Container type data machine room inspection method and system based on robot

Abstract

The invention discloses a robot-based container type data machine room inspection method and system, and relates to the technical field of inspection, and the method comprises the steps: calculating the fault value of each inspection robot, and dividing the inspection robots into fault robots and non-fault robots according to the fault values and a preset fault value threshold; calculating the bearable value of each non-fault robot for each non-fault robot; the loadable value is compared with a preset loadable value threshold value, and a robot capable of bearing an additional inspection target is screened out from the non-fault robots according to a comparison result and recorded as a normal robot; and the to-be-inspected target of the fault robot is allocated to each normal robot for inspection, so that when one or more robots break down and the inspection result of the inspection target goes wrong, the remaining inspection targets can be reasonably allocated to the robots without faults, the waste of inspection resources is reduced, and the inspection efficiency is improved. And hidden dangers of normal robot faults are weakened.

Description

Technical Field

[0001] The present invention relates to the field of inspection technology, and in particular to a robot-based containerized data room inspection method and system. Background Art

[0002] Containerized data room inspection refers to the use of automated robot systems to conduct regular inspections and maintenance of equipment in containerized data rooms. Due to the modularity, flexibility, and high space utilization of containerized data rooms, they are suitable for wide application in data centers, communication base stations, and other places. Robot inspections use a variety of sensors and cameras installed on the robots to monitor key parameters such as temperature and humidity, power load, and equipment status in the room in real time, and upload them to the central control system for analysis to determine whether there are any problems with the containerized data room. The robot can efficiently perform tasks such as environmental monitoring, equipment inspection, and troubleshooting, reducing the frequency of manual inspections and improving work efficiency. At the same time, through data collection and analysis, it ensures the stable operation and preventive maintenance of the room equipment, thereby improving the operational safety and reliability of the containerized data room.

[0003] In the process of inspecting the equipment in the container data room by multiple robots, generally a corresponding fixed inspection target is set for each robot, and the inspection is carried out according to the inspection target; if one or more robots fail, resulting in problems with the inspection results of the inspection targets, and there are no extra spare inspection robots, in order to successfully complete the inspection task; generally, the remaining inspection targets of the faulty robot will be allocated to the normal robot for inspection; however, if the remaining inspection targets are not allocated reasonably, it may lead to a waste of inspection resources and even increase the hidden dangers of failure of the remaining normal robots. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems and to provide a robot-based containerized data room inspection method and system.

[0005] In a first aspect of the present invention, a robot-based containerized data center inspection method is first proposed, the method comprising: The number of target points that each inspection robot has inspected, the number of times the robot has been disconnected from the central control system, and the packet loss rate of the robot's sensor data uploaded to the central control system are obtained to calculate the fault value of each inspection robot, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; For each non-faulty robot, obtain the initial power of the robot and the number of target points to be inspected, and calculate the load-bearing value of each non-faulty robot based on the initial power, the number of target points to be inspected and the corresponding fault value; The load-bearing value is compared with the preset load-bearing value threshold, and the robots that can undertake the additional inspection targets are selected from the non-faulty robots according to the comparison result, and the non-faulty robots that can undertake the additional inspection targets are recorded as normal robots; The targets to be inspected of the faulty robot are obtained and recorded as targets to be assigned, and the targets to be assigned are assigned to each normal robot for inspection in combination with the current targets to be inspected of all normal robots.

[0006] Optionally, the steps of obtaining the number of target points that each inspection robot has inspected, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system to calculate the fault value of each inspection robot are: For each inspection robot, obtain the target points that the robot has inspected, and obtain the actual stay time of each target point, to obtain an actual stay time sequence based on the time sequence; The similarity value between the actual stay time sequence of each inspection robot and the preset stay time sequence is calculated by the dynamic time warping algorithm, and the similarity value is used as the execution time value of the robot; Obtain the inspection path actually traveled by each robot, match the actual inspection path traveled with the preset inspection path traveled, and calculate the deviation ratio of the inspection robot from the preset inspection path; Obtain the difference between the number of target points that the robot is scheduled to inspect and the number of target points that have already been inspected, and divide the difference by the number of target points that are scheduled to be inspected as the execution progress difference; Calculate the execution deviation value of each inspection robot according to the execution time value, deviation distance ratio and execution progress difference; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system.

[0007] Optionally, the steps of calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system are: Obtain the time interval from the time each inspection robot starts inspection to the current time as the target time interval; obtain the number of disconnections between the robot and the central control system within the target time interval, and record the duration of each disconnection; The duration of each disconnection is compared with the preset maximum duration of disconnection. If the duration of the disconnection is not less than the preset maximum duration of disconnection, the corresponding disconnection state is recorded as an abnormal disconnection; Calculate the sum of the robot disconnection durations corresponding to all abnormal disconnections as the total communication failure time, and divide the total communication failure time by the target time interval to obtain the robot's communication failure value; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system.

[0008] Optionally, the steps of calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system are: Obtain the packet loss rate of sensor data uploaded by each inspection robot to the central control system, as well as the distortion rate of the image uploaded to the central control system, add the packet loss rate and the distortion rate to obtain the data upload abnormal value of the inspection robot; The execution deviation value, communication fault value and data upload abnormal value of each inspection robot are used to calculate the fault value of each inspection robot. The calculation formula is: , Where FDX is the fault value, GF, GY and KU are the execution deviation value, communication fault value and data upload abnormal value respectively, b1, b2 and b3 are the preset proportion values ​​of GF, GY and KU respectively, and b1, b2 and b3 are all greater than 0.

[0009] Optionally, the load-carrying value is compared with a preset load-carrying value threshold, and robots that can undertake additional inspection targets are selected from non-faulty robots according to the comparison result, including: Compare the load-bearing value of the non-faulty robot with the preset load-bearing value threshold. If the load-bearing value is not less than the preset load-bearing value threshold, the corresponding non-faulty robot can undertake the additional inspection target, and the corresponding non-faulty robot is recorded as a normal robot; If the load-bearing value is less than the preset load-bearing value threshold, the corresponding non-faulty robot cannot undertake the additional inspection target, and the non-faulty robot corresponding to the additional inspection target still performs the originally set inspection task.

[0010] Optionally, the steps of obtaining the initial power of the robot and the number of target points to be inspected, and calculating the carrying value of each non-faulty robot according to the initial power, the number of target points to be inspected and the corresponding fault value are: For each non-faulty robot, obtain the preset power consumption of the robot when completing the originally set inspection target, and obtain the initial power of the robot before starting the inspection, subtract the power consumption from the initial power to obtain the remaining power of the robot; From the robot's historical inspection records, obtain the probability of the robot adding additional inspection tasks, obtain the inspection completion degree of each additional inspection, calculate the average of the inspection completion degree, and multiply the average by the probability of adding additional inspection tasks to obtain the robot's additional inspection experience value; The carrying capacity of each non-faulty robot is calculated based on the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value.

[0011] Optionally, the carrying capacity of each non-faulty robot is calculated according to the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value, including: , Where HJU is the loadable value, ER, EW, ED and FDX are the remaining power, additional inspection experience value, the current number of targets to be inspected and the fault value, respectively; c1, c2, c3 and c4 are the preset proportion values ​​of ER, EW, ED and FDX, respectively, and c1, c2, c3 and c4 are all greater than 0.

[0012] Optionally, the steps of obtaining the to-be-inspected target of the faulty robot and recording it as the to-be-allocated target, and allocating it to each normal robot in combination with the current to-be-inspected targets of all normal robots are as follows: Set the optimal conditions for inspection target allocation, including the maximum load balance of normal robot inspections and the shortest inspection path; The objective function corresponding to the maximum load balancing degree is expressed as: , Where N is the total number of normal robots, i is the order number of the normal robot, gf is the number of remaining inspection targets originally set by the i-th normal robot, gt is the number of inspection targets of the faulty robot assigned to the i-th normal robot; μ represents the mean number of new inspection targets of the normal robot after the assignment is completed; The objective function corresponding to the shortest inspection path is expressed as: , In the formula, tr represents the remaining path to be driven in the inspection task originally set by the i-th normal robot, and te is the shortest driving path of the i-th normal robot after allocating the number of inspection targets to be inspected by the faulty robot; and the shortest driving path is calculated by the dynamic path planning algorithm, and when calculating, the starting point is the position corresponding to the last target point in the inspection task originally set by the robot; By solving the problem based on genetic algorithm, the number of inspection targets of the faulty robots assigned to each normal robot is output, and the containerized data room is inspected according to the inspection plan.

[0013] Optionally, the steps of outputting the number of inspection targets of the faulty robots to which each normal robot is assigned by solving the problem based on a genetic algorithm are as follows: Initialize the population, where each individual in the population represents the number of inspection targets of the faulty robot to which each normal robot is assigned; The fitness of each individual is calculated according to the objective functions H and R: F=ω1*H+ω2*R; F is the fitness, ω1 and ω2 are the preset weight influence values ​​corresponding to the objective functions H and R respectively, and the sum of ω1 and ω2 is 1; Select individuals with higher fitness from the population and use them as parents to enter the next generation; exchange the genes of the parent individuals through the crossover operation to generate new offspring individuals, and randomly change the genes of some individuals; By comparing the fitness of the offspring and the existing population, the best individuals are retained as the new generation of population; The algorithm is stopped when the fitness value converges or reaches the maximum number of iterations. The genetic algorithm finally outputs the individual with the highest fitness value as the final inspection plan. The final inspection plan includes the inspection target number of the faulty robot to which each normal robot is assigned.

[0014] In a second aspect of the present invention, a robot-based containerized data center inspection system is provided, the system comprising: Judgment module: The fault value of each inspection robot is calculated by obtaining the number of target points that each inspection robot has inspected, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; Calculation module: for each non-faulty robot, the initial power of the robot and the number of target points to be inspected are obtained, and the carrying value of each non-faulty robot is calculated based on the initial power, the number of target points to be inspected and the corresponding fault value; Screening module: compares the load-carrying value with the preset load-carrying value threshold, and selects robots that can undertake additional inspection targets from non-faulty robots based on the comparison result, and records the non-faulty robots that can undertake additional inspection targets as normal robots; Allocation inspection module: obtain the inspection targets of the faulty robot and record them as the targets to be allocated, and allocate the targets to be allocated to each normal robot for inspection in combination with the current inspection targets of all normal robots.

[0015] Beneficial effects of the present invention: The present invention proposes a robot-based container data room inspection method and system. When one or more robots fail, resulting in problems in the inspection results of the inspection targets, and the remaining inspection targets of the failed robots need to be allocated to robots without failures for inspection, the remaining inspection targets can be reasonably allocated to robots without failures, thereby reducing the waste of inspection resources and weakening the hidden dangers of normal robot failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below in conjunction with the accompanying drawings.

[0017] Figure 1 It is a flow chart of a robot-based containerized data center inspection method; Figure 2 This is a framework diagram of a robot-based containerized data center inspection system. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making any creative work shall fall within the scope of protection of the present invention.

[0020] The embodiment of the present invention provides a robot-based containerized data center inspection method. Figure 1 , Figure 1 A flowchart of a robot-based containerized data center inspection method provided in an embodiment of the present invention. The method comprises the following steps: The number of target points that each inspection robot has inspected, the number of times the robot has been disconnected from the central control system, and the packet loss rate of the robot's sensor data uploaded to the central control system are obtained to calculate the fault value of each inspection robot, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; For each non-faulty robot, obtain the initial power of the robot and the number of target points to be inspected, and calculate the load-bearing value of each non-faulty robot based on the initial power, the number of target points to be inspected and the corresponding fault value; The load-bearing value is compared with the preset load-bearing value threshold, and the robots that can undertake the additional inspection targets are selected from the non-faulty robots according to the comparison result, and the non-faulty robots that can undertake the additional inspection targets are recorded as normal robots; The targets to be inspected of the faulty robot are obtained and recorded as targets to be assigned, and the targets to be assigned are assigned to each normal robot for inspection in combination with the current targets to be inspected of all normal robots.

[0021] Based on a robot-based container data room inspection method provided by an embodiment of the present invention, through the above-mentioned method, when one or more robots fail resulting in problems in the inspection results of the inspection targets, and the remaining inspection targets of the failed robots need to be allocated to robots without failures for inspection, the remaining inspection targets can be reasonably allocated to robots without failures, thereby reducing the waste of inspection resources and weakening the hidden dangers of normal robot failures.

[0022] In one embodiment, the steps of obtaining the number of target points that each inspection robot has inspected, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system to calculate the fault value of each inspection robot are: For each inspection robot, obtain the target points that the robot has inspected, and obtain the actual stay time of each target point, to obtain an actual stay time sequence based on the time sequence; The similarity value between the actual stay time sequence of each inspection robot and the preset stay time sequence is calculated by the dynamic time warping algorithm, and the similarity value is used as the execution time value of the robot; Obtain the inspection path actually traveled by each robot, match the actual inspection path traveled with the preset inspection path traveled, and calculate the deviation ratio of the inspection robot from the preset inspection path; Obtain the difference between the number of target points that the robot is scheduled to inspect and the number of target points that have already been inspected, and divide the difference by the number of target points that are scheduled to be inspected as the execution progress difference; Calculate the execution deviation value of each inspection robot according to the execution time value, deviation distance ratio and execution progress difference; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system.

[0023] It should be noted that the formula for calculating the execution deviation value of each inspection robot according to the execution time value, deviation distance ratio and execution progress difference is: , where GF is the execution deviation value, mn, bh and tg are the execution time value, the deviation distance ratio and the execution progress difference value respectively, a1, a2 and a3 are the preset ratio values ​​of mn, bh and tg respectively, and a1, a2 and a3 are all greater than 0; in addition, a1, a2 and a3 are set by professionals according to the actual situation. Generally, the sum of a1, a2 and a3 is 1. For example, a1, a2 and a3 can be 0.3, 0.3 and 0.4 respectively, or other numbers, which are not limited to specific ones.

[0024] It should be noted that the actual stay time sequence is recorded in real time by the built-in sensors and clock modules of the inspection robot, and the stay time is automatically recorded each time the target point is reached; the preset stay time sequence is preset by the system and determined according to the standard inspection requirements of each target point; the inspection path data is acquired and recorded in real time by the robot's positioning system (such as GPS, inertial navigation system, etc.), and analyzed in combination with the robot's driving trajectory; the number of target points and the execution progress are obtained through the task management system, and the inspection task of each target point will be marked when it is executed, and the system will track the completed tasks and the remaining tasks. Through these data, combined with the real-time feedback of the robot, the execution deviation value can be effectively calculated and the execution status of the robot can be evaluated.

[0025] It should be noted that the larger the execution deviation value of the inspection robot, the greater the probability that the robot will fail when inspecting the container-type data room, resulting in problems in the inspection results of the inspection target. The reason is that when the execution deviation value of the inspection robot is large, it means that there are significant deviations in the robot's execution of the inspection task; these deviations may be reflected in many aspects, such as the deviation of the dwell time from the expectation, the deviation of the driving path, and the delay in the execution progress of the inspection task; a large execution deviation value usually indicates that the robot fails to complete the task in the predetermined manner and speed; as these deviations accumulate, the robot may miss certain target points during the inspection process, or fail to return in time for supplementary inspection while deviating from the preset path, thereby affecting the quality and accuracy of the entire inspection task; therefore, a large execution deviation value means that the robot's execution ability has been greatly disturbed, thereby increasing the risk of the robot failing and failing to complete the inspection task on time and in quality, affecting the overall safety and stability of the data room.

[0026] It should be noted that the dynamic time warping (DTW) algorithm is used to calculate the similarity between the actual stay time series and the preset stay time series of each inspection robot. First, the actual stay time series and the preset stay time series of the robot need to be converted into time series data, and then the two sets of sequences are compared by the DTW algorithm. DTW finds the optimal alignment path of two time series through dynamic programming, which can handle the time offset problem in the time series. That is, it can make up for the time difference caused by factors such as the robot's execution speed fluctuation or task delay, so as to achieve the best match of the sequence. This process measures the degree of deviation between the actual execution and the expected execution by calculating a numerical value representing the similarity. The advantage of using the DTW algorithm is that it can effectively handle nonlinearly aligned time series data, especially when the robot performs inspection tasks. The actual stay time may change due to various external factors (such as environmental changes, robot state fluctuations, etc.). DTW can automatically find the most appropriate alignment method, rather than relying solely on simple time synchronization, thereby improving the accuracy of inspection task evaluation. This allows us to detect the execution status of the robot more accurately, so as to discover potential problems and make adjustments in time to ensure the smooth completion of the inspection task; the execution time value calculates the similarity between the actual stay time sequence and the preset stay time sequence through the dynamic time warping (DTW) algorithm. The similarity output by the DTW algorithm is a dimensionless value (range 0~1) used to quantify the degree of match between the actual and preset sequences. For example, a similarity value of 0.9 indicates that the actual stay time is highly matched with the preset time. The execution time value is the similarity value output by the DTW algorithm and is directly used for subsequent deviation calculations.

[0027] In one embodiment, the steps of calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system are: Obtain the time interval from the time each inspection robot starts inspection to the current time as the target time interval; obtain the number of disconnections between the robot and the central control system within the target time interval, and record the duration of each disconnection; The duration of each disconnection is compared with the preset maximum duration of disconnection. If the duration of the disconnection is less than the preset maximum duration of disconnection, the corresponding disconnection state is recorded as a normal disconnection; if the duration of the disconnection is not less than the preset maximum duration of disconnection, the corresponding disconnection state is recorded as an abnormal disconnection; Calculate the sum of the robot disconnection durations corresponding to all abnormal disconnections as the total communication failure time, and divide the total communication failure time by the target time interval to obtain the robot's communication failure value; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system.

[0028] It should be noted that the preset maximum duration of disconnection is set by professionals based on actual circumstances, and no specific restrictions or elaboration are made; in addition, the target time interval is obtained through the start inspection time and the current time recorded by the internal clock of the inspection robot and the central control system; the number of disconnections and duration are monitored in real time through the communication module between the robot and the central control system, and the communication module will automatically record the time and duration of the disconnection event each time the communication is interrupted.

[0029] It should be noted that when the communication fault value of the inspection robot is larger, it means that when the robot is inspecting the containerized data room, the probability of a fault causing problems in the inspection results of the inspection target is greater. The reason is that the communication fault value reflects the connection stability and communication quality between the robot and the central control system. If the communication fault value is high, it means that the real-time data exchange between the robot and the system is frequently interrupted for a long time, which may cause the key tasks such as the execution progress of the inspection task, the transmission of the inspection data, and the generation of the fault report to be unable to be synchronized to the control system in time. This interruption not only increases the delay in the execution of tasks during the inspection process, but may also cause erroneous inspection data or missing tasks, thereby affecting the accurate assessment of the status of the equipment in the room. In addition, long-term communication problems may also cause the robot to be unable to obtain remote support or command adjustments in time when encountering a fault, further exacerbating the risk of its fault spreading. Therefore, when the communication fault value is too large, the reliability of the robot's task execution is reduced, the accuracy of the inspection results is seriously affected, and the hidden dangers of system problems are increased.

[0030] In one implementation, the benefits of analyzing the communication fault value for determining whether the inspection robot has a fault and whether it can continue to perform the next inspection task are as follows: the communication fault value can effectively reflect the connection stability and data transmission quality between the robot and the central control system, thereby promptly detecting communication interruptions or abnormal conditions; if the fault value is high, it indicates that the robot may face frequent communication interruptions, resulting in the inability to transmit data or receive instructions in a timely manner, affecting the execution efficiency and accuracy of the inspection task; therefore, based on the communication fault value, the robot's status can be evaluated in real time to determine whether it has the ability to continue to perform the task, thereby avoiding inspection task failures or data loss due to poor communication, thereby improving the reliability and safety of the inspection task.

[0031] In one embodiment, the steps of calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system are: Obtain the packet loss rate of sensor data uploaded by each inspection robot to the central control system, as well as the distortion rate of the image uploaded to the central control system, add the packet loss rate and the distortion rate to obtain the data upload abnormal value of the inspection robot; The execution deviation value, communication fault value and data upload abnormal value of each inspection robot are used to calculate the fault value of each inspection robot. The calculation formula is: , In the formula, FDX is the fault value, GF, GY and KU are the execution deviation value, communication fault value and data upload abnormal value respectively, b1, b2 and b3 are the preset proportional values ​​of GF, GY and KU respectively, and b1, b2 and b3 are all greater than 0. In addition, b1, b2 and b3 are set by professionals according to the actual situation. Generally, the sum of b1, b2 and b3 is 1. For example, b1, b2 and b3 can be 0.3, 0.3 and 0.4 respectively, or other numbers, which are not limited to the specific ones.

[0032] It should be noted that the packet loss rate of sensor data uploaded by the inspection robot can be obtained by monitoring the loss of data packets during network transmission, which is usually recorded by the network communication module and fed back to the central control system. The image distortion rate is calculated by comparing the original quality of the uploaded image with the quality of the received image. Image processing algorithms (such as peak signal-to-noise ratio, structural similarity, etc.) are usually used to measure the degree of image distortion. The data on packet loss rate and distortion rate will be monitored in real time and collected and analyzed by the central control system to provide basic data for the calculation of data upload abnormal values; It should be noted that when the data upload abnormal value of the inspection robot is larger, it means that when the robot is inspecting the container data room, the probability of a fault causing problems in the inspection results of the inspection target is greater. The reason is that: the increase in data upload abnormal values ​​usually means that there are major problems in the communication between the robot and the central control system during the inspection process, especially sensor data loss and image data distortion; sensor data packet loss will cause the robot to be unable to accurately perceive changes in the environment, thereby affecting the monitoring of key indicators such as equipment status, temperature and humidity, and power load, and unable to timely detect equipment abnormalities or failures. Image data distortion may make it impossible to accurately perform tasks such as image recognition and equipment inspection, resulting in the inability to capture subtle equipment failures or potential risks; the accumulation of these problems will increase the risk of misjudgment and missed inspections in the inspection task, thereby affecting the effectiveness and accuracy of the inspection results; therefore, a higher upload abnormal value is directly related to the reliability of the robot for the inspection target and the quality of the inspection task. Especially when inspecting key equipment, if there is a problem with data upload, it will greatly increase the probability that the failure is not discovered or correctly evaluated in time.

[0033] In one embodiment, classifying the inspection robot into a faulty robot and a non-faulty robot according to the fault value and the preset fault value threshold comprises: Compare the fault value of the inspection robot with the preset fault value threshold, and if the fault value is not less than the preset fault value threshold, the corresponding inspection robot is recorded as a faulty robot; If the fault value is less than the preset fault value threshold, the corresponding inspection robot is recorded as a non-faulty robot.

[0034] It should be noted that the preset fault value threshold is set by professionals according to actual conditions and will not be limited or elaborated on in detail.

[0035] It should be noted that the inspection robots are divided into faulty robots and non-faulty robots. For faulty robots, there is no way to perform the remaining inspection tasks of the containerized data center. Inspections by faulty robots may result in incomplete task execution, or even in the inability to provide real-time feedback on key equipment status or environmental change information. The sensor data of the faulty robot may be biased or lost, and the image data may be distorted, seriously affecting the quality of the inspection. In addition, if the faulty robot continues to participate in the inspection, it may delay the execution of other inspection targets due to the inability to complete the task on time, further delaying the inspection progress, and even increasing the risk of more serious failures of the robot due to excessive load; in order to ensure the smooth progress of the inspection task, the faulty robot must be removed from the task execution list, and the task must be promptly assigned to the robot with normal performance to ensure the accuracy of the inspection results and the efficient completion of the task.

[0036] In one embodiment, the steps of obtaining the initial power of the robot and the number of target points to be inspected, and calculating the carrying value of each non-faulty robot according to the initial power, the number of target points to be inspected and the corresponding fault value are as follows: For each non-faulty robot, obtain the preset power consumption of the robot when completing the originally set inspection target, and obtain the initial power of the robot before starting the inspection, subtract the power consumption from the initial power to obtain the remaining power of the robot; From the robot's historical inspection records, obtain the probability of the robot adding additional inspection tasks, obtain the inspection completion degree of each additional inspection, calculate the average of the inspection completion degree, and multiply the average by the probability of adding additional inspection tasks to obtain the robot's additional inspection experience value; The carrying capacity of each non-faulty robot is calculated based on the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value.

[0037] It should be noted that the data involved in the above calculations are obtained in the following ways: first, the power of the inspection robot is monitored in real time through the control system of the inspection robot, and the power consumption of each inspection task is recorded, so that the remaining power of the robot can be obtained; second, the robot's historical inspection records can be obtained through database query, including the probability of the robot adding additional inspection tasks and the completion degree of each additional inspection task. These data are obtained through the robot's operation log and task execution results; the number of targets to be inspected can be obtained through the inspection task management system, and the system will update the current number of targets to be executed according to task allocation and robot status; It should be noted that, when the remaining power of the non-faulty robot is larger, the additional inspection experience value is larger, the current number of targets to be inspected is smaller, and the fault value is smaller, the corresponding non-faulty robot is more selective as a normal robot to carry the remaining inspection tasks of the faulty robot, because a higher remaining power means that the robot has more energy to complete additional inspection tasks without having to worry about shutting down or failing to complete the task on time due to insufficient power. When a robot with sufficient power undertakes additional inspection tasks, it can better ensure the smooth progress of the inspection process and avoid affecting the execution efficiency and quality of the inspection tasks due to insufficient power, thereby improving the stability and efficiency of the entire inspection system. In addition, generally speaking, the non-faulty robot has designed the inspection plan before the inspection, so under normal circumstances, there will be no shortage of power in the middle of the inspection, and the robot's power is generally sufficient to support the robot to complete the preset inspection tasks.

[0038] Robots with higher experience values ​​show higher task completion when handling additional tasks, indicating that they have stronger inspection capabilities under similar circumstances and can efficiently and stably complete additional inspection tasks. The mean and probability of completion of additional inspection tasks reflect the robot's adaptability and execution ability when faced with increased tasks. Experienced robots can better allocate resources and complete inspection goals when taking over other tasks, reducing the risk of task interruption or delay. Fewer inspection targets mean that the robot has more free resources to allocate to additional tasks, and lower fault values ​​mean that the robot is more reliable and has a lower probability of failure when performing tasks; such robots can take over additional inspection tasks more efficiently without negatively affecting the completion of existing tasks; robots with smaller fault values ​​can operate stably, avoid affecting the inspection progress due to equipment failure, and ensure the continuity and timeliness of inspection tasks.

[0039] In one embodiment, the carrying capacity of each non-faulty robot is calculated based on the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value, including: , Where HJU is the loadable value, ER, EW, ED and FDX are the remaining power, additional inspection experience value, the current number of targets to be inspected and the fault value, respectively; c1, c2, c3 and c4 are the preset proportion values ​​of ER, EW, ED and FDX, respectively, and c1, c2, c3 and c4 are all greater than 0. In addition, c1, c2, c3, and c4 are set by professionals according to actual conditions. Generally, the sum of c1, c2, c3, and c4 is 1. For example, c1, c2, c3, and c4 can be 0.25, 0.25, 0.25, and 0.25, respectively. They can also be other numbers without specific limitation. In addition, before calculating the carrying capacity of the non-faulty robot, it is necessary to normalize the remaining power, additional inspection experience value, current number of targets to be inspected, and fault value of the non-faulty robot. Commonly used normalization methods include Min-Max normalization, Z-Score standardization, etc. The specific method selected depends on the actual situation and is not limited or elaborated.

[0040] In one embodiment, the load-bearing value is compared with a preset load-bearing value threshold, and the robot that can undertake the additional inspection target is screened out from the non-faulty robots according to the comparison result, including: Compare the load-bearing value of the non-faulty robot with the preset load-bearing value threshold. If the load-bearing value is not less than the preset load-bearing value threshold, the corresponding non-faulty robot can undertake the additional inspection target, and the corresponding non-faulty robot is recorded as a normal robot; If the load-bearing value is less than the preset load-bearing value threshold, the corresponding non-faulty robot cannot undertake the additional inspection target, and the non-faulty robot corresponding to the additional inspection target still performs the originally set inspection task.

[0041] It should be noted that the preset load-bearing value threshold is set by professionals based on actual conditions and will not be limited or elaborated on in detail.

[0042] In one implementation, the advantage of the robots selected in this way that can take over the remaining inspection tasks of the faulty robots is that they can be finely screened based on the actual working status of the non-faulty robots (such as remaining power, additional inspection experience value, etc.), ensuring that only robots with sufficient capabilities and resources can take over additional tasks, thereby improving the efficiency and accuracy of completing the inspection tasks; compared with simply allocating the remaining inspection targets of the faulty robots to non-faulty robots for inspection, this method can more scientifically evaluate the actual carrying capacity of each non-faulty robot, avoiding the decline in inspection quality or the inability of non-faulty robots to complete the original tasks on time due to excessive task allocation; by calculating the carrying value and comparing it with the preset threshold, the most suitable robot can be effectively selected, maximizing the efficiency of resource utilization, reducing potential overload risks, and ensuring the smooth progress of the inspection tasks; this method can not only improve the receiving capacity of faulty tasks, but also improve the stability and sustainability of the entire inspection system.

[0043] In one embodiment, the steps of obtaining the to-be-inspected target of the faulty robot and recording it as the to-be-assigned target, and assigning the to-be-assigned target to each normal robot in combination with the current to-be-inspected targets of all normal robots are as follows: Set the optimal conditions for inspection target allocation, including the maximum load balance of normal robot inspections and the shortest inspection path; The objective function corresponding to the maximum load balancing degree is expressed as: , Where N is the total number of normal robots, i is the ordinal number of the normal robot, gf is the number of remaining inspection targets originally set by the i-th normal robot, gt is the number of inspection targets of the faulty robot assigned to the i-th normal robot; μ represents the mean number of new inspection targets of the normal robot after the allocation is completed; ensure that the load is evenly distributed among the robots, and the goal is to make the new number of inspection targets of each robot close to the mean, so as to achieve load balancing.

[0044] , where tr represents the remaining path to be driven in the inspection task originally set by the i-th normal robot, and te is the shortest driving path of the i-th normal robot after allocating the number of inspection targets to be inspected by the faulty robot; and this shortest driving path can be calculated by the dynamic path planning algorithm, and when calculating, the starting point is the position corresponding to the last target point in the inspection task originally set by the robot; through the dynamic path planning algorithm, according to the position of the last target point in the inspection task originally set by each normal robot, the shortest inspection path after being allocated to the inspection target is calculated, and the allocation scheme with the smallest total driving distance is selected; By solving the problem based on genetic algorithm, the number of inspection targets of the faulty robots assigned to each normal robot is output, and the containerized data room is inspected according to the inspection plan.

[0045] It should be noted that the remaining paths to be traveled in the inspection tasks originally set by each normal robot can be directly obtained through the robot's historical inspection records and the route planning of the current task; specifically, the distance from the last inspection target point of the current task to the remaining uninspected target points can be calculated by combining the robot's inspection target point position and path planning information. This information usually comes from the robot's built-in positioning system (such as GPS, laser radar, etc.) and the path planning system of the inspection task, and the remaining paths to be traveled are accurately obtained through dynamically updated position and path data.

[0046] The solution is based on genetic algorithm, which is an optimization method based on biological evolution, which continuously improves the solution through selection, crossover and mutation.

[0047] An initial solution is randomly generated, and a population is initialized. Each individual in the population represents a possible solution. Each individual is composed of multiple genes, and the gene represents a patrol task allocation scheme. The gene sequence of the individual can be an array, in which each element corresponds to the number of patrol targets of the faulty robot assigned to a normal robot. Among them, each individual represents the number of patrol targets of the faulty robot assigned to each normal robot. The fitness of each individual is calculated according to the objective functions H and R: F=ω1*H+ω2*R; F is the fitness, ω1 and ω2 are the preset weight influence values ​​corresponding to the objective functions H and R respectively, and the sum of ω1 and ω2 is 1; the specific values ​​are set by professionals according to the actual situation and are not limited or elaborated here.

[0048] Select excellent individuals according to their fitness values ​​and have them enter the next generation as parents; commonly used roulette or tournament selection strategies; roulette selection: randomly select individuals according to the fitness ratio, and the individuals with higher fitness have a greater probability of being selected. Specifically, by mapping the fitness value to a roulette wheel, random selection is performed according to the fitness ratio; tournament selection: randomly select a certain number of individuals from the population, and select the individuals with the best fitness to enter the next generation; this method can better avoid over-selection of individuals with similar fitness, thereby increasing the diversity of the population.

[0049] Selection results: Through the selection operation, the parent individuals are obtained and the next crossover operation is entered; through the crossover operation, the genes of the parent individuals are exchanged to generate new offspring individuals; among the selected parent individuals, a new generation of individuals is generated through the crossover operation; the crossover operation generates new individuals by exchanging the gene parts of the parent individuals, which can explore new solution spaces and increase population diversity; the position of the intersection point can be randomly selected, and the offspring individuals generated by crossover will inherit some advantages of the parent individuals; therefore, the new individuals generated by this process will become the basis for the next mutation operation.

[0050] Randomly change some individual genes (such as the number of inspection targets assigned to each normal robot) to increase the diversity of the population and prevent local optimality; By comparing the fitness of the offspring and the existing population, the best individuals are retained as the new generation of population; The algorithm is stopped when the fitness value converges or reaches the maximum number of iterations. The genetic algorithm finally outputs the individual with the highest fitness value as the final inspection plan. The final inspection plan includes the inspection target number of the faulty robot assigned to each normal robot, and the container data room is inspected according to the inspection plan.

[0051] In one implementation, the benefits of doing so are mainly reflected in optimizing resource allocation, improving inspection efficiency, reducing task execution costs and enhancing adaptability; through continuous evolution and optimization, the genetic algorithm enables the inspection task allocation plan to find the best balance between load balancing and the shortest inspection path, avoiding the situation where some normal robots are overloaded or some robots have too few tasks, thereby improving the overall inspection efficiency; at the same time, the optimization of the inspection path reduces the total distance traveled by the robot, which helps to save energy consumption, reduce equipment wear and tear, and extend the service life of the robot; in addition, the genetic algorithm can find a better solution in a complex, multi-constrained environment, and is particularly suitable for situations where inspection tasks change dynamically, such as when adding or reducing faulty robots, the inspection plan can still be quickly adjusted to ensure the continuity and efficiency of the inspection work; at the same time, it provides an intelligent and automated solution for inspection tasks, greatly reducing manual intervention and planning costs, and improving the intelligence level and operational stability of the entire inspection system.

[0052] Based on the same inventive concept, the embodiment of the present invention also provides a robot-based containerized data center inspection system. Figure 2 , Figure 2 A framework diagram of a robot-based containerized data center inspection system provided in an embodiment of the present invention, the system comprising: Judgment module: The fault value of each inspection robot is calculated by obtaining the number of target points that each inspection robot has inspected, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; Calculation module: for each non-faulty robot, the initial power of the robot and the number of target points to be inspected are obtained, and the carrying value of each non-faulty robot is calculated based on the initial power, the number of target points to be inspected and the corresponding fault value; Screening module: compares the load-carrying value with the preset load-carrying value threshold, and selects robots that can undertake additional inspection targets from non-faulty robots based on the comparison result, and records the non-faulty robots that can undertake additional inspection targets as normal robots; Allocation inspection module: obtain the inspection targets of the faulty robot and record them as the targets to be allocated, and allocate the targets to be allocated to each normal robot for inspection in combination with the current inspection targets of all normal robots.

[0053] Based on a robot-based containerized data room inspection system provided by an embodiment of the present invention, through the above-mentioned method, when one or more robots fail resulting in problems with the inspection results of the inspection targets, and the remaining inspection targets of the failed robots need to be allocated to robots without failures for inspection, the remaining inspection targets can be reasonably allocated to robots without failures, thereby reducing the waste of inspection resources and weakening the hidden dangers of normal robot failures.

[0054] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be used to artificially limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A robot-based containerized data center inspection method, characterized in that: The following steps are involved: The number of target points that each inspection robot has inspected, the number of times the robot has been disconnected from the central control system, and the packet loss rate of the robot's sensor data uploaded to the central control system are obtained to calculate the fault value of each inspection robot, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; For each non-faulty robot, obtain the initial power of the robot and the number of target points to be inspected, and calculate the load-bearing value of each non-faulty robot based on the initial power, the number of target points to be inspected and the corresponding fault value; The load-bearing value is compared with the preset load-bearing value threshold, and the robots that can undertake the additional inspection targets are selected from the non-faulty robots according to the comparison result, and the non-faulty robots that can undertake the additional inspection targets are recorded as normal robots; The targets to be inspected of the faulty robot are obtained and recorded as targets to be assigned, and the targets to be assigned are assigned to each normal robot for inspection in combination with the current targets to be inspected of all normal robots.

2. A robot-based containerized data center inspection method according to claim 1, characterized in that: The steps for calculating the fault value of each inspection robot by obtaining the number of target points that each inspection robot has inspected, the number of times the robot has been disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system are as follows: For each inspection robot, obtain the target points that the robot has inspected, and obtain the actual stay time of each target point, to obtain an actual stay time sequence based on the time sequence; The similarity value between the actual stay time sequence of each inspection robot and the preset stay time sequence is calculated by the dynamic time warping algorithm, and the similarity value is used as the execution time value of the robot; Obtain the inspection path actually traveled by each robot, match the actual inspection path traveled with the preset inspection path traveled, and calculate the deviation ratio of the inspection robot from the preset inspection path; Obtain the difference between the number of target points that the robot is scheduled to inspect and the number of target points that have already been inspected, and divide the difference by the number of target points that are scheduled to be inspected as the execution progress difference; Calculate the execution deviation value of each inspection robot according to the execution time value, deviation distance ratio and execution progress difference; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system.

3. A robot-based containerized data center inspection method according to claim 2, characterized in that: The steps for calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system are as follows: Obtain the time interval from the time each inspection robot starts inspection to the current time as the target time interval; obtain the number of disconnections between the robot and the central control system within the target time interval, and record the duration of each disconnection; The duration of each disconnection is compared with the preset maximum duration of disconnection. If the duration of the disconnection is not less than the preset maximum duration of disconnection, the corresponding disconnection state is recorded as an abnormal disconnection. Calculate the sum of the robot disconnection durations corresponding to all abnormal disconnections as the total communication failure time, and divide the total communication failure time by the target time interval to obtain the robot's communication failure value; The fault value of each inspection robot is calculated based on the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system.

4. A robot-based containerized data center inspection method according to claim 3, characterized in that: The steps for calculating the fault value of each inspection robot according to the execution deviation value of each inspection robot, the communication fault value and the packet loss rate of the robot sensor data uploaded to the central control system are as follows: Obtain the packet loss rate of sensor data uploaded by each inspection robot to the central control system, as well as the distortion rate of the image uploaded to the central control system, add the packet loss rate and the distortion rate to obtain the data upload abnormal value of the inspection robot; The execution deviation value, communication fault value and data upload abnormal value of each inspection robot are used to calculate the fault value of each inspection robot. The calculation formula is: ， Where FDX is the fault value, GF, GY and KU are the execution deviation value, communication fault value and data upload abnormal value respectively, b1, b2 and b3 are the preset proportion values ​​of GF, GY and KU respectively, and b1, b2 and b3 are all greater than 0.

5. The robot-based containerized data center inspection method according to claim 1 is characterized in that: The load-carrying value is compared with the preset load-carrying value threshold, and the robots that can undertake additional inspection targets are selected from the non-faulty robots according to the comparison results, including: Compare the load-bearing value of the non-faulty robot with the preset load-bearing value threshold. If the load-bearing value is not less than the preset load-bearing value threshold, the corresponding non-faulty robot can undertake the additional inspection target, and the corresponding non-faulty robot is recorded as a normal robot; If the load-bearing value is less than the preset load-bearing value threshold, the corresponding non-faulty robot cannot undertake the additional inspection target, and the non-faulty robot corresponding to the additional inspection target still performs the originally set inspection task.

6. The robot-based containerized data center inspection method according to claim 1 is characterized in that: The steps of obtaining the initial power of the robot and the number of target points to be inspected, and calculating the carrying value of each non-faulty robot according to the initial power, the number of target points to be inspected and the corresponding fault value are as follows: For each non-faulty robot, obtain the preset power consumption of the robot when completing the originally set inspection target, and obtain the initial power of the robot before starting the inspection, subtract the power consumption from the initial power to obtain the remaining power of the robot; From the robot's historical inspection records, obtain the probability of the robot adding additional inspection tasks, obtain the inspection completion degree of each additional inspection, calculate the average of the inspection completion degree, and multiply the average by the probability of adding additional inspection tasks to obtain the robot's additional inspection experience value; The carrying capacity of each non-faulty robot is calculated based on the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value.

7. The robot-based containerized data center inspection method according to claim 1 is characterized in that: The load capacity of each non-faulty robot is calculated based on the remaining power of the non-faulty robot, the additional inspection experience value, the current number of targets to be inspected, and the fault value, including: ， Where HJU is the loadable value, ER, EW, ED and FDX are the remaining power, additional inspection experience value, the current number of targets to be inspected and the fault value, respectively; c1, c2, c3 and c4 are the preset proportion values ​​of ER, EW, ED and FDX, respectively, and c1, c2, c3 and c4 are all greater than 0.

8. The robot-based containerized data center inspection method according to claim 1 is characterized in that: The steps of obtaining the inspection target of the faulty robot as the target to be assigned, and assigning the target to be assigned to each normal robot in combination with the current inspection targets of all normal robots are as follows: Set the optimal conditions for inspection target allocation, including the maximum load balance of normal robot inspections and the shortest inspection path; The objective function corresponding to the maximum load balancing degree is expressed as: ， Where N is the total number of normal robots, i is the order number of the normal robot, gf is the number of remaining inspection targets originally set by the i-th normal robot, gt is the number of inspection targets of the faulty robot assigned to the i-th normal robot; μ represents the mean number of new inspection targets of the normal robot after the assignment is completed; The objective function corresponding to the shortest inspection path is expressed as: ， In the formula, tr represents the remaining path to be driven in the inspection task originally set by the i-th normal robot, and te is the shortest driving path of the i-th normal robot after allocating the number of inspection targets to be inspected by the faulty robot; and the shortest driving path is calculated by the dynamic path planning algorithm, and when calculating, the starting point is the position corresponding to the last target point in the inspection task originally set by the robot; By solving the problem based on genetic algorithm, the number of inspection targets of the faulty robots assigned to each normal robot is output, and the containerized data room is inspected according to the inspection plan.

9. The robot-based containerized data center inspection method according to claim 1 is characterized in that: By solving the problem based on the genetic algorithm, the steps to output the number of inspection targets of the faulty robot assigned to each normal robot are as follows: Initialize the population, where each individual in the population represents the number of inspection targets of the faulty robot to which each normal robot is assigned; The fitness of each individual is calculated according to the objective functions H and R: F=ω1*H+ω2*R; F is the fitness, ω1 and ω2 are the preset weight influence values ​​corresponding to the objective functions H and R respectively, and the sum of ω1 and ω2 is 1; Select individuals with higher fitness from the population and use them as parents to enter the next generation; exchange the genes of the parent individuals through the crossover operation to generate new offspring individuals, and randomly change the genes of some individuals; By comparing the fitness of the offspring and the existing population, the best individuals are retained as the new generation of population; The algorithm is stopped when the fitness value converges or reaches the maximum number of iterations. The genetic algorithm finally outputs the individual with the highest fitness value as the final inspection plan. The final inspection plan includes the inspection target number of the faulty robot to which each normal robot is assigned.

10. A robot-based containerized data room inspection system, used to implement a robot-based containerized data room inspection method as described in any one of claims 1 to 9, characterized in that: The system comprises: Judgment module: The fault value of each inspection robot is calculated by obtaining the number of target points that each inspection robot has inspected, the number of times the robot is disconnected from the central control system, and the packet loss rate of the robot sensor data uploaded to the central control system, and the inspection robots are divided into faulty robots and non-faulty robots according to the fault value and the preset fault value threshold; Calculation module: for each non-faulty robot, the initial power of the robot and the number of target points to be inspected are obtained, and the carrying value of each non-faulty robot is calculated based on the initial power, the number of target points to be inspected and the corresponding fault value; Screening module: compares the load-carrying value with the preset load-carrying value threshold, and selects robots that can undertake additional inspection targets from non-faulty robots based on the comparison result, and records the non-faulty robots that can undertake additional inspection targets as normal robots; Allocation inspection module: obtain the inspection targets of the faulty robot and record them as the targets to be allocated, and allocate the targets to be allocated to each normal robot for inspection in combination with the current inspection targets of all normal robots.

Images