A remote control system for an electric chassis vehicle used in intelligent switchgear

Abstract

This invention discloses a remote control system for an electric chassis vehicle used in intelligent switchgear, relating to the field of intelligent scheduling and control technology, and is used to solve the problem of poor task conflict control of chassis vehicles. This invention constructs a closed-loop control system from state recognition to task freezing and then to path reordering, achieving rapid identification and processing of disturbances such as access control changes, interlocking actions, and sudden environmental changes. When the disturbance exceeds a threshold, the system automatically triggers a freezing process, executing deceleration, parking, and braking. It uses branch growth to generate candidate paths, marking the change magnitude, expected entry time window, and preconditions. Then, it calculates and sorts the path distribution sparsity using a regional task density map, seamlessly connecting at the takeover point, and resuming execution with minimal disturbance. Compared to global reordering, it has the ability to quickly update and continuously correct local conditions, significantly improving the stability, flexibility, and security of scheduling electric chassis vehicles for intelligent switchgear.

Description

Technical Field

[0001] This invention relates to the field of intelligent scheduling and control technology, and more specifically, to a remote control system for an electric chassis vehicle used in intelligent switchgear. Background Technology

[0002] Intelligent switchgear is widely used in medium and low voltage power distribution scenarios, including substations, data centers, rail transit, and power industrial parks, for circuit control, fault isolation, and operation monitoring. Due to factors such as limited site space, dense concentration of live equipment, restricted working hours, and the risk of arc flashover, on-site equipment inspection, maintenance, and emergency response require "less manpower, remote operation, and recordability." To reduce the risk of personnel close contact with live equipment, the industry has begun using electric chassis vehicles that can move freely on the ground in the distribution room as mobile operation and handling platforms to perform tasks such as inspection and photography, alignment and docking, material transfer, and emergency evacuation. Corresponding remote control systems have become crucial to ensuring safe, controllable, accurate docking, and stable communication.

[0003] The shortcomings of existing technologies are as follows: Electric chassis vehicles do not make sufficient use of multi-source state information such as station control system, five-prevention interlocking system, access control system and environmental sensors during task execution. They often make task judgments based on a single channel or simple logic, and lack comprehensive modeling of the dynamic coupling relationship between task state and environmental state. This results in poor real-time performance and adaptability of task execution. When the chassis vehicle encounters sudden state changes during execution (such as changes in access control permissions, interlocking actions or environmental interference), it can only interrupt the task or rely on manual rescheduling. It lacks a quantitative judgment mechanism for the stability of task state and automatic freeze recovery capability, which can easily cause the task to be stalled for a long time. It is difficult to balance task priority and path conflict in the case of multiple concurrent tasks and complex state changes, resulting in low chassis vehicle scheduling efficiency and even possible task conflicts or resource competition. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the following solution is proposed to solve the problem of poor chassis vehicle task conflict control in the above-mentioned background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A remote control system for an electric chassis vehicle used in intelligent switchgear includes a task status establishment module, a task execution evaluation module, an execution process analysis module, a path reordering module, and a path update and adjustment module, with each module connected by a signal. The task status establishment module is used to construct a task association status graph based on the operating status collected by the station control system, the five-prevention interlocking system, the access control system and environmental sensors. It divides the status nodes into constraint nodes, interference nodes and trigger nodes, and establishes a mapping table between tasks and status paths in a graph structure. The task execution evaluation module is used to extract the corresponding state path for each task to be executed, construct the state path perturbation factor by the minimum effective duration and the maximum state disturbance frequency in the state path, evaluate the state stability of the current task with a logical discriminant function, and generate a task executability label. The execution process analysis module is used to continuously track the changes in the bound state path during the execution of tasks by the electric chassis vehicle. When the state path disturbance factor exceeds the preset stability threshold, the task freeze operation is triggered, and the control command sequence, path identifier and state snapshot before freezing are recorded, and the current task is marked as frozen. The path reordering module is used to inject tasks in a frozen state into the task reconstruction pool. Based on the current electric chassis vehicle position and pose, the state transition trend of the target area and the path conflict structure diagram, it performs task path reordering operations and uses a branch growth algorithm to construct a candidate path set. The path update and adjustment module is used to select sparse paths from the candidate path set as the priority index for task insertion, generate a new sequence of control commands, and send them to the electric chassis vehicle control unit to complete the task switching and path update, and report the execution results.

[0006] Furthermore, the task status establishment module includes: The operating status of the station control system, the five-prevention interlocking system, the access control system, and environmental sensors is collected and semantically standardized to form a set of state events for modeling. According to the preset judgment rules, the state nodes corresponding to the state events are classified and labeled as constraint nodes, interference nodes and trigger nodes respectively, and the effective conditions, failure conditions and time attributes are recorded for each node. Generate edge sets of graph structures based on causal, mutual exclusion and temporal constraints between nodes, and construct a job task association state graph. Based on the task template, the preconditions, prohibitions and triggering conditions of each task to be executed are analyzed, and the state path that satisfies the constraints is retrieved along the job task association state graph. For each task, generate a mapping table entry between the task and the state path. The mapping table entry shall include at least the task identifier, the state path identifier, the node category sequence, the time base and synchronization identifier, and the failure trigger condition. When the state node category or edge set changes, the corresponding table entry is rebuilt and updated, and the binding results of the currently effective task and state path are output.

[0007] Furthermore, the job task association status graph constructed by the task status establishment module has a hierarchical and version control mechanism, including: The task status establishment module constructs a task association status map with a hierarchical structure, which is divided into station control status layer, five-prevention interlocking layer, access control status layer and environmental status layer. Version number and effective time window are configured for the status map. When the job task-related status graph is updated, the new version replaces the old version within the effective time window. The old version is retained for backtracking and auditing, and the mapping table of tasks and status paths is automatically rebuilt and bound to the current effective version.

[0008] Furthermore, the task execution evaluation module includes: Read the mapping table between tasks and state paths, determine the state path corresponding to the target task, and sample and time-align each state node on the path within a preset sliding time window. Within the sliding time window, the duration for which each constraint node in the state path continuously meets the effective conditions is counted, and the minimum value among them is taken as the minimum effective duration. Within the sliding time window, the number of state transitions and the frequency of transitions per unit time are counted for each interfering node in the state path, and the maximum value is taken as the maximum state interference frequency. Based on the short-board criterion and segmented logic rules, the minimum duration effective time and the maximum state disturbance frequency are combined to construct the state path disturbance factor. When the minimum duration effective time is lower than the first judgment threshold or the maximum state disturbance frequency is higher than the second judgment threshold, the state path disturbance factor is marked as unstable or boundary unstable. A logic discriminant function with hysteresis characteristics is used to determine the stability of the state path disturbance factor, and the task stability result is output by combining the real-time status of the current task stage and key trigger nodes. Based on the task stability results, a task executability label is generated. The task executability label includes at least three states: allowed, restricted, and prohibited. When the task is changed from restricted or prohibited to permitted, the recovery conditions of the stabilization threshold and minimum dwell time must be met, and the task executability tag and timestamp must be recorded in the task scheduling record.

[0009] Furthermore, a logic discriminant function with hysteresis characteristics is used to determine the stability of the state path disturbance factor, and the task stability result is output by combining the real-time state of the current task stage and key trigger nodes, including: Configure the instability threshold, the stability recovery threshold, and boundary band parameters for each task phase to form a phased hysteresis determination threshold group; Obtain the state path disturbance factor corresponding to the target task, and select the corresponding hysteresis judgment threshold group according to the current task stage; Read the real-time status of key trigger nodes. If any key trigger node reaches the entry restriction or alarm condition, the task stability result will be directly set to unstable and a warning sign will be recorded. When the state path disturbance factor is not lower than the instability threshold, the task stability result is determined to be unstable; when it is between the recovery to stability threshold and the instability threshold and no entry prohibition condition is triggered, it is determined to be boundary unstable; when it is lower than the recovery to stability threshold and the key triggering node is in a normal state, it is determined to be stable. The stability result of the previous moment is saved by the hysteresis memory unit. Only when the threshold for recovery to stability is met continuously and the minimum dwell time count condition is reached, it is allowed to recover from instability or boundary instability to stability. Otherwise, the original judgment result is maintained. The output includes task stability results with three possible values: stable, boundary unstable, and unstable.

[0010] Furthermore, the execution process analysis module includes: Call the mapping table between tasks and state paths, establish continuous monitoring of the state path corresponding to the target task, periodically sample each state node on the path according to a unified time base and update the state path disturbance factor. The updated state path disturbance factor is compared with the preset stability threshold, and the entry prohibition or alarm conditions of key trigger nodes are detected in parallel. If any condition is met, a freeze trigger signal is generated. Upon receiving the freeze trigger signal, the safety freeze process is executed, and deceleration, parking and braking control commands are issued in sequence to block new motion control commands from entering the execution queue, and the last valid control segment in the current execution queue is intercepted as the control command sequence before freezing. Generate a freeze snapshot, which records the control command sequence, path identifier and status snapshot before freezing. The status snapshot includes at least the current value of the status path node and the timestamp. Mark the target task as frozen and output the freeze event.

[0011] Furthermore, tasks in a frozen state are injected into the task reconstruction pool. Based on the current electric chassis pose, the state transition trend of the target area, and the path conflict structure diagram, a task path rearrangement operation is performed, including: Receive the freeze event output by the execution process analysis module, parse the freeze reason, path identifier before freeze, state snapshot and current pose of electric chassis vehicle, and generate reconstruction task descriptor; The task descriptors are partitioned and pooled according to the reason for freezing, target area and risk level. A keep-alive timer and elimination conditions are set, and they are placed in the task reconstruction pool queue to be scheduled. Based on the historical sampling sequence of the state path corresponding to the target area, continuously effective segments, continuously failed segments and abrupt events are identified within the sliding time window. The state transition trend is determined to be recovery, deterioration or oscillation, and the recoverable time window and the forbidden time window are determined accordingly. Extract channel occupancy information for currently executing and pending tasks, combine it with restricted areas and bottleneck sections within the station, construct a path conflict structure diagram, mark mutually exclusive sections, passing sections and priority channels, and classify the conflict level. Within the recoverable time window, starting from the current pose of the electric chassis vehicle, search for reachable state path segments along the task-related state graph, generate a set of feasible path segments, and calculate the change index of each path segment relative to the path before freezing. Based on the state transition trend and path conflict level, the rearrangement rules are executed, path segments are selected for combination, and a reconstruction task path is generated; when the preferred conditions are not met, alternative path segments are selected to complete the combination according to the retreat or detour strategy. Output the rearrangement results, including the new path identifier, the expected entry time window and preconditions, and update the relative order in the task queue.

[0012] Furthermore, a candidate path set is constructed using a branch-growth algorithm, including: From the task association status map, select feasible path segments that are adjacent and meet the preconditions as seeds to form a seed set for branch growth, and record the starting node, allowed entry time window and baseline of change of the path relative to the freezing point for each seed. For each seed, perform iterative growth, expand branches along the reachable edges of the task-related state graph, and add path segments only when the constraint node meets the effective conditions, has not entered the restricted area, and does not overlap with the high-conflict level area in the path conflict structure graph. Simultaneously update the path identifier, time window, and change magnitude index of the branch. At each step of branch expansion, gating checks are performed: if a trigger node reaches a prohibited or alarm state, the time window constraint is violated, or the path continuity is interrupted, the current branch expansion in this round is terminated. Apply pruning rules to active branches. Pruning rules should include at least the following: branches that enter restricted areas, branches that overlap with high conflict level areas, branches that form loops, and branches whose change range exceeds the preset limit or whose length exceeds the limit should be removed. When a branch encounters an obstacle or is pruned, it backtracks to the previous branch node to select an alternative path segment to continue growing until the seed set is exhausted or the preset growth termination condition is reached. Branches that successfully reach the target area and meet the prerequisites are collected as a candidate path set. Paths in the set are deduplicated according to their path identifiers, and each path is accompanied by an expected entry time window, necessary prerequisites, and change magnitude indicators. Output a set of candidate paths and record the branch growth process log, and generate a sequence of control instructions.

[0013] Furthermore, a candidate path set is constructed using a branch-growth algorithm, including: From the task association status map, select feasible path segments that are adjacent and meet the preconditions as seeds to form a seed set for branch growth, and record the starting node, allowed entry time window and baseline of change of the path relative to the freezing point for each seed. For each seed, perform iterative growth, expand branches along the reachable edges of the task-related state graph, and add path segments only when the constraint node meets the effective conditions, has not entered the restricted area, and does not overlap with the high-conflict level area in the path conflict structure graph. Simultaneously update the path identifier, time window, and change magnitude index of the branch. At each step of branch expansion, gating checks are performed: if a trigger node reaches a prohibited or alarm state, the time window constraint is violated, or the path continuity is interrupted, the current branch expansion in this round is terminated. Apply pruning rules to active branches. Pruning rules should include at least the following: branches that enter restricted areas, branches that overlap with high conflict level areas, branches that form loops, and branches whose change range exceeds the preset limit or whose length exceeds the limit should be removed. When a branch encounters an obstacle or is pruned, it backtracks to the previous branch node to select an alternative path segment to continue growing until the seed set is exhausted or the preset growth termination condition is reached. Branches that successfully reach the target area and meet the prerequisites are collected as a candidate path set. Paths in the set are deduplicated according to their path identifiers, and each path is accompanied by an expected entry time window, necessary prerequisites, and change magnitude indicators. Output a set of candidate paths and record the branch growth process log, and generate a sequence of control instructions.

[0014] Furthermore, the calculation process for path distribution sparsity includes: A regional task density map is created based on the channel topology, and historical, currently executing, and pending paths are uniformly rasterized. For each grid cell, the historical occupancy duration is divided by the preset baseline duration, the number of passages is divided by the preset baseline number of passages, the mutual exclusion flag is multiplied by the preset penalty coefficient and then divided by the preset baseline weight. The three results are then added together to obtain the occupancy intensity of the grid cell. The candidate path is rasterized, and the intensity of the neighboring raster is first multiplied by a preset attenuation coefficient and then added to the intensity of the covering raster. The resulting values ​​are added along the path to obtain the cumulative overlap. Divide the cumulative overlap by the path length or the number of covered grids to get the overlap degree, and subtract the overlap degree from the constant 1 to get the distribution sparsity.

[0015] The technical effects and advantages of the remote control system for an electric chassis vehicle used in intelligent switchgear according to the present invention are as follows: This invention constructs a closed-loop control system from state recognition to task freezing and then to path reordering, which enables rapid identification and processing of disturbances such as access control changes, interlocking actions and sudden environmental changes. When the disturbance exceeds the threshold, the system automatically triggers the freezing process, performs deceleration, parking and braking, and solidifies the control command sequence and state snapshot to ensure the safe stopping of the electric chassis vehicle and the traceability of the task.

[0016] In the task reconstruction after freezing, the system generates a set of candidate paths based on the state transition trend and path conflict structure diagram, and marks the path change magnitude, expected entry time window and necessary preconditions. The system calculates the sparsity of path distribution through the regional task density map, sorts the candidate paths, and prioritizes the selection of paths with high sparsity and that meet the preconditions. The system completes the instruction access at the takeover point to ensure that the task resumes execution with minimal disturbance.

[0017] Compared to traditional global rescheduling methods, this approach can quickly update paths within a local scope, avoiding large-scale scheduling chaos, significantly improving the system's adaptability to dynamic environments, and enabling continuous correction of control strategies and environmental information. This, in turn, enhances the stability, flexibility, and security of electric chassis vehicle scheduling in intelligent switchgear scenarios. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of a remote control system for an electric chassis vehicle used in an intelligent switchgear according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] In order to achieve the above objectives, Figure 1 A schematic diagram of the structure of a remote control system for an electric chassis vehicle for an intelligent switchgear is provided. Specifically, it includes a task status establishment module, a task execution evaluation module, an execution process analysis module, a path reordering module, and a path update and adjustment module. The modules are connected by signals. The task status establishment module is used to construct a task association status graph based on the operating status collected by the station control system, the five-prevention interlocking system, the access control system and environmental sensors. It divides the status nodes into constraint nodes, interference nodes and trigger nodes, and establishes a mapping table between tasks and status paths in a graph structure. The task execution evaluation module is used to extract the corresponding state path for each task to be executed, construct the state path perturbation factor by the minimum effective duration and the maximum state disturbance frequency in the state path, evaluate the state stability of the current task with a logical discriminant function, and generate a task executability label. The execution process analysis module is used to continuously track the changes in the bound state path during the execution of tasks by the electric chassis vehicle. When the state path disturbance factor exceeds the preset stability threshold, the task freeze operation is triggered, and the control command sequence, path identifier and state snapshot before freezing are recorded, and the current task is marked as frozen. The path reordering module is used to inject tasks in a frozen state into the task reconstruction pool. Based on the current electric chassis vehicle position and pose, the state transition trend of the target area and the path conflict structure diagram, it performs task path reordering operations and uses a branch growth algorithm to construct a candidate path set. The path update and adjustment module is used to select sparse paths from the candidate path set as the priority index for task insertion, generate a new sequence of control commands, and send them to the electric chassis vehicle control unit to complete the task switching and path update, and report the execution results.

[0021] The specific implementation details of the task status establishment module are as follows: The system obtains the original operating status from the station control system, the five-prevention interlocking system, the access control system, and environmental sensors, and aligns the time with a unified time base. Then, semantic standardization is performed: a status event is generated for each original status. The status event includes fields such as event source, event name, event value, occurrence time, duration, sampling period, data quality identifier, and unique event number. These are collected into a status event set. The unified time base is provided by the station time synchronization. The occurrence time and duration in the status event set are all measured with the unified time base. After semantic standardization, the task state establishment module classifies the state nodes corresponding to state events according to preset judgment rules: when a state constitutes a necessary entry or hold condition for task execution, the corresponding state node is marked as a constraint node; when a state has an uncertain impact on task execution and its fluctuations need to be monitored but do not directly determine access, the corresponding state node is marked as an interference node; when a state, once triggered, directly changes task access or triggers a safety action, the corresponding state node is marked as a trigger node. For each state node, the effective condition, ineffective condition, and time attribute are recorded synchronously: the effective condition indicates when the node is considered satisfied (e.g., an unlocked access control is considered effective if it remains unlocked for more than a preset minimum duration); the ineffective condition indicates when the node is considered unsatisfied (e.g., an access control is considered ineffective if it returns to locked or the hold time is insufficient); and the time attribute records the occurrence time, duration, and maximum allowed sampling interval. Subsequently, based on the causal, mutual exclusion, and temporal constraints between nodes, an edge set of the graph structure is generated: causal edges are used to describe the dependency from predecessor to successor, mutual exclusion edges are used to describe the constraint that cannot be satisfied simultaneously, and temporal edges are used to describe the order constraint from the first occurrence to the next occurrence, thereby constructing a job task association state graph.

[0022] The task-related state graph is organized into a hierarchical structure: the station control state layer, the five-prevention interlocking layer, the access control state layer, and the environmental state layer each carry the state nodes and edge sets of their respective layers; version numbers and effective time windows are also configured. Version numbers are compiled in "year-month-serial number", and the effective time window consists of two parts: the effective start time and the effective end time. When the task-related state graph needs to be updated (e.g., adding a mutual exclusion edge or adjusting the effective conditions of a node), a new version is generated and a new effective time window is set; the new version automatically replaces the old version in the judgment within its effective time window, and the old version is read-only and used for backtracking and auditing. The mapping table of tasks and state paths is bound to the currently effective version: once the version is switched, the mapping table is automatically reconstructed according to the nodes and edge sets of the new version to ensure that subsequent retrieval and judgment are based on a consistent structure and constraints.

[0023] For each task to be executed, the system parses the preconditions, prohibitions and trigger conditions from the task template: the preconditions are used to declare the set of constraint nodes that must be met before the task starts, the prohibitions are used to declare the set of states that must not be entered or must be exited once they are met, and the trigger conditions are used to declare the actions that need to be taken immediately after the trigger node is met (such as freezing or evacuation). The task status establishment module searches for state paths that meet the above conditions along the task association status graph. The following textual judgment is used during the search: only when all the prerequisite allowable conditions are satisfied under the unified time base, and no prohibitive conditions are satisfied, and no triggering conditions have been triggered, can the state nodes be connected to form a candidate state path; if a conflicting node pointed to by a mutual exclusion edge is encountered, the branch is abandoned and backtracking to the previous feasible node to continue the search. Each retrieved state path generates a mapping entry between the task and the state path. This mapping entry includes at least a task identifier, a state path identifier, a node category sequence, a time base and synchronization identifier, and a failure trigger condition. The task identifier is obtained by concatenating the task template number and the task instance number; the state path identifier is obtained by concatenating the layer number, the start node number, and the end node number; the node category sequence is a list of constraint nodes, interference nodes, and trigger nodes arranged sequentially on the state path; the time base and synchronization identifier declare the unified time base source used by the mapping entry and the timestamp of the most recent successful synchronization; the failure trigger condition declares the conditions under which the mapping entry will automatically fail, specifically including any constraint node satisfying its failure condition, any prohibition condition being satisfied, any trigger node being triggered, or the other end node corresponding to a mutual exclusion edge entering a satisfied state for a preset minimum duration.

[0024] When the status node category or edge set changes (e.g., a maintenance mode is added to the access control status and defined as a prohibition condition, or a mutual exclusion edge is added to the five-prevention interlocking system), the task status establishment module immediately triggers the reconstruction and version update process of the corresponding mapping table entries. Specifically: First, a new version number is generated for the current job task associated status graph, and the effective start time of the new version is set to the start of the next sampling period of the current unified time base, with an empty value indicating continuous effectiveness; second, based on the nodes and edge set of the new version, the above retrieval steps are re-executed to generate a new status path, and the original task and status path mapping table entries are replaced; finally, the old version along with the old mapping table entries is frozen as read-only, retaining its version number and effective time window for subsequent auditing and backtracking.

[0025] Through this implementation method, the system can ensure that all newly generated judgments are based on the same version of the job task association status map after structural adjustment, while ensuring that historical task records are consistent with the structure at that time, thus meeting the requirements of engineering consistency and audit traceability.

[0026] The specific implementation details of the task execution evaluation module are as follows: First, the mapping table between tasks and state paths is read to determine the state path and node category sequence corresponding to the target task. A sliding time window and sampling period are then set under a unified time base (e.g., window length and window sliding step size are configured by the task template). Each state node on the state path is sampled according to the sampling period, aligning timestamps from different sources to the unified time base. For any node, if the time interval between two adjacent samples exceeds the maximum allowed sampling interval recorded in its time attribute, this gap is marked as a missing measurement. Missing measurement segments are not used for continuous duration statistics and frequency statistics. If multiple source values ​​exist at the same sampling scale, the one with a timestamp no later than that scale and the closest distance is taken as the valid value. Within the sliding time window, two types of metrics are calculated separately. For constraint nodes, the duration for which each node continuously meets its effective conditions is counted. Specifically, starting from the end of the window and proceeding backwards on each sampling scale, if a node shows that it meets its effective conditions in adjacent scales without any missing measurements, the time intervals of adjacent scales are accumulated and added together. Once a node shows that it meets its failure conditions, has a missing measurement, or reaches the beginning of the window, the accumulation stops and the continuous duration of the node is recorded. After completing the above statistics for all constraint nodes on the state path, the minimum value of the continuous duration of each node is selected as the minimum continuous effective time of this window. For interference nodes, the number of state transitions of each node within the window is counted, that is, the switch count from satisfying to not satisfying or from not satisfying to satisfying. Then, the number of transitions of the node is divided by the window length to obtain the frequency of transitions per unit time. Finally, the maximum value among all interference nodes is selected as the maximum state interference frequency of this window.

[0027] After obtaining the minimum effective duration and maximum state disturbance frequency, a state path disturbance factor is constructed according to the shortest-board criterion and segmentation logic rules, and a preliminary judgment is made: when the minimum effective duration is lower than the first judgment threshold given by the task template, the state path disturbance factor is directly judged as unstable; when the minimum effective duration is not lower than the first judgment threshold but the maximum state disturbance frequency is higher than the second judgment threshold given by the task template, the state path disturbance factor is also judged as unstable; when neither exceeds the corresponding threshold but either metric falls into the boundary band defined by the template (i.e., the buffer zone close to the threshold), it is judged as boundary unstable; only when the minimum effective duration is within the safe range and the maximum state disturbance frequency is within the safe range is it judged as stable. The first judgment threshold, the second judgment threshold, and the boundary band are configured by the operation and maintenance unit based on historical data and security procedures when the site goes online, and recorded in the task template for auditing purposes.

[0028] A logic discriminant function with hysteresis characteristics is used to determine the stability of state path disturbance factors, and the task stability result is output by combining the real-time status of the current task stage and key trigger nodes. Specific details include: The task template clearly defines the task stages, using the sequence of departure stage, en route stage, arrival stage, and operation stage. For each task stage, during site deployment and debugging, maintenance personnel configure an instability threshold, a stability recovery threshold, and boundary band parameters based on historical path disturbance factor distribution, on-site safety procedures, and false alarm tolerance. These three parameters together constitute the hysteresis judgment threshold group for that stage. The instability threshold provides the triggering condition for transitioning from stable or boundary unstable states to instability; the stability recovery threshold provides the stabilization condition for recovering from boundary unstable or unstable states to stable states; and the boundary band parameters define the buffer zone between the stability recovery threshold and the instability threshold. All threshold groups are stored in the task template in a versioned manner, including the threshold source, setting time, and applicable scope, ensuring traceability for subsequent audits.

[0029] During runtime, the output state path disturbance factor is read in each sampling period to obtain the current task stage and select the corresponding hysteresis judgment threshold group accordingly; at the same time, the set of key trigger nodes is read from the task-state path mapping table, and the real-time status of the key trigger nodes is retrieved.

[0030] If any critical trigger node reaches the prohibited or alarm conditions defined in the template, the module will no longer compare the threshold, but will directly set the task stability result to unstable, and write the corresponding critical trigger node identifier, trigger time and trigger reason into the task scheduling record as a prerequisite for subsequent task freezing.

[0031] When the critical trigger node's entry restriction or alarm conditions are not triggered, the state path disturbance factor and hysteresis judgment threshold group are judged: if the current value of the state path disturbance factor is not lower than the entry instability threshold, the task stability result is judged as unstable; if the current value of the state path disturbance factor is lower than the entry instability threshold but not lower than the recovery to stability threshold, and the critical trigger node is in a normal state, it is judged as boundary instability; if the current value of the state path disturbance factor is lower than the recovery to stability threshold, and the critical trigger node is in a normal state, it is judged as stable. The comparison relationships of "lower than" and "not lower than" are all performed using values ​​of the same dimension as the state path disturbance factor, ensuring direct comparison between thresholds at different stages; the boundary band parameter is only used to explicitly define the boundary instability interval and does not participate in the threshold numerical calculation.

[0032] To suppress decision jitter, a hysteresis memory unit is set up to store the task stability result and its start time from the previous moment. When the previous moment is unstable or boundary unstable, the task stability result is allowed to be restored to stable only if the value of the state path disturbance factor is consistently lower than the recovery threshold for stability over a number of consecutive sampling periods, and the cumulative duration of this consecutive period reaches the minimum dwell time given by the task template. If, during the aforementioned continuous timing process, the state path disturbance factor is not lower than the recovery threshold for stability in any sampling period, or if an anomaly occurs at a critical trigger node, the timing is immediately stopped and the previous decision result remains unchanged. The unit for setting the minimum dwell time is consistent with the sampling period of the state path disturbance factor, and its source and setting rationale are recorded in the task template.

[0033] The output includes a task stability result with one of three values: stable, boundary unstable, or unstable. The current task stage, the adopted hysteresis judgment threshold group number, the current value of the state path disturbance factor, the real-time state snapshot of the key trigger node, whether the minimum dwell time is met, and the judgment timestamp are written into the task scheduling record. This task stability result is synchronously provided to the task executability label generation logic to generate allowed, restricted, or prohibited task executability labels, and serves as the direct input for the subsequent execution process analysis module to trigger task freezing and path reordering.

[0034] Finally, based on the above task stability results, task executability tags are generated and written to the task scheduling record. The mapping relationship is: stable to allowed, unstable boundary to restricted, unstable or critical trigger node alarm to prohibited. If the tag changes from restricted or prohibited back to allowed, two recovery conditions must be met simultaneously: First, the state path disturbance factor continuously meets the stabilization threshold (i.e., continuously below the recovery to stability threshold) and reaches the minimum dwell time; second, the critical trigger node is always in a normal state during this continuous period. When recording the task executability tag, the timestamp, the task stage threshold group number used, the value of the minimum continuous effective time, the value of the maximum state disturbance frequency, whether the critical trigger node is triggered, and the sliding time window boundary used for judgment are written simultaneously to ensure that review and maintenance personnel can reproduce the actual test results and judgment process accordingly.

[0035] The specific implementation details of the execution process analysis module are as follows: The target task's state path and path identifier are read from the task-state path mapping table, and periodic sampling is set using station-side time synchronization as a unified time reference. The periodic sampling value is configured by the operations and maintenance personnel when the site goes online, and is used as a fixed sampling period to sample and align the state nodes on the state path. At each sampling period scale, the current values ​​of constraint nodes, interference nodes, and trigger nodes are sequentially pulled and written into the sampling cache of that scale. At the same time, the publicly disclosed calculation process of the task execution evaluation module is called to update the state path disturbance factor at this scale. If any state node is missing in this scale, the missing state of this scale is marked and used for conservative handling in subsequent freeze determination. Subsequently, two types of judgments are completed simultaneously in each sampling period: one is to compare the updated state path disturbance factor with the preset stability threshold; the other is to detect in parallel whether the key triggering node has reached the prohibition or alarm conditions.

[0036] The preset stability threshold is provided by the task template and is used to define the boundaries between stable, boundary unstable, and unstable conditions. The preset stability threshold corresponds to the entry instability threshold and the recovery to stability threshold configured in the task template (the boundary band is used to define the buffer between the two). The key trigger node is provided by the mapping table between the task and the state path and is used to directly enter the instability determination when triggered. When any of the following conditions occur, a freeze trigger signal is immediately generated: First, the state path disturbance factor is not lower than the threshold used to enter instability in the preset stability threshold; second, any key trigger node is determined to meet the prohibition or alarm conditions in the current sampling period; the freeze trigger signal is sent to the subsequent actions in the same control loop in the form of a message, without waiting for the next sampling period, so as to reduce lag.

[0037] Upon receiving the freeze trigger signal, the execution process analysis module enters the safety freeze procedure: First, a deceleration control command is issued to gradually reduce the target linear velocity and angular velocity according to the pre-set deceleration step size until the vehicle speed is below the safe parking threshold; then, a parking control command is issued to keep the electric chassis vehicle in its current position and prevent further planned advancement; finally, a braking control command is issued to lock the drive actuator to ensure that external disturbances do not trigger secondary movement. During the above three steps, the action of blocking new motion control commands from entering the execution queue is performed simultaneously: specifically, the queuing switch for the control command sequence is turned off, allowing only commands that have already been queued and confirmed by the underlying control unit to continue until the end of the current control cycle. To record the control history before the freeze, the module reads back the execution queue from the control unit and extracts the last complete control segment that has been confirmed to be executed and has not expired as the control command sequence before the freeze; the criteria for identifying a complete control segment are that it has been confirmed by the underlying unit and covers a complete planning time slice to ensure reproducibility.

[0038] After the security freeze process is completed, the execution process analysis module generates a freeze snapshot and marks the freeze status. The freeze snapshot contains three items: first, the sequence of control instructions before the freeze (listing the instruction type, issuance time, and confirmation time in chronological order); second, the path identifier (indicating the state path corresponding to this freeze); and third, the state snapshot, which at least includes the node name, node value, and timestamp of all state nodes on the state path in the current sampling period. The freeze snapshot is written to the task scheduling record and bound to a unified time base. Subsequently, the module marks the target task as frozen and outputs a freeze event. The freeze event carries the freeze reason (originating from the state path disturbance factor exceeding the threshold or being triggered by a critical trigger node), freeze time, path identifier, and freeze snapshot number, serving as the input prerequisites for the path reordering module.

[0039] Through the above implementation methods, when the execution process analysis module detects unstable or prohibited conditions, it can complete the freeze trigger, vehicle safe parking, control command interception and status archiving in a determined order, with replayable records and traceable time references, to ensure that subsequent reordering and recovery have sufficient contextual basis.

[0040] The specific implementation details of the path reordering module are as follows: Upon receiving the freeze event output by the execution process analysis module, the path reordering module first generates a reconstruction task descriptor: it reads the freeze reason, the path identifier before the freeze, the status snapshot, and the current pose of the electric chassis vehicle from the freeze event, and supplements it with the timestamp of the current unified time base, the identifier of the assigned task, and the executable tag of the most recent task; then, it partitions the reconstruction task descriptor into the task reconstruction pool according to the freeze reason, target area, and risk level, and writes it into the corresponding partition of the task reconstruction pool; each record entering the pool simultaneously starts a keep-alive timer and an elimination condition: the keep-alive timer is based on the unified time base, and if the record is not scheduled after the preset keep-alive time exceeds the time limit, the priority of the record is reduced; the elimination condition is triggered by two types of events, one is that the target area is marked as continuously prohibited, and the other is that the upper system cancels the task. If either condition is met, the record is removed from the task reconstruction pool; the queue to be scheduled in the pool adopts the first-in-first-out strategy within the same partition, and different partitions are dequeued according to the strategy of higher risk level. To determine state transition trends and time windows, the path reordering module extracts key state node records for the target area from state snapshots and historical state path sampling sequences, and performs judgments within the configured sliding time window: for each constraint node, the duration for which the effective conditions are continuously met is counted from the end of the window backwards; for each interference node, the number of times it switches from met to unmet or from unmet to met is accumulated within the window, and the number of switches is compared with the window duration to determine the frequency (frequent or sparse is determined by comparing the number of switches with the window duration); for each trigger node, once an alarm or access restriction occurs, it is recorded as a sudden event. If the proportion of consecutive effective segments within the window is high, and the switching of interfering nodes is sparse with no sudden triggering nodes, it is judged as recovery; if the proportion of consecutive failure segments is high, or the switching of interfering nodes is frequent, or there are triggering node alarms, it is judged as deterioration; if consecutive effectiveness and consecutive failure alternate and the switching of interfering nodes is between the two, it is judged as oscillation; during recovery, the starting point of the recoverable time window is obtained by adding the minimum holding time to the most recent start time of consecutive effectiveness, and the end of the current window is taken as the end point; where the minimum holding time is taken as the larger value among the minimum durations required for the effectiveness conditions of each relevant constraint node recorded in the task template; when deterioration occurs or a triggering node alarm occurs, the preset prohibition time window is obtained by extending the preset prohibition time backward from the start time of the alarm or failure. The preset prohibition time is configured and versioned in the task template by the security procedures in the operation and maintenance station (corresponding to the version number and effectiveness time window); during oscillation, multiple recoverable sub-windows formed by consecutive effective segments are taken in segments so that subsequent exploration can be carried out by sub-window.

[0041] Construct a path conflict structure diagram: First, read the channel occupancy information of currently executing tasks and pending tasks from the scheduling platform, and mark the expected passage segments of each task under a unified time base to the station channel topology; second, merge the information of restricted areas and bottleneck segments within the station; third, mark mutually exclusive segments (segments that only allow single-vehicle passage), meeting segments (segments that allow two-way slow traffic), and priority channels (channels that are prioritized by the operation and maintenance strategy). Conflict levels are classified according to the following process: First, determine if the conflict overlaps with a restricted area. If it overlaps, it is classified as the highest conflict level. If it does not overlap, count the number of tasks simultaneously occupying the segment under a unified time base, and compare the number of tasks with a preset threshold. The more tasks, the higher the level. If the segment is a bottleneck segment or a mutually exclusive segment, the level is increased by one. If the segment is a priority channel and the current task has no emergency marker, the level is increased by another one. Conflict levels are recorded as discrete levels for direct comparison between gating and pruning.

[0042] Within the recoverable time window, the path rearrangement module starts from the current pose of the electric chassis vehicle and searches for reachable state path segments along the task-related state graph to generate a set of feasible path segments. For each feasible path segment, the change magnitude index relative to the path before freezing is calculated: specifically, the node sequence, time window sequence, and control command sequence are extracted from the path before freezing and the path segment respectively, and aligned under a unified time reference; firstly, an editing script is generated on the node sequence, which consists of four types of operations: "keep, insert, delete, and replace". The number of steps of the editing operation is compared with the number of nodes in the path before freezing to obtain the degree of structural difference (the difference is judged by comparing the number of steps of the editing operation with the total number of nodes); then, the order and length of the entry time and dwell time offset are calculated on the time window sequence, the larger the offset, the greater the time sequence difference; finally, the newly introduced or replaced nodes in the editing script are subject to safety review. If they overlap with high-conflict-level segments or prohibited areas in the path conflict structure graph, or if a trigger node alarm occurs, the path segment is marked as high safety conflict.

[0043] When considering the above three results, we first look at the level of security conflict, then the structural differences, and then the temporal differences. Path segments with high security conflict are directly considered to have high change magnitude and are not included in the preferred sequence. The remaining path segments are sorted from smallest to largest according to structural differences and temporal differences to obtain a set of feasible path segments with change magnitude indicators.

[0044] To combine feasible path segments into complete alternative paths, the path reordering module executes a branch growth algorithm, with the following specific steps: A feasible path segment satisfying the preconditions is selected near the current pose of the electric chassis vehicle as a seed, forming a seed set. For each seed, the starting node, the allowed entry time window, and the baseline of the change magnitude relative to the path before freezing are recorded. Then, iterative expansion begins: each time, only path segments are added forward along the reachable edges. The addition action must pass three gates simultaneously: the constraint node meets its effective conditions; it does not enter the restricted area and does not overlap with high-conflict-level segments in the path conflict structure graph; the new change magnitude index does not exceed the preset upper limit; if any gate fails, the current branch's expansion in this round is terminated. During the expansion process, pruning rules are executed: any branch that enters a restricted area, overlaps with a high-conflict-level section, forms a loop, or whose change exceeds the preset limit, or whose path length exceeds the preset length limit, is immediately removed. The preset length limit is determined by the station's channel topology and vehicle power strategy and is configured in the task template. When removed, the branch returns to the previous branch point to select an alternative path segment to continue growing until the seed set is exhausted or the growth termination condition is met (e.g., reaching the target area or reaching the preset step limit). All branches that successfully reach the target area and meet the prerequisites are collected into a candidate path set; the paths in the set are deduplicated according to the path identifier, and each path is accompanied by the expected entry time window, necessary prerequisites, and change magnitude indicators.

[0045] For example, in the candidate path set, each path will be accompanied by three key pieces of information: the first is the expected entry time window, such as a path that allows electric chassis vehicles to enter the target area between 14:30 and 14:35 under a unified time base; the second is the necessary preconditions (i.e., preconditions that allow entry), such as the access control being unlocked and remaining unlocked for a certain period of time, the target interval being confirmed to be non-energized and the five-proof interlock being deactivated, and the key triggering nodes not being in an alarm state; the third is the change magnitude index, such as a path that only adds one node to the original path and delays the entry time by about 30 seconds, which can be judged as a minor change, while if the path passes through a high-conflict section or the entry window is delayed by more than 90 seconds, it can be judged as a medium or significant change.

[0046] Finally, the candidate path set is output and written to the branch growth process log (recording each expansion, gating pass status, pruning reason and backtracking path), so that the path update and adjustment module can generate a new control instruction sequence and complete subsequent task switching and path update.

[0047] The specific implementation details of the path update and adjustment module are as follows: First, a regional task density map is established based on the station's internal channel topology: historical paths, currently executing paths, and paths to be executed are rasterized under a unified time base; for each raster, a three-step accumulation process is performed to obtain the occupancy intensity: The first step is to divide the historical occupancy time by the preset baseline time to obtain the duration percentage; the second step is to divide the number of passes by the preset baseline number of passes to obtain the frequency percentage; the third step is to multiply the mutual exclusion flag by the preset penalty coefficient and then divide it by the preset baseline weight to obtain the mutual exclusion contribution; then the above three results are added together, and the sum is written into the regional task density map as the occupancy intensity of the grid; the preset baseline time, preset baseline number of passes, preset baseline weight, and preset penalty coefficient are set by the operations and maintenance personnel based on historical load and security procedures when the site goes online and are versioned and saved. For example, in practical applications, when the electric chassis vehicle system site goes online, maintenance personnel will pre-set a set of preset parameters and form a versioned configuration file based on past operating load conditions and power safety regulations. Among them, the preset baseline duration can be set to 10 minutes, which is used as a comparison benchmark for the cumulative occupancy time of different tasks in the same channel; the preset baseline number can be set to 5 times, which represents the standard passage frequency allowed in the channel within the same time period; the preset baseline weight can be set to 1, which serves as a unified scale for weighing different factors in occupancy calculation; and the preset penalty coefficient is set according to safety regulations. For example, when a channel has a mutual exclusion mark, it is given a 2x or 3x occupancy weight to amplify the occupancy impact of risk-sensitive sections.

[0048] For each candidate path output by the path rearrangement module, sparsity is evaluated: the candidate path is rasterized according to the same rules as the regional task density map to obtain a covered raster sequence, and the raster outside the covered raster is taken as the neighboring raster; the occupancy intensity of the neighboring raster is multiplied by a preset attenuation coefficient and added to the occupancy intensity of the covered raster, and then added sequentially along the candidate path to obtain the cumulative overlap of the candidate path; the cumulative overlap is then divided by the path length or by the number of covered raster to obtain the overlap degree; finally, the distribution sparsity of the candidate path is obtained by subtracting the overlap degree from a constant 1. When a candidate path covers a prohibited raster or overlaps with a high-conflict-level segment in the path conflict structure map, the distribution sparsity of the candidate path is directly determined to be the lowest level and marked with a prohibited symbol. The distribution sparsity of all candidate paths is sorted from high to low and the sorting timestamp is retained. It should be noted that the preset attenuation coefficient is used to reflect the attenuation law of the interference of neighboring grids on candidate paths with spatial distance. It is usually defined as a real number less than 1, such as 0.5, 0.7 or 0.9. The smaller the value, the faster the influence of neighboring grids on the covering grid attenuates. Before the site goes online, the operation and maintenance personnel use the operation records of historical task paths and channel conflict situation to statistically analyze the grid overlap probability within different distance ranges, and then fit the law of the decrease of neighboring interference with distance, and set the attenuation coefficient accordingly.

[0049] After sorting, a new control command sequence is generated by selecting sparse paths that meet the preconditions. The control command sequence includes a takeover point and a rollback point: the takeover point is set at the boundary of the first safe section where the electric chassis vehicle will enter the sparse path, and the rollback point is set at the nearest stable stopping section before the takeover point (a section where parking is permitted, there are no entry restrictions, the conflict level is no higher than medium, and there is sufficient parking space). The sequence includes speed commands, steering commands, path segment identifiers, and the effective time, and the takeover point and rollback point are explicitly declared at the beginning of the sequence. Upon reaching the takeover point, the control command sequence is sent to the electric chassis vehicle control unit to complete the task switching and path update. If an entry restriction is triggered or the path is discontinuous during switching or execution, a rollback is executed according to the rollback point, and the reason and time of the rollback are recorded in the task scheduling record. After the switchover is completed, the execution status is collected and a reporting data packet is generated. The data packet includes at least: the path identifier of the candidate path used, the corresponding distribution sparsity value (recorded as a calculation result text), the location and effective time of the takeover point and rollback point, whether the execution was successful and the reason for failure, and a timestamp of the unified time base. The data packet is synchronized to the remote scheduling platform and the station control system and associated with the current density map version number for subsequent auditing and reproduction. At the same time, the occupancy record of the current execution path is added to the regional task density map (updating the occupancy intensity of the corresponding grid according to the above rules) so as to reflect the latest channel load distribution in the next round of candidate path evaluation.

[0050] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0051] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0053] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0054] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An electric chassis car remote control system for intelligent switchgear cabinet, characterized in that: The task state establishment module, the task execution evaluation module, the execution process analysis module, the path rearrangement module, and the path update adjustment module are connected through signals; The task state establishment module is configured to establish a state graph of a work task based on the running states collected by a station control system, a five-prevention interlocking system, an access control system, and an environment sensor, divide state nodes into constraint nodes, interference nodes, and trigger nodes, and establish a mapping table of the task and the state path in a graph structure; The task execution evaluation module is configured to extract a corresponding state path for each to-be-executed task, construct a state path disturbance factor based on a minimum continuous effective time and a maximum state interference frequency in the state path, evaluate the state stability of the current task by using a logical judgment function, and generate a task executability label; The execution process analysis module is configured to continuously track changes in the bound state path during the execution of the task by the electric chassis vehicle, trigger a task freezing operation when the state path disturbance factor exceeds a preset stability threshold, record the control instruction sequence, the path identifier, and the state snapshot before freezing, and mark the current task as a frozen state; The path rearrangement module is configured to inject the task in the frozen state into a task reconstruction pool, perform a task path rearrangement operation based on the current pose of the electric chassis vehicle, the state transition trend of the target area, and a path conflict structure diagram, and construct a candidate path set by using a branch growth algorithm; The path update adjustment module is configured to select a sparse path as a task insertion priority indicator in the candidate path set, generate a new control instruction sequence, and deliver the new control instruction sequence to the electric chassis vehicle control unit to complete task switching and path updating, and report the execution result.

2. The electric chassis car remote control system for intelligent switch cabinet according to claim 1, characterized in that: The task state establishment module comprises: Collecting the running states of the station control system, the five-prevention interlocking system, the access control system, and the environment sensor and performing semantic standardization to form a state event set for modeling; According to a preset judgment rule, the state nodes corresponding to the state events are classified as constraint nodes, interference nodes, and trigger nodes, and the effective condition, the invalid condition, and the time attribute of each node are recorded; According to the causal, mutual exclusion, and time sequence constraint relationship between the nodes, an edge set of a graph structure is generated to construct a work task associated state graph; Based on the task template, the pre-allowed condition, the prohibited condition, and the trigger condition of each to-be-executed task are analyzed, and a state path that satisfies the constraints is retrieved along the work task associated state graph; A mapping table item of the task and the state path is generated for each task, and the mapping table item at least includes a task identifier, a state path identifier, a node category sequence, a time reference, a synchronization identifier, and an invalid trigger condition; When the state node category or the edge set changes, the corresponding table item is reestablished and version updated, and the current effective task and state path binding result is output.

3. The electric chassis car remote control system for intelligent switch cabinet according to claim 2, characterized in that: The work task associated state graph constructed by the task state establishment module has a hierarchical and version control mechanism, including: The work task associated state graph constructed by the task state establishment module is a hierarchical graph structure, which is hierarchically divided into a station control state layer, a five-prevention interlocking layer, an access control state layer, and an environment state layer, and is configured with a version number and an effective time window; When the job task-related status graph is updated, the new version replaces the old version within the effective time window. The old version is retained for backtracking and auditing, and the mapping table of tasks and status paths is automatically rebuilt and bound to the current effective version.

4. The electric chassis car remote control system for intelligent switch cabinet according to claim 3, characterized in that: The task execution evaluation module includes: Read the mapping table between tasks and state paths, determine the state path corresponding to the target task, and sample and time-align each state node on the path within a preset sliding time window. Within the sliding time window, the duration for which each constraint node in the state path continuously meets the effective conditions is counted, and the minimum value among them is taken as the minimum effective duration. Within the sliding time window, the number of state transitions and the frequency of transitions per unit time are counted for each interfering node in the state path, and the maximum value is taken as the maximum state interference frequency. Based on the short-board criterion and segmented logic rules, the minimum duration effective time and the maximum state disturbance frequency are combined to construct the state path disturbance factor. When the minimum duration effective time is lower than the first judgment threshold or the maximum state disturbance frequency is higher than the second judgment threshold, the state path disturbance factor is marked as unstable or boundary unstable. A logic discriminant function with hysteresis characteristics is used to determine the stability of the state path disturbance factor, and the task stability result is output by combining the real-time status of the current task stage and key trigger nodes. Based on the task stability results, a task executability label is generated. The task executability label includes at least three states: allowed, restricted, and prohibited. When the task is changed from restricted or prohibited to permitted, the recovery conditions of the stabilization threshold and minimum dwell time must be met, and the task executability tag and timestamp must be recorded in the task scheduling record.

5. The electric chassis car remote control system for intelligent switch cabinet according to claim 4, characterized in that: A logic discriminant function with hysteresis characteristics is used to determine the stability of state path disturbance factors, and the task stability results are output by combining the real-time state of the current task stage and key trigger nodes, including: Configure the instability threshold, the stability recovery threshold, and boundary band parameters for each task phase to form a phased hysteresis determination threshold group; Obtain the state path disturbance factor corresponding to the target task, and select the corresponding hysteresis judgment threshold group according to the current task stage; Read the real-time status of key trigger nodes. If any key trigger node reaches the entry restriction or alarm condition, the task stability result will be directly set to unstable and a warning sign will be recorded. When the state path disturbance factor is not lower than the instability threshold, the task stability result is determined to be unstable; when it is between the recovery to stability threshold and the instability threshold and no entry prohibition condition is triggered, it is determined to be boundary unstable; when it is lower than the recovery to stability threshold and the key triggering node is in a normal state, it is determined to be stable. The stability result of the previous moment is saved by the hysteresis memory unit. Only when the threshold for recovery to stability is met continuously and the minimum dwell time count condition is reached, it is allowed to recover from instability or boundary instability to stability. Otherwise, the original judgment result is maintained. The output includes task stability results with three possible values: stable, boundary unstable, and unstable.

6. The electric chassis car remote control system for intelligent switch cabinet according to claim 5, characterized in that: The execution process analysis module includes: Call the mapping table between tasks and state paths, establish continuous monitoring of the state path corresponding to the target task, periodically sample each state node on the path according to a unified time base and update the state path disturbance factor. The updated state path disturbance factor is compared with the preset stability threshold, and the entry prohibition or alarm conditions of key trigger nodes are detected in parallel. If any condition is met, a freeze trigger signal is generated. Upon receiving the freeze trigger signal, the safety freeze process is executed, and deceleration, parking and braking control commands are issued in sequence to block new motion control commands from entering the execution queue, and the last valid control segment in the current execution queue is intercepted as the control command sequence before freezing. Generate a freeze snapshot, which records the control command sequence, path identifier and status snapshot before freezing. The status snapshot includes at least the current value of the status path node and the timestamp. Mark the target task as frozen and output the freeze event.

7. The electric chassis car remote control system for intelligent switch cabinet according to claim 6, characterized in that: This is used to inject frozen tasks into the task reconfiguration pool. Based on the current electric chassis vehicle position and pose, the state transition trend of the target area, and the path conflict structure diagram, it performs task path rearrangement operations, including: Receive the freeze event output by the execution process analysis module, parse the freeze reason, path identifier before freeze, state snapshot and current pose of electric chassis vehicle, and generate reconstruction task descriptor; The task descriptors are partitioned and pooled according to the reason for freezing, target area and risk level. A keep-alive timer and elimination conditions are set, and they are placed in the task reconstruction pool queue to be scheduled. Based on the historical sampling sequence of the state path corresponding to the target area, continuously effective segments, continuously failed segments and abrupt events are identified within the sliding time window. The state transition trend is determined to be recovery, deterioration or oscillation, and the recoverable time window and the forbidden time window are determined accordingly. Extract channel occupancy information for currently executing and pending tasks, combine it with restricted areas and bottleneck sections within the station, construct a path conflict structure diagram, mark mutually exclusive sections, passing sections and priority channels, and classify the conflict level. Within the recoverable time window, starting from the current pose of the electric chassis vehicle, search for reachable state path segments along the task-related state graph, generate a set of feasible path segments, and calculate the change index of each path segment relative to the path before freezing. Based on the state transition trend and path conflict level, the rearrangement rules are executed, path segments are selected for combination, and a reconstruction task path is generated; when the preferred conditions are not met, alternative path segments are selected to complete the combination according to the retreat or detour strategy. Output the rearrangement results, including the new path identifier, the expected entry time window and preconditions, and update the relative order in the task queue.

8. The electric chassis car remote control system for intelligent switch cabinet according to claim 7, characterized in that: A candidate path set is constructed using a branch-growth algorithm, including: From the task association status map, select feasible path segments that are adjacent and meet the preconditions as seeds to form a seed set for branch growth, and record the starting node, allowed entry time window and baseline of change of the path relative to the freezing point for each seed. For each seed, perform iterative growth, expand branches along the reachable edges of the task-related state graph, and add path segments only when the constraint node meets the effective conditions, has not entered the restricted area, and does not overlap with the high-conflict level area in the path conflict structure graph. Simultaneously update the path identifier, time window, and change magnitude index of the branch. At each step of branch expansion, gating checks are performed: if a trigger node reaches a prohibited or alarm state, the time window constraint is violated, or the path continuity is interrupted, the current branch expansion in this round is terminated. Apply pruning rules to active branches. Pruning rules should include at least the following: branches that enter restricted areas, branches that overlap with high conflict level areas, branches that form loops, and branches whose change range exceeds the preset limit or whose length exceeds the limit should be removed. When a branch encounters an obstacle or is pruned, it backtracks to the previous branch node to select an alternative path segment to continue growing until the seed set is exhausted or the preset growth termination condition is reached. Branches that successfully reach the target area and meet the prerequisites are collected as a candidate path set. Paths in the set are deduplicated according to their path identifiers, and each path is accompanied by an expected entry time window, necessary prerequisites, and change magnitude indicators. Output a set of candidate paths and record the branch growth process log, and generate a sequence of control instructions.

9. The electric chassis car remote control system for intelligent switch cabinet according to claim 8, characterized in that: The path update and adjustment module includes: The distribution sparsity of each candidate path is calculated based on the regional task density map and sorted from high to low. Select a sparse path that meets the preconditions and generate a new control instruction sequence that includes the take-off point and the rollback point; At the takeover point, the control command sequence is sent to the electric chassis vehicle control unit to complete the task switching and path update; The execution status is collected and reported as a data packet, which is then synchronized to the remote scheduling platform and the station control system.

10. The electric chassis car remote control system for intelligent switch cabinet according to claim 9, characterized in that: The calculation process for path distribution sparsity includes: A regional task density map is created based on the channel topology, and historical, running, and pending paths are uniformly rasterized. For each grid cell, the historical occupancy duration is divided by the preset baseline duration, the number of passages is divided by the preset baseline number of passages, the mutual exclusion flag is multiplied by the preset penalty coefficient and then divided by the preset baseline weight. The three results are then added together to obtain the occupancy intensity of the grid cell. The candidate path is rasterized, and the intensity of the neighboring raster is first multiplied by a preset attenuation coefficient and then added to the intensity of the covering raster. The resulting values ​​are added along the path to obtain the cumulative overlap. Divide the cumulative overlap by the path length or the number of covered grids to get the overlap degree, and subtract the overlap degree from the constant 1 to get the distribution sparsity.

Images