Data center operation and maintenance inspection robot motion control method and system

Abstract

The embodiment of the invention provides a data center operation and maintenance inspection robot motion control method and system, and relates to the technical field of robot motion control, and the method comprises the steps: obtaining the operation data of a robot, and the operation data comprises the real-time torque and real-time axial acceleration of a drive motor; identifying whether the robot has an abnormal event according to the operation data; when the abnormal event is identified, generating an abnormal event node; obtaining an abnormal alarm outside the robot; generating an abnormal alarm node according to the abnormal alarm, wherein the abnormal alarm node comprises an alarm type, alarm time and an alarm equipment position; matching the abnormal event node with the abnormal alarm node, and determining whether causal association exists between the abnormal event and the abnormal alarm; and when causal association exists between the abnormal event and the abnormal alarm, a motion control strategy of the robot is adjusted. The operation and maintenance efficiency and safety of the data center can be guaranteed.

Description

Technical Field

[0001] This application relates to the field of robot motion control technology, and more specifically, to a motion control method and system for a data center operation and maintenance inspection robot. Background Technology

[0002] Modern data center inspection robots require high-precision motion control and environmental perception. However, under high load and long-term operation, equipment vibration and temperature changes can have a long-term, hidden impact on core sensors. Accumulated deviations and temporary interference can easily lead to a decrease in positioning and operational accuracy. Specifically, stress fatigue of the lidar mirror surface can cause periodic deviations in the laser beam, resulting in local geometric distortion of the point cloud map and distorting the robot's environmental perception. If encountering temporary obstacles with low reflectivity, the robot may fail to recognize them and cause minor collisions. Although such collisions do not cause external damage, they can cause a tiny zero-point offset of the gyroscope's single axis exceeding the self-test threshold. The system may not alarm, but it will continuously generate angular velocity measurement errors. This deviation accumulates over long distances, causing a heading angle shift that is difficult to detect at short distances. Ultimately, when performing high-precision tasks such as close-up shots of cabinet indicator lights and precise robotic arm operations, angular deviations can lead to inaccurate positioning, failure to align with targets, or even accidental contact with non-target areas, affecting operational efficiency and safety, and in severe cases, causing equipment damage. Existing technologies urgently need optimization. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a motion control method and system for a data center operation and maintenance inspection robot, aiming to solve the problem that in the prior art, the motion control accuracy of data center operation and maintenance inspection robots decreases due to environmental influences and the accumulation of small deviations in their own sensors, resulting in the inability to accurately perform high-precision operations, and may even lead to equipment damage.

[0004] In a first aspect, embodiments of this application provide a motion control method for a data center operation and maintenance inspection robot, including: Acquire the robot's operational data, which includes the real-time torque and real-time axis acceleration of the drive motor; Based on the operational data, identify whether the robot has experienced an abnormal event, wherein the abnormal event is manifested as the operational data exceeding the dynamic threshold of the operational data. When the abnormal event is detected, an abnormal event node is generated. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. Receive abnormal alarms from outside the robot; An abnormal alarm node is generated based on the abnormal alarm, and the abnormal alarm node includes the alarm type, alarm time, and alarm device location; The abnormal event nodes are matched with the abnormal alarm nodes to determine whether there is a causal relationship between the abnormal events and the abnormal alarms; When there is a causal relationship between the abnormal event and the abnormal alarm, the robot's motion control strategy is adjusted.

[0005] According to some embodiments of this application, the step of identifying whether an abnormal event has occurred in the robot based on the operational data includes: Obtain the real-time torque and real-time shaft acceleration of the drive motor; Within the sliding window, the average torque and average axial acceleration are calculated based on the real-time torque and the real-time axial acceleration. A dynamic torque threshold is set based on the average torque, and a dynamic shaft acceleration threshold is set based on the average shaft acceleration. When the real-time torque exceeds the dynamic torque threshold or when the real-time axis acceleration exceeds the dynamic axis acceleration threshold, an abnormal event is identified in the robot.

[0006] According to some embodiments of this application, the step of matching the abnormal event node with the abnormal alarm node to determine whether there is a causal relationship between the abnormal event and the abnormal alarm includes: According to the preset causal chain database, the abnormal event nodes and the abnormal alarm nodes are matched to determine the potential causal chain, wherein the potential causal chain includes the type of preceding event, the type of subsequent event, the time window, the spatial range, and the causal strength between nodes. The initial confidence level of the potential causal chain is calculated based on the causal strength between all nodes from the abnormal event node to the abnormal alarm node in the potential causal chain. Based on the robot's current task type and environmental load, the initial confidence level is adjusted to obtain the comprehensive confidence level; When the overall confidence level reaches a preset threshold, it is determined that there is a causal relationship between the abnormal event and the abnormal alarm; When the overall confidence level does not reach the preset threshold, the causal relationship between the abnormal event and the abnormal alarm is determined to be undetermined.

[0007] According to some embodiments of this application, the step of matching the abnormal event nodes and the abnormal alarm nodes according to a preset causal chain database to determine potential causal chains includes: Obtain the time of occurrence of the abnormal event node, the robot's current position at the time of the abnormality, and the type of abnormality; Obtain the alarm type, alarm time, and alarm device location of the abnormal alarm node; Extract causal chains from the preset causal chain database that simultaneously contain the exception type and the alarm type to obtain a preliminary causal chain; When the time of the anomaly occurrence and the time of the alarm match the time window in the preliminary causal chain, and the current position of the robot and the position of the alarm device at the time of the anomaly occurrence match the spatial range in the preliminary causal chain, the preliminary causal chain is determined to be a potential causal chain.

[0008] According to some embodiments of this application, the step of calculating the initial confidence level of the potential causal chain based on the causal strength among all nodes from the anomalous event node to the anomalous alarm node in the potential causal chain includes: Obtain the causal strength between all nodes in the potential causal chain, from the abnormal event node to the abnormal alarm node; The initial confidence level of the potential causal chain is calculated based on the causal strength among all nodes. The initial confidence level is the geometric mean of the causal strength among all nodes.

[0009] According to some embodiments of this application, the step of calculating the initial confidence level of the potential causal chain based on the causal strength among all nodes from the anomalous event node to the anomalous alarm node in the potential causal chain includes: Obtain information on the robot's current operating status and ambient temperature. Based on the operating status information and the ambient temperature information, the causal strength between the nodes in the potential causal chain is adjusted to obtain the adjusted causal strength value between the nodes. The initial confidence level of the potential causal chain is calculated based on the adjusted causal strength values ​​between nodes.

[0010] According to some embodiments of this application, the step of determining the causal relationship between the abnormal event and the abnormal alarm is pending when the overall confidence level does not reach a preset threshold includes: When the overall confidence level does not reach the preset threshold, new abnormal nodes and new alarm nodes are monitored in real time. When the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain, the overall confidence is increased to obtain the reset confidence. When the reset confidence level reaches a preset threshold, a causal relationship between the abnormal event and the abnormal alarm is determined.

[0011] According to some embodiments of this application, the step of increasing the overall confidence level to obtain a reset confidence level when the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain includes: When the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain, determine the weighting coefficient of the impact of the new abnormal node or the new alarm node on the overall confidence level. The reset confidence level is obtained by increasing the overall confidence level based on the influence weighting coefficient.

[0012] According to some embodiments of this application, the step of correcting the initial confidence level based on the robot's current task type and environmental load to obtain a comprehensive confidence level includes: Get the robot's current task type; The environmental load of the area where the robot is located is obtained, including the average utilization of the server CPU, network traffic, and cooling system operating power. Based on the task type and the environmental load, determine the influence coefficient of the abnormal event node or the abnormal alarm node on the confidence of the causal chain; Based on the influence coefficient, the initial confidence level is adjusted to obtain the comprehensive confidence level.

[0013] Secondly, this application also discloses a motion control system for a data center operation and maintenance inspection robot, including: The data acquisition module is used to acquire the robot's operating data, which includes the real-time torque and real-time axis acceleration of the drive motor. An anomaly detection module is used to identify whether an abnormal event has occurred in the robot based on the operating data. The abnormal event is manifested when the operating data exceeds the dynamic threshold of the operating data. An abnormal event node generation module is used to generate an abnormal event node when the abnormal event is detected. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. The abnormal alarm acquisition module is used to acquire abnormal alarms from outside the robot; An abnormal alarm node generation module is used to generate abnormal alarm nodes based on the abnormal alarms. The abnormal alarm node includes alarm type, alarm time, and alarm device location. The causal relationship determination module is used to match the abnormal event node with the abnormal alarm node to determine whether there is a causal relationship between the abnormal event and the abnormal alarm. The strategy adjustment module is used to adjust the robot's motion control strategy when there is a causal relationship between the abnormal event and the abnormal alarm.

[0014] The technical solution according to the embodiments of this application has at least the following beneficial effects: This application discloses a motion control method for a data center operation and maintenance inspection robot. By acquiring the robot's operating data and identifying abnormal events based on this data, and combining it with external abnormal alarms, a causal correlation analysis is performed between the two. When a causal correlation is determined, the system can adjust the robot's motion control strategy in a timely manner. This method effectively solves the problem in the prior art where data center operation and maintenance inspection robots suffer from decreased motion control accuracy due to long-term high-load operation, environmental interference, and the accumulation of small sensor deviations, resulting in the inability to accurately perform high-precision operations and even potential equipment damage. By real-time monitoring of the drive motor's real-time torque and real-time shaft acceleration, and dynamically identifying abnormal events, this application can promptly detect potential internal faults or external interference in the robot. Furthermore, by matching and performing causal correlation analysis between the robot's own abnormal events and external alarms, the root cause of the abnormality can be deeply explored, such as identifying the zero-point offset of the inertial measurement unit gyroscope caused by a slight collision, which leads to the cumulative deviation of the heading angle. This method avoids the lag of traditional passive fault diagnosis and achieves proactive prevention and precise intervention. Ultimately, by specifically adjusting the motion control strategy, this application can effectively improve the robot's operational stability and task execution accuracy in complex data center environments, ensuring that the robot can accurately complete high-precision inspection and operation tasks, thereby guaranteeing the efficiency and safety of data center operation and maintenance, and avoiding the risk of equipment damage caused by positioning errors.

[0015] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0016] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0017] Figure 1 A flowchart illustrating a motion control method for a data center operation and maintenance inspection robot provided in one embodiment of this application; Figure 2 This is a schematic diagram of a motion control system for a data center operation and maintenance inspection robot provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical methods, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] It should be noted that the meaning of "multiple" (or "more than") in the description of the embodiments of this application refers to two or more, and "greater than," "less than," "exceeding," etc. are understood to exclude the number itself, while "above," "below," "within," etc. are understood to include the number itself. If "first," "second," etc. are used in the description, they are only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated.

[0020] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, or the existence of B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: the existence of a alone, the existence of b alone, the existence of c alone, the simultaneous existence of a and b, the simultaneous existence of a and c, the simultaneous existence of b and c, or the simultaneous existence of a, b, and c, where a, b, and c can be single or multiple.

[0021] In the description of this application, unless otherwise expressly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0022] Based on the above, this application proposes a motion control method and system for a data center operation and maintenance inspection robot, aiming to ensure the efficiency and security of data center operation and maintenance.

[0023] The motion control method for data center operation and maintenance inspection robots provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms; the software can be an application that implements the motion control method for data center operation and maintenance inspection robots, but is not limited to the above forms.

[0024] This application can be applied to numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via communication networks. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices. It should be noted that in various specific embodiments of this invention, when processing is required based on data related to the characteristics of an object (e.g., user attributes or sets of attribute information), permission or consent from the corresponding object is obtained first, and the collection, use, and processing of this data comply with relevant laws and standards. Furthermore, when the embodiments of the present invention need to obtain the attribute information of an object, they will obtain the separate permission or separate consent of the corresponding object through pop-up windows or redirection to a confirmation page. After obtaining the separate permission or separate consent of the corresponding object, they will then obtain the relevant data of the object necessary for the embodiments of the present invention to operate normally.

[0025] See Figure 1 , Figure 1 This is a flowchart illustrating a motion control method for a data center operation and maintenance inspection robot according to an embodiment of this application. The motion control method for a data center operation and maintenance inspection robot provided in this embodiment includes, but is not limited to, steps S110 to S170, which are described in detail below.

[0026] Step S110: Obtain the robot's operating data, which includes the real-time torque and real-time axis acceleration of the drive motor; Step S120: Based on the running data, identify whether the robot has experienced any abnormal events. Abnormal events are manifested as running data exceeding the dynamic threshold of running data. Step S130: When an abnormal event is detected, an abnormal event node is generated. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. Step S140: Obtain abnormal alarms from outside the robot; Step S150: Generate abnormal alarm nodes based on abnormal alarms. The abnormal alarm nodes include alarm type, alarm time, and alarm device location. Step S160: Match the abnormal event nodes with the abnormal alarm nodes to determine whether there is a causal relationship between the abnormal events and the abnormal alarms; Step S170: When there is a causal relationship between abnormal events and abnormal alarms, adjust the robot's motion control strategy.

[0027] It's important to note that "operational data" refers to various real-time performance indicators generated during robot operation, with the core being the "real-time torque" and "real-time axis acceleration" of the drive motor. Real-time torque reflects the magnitude of the motor's output torque and is closely related to the robot's resistance, load, and motion state; real-time axis acceleration represents the rate of change of the robot's velocity along its motion axis, directly reflecting the robot's dynamic response and force conditions. These data are crucial for evaluating the robot's motion state and potential anomalies. "Abnormal events" refer to situations where the robot's operating state deviates from the normal range, manifested as operational data (such as real-time torque and real-time axis acceleration) exceeding preset "dynamic thresholds for operational data." Such abnormal events may indicate wear, loosening, or jamming of internal robot components, or sudden changes in the external environment. "Abnormal event nodes" are data units that provide a structured description of identified abnormal events, including "time of occurrence," "the robot's current position at the time of occurrence," and "type of abnormality." "Abnormal alarms" are signals issued by external systems (such as data center monitoring systems, equipment sensor networks, etc.) indicating abnormalities in data center equipment or the environment. An "abnormal alarm node" is a data unit that provides a structured description of the acquired abnormal alarms, including "alarm type," "alarm time," and "alarm device location." "Causal correlation" refers to a logical cause-and-effect relationship between abnormal events and abnormal alarms; that is, the occurrence of one event is the cause or result of another. "Motion control strategy" refers to the set of algorithms and parameters used by the robot to plan and execute its movements.

[0028] Specifically, this method first requires "acquiring the robot's operational data," which forms the basis for evaluating the robot's own state. Operational data can include the real-time torque and real-time axis acceleration of the drive motors. These sensors continuously measure the torque output of the motors and the acceleration values ​​on the robot's motion axes, converting these analog signals into digital signals and transmitting them to the control unit for processing via the robot's internal bus or wireless communication module. Another implementation involves estimating the real-time torque using current and voltage sensors within the motor controller, combined with a motor model, and acquiring the real-time axis acceleration via the robot's inertial measurement unit (IMU). After acquiring the operational data, the next step is "identifying whether the robot has experienced an abnormal event based on the operational data." Abnormal event identification is based on whether the operational data exceeds preset dynamic thresholds. For example, a fixed upper limit for torque and acceleration can be set. When the real-time torque or real-time axis acceleration exceeds these fixed thresholds, it is identified as an abnormal event. The advantage of this approach is its simplicity and low computational cost. When an abnormal event is identified, the system "generates an abnormal event node." An abnormal event node is a structured description of the abnormal event, including the time of the abnormality, the robot's current position at the time of the abnormality, and the type of abnormality. For example, when a sudden increase in real-time torque is detected and exceeds a threshold, the system records the current timestamp, the precise coordinates obtained by the robot through a positioning system (such as LiDAR SLAM), and marks the anomaly as "high torque anomaly." This information can be encapsulated into a data packet, stored in the robot's local memory, and optionally uploaded to a cloud server for further analysis. Simultaneously, this method also needs to "acquire external anomaly alarms." These alarms typically originate from other monitoring systems in the data center. These alarms can be received from the central monitoring platform of the data center via standardized communication protocols (such as MQTT and RESTful APIs). Another method is that the robot itself is equipped with environmental sensors (such as temperature sensors and smoke sensors), which can also serve as sources of external alarms when they detect anomalies. After acquiring external anomaly alarms, the system "generates anomaly alarm nodes based on the anomaly alarms." Anomaly alarm nodes are also structured descriptions of alarm information, including alarm type, alarm time, and alarm device location.

[0029] In one embodiment, "matching abnormal event nodes with abnormal alarm nodes to determine whether a causal relationship exists between the abnormal event and the abnormal alarm." This step aims to identify potential connections between robot anomalies and external environmental anomalies. If an abnormal event node and an abnormal alarm node are close in time (e.g., occurring within 5 minutes) and spatially adjacent (e.g., the robot's abnormal location and the alarm device's location are within 5 meters), they are considered to be potentially causally related. While this matching method is simple, it can quickly filter out a large number of irrelevant events and alarms. When a causal relationship is determined between an abnormal event and an abnormal alarm, the system will "adjust the robot's motion control strategy." For example, if a causal relationship is identified between an abnormal increase in the robot's drive motor torque and an excessively high temperature of a nearby server (e.g., the server overheating causes abnormal cooling fan speed, resulting in vibration that affects the robot), the robot's motion control strategy can be adjusted, such as reducing the robot's inspection speed, avoiding the area, or enabling a more refined obstacle avoidance algorithm when passing through the area to reduce the impact of vibration on the robot's sensors. Another adjustment method is to adjust the robot's drive torque distribution to increase friction or plan a detour if the causal relationship indicates that the robot's malfunction is due to slippery ground.

[0030] In this regard, this application further proposes that the steps for identifying whether an abnormal event has occurred in the robot based on operational data include: Obtain the real-time torque and real-time shaft acceleration of the drive motor; Within the sliding window, the average torque and average shaft acceleration are calculated based on the real-time torque and real-time shaft acceleration. Set a dynamic torque threshold based on the average torque and a dynamic shaft acceleration threshold based on the average shaft acceleration; When the real-time torque exceeds the dynamic torque threshold or when the real-time axis acceleration exceeds the dynamic axis acceleration threshold, an abnormal event is identified in the robot.

[0031] Specifically, the real-time torque and real-time axis acceleration of the drive motor refer to the actual operating parameters of the robot's drive system at the current moment. These data can be acquired in real time through sensors inside the robot. A sliding window can be understood as a data processing mechanism that slides across time-series data in a fixed-size window. Each slide includes the latest data point and removes the oldest, ensuring that calculations are always based on data from the most recent period. The sliding window can be set to include the most recent 5, 10, or more seconds of operating data, and its size can be adjusted according to the actual application scenario and the robot's operating characteristics. Average torque and average axis acceleration are the values ​​obtained by arithmetically averaging all real-time torque and axis acceleration data within the current sliding window. Their purpose is to smooth data fluctuations and reflect the trend of the robot's recent operating state. Dynamic torque thresholds and dynamic axis acceleration thresholds are dynamically set upper limits based on the calculated average torque and average axis acceleration. Their purpose is to allow the anomaly detection thresholds to adaptively follow changes in the robot's normal operating state, avoiding misjudgments caused by fixed thresholds.

[0032] This application's solution introduces a sliding window mechanism, enabling anomaly identification to move beyond simply comparing instantaneous data points with static thresholds. Instead, it dynamically adjusts thresholds based on the average operating state over a period of time. By calculating average torque and average axis acceleration within the sliding window, instantaneous noise and short-term fluctuations in the data are effectively filtered out, allowing the set dynamic torque and dynamic axis acceleration thresholds to more accurately reflect the robot's current baseline operating state. Only when real-time torque or real-time axis acceleration exceeds these adaptive dynamic thresholds is it identified as an anomaly, thus avoiding false alarms caused by normal fluctuations and improving sensitivity to genuine anomalies. This technical solution effectively addresses the false alarm or missed alarm problems caused by insufficiently precise dynamic threshold settings or failure to fully consider short-term fluctuations in the robot's operating state in traditional methods. By calculating average torque and average axis acceleration within a sliding window, this application enables the set dynamic thresholds to more accurately reflect the robot's current operating state, significantly improving the accuracy and robustness of anomaly identification. This adaptive threshold setting method enables the system to better adapt to changes in the robot's operating environment and task load, reducing the false alarm rate while ensuring the timely detection of real anomalies, providing a more reliable basis for subsequent anomaly handling and motion control strategy adjustments.

[0033] In some preferred embodiments, it is assumed that a data center maintenance and inspection robot is performing an inspection task. The system continuously acquires real-time torque and real-time shaft acceleration data of the drive motor. To identify anomalies, the system sets up a sliding window, for example, containing data from the most recent 10 seconds. Whenever new real-time data enters, the oldest data is removed from the window. Within each sliding window, the system calculates the average real-time torque and the average real-time shaft acceleration over those 10 seconds. For example, if the average torque over the most recent 10 seconds is 5 Nm and the average shaft acceleration is 2 m / s², the system can dynamically set thresholds based on these averages. For example, the dynamic torque threshold can be set to 1.2 times the average torque (i.e., 6 Nm), and the dynamic shaft acceleration threshold can be set to 1.5 times the average shaft acceleration (i.e., 3 m / s²). When, at a certain moment, the real-time torque suddenly reaches 6.5 Nm, exceeding the dynamic torque threshold of 6 Nm, or the real-time shaft acceleration reaches 3.2 m / s², exceeding the dynamic shaft acceleration threshold of 3 m / s², the system will identify an abnormal event in the robot. This approach allows the threshold to be adaptively adjusted as the robot's operating state changes smoothly, avoiding false alarms caused by instantaneous data fluctuations when the robot is accelerating or decelerating normally, while also being able to promptly capture real anomalies that exceed the normal fluctuation range.

[0034] In response, this application further proposes the following steps for matching abnormal event nodes with abnormal alarm nodes to determine whether there is a causal relationship between abnormal events and abnormal alarms: Based on the pre-set causal chain database, abnormal event nodes and abnormal alarm nodes are matched to determine potential causal chains, which include the type of preceding events, the type of subsequent events, the time window, the spatial range, and the causal strength between nodes. Calculate the initial confidence level of the potential causal chain based on the causal strength between all nodes from the abnormal event node to the abnormal alarm node in the potential causal chain. Based on the robot's current task type and environmental load, the initial confidence level is adjusted to obtain the overall confidence level; When the overall confidence level reaches a preset threshold, it is determined that there is a causal relationship between the abnormal event and the abnormal alarm. When the overall confidence level does not reach the preset threshold, the causal relationship between the abnormal event and the abnormal alarm is to be determined.

[0035] Specifically, when determining whether a causal relationship exists between anomaly events and anomaly alarms, the process begins by matching anomaly event nodes and anomaly alarm nodes against a pre-defined causal chain database to identify potential causal chains. This database is a knowledge base storing various known or inferred causal relationships between events. Each causal chain within the database defines in detail the type of preceding events, the type of succeeding events, the time window of the event, the spatial range of the event, and the causal strength between event nodes. By comparing the anomaly type, occurrence time, and location of anomaly event nodes with the alarm type, alarm time, and alarm device location of anomaly alarm nodes against the causal chains in the database, potential causal chains matching the current anomaly event and anomaly alarm can be filtered out. Further, after identifying potential causal chains, the initial confidence level of the potential causal chain is calculated based on the causal strength between all nodes from the anomaly event node to the anomaly alarm node within that potential causal chain. The causal strength between nodes represents the strength of the causal relationship between adjacent events in the causal chain, and its value can be preset based on historical data, expert experience, or machine learning models. The initial confidence level can be understood as the inherent credibility of the causal chain without considering real-time environmental factors. For example, it can be calculated using the geometric mean of the causal strengths among all nodes or other aggregation methods. Based on this, to improve the accuracy and adaptability of causal association judgment, the initial confidence level needs to be corrected according to the robot's current task type and environmental load to obtain a comprehensive confidence level. The robot's current task type refers to the specific task the robot is performing, such as inspection, handling, or charging. Different task types may have different impacts on the robot's operating state and the surrounding environment. Environmental load refers to the external environmental pressure of the robot's location, such as the average CPU utilization of data center servers, network traffic, and cooling system operating power. These factors affect the probability and correlation strength of abnormal events and alarms. By considering this real-time contextual information, the actual credibility of the causal chain can be assessed more accurately. Finally, when the comprehensive confidence level reaches a preset threshold, a causal relationship between the abnormal event and the alarm is determined. This preset threshold can be set according to actual application requirements and the system's fault tolerance for causal judgment. If the overall confidence level does not reach the preset threshold, the causal relationship between the abnormal event and the abnormal alarm is determined to be undetermined. This means that the system will continue to monitor the relevant events in order to obtain more information for further judgment.

[0036] This application's solution, by introducing a pre-defined causal chain database, can systematically identify the potential complex causal relationships between abnormal events and abnormal alarms, rather than simply performing event type or temporal-spatial matching. By calculating the initial confidence level of the potential causal chain, the inherent reliability of the causal chain itself can be quantified. More importantly, by combining the robot's current task type and environmental load to correct the initial confidence level, the determination of causal relationships can fully consider the real-time dynamic operating environment and working status, thereby avoiding misjudgments or omissions caused by judging a single factor in complex and ever-changing data center environments. This multi-dimensional and dynamic evaluation mechanism makes the determination of causal relationships more accurate and reliable.

[0037] Specifically, the steps mentioned above, which involve matching abnormal event nodes and abnormal alarm nodes against a pre-defined causal chain database to determine potential causal chains, include: Obtain the time of occurrence of the abnormal event node, the robot's current position at the time of the abnormality, and the type of abnormality; Obtain the alarm type, alarm time, and alarm device location of the abnormal alarm node; Extract causal chains from the preset causal chain database that contain both exception types and alarm types to obtain preliminary causal chains; When the time of the anomaly occurrence and the time of the alarm match the time window in the preliminary causal chain, and the robot's current position and the alarm device's position at the time of the anomaly occurrence match the spatial range in the preliminary causal chain, the preliminary causal chain is determined to be a potential causal chain.

[0038] Among them, abnormal event nodes are records of abnormal data in the robot's own operation, containing key information such as the time of the abnormality, the robot's current position at the time of the abnormality, and the type of abnormality. Abnormal alarm nodes are alarm information issued by external devices of the robot, containing alarm type, alarm time, and alarm device location. The pre-built causal chain database is a pre-established knowledge base that stores various known or inferred causal relationship patterns between abnormal events and alarm events. Each causal chain defines the type of preceding event, the type of succeeding event, the time window, the spatial range, and the causal strength between nodes. Specifically, the process of extracting preliminary causal chains refers to filtering from the pre-built causal chain database those causal chains whose preceding event type matches the abnormality type of the current abnormal event node, and whose succeeding event type matches the alarm type of the current abnormal alarm node. These filtered causal chains constitute the preliminary causal chain set. Furthermore, in order to accurately determine the preliminary causal chains as potential causal chains, temporal and spatial matching is also required. This means that a preliminary causal chain is only ultimately identified as a potential causal chain if the occurrence times of the abnormal event and the alarm event fall within the time window defined by the preliminary causal chain, and the robot's current position and the alarm device's position at the time of the abnormal event fall within the spatial range defined by the preliminary causal chain. The time window is used to limit the order and duration of the causal relationship, while the spatial range is used to limit the geographical area where the causal relationship occurs.

[0039] This application's solution employs multi-dimensional matching of abnormal event nodes and abnormal alarm nodes, including type matching, time matching, and spatial matching, to accurately identify the most likely potential causal chains from a pre-defined causal chain database. First, matching anomaly types and alarm types quickly narrows the search scope, eliminating irrelevant causal chains. Second, the introduction of a time window and spatial range matching mechanism further ensures that the identified causal chains are highly consistent with the actual abnormal and alarm events in both time and space. This avoids misclassifying events without causal relationships as causal, improving the accuracy of causal chain identification.

[0040] According to the above method, the steps for calculating the initial confidence of a potential causal chain based on the causal strength among all nodes from the anomalous event node to the anomalous alarm node in the potential causal chain include: Obtain the causal strength between all nodes in a potential causal chain, from the abnormal event node to the abnormal alarm node; The initial confidence level of the potential causal chain is calculated based on the causal strength among all nodes. The initial confidence level is the geometric mean of the causal strength among all nodes.

[0041] Specifically, after identifying the potential causal chain, it is necessary to extract the causal strength between all nodes from the anomalous event node to the anomalous alarm node. These inter-node causal strengths are defined in a pre-defined causal chain database and are used to quantify the correlation strength between adjacent events or nodes in the causal chain. For example, if the potential causal chain is A -> B -> C, where A is the anomalous event node and C is the anomalous alarm node, then the causal strength from A to B and the causal strength from B to C need to be obtained. The initial confidence level is calculated as the geometric mean of the causal strengths between all nodes. The geometric mean is a statistical method, particularly suitable for handling multiplicative relationships or ratio data. In the causal chain scenario, the causal strength between each node can be considered as the probability or influencing factor of chain transmission. Using the geometric mean better reflects the overall strength of the entire causal chain, avoiding the potential bias of the arithmetic mean due to an extreme value, especially when there may be significant differences in causal strength values. The geometric mean ensures that even if there are weak links in the chain, their impact can be reasonably represented, thus allowing the initial confidence level to more accurately represent the reliability of the entire causal chain. The proposed solution obtains the causal strength between all nodes in a potential causal chain, from the anomalous event node to the anomalous alarm node, and calculates the initial confidence level using the geometric mean. This effectively assesses the cumulative strength of the entire causal chain. This calculation method considers the mutual influence of each link in the causal chain, ensuring that the final initial confidence level comprehensively reflects the reliability of the entire transmission path from the anomalous event to the anomalous alarm. The geometric mean is advantageous when handling multiple multiplicative factors, preventing excessive dilution of a single weak link by a strong link, thus providing a more robust and realistic measure of causal association strength.

[0042] The steps described above for calculating the initial confidence of a potential causal chain based on the causal strength between all nodes from the anomalous event node to the anomalous alarm node include: Obtain information on the robot's current operating status and ambient temperature. Based on the operating status information and ambient temperature information, the causal strength between nodes in the potential causal chain is adjusted to obtain the adjusted causal strength value between nodes. Calculate the initial confidence level of the potential causal chain based on the adjusted causal strength values ​​between nodes.

[0043] Specifically, the robot's current operating status information may include, but is not limited to, the robot's real-time speed, load, battery level, motor power consumption, and sensor operating status. This information reflects the robot's current operating load and health status. Ambient temperature information refers to the real-time ambient temperature of the area where the robot is located, such as temperature data obtained through the robot's onboard temperature sensors or data center environmental monitoring systems. Adjusting the causal strength between nodes in the potential causal chain refers to correcting the preset causal strength between nodes based on the obtained robot's current operating status information and ambient temperature information. For example, when the robot is operating under high load, the causal strength between an overload abnormal event in its drive motor and related equipment alarms may increase; when the ambient temperature is too high, the causal strength between internal sensor failures or heat dissipation system malfunctions and related alarms may also increase. This adjustment can be based on preset adjustment rules, machine learning models, or expert experience systems. Subsequently, based on the adjusted causal strength between all nodes in the potential causal chain from the abnormal event node to the abnormal alarm node, the initial confidence level of the potential causal chain is calculated. This calculation method can use geometric mean, weighted average, or other suitable aggregation methods to comprehensively reflect the adjusted causal strength.

[0044] This application's solution dynamically adjusts the causal strength between nodes by incorporating the robot's current operating state information and ambient temperature information before calculating the initial confidence level. This solves the problem of traditional methods where statically preset causal strength fails to fully reflect the real-time dynamic environment. Specifically, the robot's operating state information provides real-time operating conditions of the robot itself; for example, high loads may accelerate wear of mechanical parts, thereby increasing the correlation strength between certain abnormal events and alarms. Ambient temperature information reflects the physical conditions of the robot's external environment; for example, high temperatures may degrade the performance of electronic components, increasing the probability of sensor failure or overheating alarms. By integrating this real-time, dynamic contextual information into the adjustment of causal strength, the causal strength between each node is no longer fixed but can be adaptively corrected according to the actual operating environment. This dynamic adjustment allows the subsequently calculated initial confidence level to more accurately quantify the causal correlation between abnormal events and abnormal alarms, providing a more solid and reliable foundation for subsequent comprehensive confidence level correction and final causal correlation judgment.

[0045] In response, this application further proposes that after determining the causal relationship between the abnormal event and the abnormal alarm is pending when the overall confidence level does not reach the preset threshold, the following steps are included: When the overall confidence level does not reach the preset threshold, new abnormal nodes and new alarm nodes are monitored in real time. When the anomaly type of a new abnormal node or the alarm type of a new alarm node appears in a potential causal chain, the overall confidence is increased to obtain the reset confidence. When the reset confidence level reaches a preset threshold, the causal relationship between the abnormal event and the abnormal alarm is determined.

[0046] Specifically, "real-time monitoring of new anomaly nodes and new alarm nodes" means that after initially determining that the causal relationship is pending, the system will not immediately abandon the evaluation of the potential causal chain, but will continuously and dynamically monitor subsequent new anomaly events and alarms that may occur. These new nodes may be related to the currently pending causal chain, providing new evidence for further judgment. A new anomaly node can be understood as an anomaly event newly detected inside or outside the robot system after the original anomaly event node occurs, and during the period when the causal relationship is pending. A new alarm node can be understood as an anomaly alarm newly generated by external devices of the robot after the original anomaly alarm node occurs, and during the period when the causal relationship is pending. "When the anomaly type of a new anomaly node or the alarm type of a new alarm node appears in the potential causal chain" means that the system will check whether these newly detected anomaly events or alarms match the event types or alarm types in the previously identified potential causal chains. If the new node matches any link in the potential causal chain, it indicates that this new evidence may be related to the pending causal relationship. "Increasing the overall confidence level to obtain a reset confidence level" means that when new evidence related to a potential causal chain is discovered, the system will revise and enhance the original overall confidence level based on this new evidence. This increase can be based on a preset influence weighting coefficient or dynamically adjusted according to the degree of matching between the new node and the potential causal chain. In this way, the overall confidence level is reassessed to reflect the support of new evidence. "When the reset confidence level reaches a preset threshold, the causal relationship between the anomalous event and the anomalous alarm is determined" means that after the overall confidence level is revised by new evidence, if its value reaches a preset threshold, a causal relationship between the original anomalous event and the anomalous alarm can be finally confirmed. This indicates that through continuous monitoring and evidence accumulation, the system can transition from a "pending" state to a "determined" state, thereby more accurately identifying causal relationships.

[0047] Specifically, the steps to increase the overall confidence level and reset the confidence level when the anomaly type of a new abnormal node or the alarm type of a new alarm node appears in a potential causal chain include: When the anomaly type of a new abnormal node or the alarm type of a new alarm node appears in a potential causal chain, determine the weighting coefficient of the impact of the new abnormal node or the new alarm node on the overall confidence level. The reset confidence level is obtained by increasing the overall confidence level based on the influence weight coefficients.

[0048] The influence weighting coefficient is a quantitative indicator used to measure the degree of influence of a new anomalous node or alarm node on the confidence level of a potential causal chain. This coefficient can be determined based on various factors, such as the degree of correlation between the type of the new node and existing nodes in the potential causal chain, the frequency of the new node's occurrence, and the temporal or spatial proximity of the new node to the anomalous event or alarm. A weighting coefficient table can be pre-defined, and the corresponding influence weighting coefficient can be found in the table based on the specific type of the new anomalous node or alarm node. For example, if a new anomalous node is highly correlated with the core event type in the potential causal chain, it can be assigned a higher influence weighting coefficient; conversely, if the correlation is low, a lower weighting coefficient can be assigned. In practical applications, various mathematical models or algorithms can be used to increase the overall confidence level based on the influence weighting coefficient. For example, the overall confidence level and the influence weighting coefficient can be weighted and summed, or adjusted using a multiplication factor. The purpose is to ensure that the appearance of a new node is reflected in the adjustment of the overall confidence level in a quantitative and reasonable way, avoiding blind or empirical increases in confidence level, thereby improving the accuracy and reliability of causal association judgments.

[0049] In some preferred embodiments, a specific example is given below. Suppose that during a data center maintenance inspection, a robot detects a slight abnormal noise from a drive motor, but the overall confidence level does not reach a preset threshold, and the causal relationship is determined to be pending. The system then monitors new abnormal nodes and alarm nodes in real time. For example, shortly after the abnormal noise occurs, the system detects a new abnormal node of "increased motor temperature," whose abnormality type appears in the potential causal chain. At this time, the system determines a high influence weight coefficient based on the closeness of the correlation between the "increased motor temperature" abnormality type and the potential causal chain (e.g., abnormal motor noise -> increased motor temperature -> bearing wear -> drive motor failure). For example, it can be set to 0.2. Subsequently, the current overall confidence level is increased by this influence weight coefficient (e.g., overall confidence level = overall confidence level + 0.2). Furthermore, if the system detects a new alarm node of "intensified bearing vibration," whose alarm type also appears in the potential causal chain. Because the correlation between "increased bearing vibration" and drive motor failure is more direct, the system may assign it a higher influence weighting coefficient, such as 0.3. At this point, the overall confidence level will be increased again by this influence weighting coefficient (e.g., Overall Confidence Level = Overall Confidence Level + 0.3). Through this gradual increase in influence weighting coefficients, the overall confidence level can more accurately reflect the cumulative effect of new evidence, potentially reaching a preset threshold. This establishes a causal relationship between the abnormal noise from the drive motor and the potential drive motor failure, allowing for timely adjustments to the robot's motion control strategy, such as reducing operating speed or scheduling maintenance, to prevent more serious failures. This approach makes the confidence level adjustment more refined and reasonable, improving the accuracy of causal relationship determination.

[0050] In this regard, this application further proposes the following steps for correcting the initial confidence level and obtaining the comprehensive confidence level based on the robot's current task type and environmental load: Get the robot's current task type; The environmental load of the area where the robot is located is obtained. The environmental load includes the average utilization of the server CPU, network traffic, and the operating power of the cooling system. Based on the task type and environmental load, determine the impact coefficient of abnormal event nodes or abnormal alarm nodes on the confidence of the causal chain; Based on the influence coefficient, the initial confidence level is adjusted to obtain the overall confidence level.

[0051] Specifically, acquiring the robot's current task type refers to the system acquiring real-time information about the specific tasks the robot is performing, such as inspection, handling, and charging. These task types may have different impacts on the robot's motion state and the data center environment, thus affecting the strength of the causal relationship between abnormal events and alarms. For example, when performing high-load handling tasks, the torque fluctuation of the robot's drive motor may be greater, and the tolerance for torque anomalies or the strength of their correlation with specific alarms may need to be reassessed. Acquiring the environmental load of the robot's area refers to the system collecting operational status data of the data center infrastructure within the robot's current working area. Environmental load can be understood as the operational pressure or resource consumption within the data center, specifically including server CPU average utilization, network traffic, and cooling system operating power. This environmental load data reflects the overall operational status of the data center. In practical applications, based on the task type and environmental load, the influence coefficient of abnormal event nodes or abnormal alarm nodes on the confidence of the causal chain is determined. The purpose is to quantify the moderating effect of these external factors on the strength of potential causal chains. This influence coefficient can be a dynamically adjusted value, determined through a pre-established rule base, machine learning model, or expert experience system. For example, when a robot performs a high-precision inspection task, even a slight abnormality in motor torque may be assigned a high impact coefficient because it may indicate a decrease in positioning accuracy; while in a low-load standby state, the same torque abnormality may have a lower impact coefficient. Similarly, when the data center environment is under high load, the correlation between certain alarms (such as temperature alarms) and abnormal robot movement may be amplified because high load may cause uneven heat dissipation, which in turn affects the robot's sensors or drive components. Further, based on the impact coefficient, the initial confidence level is corrected to obtain the overall confidence level. The initial confidence level can be understood as the initial confidence level before considering the task type and environmental load. The correction process can be achieved by multiplying, adding, or performing other function operations on the initial confidence level and the impact coefficient to obtain an overall confidence level that better reflects the current actual operating situation. For example, overall confidence level = initial confidence level × (1 + impact coefficient), where the impact coefficient can be positive or negative to strengthen or weaken the initial confidence level. In this way, the overall confidence level can more accurately reflect the true strength of the causal relationship between abnormal events and abnormal alarms, thus providing a more reliable basis for subsequent adjustments to motion control strategies.

[0052] In some preferred embodiments, assume a robot is performing an "equipment inspection" task in a data center, currently located in a server rack area. The system identifies the robot's task type as "equipment inspection." Simultaneously, the system monitors the environmental load information of the robot's location: server CPU average utilization is 85% (high load), network traffic is 900Mbps (high traffic), and cooling system operating power is 70% (medium-high power). At this point, an abnormal event occurs where the robot's drive motor experiences real-time torque exceeding the dynamic torque threshold, generating an abnormal event node. Simultaneously, the system receives an abnormal alarm from the server rack area indicating "rack temperature too high." Based on a pre-set causal chain database, the system identifies a potential causal chain: "abnormal motor torque" may lead to "rack temperature too high" (e.g., motor overheating causing uneven heat dissipation). The initial confidence level of this potential causal chain is calculated to be 0.6. To correct this initial confidence level, the system determines an influence coefficient based on the current task type "equipment inspection" and the environmental load (high CPU utilization, high network traffic, medium-high cooling power). Since this is an "equipment inspection" task, the stability requirements for robot operation are high. Furthermore, the current environmental load is high, which may exacerbate localized heat dissipation issues. Therefore, the system calculates a positive influence coefficient, for example, 0.2, using preset rules or models. Subsequently, the initial confidence level of 0.6 and the influence coefficient of 0.2 are adjusted, for example, using the formula: Overall Confidence Level = Initial Confidence Level × (1 + Influence Coefficient) = 0.6 × (1 + 0.2) = 0.72. Ultimately, the overall confidence level is 0.72. If the preset threshold is 0.7, then this overall confidence level reaches the preset threshold, and the system determines that there is a causal relationship between "abnormal motor torque" and "excessive rack temperature." Therefore, the system will adjust the robot's motion control strategy, such as reducing the robot's operating speed or guiding the robot away from the high-load area, and notify maintenance personnel to conduct an inspection to prevent further malfunctions.

[0053] See Figure 2 , Figure 2 This is a schematic diagram of a motion control system for a data center operation and maintenance inspection robot according to one embodiment of this application. The data center operation and maintenance inspection robot motion control system 200 includes: The data acquisition module 210 is used to acquire the robot's operating data, including the real-time torque and real-time axis acceleration of the drive motor. The anomaly detection module 220 is used to identify whether the robot has experienced an abnormal event based on the running data. An abnormal event is manifested as the running data exceeding the dynamic threshold of the running data. The abnormal event node generation module 230 is used to generate an abnormal event node when an abnormal event is detected. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. The abnormal alarm acquisition module 240 is used to acquire abnormal alarms from outside the robot; An abnormal alarm node generation module 250 is used to generate abnormal alarm nodes based on abnormal alarms. The abnormal alarm node includes the alarm type, alarm time, and alarm device location. The causal relationship determination module 260 is used to match abnormal event nodes with abnormal alarm nodes to determine whether there is a causal relationship between abnormal events and abnormal alarms. The strategy adjustment module 270 is used to adjust the robot's motion control strategy when there is a causal relationship between abnormal events and abnormal alarms.

[0054] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0055] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0056] The foregoing has provided a detailed description of the preferred embodiments of this application. However, this application is not limited to the above-described embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined in this application.

Claims

1. A motion control method for a data center operation and maintenance inspection robot, characterized in that, include: Acquire the robot's operational data, which includes the real-time torque and real-time axis acceleration of the drive motor; Based on the operational data, identify whether the robot has experienced an abnormal event, wherein the abnormal event is manifested as the operational data exceeding the dynamic threshold of the operational data. When the abnormal event is detected, an abnormal event node is generated. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. Receive abnormal alarms from outside the robot; An abnormal alarm node is generated based on the abnormal alarm, and the abnormal alarm node includes the alarm type, alarm time, and alarm device location; The abnormal event nodes are matched with the abnormal alarm nodes to determine whether there is a causal relationship between the abnormal events and the abnormal alarms; When there is a causal relationship between the abnormal event and the abnormal alarm, the robot's motion control strategy is adjusted.

2. The method according to claim 1, characterized in that, The step of identifying whether an abnormal event has occurred in the robot based on the operational data includes: Obtain the real-time torque and real-time shaft acceleration of the drive motor; Within the sliding window, the average torque and average axial acceleration are calculated based on the real-time torque and the real-time axial acceleration. A dynamic torque threshold is set based on the average torque, and a dynamic shaft acceleration threshold is set based on the average shaft acceleration. When the real-time torque exceeds the dynamic torque threshold or when the real-time axis acceleration exceeds the dynamic axis acceleration threshold, an abnormal event is identified in the robot.

3. The method according to claim 1, characterized in that, The step of matching the abnormal event node with the abnormal alarm node to determine whether there is a causal relationship between the abnormal event and the abnormal alarm includes: According to the preset causal chain database, the abnormal event nodes and the abnormal alarm nodes are matched to determine the potential causal chain, wherein the potential causal chain includes the type of preceding event, the type of subsequent event, the time window, the spatial range, and the causal strength between nodes. The initial confidence level of the potential causal chain is calculated based on the causal strength between all nodes from the abnormal event node to the abnormal alarm node in the potential causal chain. Based on the robot's current task type and environmental load, the initial confidence level is adjusted to obtain the comprehensive confidence level; When the overall confidence level reaches a preset threshold, it is determined that there is a causal relationship between the abnormal event and the abnormal alarm; When the overall confidence level does not reach the preset threshold, the causal relationship between the abnormal event and the abnormal alarm is determined to be undetermined.

4. The method according to claim 3, characterized in that, The step of matching the abnormal event nodes and the abnormal alarm nodes according to the preset causal chain database to determine the potential causal chain includes: Obtain the time of occurrence of the abnormal event node, the robot's current position at the time of the abnormality, and the type of abnormality; Obtain the alarm type, alarm time, and alarm device location of the abnormal alarm node; Extract causal chains from the preset causal chain database that simultaneously contain the exception type and the alarm type to obtain a preliminary causal chain; When the time of the anomaly occurrence and the time of the alarm match the time window in the preliminary causal chain, and the current position of the robot and the position of the alarm device at the time of the anomaly occurrence match the spatial range in the preliminary causal chain, the preliminary causal chain is determined to be a potential causal chain.

5. The method according to claim 3, characterized in that, The step of calculating the initial confidence of the potential causal chain based on the causal strength between all nodes from the anomalous event node to the anomalous alarm node in the potential causal chain includes: Obtain the causal strength between all nodes in the potential causal chain, from the abnormal event node to the abnormal alarm node; The initial confidence level of the potential causal chain is calculated based on the causal strength among all nodes. The initial confidence level is the geometric mean of the causal strength among all nodes.

6. The method according to claim 3, characterized in that, The step of calculating the initial confidence of the potential causal chain based on the causal strength between all nodes from the anomalous event node to the anomalous alarm node in the potential causal chain includes: Obtain information on the robot's current operating status and ambient temperature. Based on the operating status information and the ambient temperature information, the causal strength between the nodes in the potential causal chain is adjusted to obtain the adjusted causal strength value between the nodes. The initial confidence level of the potential causal chain is calculated based on the adjusted causal strength values ​​between nodes.

7. The method according to claim 3, characterized in that, The step of determining the causal relationship between the abnormal event and the abnormal alarm is pending when the overall confidence level does not reach the preset threshold includes: When the overall confidence level does not reach the preset threshold, new abnormal nodes and new alarm nodes are monitored in real time. When the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain, the overall confidence is increased to obtain the reset confidence. When the reset confidence level reaches a preset threshold, a causal relationship between the abnormal event and the abnormal alarm is determined.

8. The method according to claim 7, characterized in that, The step of increasing the overall confidence level to obtain the reset confidence level when the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain includes: When the anomaly type of the new abnormal node or the alarm type of the new alarm node appears in the potential causal chain, determine the weighting coefficient of the impact of the new abnormal node or the new alarm node on the overall confidence level. The reset confidence level is obtained by increasing the overall confidence level based on the influence weighting coefficient.

9. The method according to claim 3, characterized in that, The step of correcting the initial confidence level based on the robot's current task type and environmental load to obtain the comprehensive confidence level includes: Get the robot's current task type; The environmental load of the area where the robot is located is obtained, including the average utilization of the server CPU, network traffic, and cooling system operating power. Based on the task type and the environmental load, determine the influence coefficient of the abnormal event node or the abnormal alarm node on the confidence of the causal chain; Based on the influence coefficient, the initial confidence level is adjusted to obtain the comprehensive confidence level.

10. A motion control system for a data center operation and maintenance inspection robot, characterized in that, The system includes: The data acquisition module is used to acquire the robot's operating data, which includes the real-time torque and real-time axis acceleration of the drive motor. An anomaly detection module is used to identify whether an abnormal event has occurred in the robot based on the operating data. The abnormal event is manifested when the operating data exceeds the dynamic threshold of the operating data. An abnormal event node generation module is used to generate an abnormal event node when the abnormal event is detected. The abnormal event node includes the time of the abnormal event, the robot's current position at the time of the abnormal event, and the type of abnormality. The abnormal alarm acquisition module is used to acquire abnormal alarms from outside the robot; An abnormal alarm node generation module is used to generate abnormal alarm nodes based on the abnormal alarms. The abnormal alarm node includes alarm type, alarm time, and alarm device location. The causal relationship determination module is used to match the abnormal event node with the abnormal alarm node to determine whether there is a causal relationship between the abnormal event and the abnormal alarm. The strategy adjustment module is used to adjust the robot's motion control strategy when there is a causal relationship between the abnormal event and the abnormal alarm.

Images