consistency verification unit reads the candidate compensated SOC status and compensation execution record, retrieves a unique reference fragment from the bound sequence based on the structured fragment unit identifier, and checks whether there is a mutual explanatory relationship between the compensated SOC change trend and the power performance of the reference fragment; this includes two steps:

[0078] Directional consistency check: If the power performance of the reference segment is in a continuous state of depletion, the compensated SOC should not show a reverse change that contradicts the state of depletion in adjacent time periods.

[0079] Continuity check: If the power performance of the reference segment is stable, the compensated SOC should not exhibit abrupt jumps within the corresponding time span of the reference segment;

[0080] If the direction consistency fails, it will be marked as a direction backtracking indicator, indicating that the compensation direction and the power performance of the reference segment are inconsistent;

[0081] If the continuity fails, it will fall under the continuity regression indicator, indicating that the compensation has introduced a jump that does not conform to the stationary characteristics of the reference segment;

[0082] If the credibility level of the reference fragment itself is insufficient, it will fall into the insufficient reference back-off mark, indicating that the verification evidence is insufficient rather than the compensation necessarily being wrong;

[0083] After triggering the rollback, the compensation effect already executed within this structured segment unit is undone, the SOC output is restored to the path before compensation, and the undone action is written to the end of the compensation execution record, forming a closed-loop evidence chain.

[0084] If the rollback verification passes: Output the compensated SOC status, effective status, and rollback flag is empty;

[0085] If the rollback verification fails: output the compensated SOC status, rollback status, which is equivalent to the uncompensated state, and output the corresponding rollback flag.

[0086] Preferably, the power supply control linkage module includes a SOC path selection and power supply strategy reorganization unit and a compensation backoff and power supply recovery and start / stop boundary rule adjustment unit;

[0087] The SOC path selection and power supply strategy reorganization unit receives the compensated SOC status and rollback flag, reads the SOC output status corresponding to the current running segment in the structured segment order, and checks whether the SOC status is accompanied by a rollback.

[0088] When the SOC compensation result is in a stable state and no rollback occurs, the compensated SOC will be used as the basis for current power supply judgment, and the power supply control process will be reorganized based on the SOC state.

[0089] During this process, power supply control does not directly change equipment actions. Instead, it first restructures the existing control strategy to ensure that the power limit range, permissible conditions for operation actions, and task execution priority are consistent with the compensated State of Charge (SOC) state. For example, when the SOC is in a lower range, the system will prioritize limiting high-load operation and gradually transition the operation process to a low-energy-consumption state.

[0090] When a rollback is detected, the current compensation result fails the consistency verification. The compensated SOC path is stopped, and the original SOC judgment method is restored. The power supply control returns to the original operating logic and generates the power supply control result for the current cycle, outputting the updated power supply control strategy.

[0091] To avoid repeated switching of power supply strategies during frequent start-stop processes, the system maintains the control state of the previous cycle for a short period during recovery, so that the power supply adjustment is a continuous change rather than an instantaneous jump.

[0092] The compensation rollback and power-on / shutdown boundary rule adjustment unit synchronously records whether SOC compensation is effective, whether rollback has occurred, and the location characteristics of rollback occurrence within the current structured segment during the power supply control strategy execution process.

[0093] When rollbacks are found to be concentrated in the beginning or end regions of start-stop segments, it indicates that the original segment boundaries contain unstable operating phases. In subsequent operations, the confirmation of start-stop boundaries should be delayed to make the segment division closer to the actual stable range.

[0094] When a rollback occurs near a job status transition, the continuous confirmation of the job's stable state should be strengthened during subsequent segment division; avoid including brief action transition processes in stable operation segments.

[0095] If no rollback occurs in multiple consecutive running segments, the existing segment division method remains unchanged, keeping the structured segment index stable. The feedback process does not immediately change the division rules but gradually corrects them through continuous running results, ensuring that the formation process of start / stop segments and reference segments continuously approaches the actual operating rhythm. After multiple runs, the structured segment index can more stably reflect the actual start / stop behavior of the device, thereby reducing the possibility of subsequent SOC compensation being revoked. The processed running feedback results are then sent to the front-end segment generation process for segment division adjustment in the next running cycle.

[0096] A method for intelligent power supply control of a forklift mobile robot based on dynamic analysis includes the following steps:

[0097] Step 1: The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment.

[0098] Step 2: The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence.

[0099] Step 3: The internal resistance oscillation identification module, based on the bound sequence and the set of internal resistance candidate segments, performs continuity judgment on the changes in internal resistance between candidate segments to obtain oscillation indicators;

[0100] Step 4: After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag.

[0101] Step 5: The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

[0102] This invention provides an intelligent power supply control method and system for a forklift mobile robot based on dynamic analysis, which has the following beneficial effects:

[0103] (1) When the system is running, by further performing hierarchical screening and structured binding on the voltage and current change process based on the start-stop segment, the extraction process of battery internal resistance change is transformed from the original overall statistical method to a fine identification method for the running stage, thereby producing the following improvement effect and technical advantages, and forming a corresponding improvement to the problem of SOC estimation under frequent start-stop conditions mentioned above.

[0104] By employing a synchronous change window identification process, the system identifies continuous intervals within the start-stop segment where voltage and current exhibit a synchronous change relationship. This limits the data source for subsequent analysis to the actual energy change phase, excluding non-corresponding change intervals caused solely by current disturbances or control fluctuations. This approach avoids incorporating numerous invalid fluctuations during low-load operation into the battery state assessment process, ensuring that internal resistance change extraction is based on physically correlated operating segments, thus reducing the impact of random fluctuations on SOC estimation from the outset.

[0105] (2) Through the process of sequential comparison and fluctuation event extraction within the unit, multiple candidate segments of internal resistance that were originally scattered are merged and sorted by time according to the structured segment units, so that the internal resistance change is transformed from discrete segments into a change sequence with a continuous relationship. On this basis, the continuous change trend in the same direction is identified, and adjacent change segments are merged into a unified fluctuation event, so that the system can observe the battery state change process from the operation stage level, rather than relying on single-point change judgment, thereby reducing the situation where random fluctuations are mistaken for abnormal changes during low-load operation.

[0106] Through an event splitting mechanism, when the direction of internal resistance change repeatedly swings back and forth, it is not directly judged as oscillation. Instead, it is split and processed according to the position of the change, so that the local fluctuations caused by control actions or transient disturbances can be independently identified. This method can effectively distinguish between the actual internal resistance drift behavior and the short-term disturbances generated during the start-stop transition phase, making subsequent analysis closer to the internal state change law of the battery.

[0107] (3) By further introducing SOC compensation generation, operation consistency verification and power supply control linkage adjustment mechanism on the basis of internal resistance oscillation identification results, the battery state correction process is transformed from the traditional single estimation correction method into a dynamic adjustment process with verification and feedback capabilities, thereby generating the following improvement effects and technical advantages corresponding to the aforementioned background problems.

[0108] (4) By constructing compensation instructions and executing them within segments, differentiated processing methods are adopted according to different oscillation types, so that SOC correction behavior has a clear triggering basis. When rhythmic oscillation is identified, local compensation is only performed within the corresponding start-stop segment, so that the compensation range is limited to the operation stage that generates internal resistance drift; for single-segment oscillation, only recording is performed without intervening in the SOC path, thereby avoiding the impact of occasional fluctuations on the power judgment; for boundary oscillation, a restrictive compensation method is adopted, so that the impact of the start-stop transition stage is controlled. This categorized processing method enables SOC correction to respond to continuous micro-fluctuations formed under frequent start-stop conditions without expanding to the overall operation process, which is conducive to reducing the short-term SOC cumulative deviation mentioned in the background. Attached Figure Description

[0109] Figure 1 This is a block diagram and flowchart of an intelligent power supply control system for a fork-type mobile robot based on dynamic analysis, according to the present invention.

[0110] Figure 2 This is a schematic diagram illustrating the steps of an intelligent power supply control method for a fork-type mobile robot based on dynamic analysis according to the present invention.

[0111] Figure 3 This is a schematic diagram of the oscillation identifier reading process of the present invention;

[0112] Figure 4 This is a schematic diagram illustrating the process for determining the compensated SOC status and rollback flag according to the present invention. Detailed Implementation

[0113] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0114] Example 1

[0115] This invention provides an intelligent power supply control system for a forklift mobile robot based on dynamic analysis. Please refer to [link / reference]. Figure 1 It includes a start / stop segment and reference segment generation module, an internal resistance candidate segment screening module, an internal resistance oscillation identification module, a SOC compensation command generation module, and a power supply control linkage module;

[0116] The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment.

[0117] The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence;

[0118] The internal resistance oscillation identification module is based on the binding sequence and the set of internal resistance candidate segments. It performs continuous discrimination on the changes in internal resistance between candidate segments to obtain oscillation indicators.

[0119] After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag.

[0120] The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

[0121] In this embodiment, by introducing a dynamic analysis mechanism based on operating rhythm into the power supply control process of the forklift mobile robot, the battery status judgment process no longer relies solely on the voltage or current performance at a single moment, but instead performs segmented organization and correlation analysis around the actual operation process of the equipment. This results in a power supply control method that is more consistent with the actual operating conditions under frequent start-stop conditions. The beneficial effects it can produce are mainly reflected in the following aspects:

[0122] The entire machine operation process is structured by generating start / stop segments and reference segments. This reorganizes the originally continuous and mixed operation data into operation segments with clear start / stop boundaries and stable reference relationships. This distinguishes and processes transient disturbances generated during frequent start / stop processes from stable operating states, preventing electrical changes at the moment of start / stop from directly participating in battery status judgment. This gives SOC estimation a clear operational background basis, which matches the intermittent operation characteristics of the forklift mobile robot.

[0123] By leveraging the synergistic effect of the internal resistance candidate segment screening module and the internal resistance oscillation identification module, the relationship between internal voltage and current changes during the start-stop phase can be selectively screened without adding sensors. Furthermore, the repeated manifestation of internal resistance changes in multiple operating segments can be continuously identified. This allows micro-fluctuations in internal resistance that are difficult to directly identify under low load conditions to be extracted separately. This addresses the deficiency in existing power supply systems that do not identify short-cycle internal resistance drift, enabling the SOC estimation process to reflect subtle changes in the internal state of the battery under frequent start-stop conditions.

[0124] The SOC compensation instruction generation module performs categorized compensation based on the oscillation identification results and introduces a consistency verification process after compensation. This ensures that the compensation behavior is not directly used as the final result, but is verified under the constraints of a reference operating segment. When the compensation result is inconsistent with the actual operating performance, it can be automatically revoked and the original judgment path restored, thereby reducing continuous erroneous corrections caused by local abnormal fluctuations and maintaining a stable and continuous SOC change trend. The power supply control linkage module directly incorporates the compensated SOC state into the power supply strategy adjustment, enabling power limiting, job execution, and operation scheduling to respond based on the verified power status. At the same time, the compensation execution results and rollback status are fed back to the front-end operating segment division process, gradually aligning the start-stop boundaries with the actual operating rhythm of the equipment, forming a closed-loop correlation between power supply judgment and operating behavior.

[0125] This embodiment enables the forklift mobile robot to segmentally identify, verify, and control the micro-changes in battery internal resistance under conditions of frequent start-stop and low-load alternating operation. This reduces the cumulative deviation of SOC estimation during short-term operation and keeps the power supply control process consistent with the actual operating state. This improves the power supply judgment deviation problem caused by the failure to consider the micro-fluctuations during start-stop in the prior art.

[0126] Example 2

[0127] Please refer to Figure 1 Specifically: the start / stop segment and reference segment generation module includes a drive job state boundary identification unit and a structured segment index construction unit;

[0128] The drive operation state boundary recognition unit reads the existing state quantities of the fork mobile robot controller, including: drive enable state, drive speed command valid / invalid state, brake release state, lifting / lowering operation state, and steering command active state; and converts them into unified discrete state words to form a continuous state sequence.

[0129] When a change in the drive enable state from off to on or from on to off is detected, the boundary verification process is initiated, which includes two consecutive actions: triggering validity verification and stability solidification.

[0130] Trigger validity check: This determines whether the change in drive enable state is accompanied by speed command activation or brake release. If it is only a control signal jitter and no actual drive action is formed, then no start / stop boundary is generated.

[0131] Stability solidification: After triggering, continuously observe the changes in the driving state. If a reverse swing or multiple rapid switches occur in a short period of time, the interval is merged into a single start-stop event, and the boundary confidence level is recorded. If the swing is frequent, it is marked as a boundary interference interval.

[0132] When the drive enable state enters a continuous stable phase and the operation state is no longer in the lifting, lowering, or steering jogging switching range, the end boundary of the start-stop segment is confirmed, a start-stop segment with start and end times and a reliable marker is formed, and the start-stop segment is output as a set of start-stop segments in chronological order.

[0133] For each start-stop segment, the structured segment index building unit first searches for candidate reference intervals in the adjacent intervals before and after it. During the search process, it excludes areas with lifting switching, braking switching, or frequent steering inching, and only retains continuous segments where the drive command direction is stable and the speed command maintains a low load interval.

[0134] When a brief disturbance occurs within a candidate reference interval, the candidate interval is not discarded directly. Instead, interval pruning is performed, using the disturbance point as the dividing point. The longest sub-intervals on both sides that satisfy the continuous stability condition are extracted, and the pruning behavior is recorded to retain the largest usable stable segment.

[0135] After obtaining the first and second candidate reference intervals, a unique matching decision process is executed: the candidate interval that is consistent with the running direction of the start and stop segments is selected first;

[0136] If the directions are consistent, select the interval that is closer to the start and stop segment time and has not been pruned;

[0137] If multiple intervals satisfy the conditions, the one with the longest continuity is selected; finally, only one reference segment is output, and a unique identifier is generated.

[0138] The unique identifier of the start / stop segment is bound to the reference segment identifier, encapsulated into a structured segment unit, and segment attribute information is attached;

[0139] Write all structured fragment units into the structured fragment index sequence in the order of start and stop, and establish a bidirectional index relationship to obtain the structured fragment index sequence.

[0140] In this embodiment, by introducing a start-stop segment identification and reference segment construction mechanism based on the driving operation state during the operation of the forklift mobile robot, the battery state analysis is no longer based on undifferentiated continuous operation data, but rather on subsequent processing based on a clear division of operation stages. This results in the following improvements and technical advantages:

[0141] By using a drive operation state boundary recognition unit to jointly determine the drive enable state, speed command state, and operation action state, the identification of start-stop behavior no longer relies solely on changes in a single control signal. Instead, it combines confirmation of whether the actual drive action has occurred, thus distinguishing between control signal jitter and genuine start-stop processes. This approach avoids the missegmentation problem caused by instantaneous changes in the controller state in frequent start-stop scenarios, making the boundaries of operational segments more closely match the actual operational rhythm of the forklift robot, and providing a stable time basis for subsequent battery state analysis.

[0142] By introducing stability solidification processing during the start-stop boundary confirmation process, short-term state swings are merged into the same start-stop event, and frequent swing regions are marked with interference, thereby reducing the repeated identification of start-stop transient disturbances. This processing method can effectively isolate the unstable operating range generated during the start-stop transition phase, allowing subsequent analysis to be mainly based on the stable operating segment, thus improving the source of bias caused by the direct participation of start-stop transients in SOC estimation in existing technologies.

[0143] By employing a unique matching decision-making process, multiple candidate reference intervals are comprehensively screened based on directional consistency, temporal proximity, and continuity. This ensures a clear one-to-one correspondence between reference segments and their corresponding start / stop segments, guaranteeing that subsequent internal resistance change analysis is always conducted under the same operational context. Compared to existing methods where the source of reference intervals is not fixed, this structured matching method makes the operational control relationship more stable.

[0144] By establishing structured segment units and constructing bidirectional index relationships, a traceable association structure is formed between start / stop segments and reference segments. This enables the system to quickly locate the source of the operational phase in subsequent modules and continuously correct the segment division during power supply control feedback. This structured organization transforms the original continuous mixed state of operational data into an operational sequence with a clear hierarchical relationship, laying the foundation for solving the problem of difficulty in identifying micro-fluctuations in battery internal resistance under frequent start / stop conditions. This provides a targeted improvement to the short-term cumulative error problem of SOC mentioned in the background.

[0145] Example 3

[0146] Please refer to Figure 1 Specifically: the internal resistance candidate segment screening module includes a synchronous change window identification and boundary interference interval elimination unit and a direction consistency confirmation and reference segment binding sequence generation unit;

[0147] The synchronous change window identification and boundary interference interval elimination unit reads the structured fragment unit one by one according to the structured fragment index sequence and enters the start and stop fragment to perform the identification action;

[0148] Within the start-stop segment, find the continuous interval where voltage and current change synchronously:

[0149] If the current changes significantly but the voltage does not change accordingly, it is judged as an asynchronous interval and is not considered as a candidate window.

[0150] If the voltage and current show a consistent trend of change within the same time period, then the continuous segments are cut out to form a preliminary draft of the candidate window.

[0151] The incomparable boundaries that may exist within the start-stop segments are located, and a list of boundary interference intervals is generated, specifically including three categories:

[0152] Contactor switching interference range: Centered on the contactor state change, a short buffer zone extends forward and backward, and the entire area is marked as unavailable;

[0153] Braking switching interference range: The buffer zone is also extended with the braking release / re-braking switching point as the center;

[0154] Command jump interference range: When a speed command or torque command experiences a sudden change and a short-duration "spiking behavior", the area before and after the spike will be merged into the interference range;

[0155] Each of the initially selected candidate windows was compared with the boundary interference interval:

[0156] If a candidate window falls entirely within the interference range, it is discarded.

[0157] If the candidate window partially overlaps with the interference interval, then perform cut-off and retention trimming: remove the overlapping part, retain only the non-overlapping sub-intervals that still maintain continuity, and use the trimming result as a new candidate window;

[0158] If the pruned subintervals are too short or fragmented, they are marked as unusable windows.

[0159] The direction consistency confirmation and reference segment binding sequence generation unit processes the candidate window. When the candidate window meets the synchronous change, but the voltage and current directions are opposite or misaligned, the direction consistency confirmation is performed on the candidate window.

[0160] The steps for consistency verification are: only retain windows where the voltage and current change in the same direction within the same window;

[0161] For windows with brief directional swings, if the swing occurs at the window edge, a secondary trim is performed to extract the main continuous segments with the same direction.

[0162] If the swing occurs in the middle of the window and causes the window to split, it is judged as an unmatchable window and is removed;

[0163] For each candidate window that has passed the directional consistency confirmation, it is not directly announced as an internal resistance candidate segment. Instead, it is returned to the structured segment index sequence to find the structured segment unit to which it belongs, read the attribute information of the corresponding reference segment, and confirm whether the window has reference significance.

[0164] If a reference fragment is marked as trimmed and has a low confidence level, then the conservative retention rule is applied, allowing only fragments with high window quality to be retained.

[0165] If the reference fragment has a high credibility level, the window is retained according to the normal rules;

[0166] For windows that meet the directional consistency and control conditions, they are solidified as internal resistance candidate segments and a unique number is generated; multiple internal resistance candidate segments are allowed within the same start-stop segment, and the sequence number is kept to increase, and it is recorded whether the source window of each candidate segment has undergone pruning;

[0167] For each candidate segment of internal resistance, establish a clear binding relationship:

[0168] Bind to the identifier of the start / stop segment; further bind to the identifier of the structured segment unit to which the start / stop segment belongs; and bind to the unique reference segment identifier within the structured segment unit; organize the binding relationships according to the structured segment index sequence to form a binding sequence.

[0169] In this embodiment, by further performing hierarchical screening and structured binding processing on the voltage and current change process based on the start-stop segment, the extraction process of battery internal resistance change is transformed from the original overall statistical method to a fine identification method for the operation stage. This can produce the following improvement effects and technical advantages, and form a corresponding improvement to the problem of SOC estimation easily generating cumulative deviation under frequent start-stop conditions mentioned in the background.

[0170] By employing a synchronous change window identification process, the system identifies continuous intervals within the start-stop segment where voltage and current exhibit a synchronous change relationship. This limits the data source for subsequent analysis to the actual energy change phase, excluding non-corresponding change intervals caused solely by current disturbances or control fluctuations. This approach avoids incorporating numerous invalid fluctuations during low-load operation into the battery state assessment process, ensuring that internal resistance change extraction is based on physically correlated operating segments, thus reducing the impact of random fluctuations on SOC estimation from the outset.

[0171] By setting multiple boundary interference intervals, such as contactor switching, braking switching, and command jumps, and performing comparison and pruning on candidate windows during the identification process, transient electrical disturbances generated during the start-stop transition phase are actively isolated, rather than passively involved in the analysis. Compared to the existing technology that directly uses continuous sampling data for state estimation, this embodiment can effectively eliminate short-term abnormal segments caused by start-stop control actions, making the extracted operating segments more stable, thereby reducing the possibility of misjudgment in internal resistance determination under frequent start-stop conditions.

[0172] By confirming directional consistency, the directions of voltage and current changes are further constrained, ensuring that only windows exhibiting consistent energy change characteristics are retained. Simultaneously, edge sway patterns are pruned to preserve the main continuous change segments. This approach frees candidate internal resistance segments from local disturbances, enabling a more accurate reflection of the battery's response characteristics under real load changes, and facilitating the identification of the micro-fluctuation range of internal resistance mentioned in the background.

[0173] Example 4

[0174] Please refer to Figure 1 Specifically: the internal resistance oscillation identification module includes an intra-unit sequence comparison and fluctuation event extraction unit and a cross-unit continuity discrimination and oscillation identifier generation unit;

[0175] The unit reads the binding sequence of the sequential comparison and fluctuation event extraction unit, merges all internal resistance candidate segments according to the structured segment unit to which they belong, and sorts the internal resistance candidate segments according to the occurrence time within each structured segment unit to form a segment sequence;

[0176] Perform order comparison on the fragment sequence within the same structured fragment unit:

[0177] First, identify whether there is a main trend of continuous change in the same direction in the segment sequence, that is, whether the direction of change of internal resistance is consistent between adjacent segments;

[0178] If the directions are consistent, adjacent segments are merged into a continuous fluctuation event, and the number of segments covered by the event and the duration are recorded.

[0179] If the direction swings back and forth repeatedly, it is not directly regarded as an oscillation, but an event splitting action is triggered: the segment sequence is divided into multiple smaller fluctuation events based on the direction switching point, and marked as internal swing-type events.

[0180] For fluctuation events, read the source pruning mark and the position attributes of candidate segments within start and stop segments;

[0181] If an event contains a large number of source trimmed segments, or if the event is mainly distributed near the start / end of start / stop segments, then add a boundary proximity marker to the event;

[0182] If the events are mainly distributed in the middle of the start-stop segment and the segment sources are all unpruned, then add a middle segment stability tag;

[0183] Read the attributes of the reference segment corresponding to the structured segment unit in the bound sequence and use them as a reliable reference marker for the fluctuation event to obtain the fluctuation event sequence;

[0184] The cross-unit continuity discrimination and oscillation identification generation unit reads oscillation events in the time sequence of structured segment units and establishes a comparison relationship between adjacent units: each structured segment unit is only compared with its immediate and adjacent units; when comparing, the main oscillation event within each unit is selected first as the representative event of that unit; the remaining events are used as auxiliary evidence and only participate in the judgment when the representative event is unclear.

[0185] Perform continuity determination on representative events of multiple adjacent structured fragment units:

[0186] When multiple consecutive structured segment units all exhibit representative events with the same direction and shape, and none of them have boundary proximity markers, a rhythmic repetition chain is formed.

[0187] When the rhythmic repetition chain reaches the preset continuity requirement, rhythmic oscillation markers are generated on all structured segment units covered by the rhythmic repetition chain.

[0188] When a structured segment unit has a representative event, but no event with the same direction and shape can be found in the adjacent units before and after it, the structured segment unit is classified as a single-segment oscillation.

[0189] When a representative event has a boundary proximity marker and the event energy is mainly concentrated near the boundary of the start-stop segment, it is marked as a boundary oscillation marker regardless of whether it repeats in adjacent units; the obtained rhythmic oscillation markers, single-segment oscillation markers and boundary oscillation markers are fitted to obtain the oscillation marker.

[0190] In this embodiment, by further constructing an oscillation identification process across operating segments based on the candidate segments of internal resistance, the change in battery internal resistance is no longer judged based on a single fluctuation, but is analyzed continuously in combination with the actual start-stop rhythm of the forklift mobile robot. This can produce the following improvement effect and specifically address the problem of SOC estimation under frequent start-stop conditions that is prone to short-term cumulative error as described in the background.

[0191] By employing intra-unit sequential comparison and fluctuation event extraction processes, multiple previously dispersed internal resistance candidate segments are grouped and time-ordered according to structured segment units, transforming internal resistance changes from discrete segments into a continuous sequence of changes. Based on this, continuous trends in the same direction are identified, and adjacent change segments are merged into unified fluctuation events. This allows the system to observe battery state changes at the operational stage level, rather than relying on single-point changes for judgment, thereby reducing the likelihood of random fluctuations being mistaken for abnormal changes during low-load operation.

[0192] Through an event splitting mechanism, when the direction of internal resistance change repeatedly swings back and forth, it is not directly judged as oscillation. Instead, it is split and processed according to the position of the change, so that the local fluctuations caused by control actions or transient disturbances can be independently identified. This method can effectively distinguish between the actual internal resistance drift behavior and the short-term disturbances generated during the start-stop transition phase, making subsequent analysis closer to the internal state change law of the battery.

[0193] By attaching boundary proximity markers, mid-segment stability markers, and reliable reference markers to fluctuation events, each fluctuation event possesses operational location and reference environment information, enabling the identification of the operational background that generates the fluctuation during subsequent judgment. This processing method allows the system to determine whether changes in internal resistance originate from the influence of start-stop boundaries or changes during stable operation, which is beneficial for identifying the micro-fluctuation ranges formed by frequent start-stops mentioned in the background. By identifying changes occurring only in a single operational segment as single-segment oscillations and separately marking changes concentrated near start-stop boundaries as boundary oscillations, fluctuations from different sources are treated differently, preventing boundary disturbances or sporadic changes from entering the subsequent SOC correction process, thereby reducing the generation of instability compensation.

[0194] This embodiment utilizes the synergistic effect of intra-cell event extraction and cross-cell continuity discrimination to enable battery internal resistance changes to be identified and classified along the operational rhythm dimension. This achieves structured identification of minute internal resistance drifts under frequent start-stop conditions, providing a reliable basis for subsequent SOC compensation. This improves the problem in existing technologies where SOC estimation gradually deviates from the actual state due to the failure to distinguish the source of fluctuations.

[0195] Example 5

[0196] Please refer to Figure 1 Specifically: the SOC compensation instruction generation module includes a compensation instruction construction and intra-fragment execution unit and a reference fragment consistency verification unit;

[0197] The compensation instruction construction and intra-fragment execution unit are based on the structured fragment index sequence, reading the structured fragment unit and oscillation identifier one by one;

[0198] When the oscillation identifier is a rhythmic oscillation identifier, a standard compensation instruction is generated to locally correct the SOC estimation path inside the start and stop segments of the structured segment unit.

[0199] When the oscillation indicator is a single-segment oscillation indicator, no compensation instruction is generated, and instead an observation and recording instruction is generated. Only the observation status and the location of occurrence are written, and the SOC path is not modified.

[0200] When the oscillation indicator is a boundary oscillation indicator, a restrictive compensation instruction is generated;

[0201] For standard compensation instructions and restrictive compensation instructions that have met the conditions, perform local correction of the SOC path within the segment, and simultaneously perform three types of boundary control:

[0202] Fragment boundary locking: Compensation is only allowed to take effect within the current start / stop fragment, and the start and end boundaries cannot be crossed;

[0203] Window continuity locking: If there are multiple internal resistance candidate segments within the start / stop segment, the compensation action is organized in the order of the candidate segments, and continuous correction is not allowed to cross the boundary interference interval.

[0204] State consistency locking: If a sudden change in job state occurs in the start / stop segment, the compensation within the start / stop segment is split into multiple sub-segments for execution, and the reason for the split is recorded.

[0205] After the oscillation flag is executed, obtain the candidate compensation SOC status and compensation execution record.

[0206] The reference fragment consistency verification unit reads the candidate compensated SOC status and compensation execution record, retrieves a unique reference fragment from the bound sequence based on the structured fragment unit identifier, and checks whether there is a mutual explanatory relationship between the compensated SOC change trend and the power performance of the reference fragment; this includes two steps:

[0207] Directional consistency check: If the power performance of the reference segment is in a continuous state of depletion, the compensated SOC should not show a reverse change that contradicts the state of depletion in adjacent time periods.

[0208] Continuity check: If the power performance of the reference segment is stable, the compensated SOC should not exhibit abrupt jumps within the corresponding time span of the reference segment;

[0209] If the direction consistency fails, it will be marked as a direction backtracking indicator, indicating that the compensation direction and the power performance of the reference segment are inconsistent;

[0210] If the continuity fails, it will fall under the continuity regression indicator, indicating that the compensation has introduced a jump that does not conform to the stationary characteristics of the reference segment;

[0211] If the credibility level of the reference fragment itself is insufficient, it will fall into the insufficient reference back-off mark, indicating that the verification evidence is insufficient rather than the compensation necessarily being wrong;

[0212] After triggering the rollback, the compensation effect already executed within this structured segment unit is undone, the SOC output is restored to the path before compensation, and the undone action is written to the end of the compensation execution record, forming a closed-loop evidence chain.

[0213] If the rollback verification passes: Output the compensated SOC status, with the rollback flag empty;

[0214] If the rollback verification fails: output the compensated SOC status and the corresponding rollback flag.

[0215] The power supply control linkage module includes a SOC path selection and power supply strategy reorganization unit and a compensation backoff and recovery and start / stop boundary rule adjustment unit;

[0216] The SOC path selection and power supply strategy reorganization unit receives the compensated SOC status and rollback flag, reads the SOC output status corresponding to the current running segment in the structured segment order, and checks whether the SOC status is accompanied by rollback.

[0217] When the SOC compensation result is in a stable state and no rollback occurs, the compensated SOC will be used as the basis for current power supply judgment, and the power supply control process will be reorganized based on the SOC state.

[0218] When a rollback is detected, the current compensation result fails the consistency verification. The compensated SOC path is stopped, and the original SOC judgment method is restored. The power supply control returns to the original operating logic and generates the power supply control result for the current cycle, outputting the updated power supply control strategy.

[0219] The compensation rollback and power-on / shutdown boundary rule adjustment unit synchronously records whether SOC compensation is effective, whether rollback has occurred, and the location characteristics of rollback occurrence within the current structured segment during the power supply control strategy execution process.

[0220] When rollbacks are found to be concentrated in the start or end regions of start / stop segments, it indicates that the original segment boundaries contain unstable operating phases, and the confirmation of start / stop boundaries should be delayed in subsequent operations.

[0221] When a rollback occurs near the start of a job status transition, the continuous confirmation of the job's stable state should be strengthened during subsequent segmentation processes.

[0222] If no rollback occurs in multiple consecutive running segments, the existing segment division method remains unchanged.

[0223] In this embodiment, by further introducing SOC compensation generation, operational consistency verification, and power supply control linkage adjustment mechanisms based on the internal resistance oscillation identification results, the battery state correction process is transformed from the traditional single-estimate correction method into a dynamic adjustment process with verification and feedback capabilities, thereby producing the following improvements and technical advantages corresponding to the aforementioned background problems.

[0224] By constructing compensation instructions and executing them within segments, differentiated processing methods are adopted based on different oscillation types, giving SOC correction behavior a clear triggering basis. When rhythmic oscillations are identified, local compensation is performed only within the corresponding start-stop segment, limiting the compensation range to the operating phase that generates internal resistance drift. For single-segment oscillations, only recording is performed without intervening in the SOC path, thus avoiding the impact of occasional fluctuations on power assessment. For boundary oscillations, a restrictive compensation method is used to control the impact of the start-stop transition phase. This categorized processing method enables SOC correction to respond to continuous micro-fluctuations formed under frequent start-stop conditions without amplifying them to the overall operating process, which helps reduce the short-term SOC cumulative deviation mentioned in the background.

[0225] By employing segment boundary locking, window continuity locking, and state consistency locking, the compensation process is always constrained by the operating segments, preventing corrections from crossing start / stop boundaries or interference intervals. This ensures that the SOC change process remains consistent with the actual operating state. This approach avoids the new estimation fluctuations introduced by excessively wide correction ranges in traditional compensation methods, making SOC adjustments more stable and continuous.

[0226] By employing a reference segment consistency verification mechanism, the SOC change trend is re-verified after compensation, ensuring that the compensation result corresponds to the power performance during stable operation before being retained. When the compensation direction is inconsistent with actual energy consumption or undergoes abrupt changes, compensation can be automatically revoked, generating a rollback flag to prevent erroneous compensation from accumulating in subsequent operations. This approach transforms SOC compensation from a one-time correction into a verifiable and revocable process, effectively addressing the problem of long-term accumulation of compensation errors in existing technologies.

[0227] Example 6

[0228] A method for intelligent power supply control of a forklift mobile robot based on dynamic analysis is described in the following reference: Figure 2 Specifically, it includes the following steps:

[0229] Step 1: The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment.

[0230] Step 2: The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence.

[0231] Step 3: The internal resistance oscillation identification module, based on the bound sequence and the set of internal resistance candidate segments, performs continuity judgment on the changes in internal resistance between candidate segments to obtain oscillation indicators;

[0232] Step 4: After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag.

[0233] Step 5: The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

[0234] In this embodiment, by changing the power supply control process of the forklift mobile robot from the traditional control method based on single power estimation to a phased processing method based on dynamic analysis of the operation process, the battery status judgment can be continuously corrected in combination with the actual start-stop rhythm of the equipment, thereby producing the following improvement effects and technical advantages corresponding to the aforementioned background problems.

[0235] Step 1 divides the operation of the forklift mobile robot into start-stop segments and reference segments, transforming the overall operating state from continuous mixed data into a structured sequence with clearly defined operating stages. This allows for the differentiation between start-stop transition phases and stable operating phases. This processing method prevents transient disturbances generated during frequent start-stop processes from directly participating in battery state assessment, establishing a stable operating benchmark for subsequent analysis and improving the problem of fluctuating SOC estimation due to mixed operating states in existing technologies.

[0236] Secondly, by screening candidate internal resistance segments within the start-stop segment in step two and establishing a binding relationship with the structured segments, voltage and current changes are extracted and analyzed only within the meaningful operating range, thereby eliminating interference from contactor switching, braking changes, and sudden changes in control commands. This method makes the acquisition of battery internal resistance changes closer to the actual load change process, which is beneficial for identifying subtle changes in internal resistance that are difficult to detect under low load conditions.

[0237] Furthermore, by performing continuity discrimination across operating segments on the candidate internal resistance segments in step three, the system can identify recurring patterns of change in multiple start-stop cycles and distinguish the sources of fluctuations caused by rhythmic changes, single disturbances, and boundary effects. This processing method transforms the micro-fluctuations of internal resistance from random changes into identifiable objects with operating rhythm characteristics, thereby compensating for the deficiency in existing power supply systems that do not specifically model frequent start-stop conditions.

[0238] Furthermore, step four generates differentiated SOC compensation instructions based on the oscillation type, and a consistency verification process is introduced after compensation to ensure that the SOC correction result is consistent with the performance during the stable operation phase before it takes effect. When the compensation result does not match the actual operating state, it can automatically revert to its previous state, thereby preventing the continuous accumulation of erroneous compensations and reducing the deviation trend of SOC estimation in short-term operation.

[0239] This embodiment improves upon the progressive coordination between start-stop segment analysis, internal resistance change screening, oscillation identification, SOC dynamic correction, and power supply strategy linkage. This enables the forklift mobile robot to continuously correct the battery status under conditions of frequent start-stop and low-load alternating operation, reducing the accumulation of SOC estimation errors in a short period of time and making the power supply control process more consistent with the actual operating state. This addresses the problem of power supply judgment deviation caused by the failure to identify micro fluctuations in battery internal resistance mentioned in the background.

[0240] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A smart power supply control system for a forklift mobile robot based on dynamic analysis, characterized in that: It includes a start / stop segment and reference segment generation module, an internal resistance candidate segment screening module, an internal resistance oscillation identification module, a SOC compensation command generation module, and a power supply control linkage module; The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment. The start / stop and reference segment generation module includes a drive job state boundary identification unit and a structured segment index construction unit; The drive operation state boundary recognition unit reads the existing state quantities of the fork mobile robot controller, including: drive enable state, drive speed command valid / invalid state, brake release state, lifting / lowering operation state, and steering command active state; and converts them into unified discrete state words to form a continuous state sequence. When a change in the drive enable state from off to on or from on to off is detected, the boundary verification process is initiated, which includes two consecutive actions: triggering validity verification and stability solidification. Trigger validity check: This determines whether the change in drive enable state is accompanied by speed command activation or brake release. If it is only a control signal jitter and no actual drive action is formed, then no start / stop boundary is generated. Stability solidification: After triggering, continuously observe the changes in the driving state. If a reverse swing or multiple rapid switches occur in a short period of time, the interval is merged into a single start-stop event, and the boundary confidence level is recorded. If the swing is frequent, it is marked as a boundary interference interval. When the drive enable state enters a continuous stable phase and the operation state is no longer in the lifting, lowering or steering jogging switching range, the start-stop segment end boundary is confirmed, a start-stop segment with start and end times and a reliable marker is formed, and the start-stop segment is output as a set of start-stop segments in chronological order. For each start-stop segment, the structured segment index building unit first searches for candidate reference intervals in the adjacent intervals before and after it. During the search process, it excludes areas with lifting switching, braking switching, or frequent steering inching, and only retains continuous segments where the drive command direction is stable and the speed command maintains a low load interval. When a brief disturbance occurs within a candidate reference interval, the candidate interval is not discarded directly. Instead, interval pruning is performed, using the disturbance point as the dividing point. The longest sub-intervals on both sides that satisfy the continuous stability condition are extracted, and the pruning behavior is recorded to retain the largest usable stable segment. After obtaining the first and second candidate reference intervals, a unique matching decision process is executed: the candidate interval that is consistent with the running direction of the start and stop segments is selected first; If the directions are consistent, select the interval that is closer to the start and stop segment time and has not been pruned; If multiple intervals satisfy the conditions, the one with the longest continuity is selected; finally, only one reference segment is output, and a unique identifier is generated. The unique identifier of the start / stop segment is bound to the reference segment identifier, encapsulated into a structured segment unit, and segment attribute information is attached; Write all structured fragment units into the structured fragment index sequence in the order of start and stop, and establish a bidirectional index relationship to obtain the structured fragment index sequence; The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence; The internal resistance oscillation identification module is based on the binding sequence and the set of internal resistance candidate segments. It performs continuous discrimination on the changes in internal resistance between candidate segments to obtain oscillation indicators. After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag. The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

2. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 1, characterized in that: The internal resistance candidate segment screening module includes a synchronous change window identification and boundary interference interval elimination unit and a direction consistency confirmation and reference segment binding sequence generation unit; The synchronous change window identification and boundary interference interval elimination unit reads the structured fragment unit one by one according to the structured fragment index sequence and enters the start and stop fragment to perform the identification action; Within the start-stop segment, find the continuous interval where voltage and current change synchronously: If the current changes significantly but the voltage does not change accordingly, it is judged as an asynchronous interval and is not considered as a candidate window. If the voltage and current show a consistent trend of change within the same time period, then the continuous segments are cut out to form a preliminary draft of the candidate window. The incomparable boundaries that may exist within the start-stop segments are located, and a list of boundary interference intervals is generated, specifically including three categories: Contactor switching interference range: Centered on the contactor state change, a short buffer zone extends forward and backward, and the entire area is marked as unavailable; Braking switching interference range: The buffer zone is also extended with the braking release / re-braking switching point as the center; Command jump interference range: When a speed command or torque command experiences a sudden change and a short-duration "peak behavior", the area before and after the peak will be merged into the interference range; Each of the initially selected candidate windows was compared with the boundary interference interval: If a candidate window falls entirely within the interference range, it is discarded. If the candidate window partially overlaps with the interference interval, then perform cut-and-hold pruning: remove the overlapping part, retain only the non-overlapping sub-intervals that still maintain continuity, and use the pruning result as a new candidate window; If the pruned subintervals are too short or fragmented, they are marked as unusable windows.

3. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 2, characterized in that: The direction consistency confirmation and reference segment binding sequence generation unit processes the candidate window. When the candidate window meets the synchronous change, but the voltage and current directions are opposite or misaligned, the direction consistency confirmation is performed on the candidate window. The steps for consistency verification are: only retain windows where the voltage and current change in the same direction within the same window; For windows with brief directional swings, if the swing occurs at the window edge, a secondary trim is performed to extract the main continuous segments with the same direction. If the swing occurs in the middle of the window and causes the window to split, it is judged as an unmatchable window and is removed; For each candidate window that has passed the directional consistency confirmation, it is not directly announced as an internal resistance candidate segment. Instead, it is returned to the structured segment index sequence to find the structured segment unit to which it belongs, read the attribute information of the corresponding reference segment, and confirm whether the window has reference significance. If a reference fragment is marked as trimmed and has a low confidence level, then the conservative retention rule is applied, allowing only fragments with high window quality to be retained. If the reference fragment has a high credibility level, the window is retained according to the normal rules; For windows that meet the directional consistency and control conditions, they are solidified as internal resistance candidate segments and a unique number is generated; multiple internal resistance candidate segments are allowed within the same start-stop segment, and the sequence number is kept to increase, and it is recorded whether the source window of each candidate segment has undergone pruning; For each candidate segment of internal resistance, establish a clear binding relationship: Bind to the identifier of the start / stop segment; further bind to the identifier of the structured segment unit to which the start / stop segment belongs; and bind to the unique reference segment identifier within the structured segment unit; organize the binding relationships according to the structured segment index sequence to form a binding sequence.

4. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 3, characterized in that: The internal resistance oscillation identification module includes an intra-unit sequence comparison and fluctuation event extraction unit and a cross-unit continuity discrimination and oscillation identifier generation unit; The unit reads the binding sequence of the sequential comparison and fluctuation event extraction unit, merges all internal resistance candidate segments according to the structured segment unit to which they belong, and sorts the internal resistance candidate segments according to the occurrence time within each structured segment unit to form a segment sequence; Perform order comparison on the fragment sequence within the same structured fragment unit: First, identify whether there is a main trend of continuous change in the same direction in the segment sequence, that is, whether the direction of change of internal resistance is consistent between adjacent segments; If the directions are consistent, adjacent segments are merged into a continuous fluctuation event, and the number of segments covered by the event and the duration are recorded. If the direction swings back and forth repeatedly, it is not directly regarded as an oscillation, but an event splitting action is triggered: the segment sequence is divided into multiple smaller fluctuation events based on the direction switching point, and marked as internal swing-type events. For fluctuation events, read the source pruning mark and the position attributes of candidate segments within start and stop segments; If an event contains a large number of source trimmed segments, or if the event is mainly distributed near the start / end of start / stop segments, then add a boundary proximity marker to the event; If the events are mainly distributed in the middle of the start-stop segment and the segment sources are all unpruned, then add a middle segment stability tag; Read the attributes of the reference segment corresponding to the structured segment unit in the bound sequence and use them as a reliable reference marker for the fluctuation event to obtain the fluctuation event sequence; The cross-unit continuity discrimination and oscillation identification generation unit reads oscillation events in the time sequence of structured segment units and establishes a comparison relationship between adjacent units: each structured segment unit is only compared with its immediate and adjacent units; when comparing, the main oscillation event within each unit is selected first as the representative event of that unit; the remaining events are used as auxiliary evidence and only participate in the judgment when the representative event is unclear. Perform continuity determination on representative events of multiple adjacent structured fragment units: When multiple consecutive structured segment units all exhibit representative events with the same direction and shape, and none of them have boundary proximity markers, a rhythmic repetition chain is formed. When the rhythmic repetition chain reaches the preset continuity requirement, rhythmic oscillation markers are generated on all structured segment units covered by the rhythmic repetition chain. When a structured segment unit has a representative event, but no event with the same direction and shape can be found in the adjacent units before and after it, the structured segment unit is classified as a single-segment oscillation. When an event is marked with a boundary proximity marker and the event energy is mainly concentrated near the boundary of the start-stop segment, it is marked as a boundary oscillation marker, regardless of whether it repeats in adjacent units. The obtained rhythmic oscillation identifiers, single-segment oscillation identifiers, and boundary oscillation identifiers are fitted to obtain oscillation identifiers.

5. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 4, characterized in that: The SOC compensation instruction generation module includes a compensation instruction construction and intra-fragment execution unit and a reference fragment consistency verification unit. The compensation instruction construction and intra-fragment execution unit are based on the structured fragment index sequence, reading the structured fragment unit and oscillation identifier one by one; When the oscillation identifier is a rhythmic oscillation identifier, a standard compensation instruction is generated to locally correct the SOC estimation path inside the start and stop segments of the structured segment unit. When the oscillation indicator is a single-segment oscillation indicator, no compensation instruction is generated, and instead an observation and recording instruction is generated. Only the observation status and the location of occurrence are written, and the SOC path is not modified. When the oscillation indicator is a boundary oscillation indicator, a restrictive compensation instruction is generated; For standard compensation instructions and restrictive compensation instructions that have met the conditions, perform local correction of the SOC path within the segment, and simultaneously perform three types of boundary control: Fragment boundary locking: Compensation is only allowed to take effect within the current start / stop fragment, and the start and end boundaries cannot be crossed; Window continuity locking: If there are multiple internal resistance candidate segments within the start / stop segment, the compensation action is organized in the order of the candidate segments, and continuous correction is not allowed to cross the boundary interference interval. State consistency locking: If a job state change occurs in the start / stop segment, the compensation within the start / stop segment is split into multiple sub-segments for execution, and the reason for the split is recorded. After the oscillation flag is executed, obtain the candidate compensation SOC status and compensation execution record.

6. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 5, characterized in that: The reference segment consistency verification unit reads the candidate compensated SOC status and compensation execution record, retrieves a unique reference segment from the bound sequence based on the structured segment unit identifier, and checks whether there is a mutual explanatory relationship between the compensated SOC change trend and the power performance of the reference segment; including Two steps: Directional consistency check: If the power performance of the reference segment is in a continuous state of depletion, the compensated SOC should not show a reverse change that contradicts the state of depletion in adjacent time periods. Continuity check: If the power performance of the reference segment is stable, the compensated SOC should not exhibit abrupt jumps within the corresponding time span of the reference segment; If the direction consistency fails, it will be marked as a direction backtracking indicator, indicating that the compensation direction and the power performance of the reference segment are inconsistent; If the continuity fails, it will fall under the continuity regression indicator, indicating that the compensation has introduced a jump that does not conform to the stationary characteristics of the reference segment; If the credibility level of the reference fragment itself is insufficient, it will fall into the insufficient reference back-off mark, indicating that the verification evidence is insufficient rather than the compensation necessarily being wrong; After triggering the rollback, the compensation effect already executed within this structured segment unit is undone, the SOC output is restored to the path before compensation, and the undone action is written to the end of the compensation execution record, forming a closed-loop evidence chain. If the rollback verification passes: Output the compensated SOC status, with the rollback flag empty; If the rollback verification fails: output the compensated SOC status and the corresponding rollback flag.

7. The intelligent power supply control system for a forklift mobile robot based on dynamic analysis according to claim 6, characterized in that: The power supply control linkage module includes a SOC path selection and power supply strategy reorganization unit and a compensation backoff and recovery and start / stop boundary rule adjustment unit; The SOC path selection and power supply strategy reorganization unit receives the compensated SOC status and rollback flag, reads the SOC output status corresponding to the current running segment in the structured segment order, and checks whether the SOC status is accompanied by rollback. When the SOC compensation result is in a stable state and no rollback occurs, the compensated SOC will be used as the basis for current power supply judgment, and the power supply control process will be reorganized based on the SOC state. When a rollback is detected, the current compensation result fails the consistency verification. The compensated SOC path is stopped, and the original SOC judgment method is restored. The power supply control returns to the original operating logic and generates the power supply control result for the current cycle, outputting the updated power supply control strategy. The compensation rollback and power-on / shutdown boundary rule adjustment unit synchronously records whether SOC compensation is effective, whether rollback has occurred, and the location characteristics of rollback occurrence within the current structured segment during the power supply control strategy execution process. When rollbacks are found to be concentrated in the start or end regions of start / stop segments, it indicates that the original segment boundaries contain unstable operating phases, and the confirmation of start / stop boundaries should be delayed in subsequent operations. When a rollback occurs near the start of a job status transition, the continuous confirmation of the job's stable state should be strengthened during subsequent segmentation processes. If no rollback occurs in multiple consecutive running segments, the existing segment division method remains unchanged.

8. A dynamic analysis-based intelligent power supply control method for a forklift mobile robot, applied to the dynamic analysis-based intelligent power supply control system for a forklift mobile robot as described in any one of claims 2 to 7, characterized in that: Includes the following steps: Step 1: The start / stop segment and reference segment generation module divides the entire operation process into multiple start / stop segments based on the drive control state and operation state of the forklift mobile robot, and matches a unique structured segment index sequence for each start / stop segment. Step 2: The internal resistance candidate segment filtering module filters the start and stop segments according to the structured segment index sequence, generates a set of internal resistance candidate segments, and binds them with the structured segment index sequence to obtain the binding sequence. Step 3: The internal resistance oscillation identification module, based on the bound sequence and the set of internal resistance candidate segments, performs continuity judgment on the changes in internal resistance between candidate segments to obtain oscillation indicators; Step 4: After receiving the oscillation flag, the SOC compensation instruction generation module generates the corresponding compensation instruction according to the oscillation type, performs consistency verification, and obtains the compensated SOC status and rollback flag. Step 5: The power supply control linkage module dynamically adjusts the original power supply control strategy based on the compensated SOC status and rollback flag, and obtains the updated power supply control strategy.

Images