Container front-mounted crane loading and unloading detection method and system based on artificial intelligence

Abstract

This application relates to the field of artificial intelligence technology, specifically to an AI-based method and system for detecting container front-end lifting and unloading operations. First, it acquires a real-time data set of container front-end lifting and unloading operations, including front-end lifting actions, container status, and operational environment data. Then, it performs operational status correlation mapping on the real-time data set and generates operation adjustment instructions based on judgment criteria and abnormal situations, sending these instructions to the front-end lifting control terminal. This improves the efficiency, safety, and intelligence of loading and unloading operations, enabling timely adjustments to loading and unloading operations and enhancing the efficiency and safety of container front-end lifting and unloading operations.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to an artificial intelligence-based method and system for detecting container front-end lifting and unloading. Background Technology

[0002] In the field of container logistics and transportation, the container front-end crane is a key loading and unloading equipment, and the efficiency and safety of its loading and unloading operations are of paramount importance. Traditional methods for inspecting container front-end crane loading and unloading mainly rely on manual on-site observation and experience. Workers need to monitor the crane's operation, the container's condition, and the working environment in real time at the work site.

[0003] However, manual inspection has many limitations. On the one hand, due to limited human attention, it is difficult to simultaneously and comprehensively monitor the complex interrelationships between the reach stacker operation, container condition, and the working environment, easily leading to missed inspections and misjudgments. On the other hand, manual inspection lacks systematicity and standardization; different workers may have different judgment standards, making it difficult to guarantee the consistency and accuracy of inspection results. Moreover, when faced with large-scale, high-intensity loading and unloading operations, manual inspection is inefficient, unable to promptly and accurately detect abnormalities in the operation process, and difficult to take effective adjustment measures in a timely manner, thereby increasing operational risks and affecting the efficiency of loading and unloading operations and the smoothness of overall logistics transportation. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a container front-end lifting and unloading detection method and system based on artificial intelligence.

[0005] According to a first aspect of this application, an artificial intelligence-based method for detecting container front-end lifting and unloading is provided, the method comprising:

[0006] Acquire a real-time data set of container front-end crane loading and unloading operations, the real-time data set including front-end crane operation action data, container status data, and operation environment data;

[0007] The real-time collected data set is processed by operation status association mapping to establish a first association relationship between the front crane operation action data and the container status data, a second association relationship between the front crane operation action data and the operation environment data, and a third association relationship between the container status data and the operation environment data, so as to obtain the operation status association mapping result.

[0008] The pre-trained artificial intelligence loading and unloading detection model is invoked to determine the operation status association mapping results, and an operation compliance determination result containing the abnormal association type and the corresponding determination basis is generated.

[0009] Based on the abnormal relationship types in the operation compliance judgment results, and combined with the segment division rules of the front lifting and unloading operation, the operation segment corresponding to each abnormal relationship is determined, and the abnormal segment matching result containing the correspondence between abnormal relationships and operation segments is obtained.

[0010] Based on the judgment criteria in the operation compliance judgment result and the operation links in the abnormal link matching result, an operation adjustment instruction for the corresponding operation link is generated, and the operation adjustment instruction is sent to the front crane control terminal to adjust the loading and unloading operation of the corresponding operation link.

[0011] According to a second aspect of this application, an artificial intelligence-based container front-end lifting and unloading detection system is provided. The artificial intelligence-based container front-end lifting and unloading detection system includes a processor and a readable storage medium. The readable storage medium stores a program that, when executed by the processor, implements the aforementioned artificial intelligence-based container front-end lifting and unloading detection method.

[0012] Based on any of the above aspects, by acquiring real-time data sets of container front-end crane operations, comprehensively covering information such as front-end crane actions, container status, and operating environment, the real-time data sets are processed for operation status correlation mapping. This establishes three correlation relationships between front-end crane actions, container status, and the operating environment, resulting in operation status correlation mapping results that clearly present the inherent connections between various elements. A pre-trained artificial intelligence loading and unloading detection model is then invoked to judge the operation status correlation mapping results. Leveraging the powerful analytical and learning capabilities of artificial intelligence, an operation compliance judgment result is generated, including abnormal correlation types and corresponding judgment criteria, improving the accuracy and objectivity of the judgment. Based on the judgment results, the corresponding operational links for abnormal correlations are determined, obtaining abnormal link matching results. Finally, based on the judgment criteria and abnormal links, operation adjustment instructions are generated and sent to the front-end crane control terminal, enabling timely adjustments to loading and unloading operations and improving the efficiency and safety of container front-end crane operations. Attached Figure Description

[0013] Figure 1 A flowchart illustrating the container front-end lifting and unloading detection method based on artificial intelligence provided in an embodiment of this application is shown.

[0014] Figure 2 A schematic diagram of the component structure of the container front-end lifting and unloading detection system provided in an embodiment of this application is shown. Detailed Implementation

[0015] Figure 1The diagram illustrates a flowchart of an AI-based container front-end lifting and unloading detection method provided in this application. It should be understood that in other embodiments, the order of some steps in the AI-based container front-end lifting and unloading detection method can be interchanged according to actual needs, or some steps can be omitted or deleted. The detailed steps of this AI-based container front-end lifting and unloading detection method are described below.

[0016] Step S110: Obtain the real-time data set of container front crane loading and unloading operations. This real-time data set includes front crane operation action data, container status data, and operation environment data.

[0017] In the daily operations of container terminals, various types of sensors can be deployed at multiple key locations to comprehensively and meticulously monitor container reach stacker operations. Multiple motion sensors can be installed on the reach stacker equipment. For example, sensors are installed on the lifting mechanism of the spreader to capture information such as vertical displacement, velocity, and acceleration; sensors are installed on the rotating mechanism of the spreader to obtain parameters such as the angle and angular velocity of rotation; and sensors are installed near the translational track of the spreader to record its horizontal movement. These sensors continuously collect data, forming reach stacker operation motion data, which reflects the various motion states of the reach stacker during loading and unloading.

[0018] The container itself is also equipped with a series of status monitoring sensors. Tilt sensors are installed at the four corners of the container to monitor the tilt angle and changes in tilt in all directions in real time; connection sensors are installed at the locking mechanism to determine whether the lock is in the correct connection state and the strength of the connection; and a center of gravity sensor is installed at the bottom of the container to detect the position of the container's center of gravity and its dynamic changes during loading and unloading. The data collected by these sensors constitutes the container status data, which helps to understand the stability and safety of the container during loading and unloading.

[0019] Regarding the working environment, various environmental monitoring sensors will be deployed around the dock. Meteorological sensors will be installed at different locations on the dock to collect meteorological information such as temperature, humidity, wind speed, and wind direction. Optical or radar sensors will be installed in the dock's passageways and working areas to detect the presence of obstacles in the surrounding area, as well as the location, size, and speed of movement of those obstacles. The data collected by these sensors constitutes the working environment data, which can reflect the external environmental conditions during operations.

[0020] Through a data transmission network, the data collected by the various sensors are aggregated into a data storage center, thereby obtaining a real-time data set that includes data collected on the actions of the reach stacker, the status of the container, and the working environment.

[0021] Step S120: Perform operation status association mapping processing on the real-time collected data set, establish the first association relationship between the front crane operation action collection data and the container status collection data, the second association relationship between the front crane operation action collection data and the operation environment collection data, and the third association relationship between the container status collection data and the operation environment collection data, and obtain the operation status association mapping result.

[0022] Step S121: Separate the front-end crane operation action data, container status data, and operation environment data from the real-time data collection to obtain three independent types of data.

[0023] Since real-time data acquisition datasets are a mixture of various data types, effective separation is necessary. First, preliminary screening is performed based on the data's source identifier. Each sensor, when collecting data, can attach a unique identifier indicating whether the data comes from a reach stack motion sensor, a container status sensor, or a monitoring sensor of the operating environment. By identifying these identifiers, the data can be broadly categorized into three types.

[0024] Then, further feature verification is performed on the initially classified data. For example, for suspected reach stacker operation data, the data characteristics are checked to see if they conform to the parameter range and variation patterns related to spreader movements; for suspected container status data, it is verified whether the data is related to container tilt, locking device connection, center of gravity, and other status information; for suspected operating environment data, it is confirmed whether the data is consistent with environmental factors such as weather conditions and obstacle detection. Through the above dual screening and verification methods, independent reach stacker operation data, container status data, and operating environment data are finally accurately separated from the real-time data collection.

[0025] Step S122: Mark the action events of the front-end crane operation action data, record the occurrence time and action attributes of each action event, and obtain the marked front-end crane operation action data.

[0026] Step S1221: Traverse the data collected for the front-end crane operation and identify the feature points in the data that represent the start and end of the operation. These feature points include abrupt change points in the operation parameters and transition points in the operation state.

[0027] When processing data collected from reach stacker operations, the data can be scanned point by point. During the scanning process, the focus should be on changes in motion parameters. Abrupt changes in motion parameters are one of the important bases for identifying the start and end of an action. For example, when the speed parameter of the spreader shows a sharp increase or decrease within a short period of time, this likely indicates the start of a new action or the end of the current action. Such a sudden change in speed may be caused by the operator starting or stopping the lifting, rotating, or translating operations of the spreader.

[0028] The transition points between motion states are also key feature points. The motion states of a spreading device can be categorized into different states such as stationary, ascending, descending, rotating, and translating. When data shows that the spreading device transitions from one state to another, such as from stationary to ascending, or from rotating to translating, it's possible to determine the start or end of an action. Through meticulous monitoring and analysis of motion parameters and states, it's possible to accurately identify the feature points in the data that characterize the start and end of an action.

[0029] Step S1222: Divide the front-end crane operation action acquisition data into multiple continuous action segments based on the feature point, with each action segment corresponding to a complete action event.

[0030] After identifying the feature points that mark the start and end of the action, the data collected for the front-end lifting operation is divided into multiple continuous action segments using these feature points as boundaries. Each action segment represents a complete action event. For example, the data from the feature point where the spreader begins to rise to the feature point where the spreader stops rising constitutes an action segment, which corresponds to the complete action event of the spreader rising.

[0031] For each action segment, the internal data is coherent and consistent, reflecting the continuous change process of the same action over time. Using this method, the complex data collected from reach stacking operations can be broken down into independent and complete action events, facilitating detailed analysis and labeling of each event later.

[0032] Step S1223: Extract the start time and end time of each action segment, and calculate the midpoint between the start time and end time as the occurrence time of the action event.

[0033] For each segmented action, its precise occurrence time needs to be determined. First, the start and end times of the action segment are extracted from the data. The start time is the point in time when the action segment begins, that is, the time corresponding to the feature point where the action begins; the end time is the point in time when the action segment ends, that is, the time corresponding to the feature point where the action ends.

[0034] To more accurately represent the occurrence time of an action event, the midpoint between the start and end times is calculated. This midpoint serves as a representative occurrence time for the action event, taking into account both the start and end times, and thus more accurately reflects the event's position on the timeline. In this way, each action event is assigned a specific occurrence time, facilitating subsequent correlation analysis and data processing.

[0035] Step S1224: Analyze the changes in motion parameters in each motion segment to determine the motion type of the motion event. The motion type includes lifting motion of the spreader, rotation motion of the spreader, and translation motion of the spreader.

[0036] After determining the occurrence time of each action event, further analysis of the changes in motion parameters within the action segment is conducted to identify the action type of the event. For lifting and lowering actions, the focus is on parameters such as vertical displacement, velocity, and acceleration. If, within an action segment, the vertical displacement continuously increases, and the velocity and acceleration exhibit a pattern consistent with the lifting or lowering operation, then the action event can be identified as a lifting and lowering action.

[0037] For the rotation of the spreader, the main observations are the changes in parameters such as rotation angle and angular velocity. When the rotation angle changes significantly in the motion segment, and the angular velocity matches the characteristics of a rotation operation, the motion event can be identified as a spreader rotation motion.

[0038] For spreader translation, focus on parameters such as horizontal displacement, velocity, and acceleration. If the horizontal displacement continuously changes, and the velocity and acceleration exhibit a trend consistent with the translation operation, then this action event is a spreader translation. Through detailed analysis and judgment of the action parameters, the action type of each action event can be accurately determined.

[0039] Step S1225: Extract the motion parameters for each motion event. These motion parameters include changes in motion amplitude, motion speed, and motion duration within the motion segment.

[0040] After determining the action type of each action event, it is necessary to extract the specific action parameters for each action event. The change in motion amplitude reflects the range of spatial variation of the action. For lifting and lowering actions of the spreader, the change in motion amplitude can be represented by the difference between the starting and ending positions in the vertical direction; for rotating actions of the spreader, the change in motion amplitude can be measured by the difference between the starting and ending angles of the rotation; for translating actions of the spreader, the change in motion amplitude can be determined by the difference between the starting and ending positions in the horizontal direction.

[0041] Changes in motion speed describe how quickly a motion changes over time. The change in motion speed can be determined by calculating the speed difference at different points in time within a motion segment, as well as the rate of change of speed over time. For example, during the lifting and lowering of a lifting device, observing whether the speed gradually increases, remains constant, or gradually decreases reveals the trend of motion speed change.

[0042] The duration of an action refers to the time elapsed from the start to the end of an action event. The duration of an action can be obtained by calculating the difference between the end time and the start time of an action segment. These action parameters can describe the characteristics of each action event in more detail.

[0043] Step S1226: Associate and store the occurrence time, action type, and action parameters of each action event, and add a unique identifier to each action event to obtain the marked front-end lifting operation action data.

[0044] To facilitate subsequent data management and analysis, the occurrence time, action type, and action parameters of each action event are stored in a related manner. This can be achieved using a database, storing this information in a single table, with one record corresponding to each action event. Within each record, the occurrence time, action type, and action parameters are stored as distinct fields to ensure data integrity and consistency.

[0045] Simultaneously, a unique identifier is added to each motion event. This identifier can be a string consisting of numbers and letters, used to uniquely identify each motion event in the data. By adding identifiers, motion events can be easily queried, retrieved, and statistically analyzed. After the above processing, the tagged reach stacker operation motion data is obtained.

[0046] Step S123: Mark the state changes of the container status data, record the time of occurrence and state attributes of each state change, and obtain the marked container status data.

[0047] Step S1231: Analyze the container status acquisition data and extract the status monitoring parameters from the container status acquisition data. The status monitoring parameters include container tilt monitoring parameters, container lock connection monitoring parameters, and container center of gravity monitoring parameters.

[0048] When processing container status data, the first step is to parse the data. Since this data is collected from different types of sensors, it may have different data formats and encoding methods. Therefore, appropriate parsing algorithms are needed to convert the data into a recognizable and processable form.

[0049] During the analysis process, the focus is on extracting state monitoring parameters from the data. For container tilt monitoring parameters, information such as the tilt angle and rate of change of tilt in various directions is extracted from the data collected by the tilt sensor. For container lock connection monitoring parameters, information such as the connection status and strength of the locks is obtained from the data collected by the lock connection sensor. For container center of gravity monitoring parameters, the position coordinates of the container's center of gravity and its dynamic changes during loading and unloading are extracted from the data collected by the center of gravity sensor. By accurately extracting these state monitoring parameters, a comprehensive understanding of the container's state changes during loading and unloading can be achieved.

[0050] Step S1232: Set a state change threshold for each state monitoring parameter. When the change of any state monitoring parameter exceeds the corresponding state change threshold, a state change is determined to have occurred.

[0051] To accurately determine whether the container's condition has changed, it is necessary to set a threshold for each condition monitoring parameter. These thresholds are determined based on the container's design standards, safety requirements, and practical operational experience. For example, for the container tilt monitoring parameter, a threshold for the tilt angle is set. When the tilt angle of the container in a certain direction changes beyond this threshold, the container's tilt state is considered to have changed.

[0052] For container lock connection monitoring parameters, a threshold for changes in connection strength is set. When the change in lock connection strength exceeds this threshold, the lock connection status is determined to have changed. For container center of gravity monitoring parameters, a threshold for changes in center of gravity position is set. When the change in the container's center of gravity position exceeds this threshold, the container's center of gravity status is considered to have changed. A status change is determined to have occurred when the change in any status monitoring parameter exceeds its corresponding status change threshold.

[0053] Step S1233: Record the initial value of the state monitoring parameter, the value of the state monitoring parameter after the change, and the duration of the state change process when each state change occurs, and determine the direction attribute of the state change, which includes the direction of parameter increase and the direction of parameter decrease.

[0054] After determining that a state change has occurred, detailed information needs to be recorded. First, record the initial values ​​of the state monitoring parameters at the time the state change begins; then, record the values ​​of the state monitoring parameters after the change, i.e., the values ​​at the time the state change ends. Simultaneously, calculate the duration of the state change process, i.e., the time elapsed from the start to the end of the state change.

[0055] Furthermore, it is necessary to determine the directional attribute of the state change. For some state monitoring parameters, such as tilt angle and center of gravity position, the changes may involve either increase or decrease. By comparing the initial and subsequent values ​​of the state monitoring parameters, it can be determined whether the state change is occurring in the direction of parameter increase or decrease. For example, if the tilt angle of the container increases from its initial value to its subsequent value, then the directional attribute of the state change is the direction of parameter increase; if the tilt angle decreases, then the directional attribute is the direction of parameter decrease. By recording this information, a more comprehensive understanding of the process and characteristics of the container's state changes can be obtained.

[0056] Step S1234: Extract the start time and end time of each state change, and calculate the midpoint between the start time and end time as the time when the state change occurs.

[0057] Similar to determining the occurrence time of an action event, it is necessary to accurately determine the occurrence time of each state change. The start and end times of the state change are extracted from the data. The start time is the point at which the state monitoring parameters begin to change, and the end time is the point at which the state monitoring parameter changes end.

[0058] Calculate the midpoint between the start and end times and take it as the moment the state change occurs. This midpoint provides a more accurate representation of the state change's position on the timeline.

[0059] Step S1235: Classify the state changes according to the type of state monitoring parameters to obtain tilt state changes, lock connection state changes and center of gravity state changes.

[0060] Based on the type of extracted condition monitoring parameters, condition changes are categorized. Condition changes related to container tilt monitoring parameters are classified as tilt condition changes, reflecting alterations in the container's tilt angle and tilt stability. Condition changes related to container lock connection monitoring parameters are classified as lock connection condition changes, reflecting changes in the strength and state of the lock connection. Condition changes related to container center of gravity monitoring parameters are classified as center of gravity condition changes, indicating shifts in the container's center of gravity position and changes in its stability. This categorization provides a clearer understanding of the impact of different types of condition changes on container loading and unloading operations.

[0061] Step S1236: Associate and store the occurrence time of each state change, the type of state monitoring parameter, the direction attribute, the initial value of the state monitoring parameter, the value of the state monitoring parameter after the change, and the duration of the state change process. Add a unique tag to each state change to obtain the tagged container state data.

[0062] To facilitate data management and analysis, the occurrence time, state monitoring parameter type, direction attribute, initial value of the state monitoring parameter, post-change value of the state monitoring parameter, and duration of the state change process are stored in a correlated manner. A database approach can be used, storing this information in a single table, with one record corresponding to each state change. Within each record, this information is stored as different fields to ensure data integrity and consistency.

[0063] Simultaneously, a unique identifier is added to each state change. This identifier can be a string consisting of numbers and letters, used to uniquely identify each state change in the data. By adding identifiers, state changes can be easily queried, retrieved, and statistically analyzed. After the above processing, the tagged container state data is obtained.

[0064] Step S124: Mark the environmental parameters of the collected data of the working environment, record the collection time and parameter attributes of each environmental parameter, and obtain the marked working environment data.

[0065] Step S1241: Analyze the work environment data and extract the environmental monitoring parameters from the work environment data. These environmental monitoring parameters include meteorological parameters and obstacle parameters.

[0066] When processing data collected from the work environment, the first step is to parse the data. Since this data is collected from different types of environmental monitoring sensors, it may have different data formats and encoding methods. Therefore, appropriate parsing algorithms are needed to convert the data into a recognizable and processable form.

[0067] During the analysis process, the focus is on extracting environmental monitoring parameters from the data. For meteorological parameters, information such as temperature, humidity, wind speed, and wind direction are extracted from data collected by meteorological sensors. For obstacle parameters, information such as the location, size, shape, and movement speed of surrounding obstacles is obtained from data collected by optical or radar sensors. By accurately extracting these environmental monitoring parameters, a comprehensive understanding of the working environment can be achieved.

[0068] Step S1242: Record the acquisition time of each environmental monitoring parameter and associate the acquisition time with the corresponding environmental monitoring parameter for storage.

[0069] To accurately record changes in environmental monitoring parameters, it is necessary to record the acquisition time for each parameter. Each sensor can record the acquisition time simultaneously with data collection. The acquisition time and corresponding environmental monitoring parameter can be associated and stored using a database, with each parameter corresponding to one record. Within each record, the acquisition time and environmental monitoring parameter are stored as separate fields to ensure the temporal correlation and completeness of the data.

[0070] Step S1243: Analyze the characteristics and variation patterns of each environmental monitoring parameter, and determine the parameter attributes, including the parameter's stability, variation range, and variation trend.

[0071] Each environmental monitoring parameter is analyzed in depth to understand its characteristics and patterns of change. For meteorological parameters, their stability is analyzed, such as whether the temperature remains relatively stable over a period of time or fluctuates significantly; their range of variation is determined, such as the maximum and minimum values ​​of wind speed; and their trends are observed, whether they gradually increase, decrease, or remain unchanged.

[0072] For obstacle parameters, analyze the stability of their position, whether the obstacle is fixed in a certain position or constantly moving; determine the range of variation in their size and shape; observe the trend of their moving speed. By analyzing environmental monitoring parameters, determine their parameter attributes, which can more comprehensively describe the characteristics of the working environment.

[0073] Step S1244: Collect the acquisition time, parameter attributes and corresponding environmental monitoring parameters of each environmental monitoring parameter and store them together. Add a unique identifier to each environmental monitoring parameter to obtain the labeled working environment data.

[0074] To facilitate subsequent data management and analysis, the collection time, parameter attributes, and corresponding environmental monitoring parameters for each environmental monitoring parameter are stored together. This can be done using a database, storing this information in a single table, with one record for each environmental monitoring parameter. Within each record, the collection time, parameter attributes, and environmental monitoring parameter are stored as separate fields to ensure data integrity and consistency.

[0075] Simultaneously, a unique identifier is added to each environmental monitoring parameter. This identifier can be a string consisting of numbers and letters, used to uniquely identify each environmental monitoring parameter in the data. By adding identifiers, it is possible to easily query, retrieve, and statistically analyze environmental monitoring parameters. After the above processing, the tagged work environment data is obtained.

[0076] Step S125: Based on the occurrence time of action events in the marked reach stacker operation action data and the occurrence time of state changes in the marked container status data, associate action events and state changes that occur at the same time or with a time difference within a preset range to establish the first association relationship between the reach stacker operation action data and the container status data.

[0077] After obtaining the tagged reach stacker operation data and the tagged container status data, the first correlation is established. First, the occurrence time of the action events in the tagged reach stacker operation data is compared with the occurrence time of the status changes in the tagged container status data.

[0078] If the timing of an action event coincides with the timing of a state change, then the action event and the state change can be directly correlated. For example, if the container's tilt state changes just as the spreader begins to rise, then the action event of the spreader rising can be correlated with the change in the container's tilt state.

[0079] If the time of an action event differs from the time of a state change, but the time difference is within a preset range, they are also correlated. This preset range is determined based on actual operational conditions and experience, taking into account the time delay that might affect the container's state change. For example, after the spreader rotates, the container's center of gravity may not change immediately, but rather after a certain period. If the time difference between the rotation and the change in center of gravity is within the preset range, these two are correlated. Through this method, the first correlation between the data collected from the reach stacker operation and the container's state data is established.

[0080] Step S126: Based on the occurrence time of the action events in the marked reach stacker operation action data and the collection time of the environmental parameters in the marked operation environment data, associate the action events at the same time or with the environmental parameters within a preset range to establish a second association relationship between the reach stacker operation action data and the operation environment data.

[0081] Similarly, after obtaining the tagged reach stacker operation motion data and the tagged operating environment data, a second correlation is established. The occurrence times of motion events in the tagged reach stacker operation motion data are compared with the acquisition times of environmental parameters in the tagged operating environment data.

[0082] When an action event occurs at the same time as the environmental parameter acquisition, the action event is associated with the environmental parameter. For example, if the wind speed changes just as the spreader begins to move, then the action event of spreader movement can be associated with the environmental parameter of wind speed change.

[0083] When the time of an action event differs from the time of environmental parameter acquisition, but the time difference is within a preset range, they are also correlated. The preset range is determined by considering the time delay that environmental factors may have on the actions of the reach stacker. For example, in strong winds, the lifting and lowering of the spreader may be affected, but this effect may not be immediately apparent. If the time difference between the lifting / lowering event and the wind speed change acquisition time is within the preset range, these two are correlated. Through this method, a second correlation is established between the data collected on the actions of the reach stacker and the data collected on the working environment.

[0084] Step S127: Based on the time of occurrence of state change in the marked container state data and the time of collection of environmental parameters in the marked operating environment data, associate the state changes at the same time or with the environmental parameters within a preset range to establish a third association between the container state data and the operating environment data.

[0085] After obtaining the tagged container status data and the tagged operating environment data, a third correlation is established. The timing of status changes in the tagged container status data is compared with the timing of environmental parameter acquisition in the tagged operating environment data.

[0086] If the state change occurs at the same time as the environmental parameter acquisition, the state change is correlated with the environmental parameter. For example, if the container's tilt changes at the same time as the wind direction changes, the container's tilt change can be correlated with the environmental parameter of wind direction change.

[0087] If the time of the change in container status differs from the time of environmental parameter acquisition, but the time difference is within a preset range, they are also correlated. The preset range takes into account the time delay that environmental factors may cause to affect the container's status changes. For example, in a humid environment, the container's locking mechanism connection status may be affected, but this effect may not be immediately apparent. If the time difference between the time of the lock connection status change and the time of humidity change acquisition is within the preset range, these two are correlated. Through this method, a third correlation is established between the container status acquisition data and the operating environment acquisition data.

[0088] Step S128: Integrate the first association, the second association, and the third association, and supplement the association time information corresponding to each association to obtain the job status association mapping result.

[0089] After establishing the first, second, and third associations, they are integrated. This can be done by merging the data from the three associations into a single data structure. During the integration process, the association timing information for each association is supplemented. This timing information can be the specific time point when an action event is associated with a state change, an action event with environmental parameters, or a state change with environmental parameters, or the time difference between them.

[0090] By supplementing the associated time information, the temporal characteristics of the relationships between operational states can be more clearly reflected. After integrating and supplementing the information, the operational state association mapping result is obtained. This operational state association mapping result comprehensively demonstrates the relationship between the reach stacker operation, container status, and operational environment.

[0091] Step S130: Call the pre-trained artificial intelligence loading and unloading detection model to determine the operation status association mapping result, and generate an operation compliance determination result containing the abnormal association type and the corresponding determination basis.

[0092] Step S131: Extract the first standard association rule for the first association relationship, the second standard association rule for the second association relationship, and the third standard association rule for the third association relationship from the preset loading and unloading specifications.

[0093] Pre-defined loading and unloading specifications are a series of rules developed based on safety standards, operating procedures, and experience summaries for container front-end lifting and unloading operations. Before determining operational compliance, it is necessary to extract the standard association rules corresponding to the three relationships from the pre-defined loading and unloading specifications.

[0094] For the first correlation, namely the correlation between the data collected from the actions of the reach stacker and the data collected from the container status, the first standard correlation rule is extracted. This rule specifies which changes in the container's status should correspond to the various actions of the reach stacker under normal operating conditions. For example, the lifting action of the spreader should be accompanied by the stabilization or slight rise of the container's center of gravity, and should not result in abnormal changes in the container's status such as excessive tilting.

[0095] For the second correlation, namely the correlation between the data collected on the reach stacker's operational actions and the data collected on the operational environment, a second standard correlation rule is extracted. This rule clarifies how the reach stacker's actions should be adjusted under different operational environmental conditions. For example, in strong winds, the traversing speed of the spreader should be appropriately reduced to ensure operational safety.

[0096] For the third correlation, namely the correlation between container status data and operational environment data, a third standard correlation rule is extracted. This rule describes the normal impact of operational environment factors on container status. For example, in a high-temperature environment, the container's locking connections may become somewhat loose, but this loosening should be within a reasonable range.

[0097] Step S132: Input the first association relationship in the operation status association mapping result into the first judgment module of the artificial intelligence loading and unloading detection model. The first judgment module loads the first standard association rule and compares the degree of fit between the first association relationship and the first standard association rule.

[0098] Step S1321: Parse the first association relationship and extract the action event information and state change information in the first association relationship. The action event information includes the action type, action parameters and associated time. The state change information includes the state monitoring parameter type, direction attribute and associated time.

[0099] In the operational status association mapping results, the first association relationship includes the association information between the data collected from the reach stacker's operational actions and the data collected from the container status. To accurately compare its fit with the first standard association rule, the first association relationship must first be parsed. During data processing, based on the structural characteristics of the data in the first association relationship, data segments related to action events and status changes are identified. From these data segments, action event information is extracted, where action types may include spreader lifting, spreader rotation, and spreader translation; action parameters cover changes in action amplitude, speed, and duration; and the association time is the specific time point when the action event occurs. Simultaneously, status change information is extracted, where status monitoring parameter types include container tilt monitoring parameters, container locking device connection monitoring parameters, and container center of gravity monitoring parameters; directional attributes are divided into parameter increase direction and parameter decrease direction; and the association time is the specific time point when the status change occurs. For example, when the reach stacker performs a spreader lifting action, the start time, speed change, and duration of the action are recorded, along with whether the container tilt angle changes, the direction of the change, and the time of the change.

[0100] Step S1322: Parse the first standard association rule, extract the correspondence between allowed action types and state monitoring parameter types, the matching relationship between allowed action parameter ranges and state change direction attributes, and the allowed association time difference range in the first standard association rule.

[0101] The first standard association rule is formulated based on preset loading and unloading specifications and is used to measure whether the first association relationship is compliant. When parsing it, the correspondence between allowed action types and status monitoring parameter types is found according to the logical structure of the rule. For example, if it is stipulated that when the spreader is rising, the container's center of gravity monitoring parameter should show a certain stable change state, this is a correspondence between action type and status monitoring parameter type. Simultaneously, the matching relationship between the allowed action parameter range and the state change direction attribute is extracted. For example, when the spreader rises at a certain speed, the container's tilt angle should change in the direction of decrease. In addition, the allowed correlation time difference range also needs to be extracted, that is, the allowed time difference range between the time of the action event and the time of the state change. If the correlation time between the action event and the state change exceeds this range, it may not comply with the standard rule.

[0102] Step S1323: Compare the correspondence between the action type and the status monitoring parameter type in the first association relationship to see if it is consistent with the correspondence between the allowed action type and the status monitoring parameter type in the first standard association rule, and count the first proportion of the number of consistent items to the total number of correspondence relationships.

[0103] After obtaining the relevant information of the first association relationship and the first standard association rule, a specific comparison begins. The correspondence between the actual action types and status monitoring parameter types in the first association relationship is compared one by one with the allowed correspondences in the first standard association rule. For each pair of correspondences, it is determined whether they are consistent. If they are consistent, they are recorded as a consistent item. The number of all consistent items is counted, and their proportion to the total number of correspondences is calculated to obtain the first proportion. For example, in a single operation, there may be multiple correspondences between action types and status monitoring parameter types in the first association relationship. After comparing these cases with the standard rule, the number of consistent correspondences is counted, and then divided by the total number of correspondences to obtain the first proportion. This proportion reflects the degree of fit between the first association relationship and the standard rule in terms of the correspondence between action types and status monitoring parameter types.

[0104] Step S1324: Compare whether the action parameters in the first association relationship are within the range of action parameters allowed by the first standard association rule, and whether the state change direction attribute is consistent with the matching relationship between the range of action parameters allowed by the first standard association rule and the state change direction attribute. Count the second proportion of the number of matching items to the total number of parameters.

[0105] In addition to the correspondence between action types and status monitoring parameter types, it is also necessary to compare the matching of action parameters and status change direction attributes. The action parameters in the first association relationship, such as changes in action amplitude, speed, and duration, are compared with the range of action parameters allowed by the first standard association rule to determine if they are within the allowed range. Simultaneously, it is checked whether the status change direction attribute matches the matching relationship between the allowed range of action parameters and the status change direction attribute in the standard rule. If the action parameters are within the allowed range and the status change direction attribute also matches, it is recorded as a matching item. The number of all matching items is counted, and their proportion to the total number of parameters is calculated to obtain the second proportion. This proportion reflects the degree of fit between the first association relationship and the standard rule in terms of matching action parameters and status change direction attributes. For example, when a spreader rotates, its rotation speed and angle must be within the range allowed by the standard rule, and the direction of the container's tilt change must be consistent with the standard matching relationship under the aforementioned rotation action for it to be considered a matching item.

[0106] Step S1325: Calculate the difference between the action event association time and the state change association time in the first association relationship, determine whether the difference is within the range of association time difference allowed by the first standard association rule, and count the third proportion of the number of matching items to the total number of association time pairs.

[0107] The timing of the association between action events and state changes is also an important aspect of the comparison. The difference between the timing of the action event association and the timing of the state change association in the first association relationship is calculated. This difference is then compared to the range of possible differences allowed by the first standard association rule. If the difference is within the allowed range, it is recorded as a conforming item. The number of all conforming items is counted, and their proportion to the total number of association time pairs is calculated to obtain the third proportion. This proportion reflects the degree of conformity between the first association relationship and the standard rule in terms of association time. For example, in a single task, there are multiple pairs of action events and state changes with corresponding timing. The difference between each pair of timings is calculated to determine if it is within the range allowed by the standard rule. The number of conforming pairings is counted, and then divided by the total number of pairings to obtain the third proportion.

[0108] Step S1326: Set the weight coefficients of the first ratio, the second ratio and the third ratio, and calculate the comprehensive fit between the first association relationship and the first standard association rule by weighted summation. The comprehensive fit is the fit between the first association relationship and the first standard association rule.

[0109] To comprehensively consider the impact of three aspects on the degree of fit—the correspondence between action type and status monitoring parameter type, the matching relationship between action parameters and status change direction attributes, and the relationship of associated time—it is necessary to set weight coefficients for the first, second, and third proportions. These weight coefficients are determined based on the importance of different aspects in operational compliance. Then, the overall degree of fit between the first association relationship and the first standard association rule is calculated by weighted summation. For example, assuming the weight coefficient for the first proportion is one value, the weight coefficient for the second proportion is another value, and the weight coefficient for the third proportion is yet another value, multiplying each of these three proportions by its respective weight coefficient and then adding them together yields the overall degree of fit between the first association relationship and the first standard association rule. This overall degree of fit can comprehensively reflect the overall fit between the first association relationship and the standard rule.

[0110] Step S133: When the degree of fit between the first association relationship and the first standard association rule is lower than the first preset fit threshold, the first association relationship is determined to be an abnormal first association relationship, and the fit degree value and the non-fit first standard association rule entries are recorded.

[0111] After obtaining the degree of fit between the first association relationship and the first standard association rule, it is compared with the first preset fit threshold. The first preset fit threshold is a standard value set based on the safety requirements and experience of the operation.

[0112] If the degree of fit between the first association relationship and the first standard association rule is lower than the first preset fit threshold, it indicates that the first association relationship does not conform to the normal operating procedures, and the first association relationship is determined to be an abnormal first association relationship.

[0113] Simultaneously, record the degree of fit value, which can intuitively reflect the degree of deviation between the first association relationship and the first standard association rule. In addition, it is also necessary to record the non-fitting first standard association rule entries, clearly indicating which rules were not satisfied. The above recorded information will serve as an important basis for subsequent analysis and processing.

[0114] Step S134: Input the second association relationship in the operation status association mapping result into the second judgment module of the artificial intelligence loading and unloading detection model. The second judgment module loads the second standard association rule and compares the degree of fit between the second association relationship and the second standard association rule.

[0115] Similarly, the second correlation in the operation status correlation mapping result is input into the second decision module of the AI ​​loading and unloading detection model. The second decision module is another sub-module in the AI ​​loading and unloading detection model, which is specifically used to handle the correlation between the reach stacker operation and the operation environment.

[0116] After loading the second standard association rule, the second judgment module begins to compare the degree of fit between the second association relationship and the second standard association rule. The specific comparison process is similar to that of the first judgment module, which first parses the second association relationship and the second standard association rule, extracts relevant information, then performs various matching and comparisons, and finally calculates the overall degree of fit through weighted summation.

[0117] Step S135: When the degree of fit between the second association relationship and the second standard association rule is lower than the second preset fit threshold, the second association relationship is determined to be an abnormal second association relationship, and the fit degree value and the non-fit second standard association rule entries are recorded.

[0118] After obtaining the degree of fit between the second association relationship and the second standard association rule, it is compared with the second preset fit threshold. The second preset fit threshold is also a standard value set based on the safety requirements and experience of the operation.

[0119] If the degree of fit between the second association and the second standard association rule is lower than the second preset fit threshold, the second association is determined to be an abnormal second association. The fit value and the non-fitting second standard association rule entries are recorded.

[0120] Step S136: Input the third association relationship in the operation status association mapping result into the third judgment module of the artificial intelligence loading and unloading detection model. The third judgment module loads the third standard association rule and compares the degree of fit between the third association relationship and the third standard association rule.

[0121] The third correlation in the operation status association mapping result is input into the third decision module of the AI ​​loading and unloading detection model. The third decision module is another sub-module in the AI ​​loading and unloading detection model, which is specifically used to handle the correlation between container status and operation environment.

[0122] After the third judgment module loads the third standard association rule, it compares the degree of fit between the third association relationship and the third standard association rule in a similar way to the first two judgment modules, and calculates the comprehensive fit.

[0123] Step S137: When the degree of fit between the third association relationship and the third standard association rule is lower than the third preset fit threshold, the third association relationship is determined to be an abnormal third association relationship, and the fit degree value and the non-fit third standard association rule entries are recorded.

[0124] After obtaining the degree of fit between the third association relationship and the third standard association rule, it is compared with the third preset fit threshold. The third preset fit threshold is set based on the safety requirements of the operation and experience.

[0125] If the degree of fit between the third association relationship and the third standard association rule is lower than the third preset fit threshold, the third association relationship is determined to be an abnormal third association relationship. The fit value and the non-fitting third standard association rule entries are recorded for subsequent analysis and processing.

[0126] Step S138: Collect all first, second, and third abnormal relationships, classify and label the abnormal relationship types, organize the matching degree values ​​and non-standard association rule entries corresponding to each abnormal relationship as the basis for judgment, and generate an operation compliance judgment result that includes the abnormal relationship type and the corresponding judgment basis.

[0127] After determining the three types of relationships, all abnormal first, second, and third relationships are collected. These abnormal relationships are then categorized and labeled to clearly distinguish whether they are abnormal relationships between the reach stacker operation and the container status, between the reach stacker operation and the operating environment, or between the container status and the operating environment.

[0128] The matching degree values ​​and non-standard association rule entries for each abnormal relationship are compiled, and this information is used as the basis for judgment. Through this method, an operational compliance judgment result is generated, containing the abnormal relationship type and the corresponding judgment basis. This operational compliance judgment result clearly indicates the abnormal relationships existing in the operation process and their causes.

[0129] Step S140: Based on the abnormal relationship type in the compliance judgment result of the operation, and combined with the segment division rules of the front lifting and unloading operation, determine the operation segment corresponding to each abnormal relationship, and obtain the abnormal segment matching result containing the correspondence between abnormal relationships and operation segments.

[0130] Step S141: Obtain the segment division rules for front-end lifting and unloading operations. These segment division rules define the time interval division standards and operation content characteristics for the spreader alignment segment, container lifting segment, container translation segment, and container placement segment.

[0131] The segmentation rules for front-end lifting and unloading operations are formulated based on the operation process and characteristics. These segmentation rules clarify the time intervals and operational characteristics for the spreader alignment, container lifting, container translation, and container placement segments.

[0132] For the spreader alignment process, the time interval can be determined based on the time it takes for the reach stacker to move from its initial position to above the container in preparation for the lifting operation. The operational characteristics include fine-tuning of the spreader, such as small-range lifting, rotation, and translation, to ensure accurate alignment of the spreader with the container's locking holes; it also includes monitoring parameters of the spreader alignment status, such as the relative position and angle between the spreader and the container.

[0133] For container lifting, the time interval is defined as the period from when the spreader locks onto the container until the container leaves the ground and reaches a certain height. The operational characteristics include the lifting action of the spreader, monitoring the container's locking status parameters to ensure a secure connection, and monitoring the container's tilt status parameters to ensure stability during lifting.

[0134] For container lateral movement, the time interval is defined as the period from when the container is lifted to a certain height and begins lateral movement until it reaches the target position. The operational characteristics include the horizontal movement of the spreader, monitoring the container's center of gravity parameters to ensure stability during lateral movement, and monitoring environmental obstacle parameters to prevent collisions with surrounding obstacles.

[0135] For the container placement process, the time interval is defined as the period from when the container begins to descend above the target position until it is fully placed and unlocked. The operational characteristics include the descent of the spreader, monitoring the container's placement status parameters to ensure accurate placement, and monitoring the container's lock unlocking status parameters to ensure successful unlocking.

[0136] Step S142: Extract the association time information corresponding to each abnormal relationship from the compliance judgment result of the operation. The association time information includes the association time of the first abnormal relationship, the association time of the second abnormal relationship, and the association time of the third abnormal relationship.

[0137] After obtaining the compliance determination results, the correlation time information corresponding to each abnormal correlation is extracted. For the first abnormal correlation, the correlation time when the action event is associated with the state change is extracted; for the second abnormal correlation, the correlation time when the action event is associated with the environmental parameters is extracted; for the third abnormal correlation, the correlation time when the state change is associated with the environmental parameters is extracted. The above correlation time information can accurately reflect the time point when the abnormal correlation occurs.

[0138] Step S143: Based on the time interval division criteria in the process division rules, determine the time interval to which the associated time of each abnormal relationship belongs, and initially match the corresponding work process.

[0139] Based on the extracted time information of abnormal relationships, and combined with the time interval division criteria in the process segmentation rules, the time interval to which the time of association of each abnormal relationship belongs is determined. For example, if the time of association of an abnormal relationship falls within the time interval of the spreader alignment process, then the initial matching of the work process corresponding to this abnormal relationship is the spreader alignment process.

[0140] Step S144: Extract action event information or state change information from each abnormal relationship. The action event information includes the action type, and the state change information includes the state monitoring parameter type.

[0141] After completing the initial matching, extract the action event information or state change information from each abnormal relationship. For abnormal relationships containing action events, extract the action type, such as spreader lifting action, spreader rotation action, or spreader translation action; for abnormal relationships containing state changes, extract the state monitoring parameter type, such as container tilt monitoring parameters, container lock connection monitoring parameters, or container center of gravity monitoring parameters.

[0142] Step S145: Based on the task content characteristics in the task segmentation rules, verify whether the initially matched task segments match the action event information or state change information. The task content characteristics define the typical action types and state monitoring parameter types for each task segment.

[0143] Step S1451: Extract the operational content features of each operational stage from the stage division rules. The operational content features of the spreader alignment stage include spreader fine-tuning actions and spreader alignment status monitoring parameters; the operational content features of the container lifting stage include spreader raising actions, container locking status monitoring parameters, and container tilting status monitoring parameters; the operational content features of the container translating stage include spreader horizontal movement actions, container center of gravity status monitoring parameters, and operational environment obstacle monitoring parameters; the operational content features of the container lowering stage include spreader lowering actions, container lowering status monitoring parameters, and container lock unlocking status monitoring parameters.

[0144] The segmentation rules detail the operational characteristics of each stage of the front-end lifting and unloading operation. When verifying the initially matched operational stages, the specific characteristics of each stage must first be extracted from these rules. For the spreader alignment stage, the operational characteristics mainly revolve around the spreader's fine-tuning movements, such as small-range lifting, rotation, and translation, to achieve accurate alignment; it also includes spreader alignment status monitoring parameters to monitor the relative position and angle between the spreader and the container. For the container lifting stage, the focus is on the spreader's upward movement and the container's locking status monitoring parameters to ensure correct locking; there are also container tilting status monitoring parameters to ensure container stability during lifting. The container traversing stage includes the spreader's horizontal movement and the container's center of gravity status monitoring parameters to prevent center of gravity shift; it also involves monitoring parameters for obstacles in the working environment to avoid collisions during traversal. The container lowering stage mainly focuses on the spreader's downward movement and the container lowering status monitoring parameters to ensure accurate container lowering; there are also container lock unlocking status monitoring parameters to ensure smooth lock unlocking.

[0145] Step S1452: For action event information in abnormal association relationships, extract the action type and determine whether the action type belongs to the typical action type in the work content characteristics of the preliminary matching work process.

[0146] After obtaining the operational characteristics of each work stage, the action event information in the abnormal correlations is analyzed. The action types are extracted and then compared with typical action types in the operational characteristics of the initially matched work stages. For example, if the initially matched work stage is container lifting, and the action type in the abnormal correlation is spreader raising, then this action type matches the typical action type in the operational characteristics of container lifting. If the action type is spreader rotation, but rotation is not a typical action type in container lifting, then it does not meet the requirements. Through this comparison, it can be preliminarily determined whether the abnormal correlations match the initially matched work stages in terms of action types.

[0147] Step S1453: For the status change information in the abnormal association, extract the status monitoring parameter type and determine whether the status monitoring parameter type belongs to the typical status monitoring parameter type in the operation content characteristics of the preliminary matching operation link.

[0148] In addition to the action type, it is also necessary to analyze the state change information in the abnormal correlation. Extract the state monitoring parameter types and then compare them with the typical state monitoring parameter types in the initial matching operation's content characteristics. For example, if the initial matching operation is a container translation, and the state monitoring parameter type in the abnormal correlation is a container center of gravity status monitoring parameter, this matches the typical state monitoring parameter type in the container translation operation's content characteristics. However, if the state monitoring parameter type is a container lock connection monitoring parameter, and lock connection monitoring is not a primary state monitoring content in the container translation operation, then it does not meet the requirements. Through the above comparison, it can be determined whether the abnormal correlation matches the initial matching operation in terms of state monitoring parameter types.

[0149] Step S1454: When the action type belongs to the typical action type in the work content characteristics of the initially matched work link, and the status monitoring parameter type belongs to the typical status monitoring parameter type in the work content characteristics of the initially matched work link, it is determined that the initially matched work link matches the action event information or status change information.

[0150] After comparing the action types and status monitoring parameter types, if the action type is a typical action type among the characteristics of the work content of the initially matched operation, and the status monitoring parameter type is also a typical status monitoring parameter type for that operation, then it can be determined that the initially matched operation matches the action event information or status change information. For example, in the container lifting operation, the action type in the abnormal correlation is the spreader raising action, and the status monitoring parameter type is the container tilting status monitoring parameter. Both of these match the characteristics of the work content of the container lifting operation, so it can be determined that the information of the initially matched operation matches the abnormal correlation.

[0151] Step S1455: When the action type is not a typical action type in the work content characteristics of the initially matched work link, or the status monitoring parameter type is not a typical status monitoring parameter type in the work content characteristics of the initially matched work link, it is determined that the initially matched work link does not match the action event information or status change information.

[0152] Conversely, if the action type is not a typical action type among the characteristics of the work content in the initially matched work segment, or if the status monitoring parameter type is not a typical status monitoring parameter type for that work segment, then the initially matched work segment is determined to be mismatched with the action event information or status change information. For example, in the container translation process, the action type in the abnormal correlation is the spreader rotation action, but rotation is not a typical action type in the container translation process; or the status monitoring parameter type is the container lock connection monitoring parameter, but lock connection monitoring is not a primary monitoring content in the translation process. In this case, the information in the initially matched work segment is determined to be mismatched with the abnormal correlation.

[0153] Step S1456: Record the judgment result of each verification and the corresponding work content characteristics.

[0154] After each verification, the judgment result and the corresponding work content feature basis need to be recorded. The judgment result clearly indicates whether the initially matched work step matches the information of the anomaly correlation; the work content feature basis details which work content features were used for the judgment. For example, if the initially matched work step is determined to match the information of the anomaly correlation, the record will explain that this is because the action type and status monitoring parameter type both conform to the typical characteristics of the work step, and list the specific feature content; if it is determined not to match, the specific reasons for the mismatch and the work content features involved will also be recorded. The above records are helpful for subsequent further analysis and processing of anomaly correlations.

[0155] Step S146: When the initially matched work segment matches the action event information or state change information, the work segment is confirmed as the work segment corresponding to the abnormal association; when they do not match, a second matching is performed based on the time interval division standard and the work content characteristics in the segment division rules until a unique corresponding work segment is determined.

[0156] If the initially matched operational step matches the action event information or status change information, then the operational step is confirmed as the operational step corresponding to the abnormal association. For example, if the initially matched container lifting step matches the spreader lifting action and container locking status monitoring parameter change information in the abnormal association, then it can be determined that the operational step corresponding to the abnormal association is the container lifting step.

[0157] If the initially matched task segment does not match the action event information or state change information, a second matching is required based on the time interval division criteria and task content characteristics in the segment division rules. During the second matching process, the time intervals to which the associated moments of abnormal relationships belong are checked again, and a more detailed analysis and judgment are made in conjunction with the task content characteristics until a unique corresponding task segment is determined.

[0158] Step S147: Associate each abnormal relationship with the determined work process, mark the time interval and work content feature used in the matching process, and obtain the abnormal process matching result containing the correspondence between abnormal relationships and work processes.

[0159] After identifying the work steps corresponding to each type of anomaly association, the anomaly associations are linked to the work steps and recorded. This can be done using a database, storing the relevant information about the anomaly associations and the corresponding work step information in a single table, with each record corresponding to one type of anomaly association.

[0160] During the recording process, the time interval and task content characteristics used in the matching process are labeled. The time interval criterion clarifies the specific time interval to which the associated moments of the abnormal relationships belong, while the task content characteristics criterion explains which task content characteristics were used to determine the task steps. Through this method, an abnormal step matching result containing the correspondence between abnormal relationships and task steps is obtained. This abnormal step matching result clearly shows the specific step in the loading and unloading operation corresponding to each abnormal relationship.

[0161] Step S150: Based on the judgment criteria in the compliance judgment result of the operation and the operation links in the abnormal link matching result, generate an operation adjustment instruction for the corresponding operation link, and send the operation adjustment instruction to the front crane control terminal to adjust the loading and unloading operation of the corresponding operation link.

[0162] Step S151: Extract the standard association rule entries and corresponding parameter deviation information for each type of abnormal association relationship from the judgment basis of the compliance judgment result of the operation. The parameter deviation information includes action parameter deviation, state parameter deviation and association time deviation.

[0163] After obtaining the compliance determination result, extract the standard association rule entries and corresponding parameter deviation information for each type of abnormal correlation from the determination criteria. For the non-compliant standard association rule entries, clearly indicate which rules were not met; for example, the lifting speed of the spreader is too fast and does not conform to the speed range specified in the standard rules.

[0164] Parameter deviation information includes motion parameter deviation, state parameter deviation, and correlation time deviation. Motion parameter deviation can be the difference between parameters such as the lifting speed and rotation angle of the spreader and the allowable range of the standard rules; state parameter deviation can be the deviation between parameters such as the tilt angle of the container and the strength of the locking mechanism and the requirements of the standard rules; correlation time deviation can be the deviation between the correlation time between the motion event and the state change, the motion event and the environmental parameters, and the state change and the environmental parameters and the correlation time allowed by the standard rules. By extracting this information, the causes and degrees of deviation of abnormal correlations can be accurately understood.

[0165] Step S152: Based on the abnormal link matching results, determine the work link corresponding to each abnormal relationship, query the preset link adjustment strategy library, and obtain the adjustment direction for different parameter deviations under the work link. The adjustment direction includes the direction of motion parameter correction, the direction of state parameter improvement, and the direction of correlation time optimization.

[0166] Based on the abnormal process matching results, the corresponding work process for each abnormal relationship is determined. Then, a pre-set process adjustment strategy library is queried. This strategy library is developed based on operational experience and safety requirements, and provides corresponding adjustment directions for different work processes and parameter deviations.

[0167] Regarding the direction of motion parameter correction, if the spreader's lifting speed is too fast during the container lifting phase, the adjustment direction may be to reduce the spreader's lifting speed; if the spreader's lateral speed is unstable during the container lateral movement phase, the adjustment direction may be to adjust the lateral speed control strategy to make it more stable.

[0168] Regarding the direction of improving the status parameters, if the container tilt angle is too large during the container lifting process, the adjustment direction may be to adjust the position or attitude of the spreader to reduce the tilt angle of the container; if the container lock connection is not firm during the container placement process, the adjustment direction may be to strengthen the connection of the lock.

[0169] Regarding the optimization direction for correlation timing, if there is a significant deviation between the correlation timing of action events and state changes during the spreader alignment process, the adjustment direction might be to optimize the operation process to make the correlation timing of action events and state changes more consistent with standard rules. This can be achieved by querying the process adjustment strategy library to obtain adjustment directions for different parameter deviations.

[0170] Step S153: Based on the degree of deviation of the parameter deviation information, determine the adjustment force corresponding to each adjustment direction. The greater the degree of deviation, the greater the adjustment force.

[0171] For example, step S1531: Extract the deviation value from the parameter deviation information, which includes the deviation value of the action parameter, the deviation value of the state parameter, and the deviation value of the associated time.

[0172] Parameter deviation information reflects the difference between actual parameters and standard rule requirements in abnormal correlations. When determining the adjustment force, the deviation value must first be extracted from this information. For motion parameter deviation values, this might be the difference between motion parameters such as the lifting speed and rotation angle of the spreader and the allowable range of the standard rule; for state parameter deviation values, this is the deviation between state parameters such as the tilt angle of the container and the strength of the locking mechanism and the requirements of the standard rule; for correlation time deviation values, this is the deviation between the correlation time between a motion event and a state change, a motion event and environmental parameters, and a state change and environmental parameters and the correlation time allowed by the standard rule. For example, when the actual lifting speed of the spreader exceeds the speed range allowed by the standard rule, the difference is the motion parameter deviation value; if the actual tilt angle of the container differs from the tilt angle required by the standard, this difference is the state parameter deviation value.

[0173] Step S1532: Obtain a preset deviation degree classification standard, which divides the deviation value into different deviation levels, and each deviation level corresponds to a deviation degree description.

[0174] To accurately determine the degree of adjustment, a pre-defined standard for classifying deviation levels is needed. This standard categorizes deviation values ​​into different levels, such as minor, moderate, and severe deviations. Each deviation level corresponds to a description of its degree, clarifying the extent to which that level of deviation affects operational compliance. For example, a minor deviation indicates a small impact on the operation, potentially requiring only minor adjustments; a moderate deviation indicates a certain impact, necessitating moderate adjustments; and a severe deviation indicates a significant impact, requiring substantial adjustments.

[0175] Step S1533: Compare the extracted deviation values ​​with the deviation degree classification standard to determine the deviation level corresponding to each parameter deviation information, and then determine the degree of deviation.

[0176] After obtaining the deviation value and the standard for classifying the degree of deviation, the extracted deviation value is compared with the standard. Based on the range of the deviation value, the deviation level corresponding to each parameter deviation information is determined. For example, if the deviation value of a motion parameter is within the range of mild deviation, then the deviation level corresponding to that motion parameter deviation information is mild deviation, and its degree of deviation is determined to be mild. Through the above comparison and determination, the severity of each parameter deviation can be accurately understood.

[0177] Step S1534: Query the preset adjustment force mapping table, which defines the adjustment force range corresponding to different work stages, different adjustment directions, and different degrees of deviation.

[0178] The pre-defined adjustment intensity mapping table is developed based on operational experience and safety requirements to determine the adjustment intensity under different conditions. After determining the deviation level and degree of each parameter deviation, the mapping table is queried based on the operational stage, adjustment direction, and deviation degree corresponding to the current abnormal correlation. This table defines the adjustment intensity range corresponding to different operational stages, different adjustment directions (such as motion parameter correction direction, state parameter improvement direction, and correlation time optimization direction), and different deviation degrees. For example, in the container lifting stage, for the motion parameter correction direction, the adjustment intensity range corresponding to a slight deviation might be a small numerical range; the adjustment intensity range corresponding to a moderate deviation might be larger; and the adjustment intensity range corresponding to a severe deviation would be even larger.

[0179] Step S1535: Based on the current abnormal correlation, the corresponding work process, adjustment direction, and determined deviation degree, query the corresponding adjustment intensity from the adjustment intensity mapping table. If the adjustment intensity is within the safe range of the operating parameters under the work process, determine the adjustment intensity as the final adjustment intensity; if the adjustment intensity exceeds the safe range of the operating parameters under the work process, correct the adjustment intensity to the boundary value of the safe range of the operating parameters under the work process, and use it as the final adjustment intensity.

[0180] Based on the retrieved adjustment intensity mapping table, and considering the current abnormal correlation, the corresponding operational stage, adjustment direction, and determined deviation degree, the corresponding adjustment intensity is determined. Then, it's necessary to consider whether this adjustment intensity falls within the safe range of the operating parameters for that operational stage. If the adjustment intensity is within the safe range, it can be directly determined as the final adjustment intensity. For example, in the container translation stage, for the direction of improving state parameters, if the adjustment intensity retrieved based on the deviation degree is within the safe range of the operating parameters, then this adjustment intensity is the final adjustment intensity. If the adjustment intensity exceeds the safe range, to ensure operational safety, the adjustment intensity needs to be corrected to the boundary value of the safe range of the operating parameters for that operational stage, and this becomes the final adjustment intensity. For example, in the container placement stage, if the retrieved adjustment intensity for the direction of optimizing the correlation moment is too large and exceeds the safe range of the operating parameters, then the adjustment intensity is adjusted to the boundary value of the safe range, and this becomes the final adjustment intensity.

[0181] Step S1536: Record the basis for the deviation degree corresponding to the final adjustment force, the basis for querying the adjustment force mapping table, and the results of the rationality check.

[0182] After determining the final adjustment level, it is necessary to record the relevant basis and inspection results. Record the basis for the deviation level, specifying which deviation grade and degree determined the adjustment level; record the basis for looking up the adjustment level mapping table, clarifying from which position in the mapping table the adjustment level was retrieved; record the results of the rationality check, explaining whether the adjustment level was within the safe range, and if corrections were made, the reasons and process for such corrections. These records help in the subsequent evaluation and optimization of the adjustment operation, ensuring that the determination of the adjustment level is reasonable, accurate, and safe.

[0183] Step S154: Combine the adjustment direction and adjustment force to generate specific operation adjustment content for this operation step. The specific operation adjustment content includes the adjustment value of the action parameter, the control target of the state parameter and the coordination scheme of the associated time.

[0184] By combining the adjustment direction and adjustment force, specific operational adjustments are generated for this operational phase. For the motion parameter adjustment value, the specific adjustment value is determined based on the adjustment direction and adjustment force of the motion parameter. For example, if the adjustment direction is to reduce the lifting speed of the spreader, and the adjustment force is to significantly reduce it, then the motion parameter adjustment value might be to reduce the lifting speed of the spreader to a lower value.

[0185] For state parameter control objectives, the target value of the state parameter to be achieved is determined based on the direction of improvement and the adjustment force. For example, if the adjustment direction is to reduce the tilt angle of the container, and the adjustment force is a relatively large adjustment, then the state parameter control objective might be to reduce the tilt angle of the container to a smaller range.

[0186] For the correlation timing coordination scheme, a specific coordination plan is formulated based on the optimization direction and adjustment intensity of the correlation timing. For example, if the adjustment direction is to optimize the correlation timing between action events and state changes, and the adjustment intensity is a relatively large adjustment, then the correlation timing coordination scheme may involve a significant optimization of the operation process to make the correlation timing between action events and state changes more accurate.

[0187] Step S155: Add a work process identifier, an abnormal relationship type identifier, and an adjustment priority identifier to the specific operation adjustment content to construct a structured operation adjustment instruction. The structure of the operation adjustment instruction includes an instruction header, an adjustment content body, and an instruction tail. The instruction header contains identification information, the adjustment content body contains the specific operation adjustment content, and the instruction tail contains the execution requirements.

[0188] To facilitate the identification and processing of operational adjustments by the reach stacker control terminal, a work phase identifier, an anomaly relationship type identifier, and an adjustment priority identifier are added to each specific adjustment. The work phase identifier clarifies which work phase the adjustment targets, such as spreader alignment or container lifting. The anomaly relationship type identifier distinguishes the type of anomaly relationship causing the adjustment, such as an anomaly between the reach stacker's operation and the container's status, or an anomaly between the reach stacker's operation and the operating environment. The adjustment priority identifier determines the execution priority of the adjustment, set based on the severity of the anomaly relationship and its impact on operational safety.

[0189] A structured operational adjustment instruction is constructed, comprising an instruction header, an adjustment content body, and an instruction tail. The instruction header includes identification information such as the work process identifier, the abnormal relationship type identifier, and the adjustment priority identifier, facilitating rapid identification of relevant information by the front-end crane control terminal. The adjustment content body contains specific operational adjustment details, such as motion parameter adjustment values, status parameter control targets, and associated timing coordination schemes. The instruction tail includes execution requirements, such as execution time and execution method. By constructing a structured operational adjustment instruction, the clarity and executability of the instructions are ensured.

[0190] Step S156: Establish a secure communication link with the front crane control terminal and send the operation adjustment command to the front crane control terminal through encrypted transmission.

[0191] To ensure that operational adjustment commands are transmitted securely and accurately to the reach stacker control terminal, a secure communication link must be established. Encryption technology can be used to protect the communication link, preventing commands from being stolen or tampered with during transmission.

[0192] Operation adjustment commands are sent to the forward crane control terminal via encrypted transmission. During transmission, the commands are encrypted and transmitted in ciphertext form. Upon receiving the ciphertext commands, the forward crane control terminal uses a corresponding decryption algorithm to restore them to plaintext. This method ensures the security and reliability of the operation adjustment commands.

[0193] Step S157: Receive instruction reception confirmation information from the front crane control terminal. This instruction reception confirmation information includes the instruction identifier and reception status.

[0194] After receiving an operation adjustment command, the reach stacker control terminal can send a command reception confirmation message back to the sending end. This confirmation message includes a command identifier and a reception status. The command identifier uniquely identifies the operation adjustment command, ensuring that the feedback message corresponds to the sent command; the reception status indicates whether the reach stacker control terminal successfully received the command, and can be categorized as either successful or failed.

[0195] Step S158: When the reception status is successful, the operation adjustment command is sent; when the reception status is unsuccessful, the secure communication link with the front crane control terminal is re-established and the operation adjustment command is sent again until the reception status is successful, so as to ensure that the front crane control terminal can adjust the loading and unloading operations of the corresponding work link according to the operation adjustment command.

[0196] The determination is made based on the reception status in the instruction reception confirmation information fed back by the front-end crane control terminal. If the reception status is successful, it means that the operation adjustment instruction has been successfully sent to the front-end crane control terminal, and the transmission of the operation adjustment instruction is completed.

[0197] If the reception status fails, it indicates a problem occurred during command transmission. A secure communication link with the reach stacker control terminal needs to be re-established, and the operation adjustment command resent. During the re-establishment of the communication link, the connection status and encryption settings are checked to ensure the link's security and stability. After resending the operation adjustment command, continue receiving feedback from the reach stacker control terminal until the reception status is successful. Through this process, the reach stacker control terminal can adjust the loading and unloading operations of the corresponding work stages according to the operation adjustment command, improving operational safety and compliance.

[0198] Furthermore, Figure 2 A schematic diagram of the hardware structure of an artificial intelligence-based container front-end lifting and unloading detection system 100 for implementing the methods provided in the embodiments of this application is shown. Figure 2 As shown, the AI-based container front-end handling and detection system 100 may include at least one processor 102 (the processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, a transmission device 106 for communication functions, and a controller 108. Those skilled in the art will understand that... Figure 2 The structure shown is for illustrative purposes only and does not limit the structure of the AI-based container front-end handling detection system 100. For example, the AI-based container front-end handling detection system 100 may also include components that are more advanced than those shown above. Figure 2 The more or fewer components shown, or having the same Figure 2 The different configurations shown.

[0199] The memory 104 can be used to store software programs and modules for application software, such as the program instructions corresponding to the method embodiments described above in this application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the above-described artificial intelligence-based container front-end lifting and unloading detection method. The transmission device 106 is used to acquire or send data via a network.

[0200] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

Claims

1. An artificial intelligence-based container front loader loading and unloading detection method, characterized by, The method includes: Acquire a real-time data set of container front-end crane loading and unloading operations, the real-time data set including front-end crane operation action data, container status data, and operation environment data; The real-time collected data set is processed by operation status association mapping to establish a first association relationship between the front crane operation action data and the container status data, a second association relationship between the front crane operation action data and the operation environment data, and a third association relationship between the container status data and the operation environment data, so as to obtain the operation status association mapping result. The pre-trained artificial intelligence loading and unloading detection model is invoked to determine the operation status association mapping results, and an operation compliance determination result containing the abnormal association type and the corresponding determination basis is generated. Based on the abnormal relationship types in the operation compliance judgment results, and combined with the segment division rules of the front lifting and unloading operation, the operation segment corresponding to each abnormal relationship is determined, and the abnormal segment matching result containing the correspondence between abnormal relationships and operation segments is obtained. Based on the judgment criteria in the operation compliance judgment result and the operation links in the abnormal link matching result, an operation adjustment instruction for the corresponding operation link is generated, and the operation adjustment instruction is sent to the front crane control terminal to adjust the loading and unloading operation of the corresponding operation link.

2. The AI-based container reach stacker detection method of claim 1, wherein, The process of performing operation status association mapping on the real-time collected data set establishes a first association relationship between the front-end crane operation action data and the container status data, a second association relationship between the front-end crane operation action data and the operation environment data, and a third association relationship between the container status data and the operation environment data, resulting in the operation status association mapping result, including: The real-time collected data set is separated into three independent types of collected data: reach stacker operation action data, container status data, and operation environment data. The data collected from the front-end crane operation is marked with action events, and the occurrence time and action attributes of each action event are recorded to obtain marked front-end crane operation action data; The container status data is marked with status changes, and the time of occurrence and status attributes of each status change are recorded to obtain marked container status data. The collected data of the working environment is labeled with environmental parameters, and the collection time and parameter attributes of each environmental parameter are recorded to obtain labeled working environment data; Based on the occurrence time of action events in the tagged reach stacker operation action data and the occurrence time of state changes in the tagged container status data, action events and state changes that occur at the same time or with a time difference within a preset range are associated to establish the first association between reach stacker operation action data and container status data. Based on the occurrence time of action events in the labeled reach stacker operation action data and the acquisition time of environmental parameters in the labeled operation environment data, action events with the same time or time difference within a preset range are associated with environmental parameters to establish a second association relationship between reach stacker operation action acquisition data and operation environment acquisition data. Based on the time of occurrence of state changes in the labeled container status data and the time of collection of environmental parameters in the labeled operating environment data, the state changes at the same time or with a time difference within a preset range are associated with the environmental parameters to establish a third association between container status data and operating environment data. By integrating the first, second, and third association relationships and supplementing the association time information corresponding to each association relationship, the job status association mapping result is obtained. 3.The AI-based container reach stacker detection method of claim 2, wherein, The process of tagging action events in the collected data of the reach stacker operation, recording the occurrence time and action attributes of each action event, yields tagged reach stacker operation action data, including: The front-end crane operation action data is traversed, and feature points in the data that represent the start and end of the action are identified. These feature points include action parameter change points and action state switching points. Based on the feature points, the front-end crane operation motion acquisition data is divided into multiple continuous motion segments, and each motion segment corresponds to a complete motion event. Extract the start and end times of each action segment, and calculate the midpoint between the start and end times as the occurrence time of the action event; Analyze the changes in motion parameters in each motion segment to determine the motion type of the motion event. The motion type includes lifting motion of the spreader, rotation motion of the spreader, and translational motion of the spreader. Extract motion parameters for each motion event, including changes in motion amplitude, changes in motion speed, and duration of motion within the motion segment; The occurrence time, action type, and action parameters of each action event are associated and stored. A unique identifier is added to each action event to obtain labeled front-end lifting operation action data. 4.The AI-based container reach stacker detection method of claim 2, wherein The process of marking state changes in the collected container status data, recording the time of occurrence and status attributes of each state change, yields marked container status data, including: The container status acquisition data is analyzed, and the status monitoring parameters in the container status acquisition data are extracted. The status monitoring parameters include container tilt monitoring parameters, container lock connection monitoring parameters, and container center of gravity monitoring parameters. Set a state change threshold for each state monitoring parameter. When the change of any state monitoring parameter exceeds the corresponding state change threshold, it is determined that a state change has occurred. Record the initial value of the state monitoring parameter, the value of the state monitoring parameter after the change, and the duration of the state change process at the time of each state change, and determine the directional attribute of the state change, which includes the direction of parameter increase and the direction of parameter decrease. Extract the start and end times of each state change, and calculate the midpoint between the start and end times as the time when the state change occurs. The state changes are classified according to the type of state monitoring parameters, resulting in tilt state changes, lock connection state changes, and center of gravity state changes. The occurrence time of each state change, the type of state monitoring parameter, the direction attribute, the initial value of the state monitoring parameter, the value of the state monitoring parameter after the change, and the duration of the state change process are associated and stored. A unique tag is added to each state change to obtain tagged container state data. 5.The AI-based container reach stacker detection method of claim 1, wherein The process of calling a pre-trained AI loading and unloading detection model to determine the operation status association mapping results generates an operation compliance determination result that includes the type of abnormal association relationship and the corresponding determination criteria, including: Extract the first standard association rule for the first relationship, the second standard association rule for the second relationship, and the third standard association rule for the third relationship from the preset loading and unloading specifications; The first association relationship in the operation status association mapping result is input into the first judgment module of the artificial intelligence loading and unloading detection model. The first judgment module loads the first standard association rule and compares the degree of fit between the first association relationship and the first standard association rule. When the degree of fit between the first association relationship and the first standard association rule is lower than the first preset fit threshold, the first association relationship is determined to be an abnormal first association relationship, and the fit degree value and the non-fit first standard association rule entries are recorded. The second association relationship in the operation status association mapping result is input into the second judgment module of the artificial intelligence loading and unloading detection model. The second judgment module loads the second standard association rule and compares the degree of fit between the second association relationship and the second standard association rule. When the degree of fit between the second association relationship and the second standard association rule is lower than the second preset fit threshold, the second association relationship is determined to be an abnormal second association relationship, and the fit degree value and the non-fit second standard association rule entries are recorded. The third association relationship in the operation status association mapping result is input into the third judgment module of the artificial intelligence loading and unloading detection model. The third judgment module loads the third standard association rule and compares the degree of fit between the third association relationship and the third standard association rule. When the degree of fit between the third association relationship and the third standard association rule is lower than the third preset fit threshold, the third association relationship is determined to be an abnormal third association relationship, and the fit degree value and the non-fit third standard association rule entries are recorded. Collect all first, second, and third abnormal relationships, classify and label the abnormal relationship types, compile the matching degree values ​​and non-compliance standard relationship rule entries corresponding to each abnormal relationship as the basis for judgment, and generate operation compliance judgment results that include abnormal relationship types and corresponding judgment criteria. 6.The AI-based container reach stacker detection method of claim 5, wherein, The first association relationship in the operation status association mapping result is input into the first judgment module of the artificial intelligence loading and unloading detection model. The first judgment module loads the first standard association rule and compares the degree of fit between the first association relationship and the first standard association rule, including: The first association relationship is parsed, and action event information and state change information in the first association relationship are extracted. The action event information includes action type, action parameters and associated time. The state change information includes state monitoring parameter type, direction attribute and associated time. Parse the first standard association rules and extract the correspondence between allowed action types and state monitoring parameter types, the matching relationship between allowed action parameter ranges and state change direction attributes, and the allowed association time difference range in the first standard association rules; Compare the correspondence between the action type and the status monitoring parameter type in the first association relationship to see if it is consistent with the correspondence between the action type and the status monitoring parameter type allowed in the first standard association rule, and count the first proportion of the number of consistent items to the total number of correspondence relationships. Compare whether the action parameters in the first association relationship are within the range of action parameters allowed by the first standard association rule, and whether the state change direction attribute is consistent with the matching relationship between the range of action parameters allowed by the first standard association rule and the state change direction attribute. Count the second proportion of the number of matching items to the total number of parameters. Calculate the difference between the action event association time and the state change association time in the first association relationship, determine whether the difference is within the range of association time difference allowed by the first standard association rule, and count the third proportion of the number of matching items to the total number of association time pairs; Weight coefficients are set for the first ratio, the second ratio, and the third ratio. The comprehensive fit between the first association relationship and the first standard association rule is calculated by weighted summation. The comprehensive fit is the degree of fit between the first association relationship and the first standard association rule. 7.The AI-based container reach stacker detection method of claim 1, wherein The step involves determining the operational steps corresponding to each abnormal relationship based on the abnormal relationship type in the operational compliance determination result, combined with the step division rules for front-end lifting and unloading operations. This yields an abnormal step matching result that includes the correspondence between abnormal relationships and operational steps, including: The process segmentation rules for front-end lifting and unloading operations are obtained. These rules define the time interval segmentation standards and operation content characteristics for the spreader alignment, container lifting, container translation, and container placement stages. Extract the association time information corresponding to each abnormal association from the operation compliance judgment result. The association time information includes the association time of the first abnormal association, the association time of the second abnormal association, and the association time of the third abnormal association. Based on the time interval division criteria in the process division rules, determine the time interval to which the associated time of each abnormal relationship belongs, and initially match the corresponding work process; Extract action event information or state change information from each abnormal association. The action event information includes the action type, and the state change information includes the state monitoring parameter type. Based on the task content characteristics in the task segmentation rules, verify whether the initially matched task segments match the action event information or state change information. The task content characteristics define the typical action types and state monitoring parameter types for each task segment. When the initially matched task step matches the action event information or status change information, the task step is confirmed as the task step corresponding to the abnormal relationship; when there is no match, a second matching is performed according to the time interval division standard and the task content characteristics in the task division rule until a unique corresponding task step is determined. Each abnormal relationship is associated with a specific work step and recorded. The time interval and work content characteristics used in the matching process are marked to obtain the abnormal step matching results that include the correspondence between abnormal relationships and work steps. 8.The AI-based container reach stacker detection method of claim 7, wherein, The step of verifying whether the initially matched work steps match action event information or state change information based on the work content characteristics in the step segmentation rules includes: The operational content characteristics of each operational stage are extracted from the stage segmentation rules. The operational content characteristics of the spreader alignment stage include spreader fine-tuning actions and spreader alignment status monitoring parameters; the operational content characteristics of the container lifting stage include spreader raising actions, container locking status monitoring parameters, and container tilting status monitoring parameters; the operational content characteristics of the container translating stage include spreader horizontal movement actions, container center of gravity status monitoring parameters, and operational environment obstacle monitoring parameters; and the operational content characteristics of the container lowering stage include spreader lowering actions, container lowering status monitoring parameters, and container lock unlocking status monitoring parameters. For action event information in abnormal relationships, extract the action type and determine whether the action type belongs to the typical action type in the work content characteristics of the preliminary matching work process. For the status change information in the abnormal correlation, extract the status monitoring parameter type and determine whether the status monitoring parameter type belongs to the typical status monitoring parameter type in the work content characteristics of the preliminary matching work link. When the action type is a typical action type in the work content characteristics of the initially matched work link, and the status monitoring parameter type is a typical status monitoring parameter type in the work content characteristics of the initially matched work link, it is determined that the initially matched work link matches the action event information or status change information. When the action type is not a typical action type in the characteristics of the work content of the initially matched work link, or the status monitoring parameter type is not a typical status monitoring parameter type in the characteristics of the work content of the initially matched work link, it is determined that the initially matched work link does not match the action event information or status change information. Record the judgment results of each verification and the corresponding task content characteristics. 9.The AI-based container reach stacker detection method of claim 1, wherein Based on the judgment criteria in the operation compliance judgment result and the operation links in the abnormal link matching result, an operation adjustment instruction for the corresponding operation link is generated, and the operation adjustment instruction is sent to the front crane control terminal to adjust the loading and unloading operation of the corresponding operation link, including: Extract the standard association rule entries and corresponding parameter deviation information for each type of abnormal association relationship from the judgment criteria of the operation compliance judgment result. The parameter deviation information includes action parameter deviation, state parameter deviation and association time deviation. Based on the abnormal link matching results, determine the work link corresponding to each abnormal relationship, query the preset link adjustment strategy library, and obtain the adjustment direction for different parameter deviations under the work link. The adjustment direction includes the action parameter correction direction, the state parameter improvement direction, and the correlation time optimization direction. Based on the degree of deviation of the parameter deviation information, the adjustment force corresponding to each adjustment direction is determined; the greater the degree of deviation, the greater the adjustment force. By combining the adjustment direction and adjustment intensity, specific operational adjustment content for this work process is generated. The specific operational adjustment content includes the adjustment value of the action parameter, the control target of the state parameter, and the coordination scheme of the associated time. Add work process identifiers, abnormal relationship type identifiers, and adjustment priority identifiers to the specific operation adjustment content to construct a structured operation adjustment instruction. The structure of the operation adjustment instruction includes an instruction header, an adjustment content body, and an instruction tail. The instruction header contains identification information, the adjustment content body contains the specific operation adjustment content, and the instruction tail contains the execution requirements. Establish a secure communication link with the front crane control terminal, and send the operation adjustment command to the front crane control terminal through encrypted transmission; The system receives instruction reception confirmation information from the front-end crane control terminal, the instruction reception confirmation information including instruction identifier and reception status; When the reception status is successful, the operation adjustment command is sent. When the reception status is unsuccessful, the secure communication link with the front crane control terminal is re-established and the operation adjustment command is sent again until the reception status is successful, so as to ensure that the front crane control terminal can adjust the loading and unloading operations of the corresponding work links according to the operation adjustment command.

10. An artificial intelligence-based container reach stacker detection system, characterized by, The device includes a processor and a readable storage medium storing a program that, when executed by the processor, implements the AI-based container front-end lifting and unloading detection method as described in any one of claims 1-9.

Images