Hard disk hot plug test method and system based on manipulator

Abstract

The invention provides a hard disk hot plug test method and system based on a manipulator, and the method comprises the steps that a test server executes test cases in sequence based on a test task issued by a test management platform and a current operation request related to the test task, and a management server issues an operation instruction to a manipulator subsystem; the PLC industrial personal computer controls the camera to collect environment visual information of the hard disk array cabinet and transmit the environment visual information to a trained deep learning model in the management server to obtain a judgment result representing a hard disk condition; and the PLC industrial personal computer controls the manipulator to perform hot plug operation on the hard disks in the hard disk array cabinet connected with the test server based on the received operation instruction and the judgment result, the steps are repeated, and the test result is fed back to the test management platform until the test task is completed. The scheme provided by the invention can be better compatible with user use scenes, various test samples are provided, and the test result has the advantages of higher universality, high automation degree, high safety and the like.

Description

Technical Field

[0001] This application belongs to the technical field of automated testing, and particularly relates to a hard disk hot plugging test method and system based on a manipulator. Background Art

[0002] The related automated hard disk hot plugging test methods mainly include two categories: electronic hot plugging method and mechanical hot plugging method.

[0003] One category is the electronic hot plugging method, represented by the Quarch storage hot plugging device that provides HDD / SSD hot plugging automation testing for 12G SAS, 6G SAS / SATA, PCIe / NVMe U.2 / U.3 / M.2 SSD. Using the supporting software, it is convenient to conduct tests and statistically analyze the test results. However, this method has problems such as limited application scenarios and insufficient usability, specifically including: 1) The Quarch device is only applicable to test scenarios with a small sample size and is difficult to apply to actual application scenarios such as hard disk arrays with a large number of hard disks and compact positions; 2) For different application scenarios, special cables need to be customized for the hard disks and backplanes of the Quarch device, and each time the hot-plugged hard disk is replaced, the device needs to be manually changed, resulting in poor usability.

[0004] The other category is the mechanical hot plugging method, which uses mechanical structures (such as push rods, rockers, bolts, etc.) to perform hard disk plugging and unplugging. However, it has problems such as poor compatibility, insufficient stability, low automation level, and poor safety, specifically including: 1) Inflexible, generally, one mechanical jig can only be used for a specific use scenario and is difficult to adapt to server hard disk slots of different specifications; 2) Poor stability. In the long-term hard disk plugging and unplugging test scenario that requires tens of thousands of times, the mechanical structure is inevitably worn, which will cause irreversible damage to the interface, resulting in a short service life and poor stability. During use, manual correction of deviations is required regularly to ensure accuracy; 3) Lack of monitoring of the mechanical jig and inability to accurately record some data (such as plugging and unplugging time intervals, etc.); 4) Difficult to ensure safety. For example, in case of a failure, it cannot be stopped in time, and there are problems with difficult traceability. Summary of the Invention

[0005] The purpose of this application is to provide a hard disk hot plugging test method and system based on a manipulator, aiming to solve the technical problems of single application scenario, poor compatibility, and low automation level existing in the related hard disk hot plugging tests.

[0006] According to the first aspect of this application, a hard disk hot plugging test method based on a manipulator is provided, including:

[0007] S1, based on the test tasks issued by the test management platform, the test server sequentially executes test cases;

[0008] S2. Based on the current operation request related to the test task issued by the test management platform, the management server issues an operation instruction to the manipulator subsystem;

[0009] S3. Based on the received operation instruction, the PLC industrial control computer in the manipulator subsystem controls the camera to collect the environmental visual information of the hard disk array cabinet connected to the test server and transmits it to the trained deep learning model in the management server to obtain a judgment result representing the hard disk condition;

[0010] S4. The PLC industrial control computer in the manipulator subsystem controls the manipulator to perform a hot plug operation on the hard disk in the hard disk array cabinet connected to the test server based on the received operation instruction and the judgment result;

[0011] S5. Repeat steps S2 - S4, and feedback the test result to the test management platform until the test task is completed.

[0012] By inputting the environmental visual information of the hard disk array cabinet collected by the camera in real time into the trained deep learning model on the management server for image recognition, this method can accurately obtain the judgment result of the hard disk condition; the PLC industrial control computer controls the manipulator to perform a hot plug operation on the hard disk in the hard disk array cabinet connected to the test server based on the received operation instruction and the judgment result. The manipulator performs the hot plug operation instead of manual labor according to the instruction, with high operation accuracy, more efficient and faster. The manipulator is easy to control, and with the visual information of the camera, it can greatly avoid damage to personnel and products, with reliable safety. The test management platform can provide various test samples to ensure that the test result is more universal.

[0013] In an alternative embodiment, after the current operation request related to the test task issued by the test management platform, it further includes:

[0014] S201. The test management platform determines whether the current test case in the test task includes a hard disk plug - and - unplug step. If it does, it determines whether the manipulator has completed the initialization operation;

[0015] S202. If it is determined that the manipulator has not completed the initialization operation, the manipulator initialization process is performed. After the initialization process ends, it is determined whether the initialization is successful. If it fails, it is determined whether the number of times of the manipulator initialization failure is greater than the preset number of times; if it is greater than the preset number of times, the test management platform aborts the test; if it is less than or equal to the preset number of times, the manipulator initialization process continues;

[0016] S203. If it is determined that the manipulator has completed the initialization operation, then it is determined whether the current test step is a hard disk hot plug step. If so, the test management platform sends a current hard disk hot plug operation request to the management server; the management server sends an operation instruction to the PLC industrial control computer, and the industrial control computer cooperates with the manipulator to perform the hot plug operation.

[0017] After the current test case in the test task includes the hard disk plugging and unplugging step, the manipulator that is determined to have not completed the initialization operation is initialized, avoiding hot plug operation failures caused by manipulator problems and ensuring the reliability of each step of the manipulator's hot plug operation.

[0018] In an alternative embodiment, before step S1, the position numbers of each hard disk are marked based on the specifications of the hard disk array cabinet connected to the test server. After that, the spatial position coordinates of the manipulator to which each position number corresponds are calibrated.

[0019] By marking the position numbers of each hard disk and then calibrating the spatial position coordinates of the corresponding manipulator based on each position number, when it is necessary to perform a hard disk hot plug test, only the position number of the hard disk needs to be input, which is convenient for the staff to operate.

[0020] In an alternative embodiment, based on the received operation instruction, the PLC industrial control computer in the manipulator subsystem controls the camera to collect the environmental visual information of the hard disk array cabinet connected to the test server and transmits it to the trained deep learning model in the management server to obtain a judgment result representing the hard disk condition, specifically including:

[0021] S301. Input the position number of the hard disk to be hot plugged, and determine whether the position number is valid. If it is invalid, an error is reported and the test process ends; if it is valid, continue to the next step:

[0022] S302. The manipulator reaches the spatial coordinate point corresponding to the position number according to the pre-set motion trajectory, adjusts the manipulator posture and controls the camera at the end of the manipulator to collect the environmental visual information of the hard disk array cabinet and send it back to the management server;

[0023] S303. The management server inputs the environmental visual information into the trained deep learning training model to obtain a judgment result representing the hard disk condition.

[0024] Inputting the environmental visual information of the hard disk array cabinet collected by the camera in real time into the trained deep learning model on the management server for image recognition can obtain accurate hard disk condition information, providing a reliable judgment basis for controlling the movement of the robotic arm to help the staff adjust the hot plug test process in real time.

[0025] In an alternative embodiment, the judgment result includes: the hard disk is not in place, the hard disk is in place and the buckle is open, or the hard disk is in place and the buckle is closed.

[0026] In an alternative embodiment, the PLC industrial control computer in the manipulator subsystem controls the manipulator to perform hot plugging operations on the hard disks in the hard disk array cabinet connected to the test server based on the received operation instruction and the judgment result, including:

[0027] If the judgment result is that the hard disk is not in place, an error is reported and the test process ends; if the judgment result is that the hard disk is in place and the buckle is open, the industrial control computer cooperates with the manipulator to close the buckle; if the judgment result is that the hard disk is in place and the buckle is closed, the manipulator completes the hot plugging operation of the hard disk according to the pre-set movement trajectory and posture.

[0028] In an alternative embodiment, if the judgment result is that the hard disk is in place and the buckle is open, the industrial control computer cooperating with the manipulator to close the buckle includes:

[0029] Judge whether the number of operations of the manipulator to close the buckle is greater than a preset threshold. If it is greater than the preset threshold, an error is reported and the test process ends; if it is not greater than the preset threshold, the manipulator reaches the spatial coordinate point corresponding to the position number according to the pre-set movement trajectory, adjusts the posture of the manipulator, and controls the camera at the end of the manipulator to collect the environmental visual information of the hard disk array cabinet and transmit it back to the management server.

[0030] For different judgment results, different operations are further performed, and error reporting is performed when the process is incorrect, which can facilitate tracing when a fault occurs during the test process and improve the test efficiency.

[0031] In an alternative embodiment, a ROS system (Robot Operating System) is set in the PLC industrial control computer, and the ROS system provides gain position compensation for the positioning of the manipulator based on the visual information collected by the camera.

[0032] The ROS system can ensure the positioning accuracy of the manipulator. It can implement a suitable and feasible solution according to different production environments, different server models, different fixed-point positions, and different hard disk enclosures (which mainly affect the plugging and unplugging methods), retain an interface that is easy for secondary development, has strong usability, is compatible with a variety of user usage scenarios, is convenient for testing products of different specifications, and feeds back the visual information collected by the camera as gain compensation to the PLC industrial control computer, which can avoid the situation of drift when the manipulator is fixed-point, making the manipulator operation more stable and with higher accuracy.

[0033] According to the second aspect of the present application, there is provided a hard disk hot plug test system based on a manipulator for implementing the hard disk hot plug test method as described in the first aspect. The hard disk hot plug test system includes: a management server, a test management platform, a manipulator subsystem, and a test server; a hard disk array cabinet is connected to the test server to form a storage system; the manipulator subsystem includes a manipulator, a PLC industrial control computer, and a camera, and the camera is disposed at the end of the manipulator;

[0034] The test management platform is used to send test tasks to the test server and send operation requests related to the test tasks to the management server;

[0035] The management server sends operation instructions to the manipulator in the manipulator subsystem based on the received operation requests, and the management server is further used to receive the environmental visual information transmitted by the PLC industrial control computer in the manipulator subsystem;

[0036] The PLC industrial control computer in the manipulator subsystem controls the manipulator to perform hot plug operations on the hard disks of the hard disk array cabinet based on the received operation instructions, and controls the camera to collect the environmental visual information of the hard disk array cabinet and transmit it to the management server;

[0037] The test server sequentially executes test cases based on the sent test tasks and feeds back the test results to the test management platform.

[0038] In an alternative embodiment, a cloud server is further included. The management server transmits the received historical environmental visual information to the cloud server, and the cloud server trains an initial deep learning model based on the historical environmental visual information collected by the camera to obtain a trained deep learning model.

[0039] Compared with the related art, the present application can better be compatible with user usage scenarios, provide diverse test samples, making the test results more universal, with high efficiency, strong reliability, strong stability, full automation, strong usability, easy for secondary development, high security, and strong scalability. Specifically, it includes:

[0040] (1) Can better be compatible with user usage scenarios

[0041] The ROS of the present application can ensure the positioning accuracy of the manipulator, and can implement a suitable and feasible solution according to different production environments, different server models, different fixed-point positions, and different hard disk enclosures (which mainly affect the plugging and unplugging methods), retain interfaces that are easy for secondary development, have strong usability, can be compatible with a variety of user usage scenarios, and are convenient for testing products of different specifications.

[0042] (2) Can provide diverse test samples, making the test results more universal

[0043] The manipulator of this application has high operating precision and can better simulate the actual customer scenarios of manually inserting and removing hard drives. By dynamically adjusting parameters such as the spatial coordinate axes of the end joint of the manipulator, the precision of the force control sensor, and the y-axis offset, the subtle differences of different testers in actual insertion and removal operations (such as insertion and removal force, insertion and removal interval time, etc.) can be replicated, making the hard drive insertion and removal operations more realistic. Therefore, this application can provide diverse test samples and ensure that the test results are more universal.

[0044] (3) High efficiency, strong reliability, and strong stability

[0045] As the test time accumulates, the fixtures used in traditional mechanical hard drive insertion and removal solutions will have varying degrees of wear, resulting in a significant reduction in test stability. However, the manipulator adopted in this application is a mature closed-loop system. The built-in spatial coordinate system of ROS can ensure the accuracy of the manipulator's operation. At the same time, the visual information obtained by the camera is fed back to the control system as gain compensation to ensure that the manipulator does not drift at a fixed point, making the operation more stable. For thousands of hard drive insertion and removal tests, the manipulator of this application can work efficiently and continuously, thereby shortening the test cycle and improving the test efficiency.

[0046] (4) High degree of automation, strong ease of use, and convenient for secondary development

[0047] This application has a high degree of automation and can realize remote control of the manipulator's operation and integrate this operation into the test cases for use. The hard drive insertion and removal operations are modularized, and easy-to-use API interfaces are provided, which are convenient to call at any stage of the test cases. Based on the logging service provided by ROS, it is more convenient to trace each posture of the manipulator according to the time point, ensuring the controllability of the hard drive insertion and removal operations and achieving start-and-stop as needed. The modular design in this application improves the ease of use to facilitate access to different test frameworks for secondary development.

[0048] (5) High security

[0049] The camera in this application ensures the reliability of each operation during the hard drive insertion and removal sub-operations. This application deploys the deep learning model obtained by remote training on the local management server, greatly improving the accuracy of image recognition and providing a reliable judgment basis for controlling the movement of the robotic arm. At the same time, the manipulator has high precision and is easy to control. Combined with the visual information of the camera, it can greatly avoid damage to personnel and products and has reliable security.

[0050] (6) Strong expandability

[0051] This application can achieve more extensive functions by replacing the gripper of the manipulator and has strong expandability.

[0052] Other features and advantages of the present application will be described in the following specification, and will be partly obvious from the specification, or will be understood by implementing the present application. The objectives and other advantages of the present application can be achieved and obtained through the structures and processes pointed out in the specification and the accompanying drawings. Description of the Drawings

[0053] To more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following briefly introduces the accompanying drawings required for use in the description of the embodiments or the related art. Obviously, the accompanying drawings in the following description are certain embodiments of the present application. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 is a schematic flowchart of a hard disk hot plug test method according to an exemplary embodiment of the present application.

[0055] Figure 2 is a structural block diagram of a hard disk hot plug test system according to an exemplary embodiment of the present application. Detailed Embodiments

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0057] As Figure 1 shown, exemplarily, this embodiment provides a hard disk hot plug test method based on a manipulator, including:

[0058] S1. Based on the test tasks issued by the test management platform, the test server sequentially executes the test cases;

[0059] S2. Based on the current operation requests related to the test tasks issued by the test management platform, the management server issues operation instructions to the manipulator subsystem;

[0060] S3. Based on the received operation instructions, the PLC industrial control computer in the manipulator subsystem controls the camera to collect the environmental visual information of the hard disk array cabinet and transmits it to the trained deep learning model in the management server to obtain a judgment result characterizing the hard disk condition;

[0061] S4. The PLC industrial control computer in the manipulator subsystem controls the manipulator to perform hot plugging operations on the hard disks in the hard disk array cabinet connected to the test server based on the received operation instructions and judgment results;

[0062] S5. Repeat steps S2 - S4, and feedback the test results to the test management platform until the test task is completed.

[0063] This method inputs the environmental visual information of the hard disk array cabinet collected by the camera in real time into the deep learning model trained on the management server for image recognition, and can accurately obtain the judgment result of the hard disk status; the PLC industrial control computer controls the manipulator to perform hot plugging operations on the hard disks in the hard disk array cabinet connected to the test server based on the received operation instructions and judgment results. The manipulator performs hot plugging operations on behalf of the manual according to the instructions, with high operation accuracy, and is more efficient and fast. This method is easy to control the manipulator, and combined with the visual information of the camera, it can greatly avoid damage to personnel and products, has reliable safety. The test management platform can provide various test samples to ensure that the test results are more universal.

[0064] In some embodiments, the management server transmits the received environmental visual information to the cloud server, and the cloud server trains the initial deep learning model based on the historical environmental visual information collected by the camera to obtain a trained deep learning model. This training includes steps such as image calibration and classification. Specifically, the cloud server is used to train the initial deep learning model based on the historical environmental visual information collected by the camera until convergence to obtain a trained deep learning model. The trained deep learning model can output the judgment result of the hard disk status of the hard disk array cabinet based on the environmental visual information of the hard disk array cabinet collected by the camera. The judgment results include: hard disk not in place, hard disk in place and the buckle open, or hard disk in place and the buckle closed. The trained deep learning model is deployed on the management server. After the management server receives the environmental visual information transmitted by the PLC industrial control computer, it inputs the environmental visual information into the trained deep learning model to obtain the judgment result of the hard disk status of the hard disk array cabinet, and transmits the judgment result to the test management platform. The test management platform issues an operation request to the management server based on the judgment result.

[0065] Exemplarily, the initial deep learning model is an image recognition model, specifically, the YOLO V5 model can be adopted. Training the initial deep learning model based on the historical environmental visual information collected by the camera until convergence specifically includes: controlling the robotic arm to take 150 pictures at each calibration point of the hard disk array cabinet as the data set, and taking the first 100 pictures as the pre-training data set (manually annotating and classifying the pre-training data set, that is, the hard disk is in place with the buckle open, the hard disk is in place with the buckle closed, and the hard disk is not in place). YOLO V5 performs preprocessing, sampling, feature extraction and other processes on the pictures in the pre-training data set to generate a training model. After the model training is completed, the last 50 pictures are used as the test set for verification. For example, for an input image in the test set, the trained model outputs a set of probability values (such as: the hard disk is in place with the buckle open 20%, the hard disk is in place with the buckle closed 70%, the hard disk is not in place 10%), and the category corresponding to the highest probability is taken as the judgment result (that is, the result is that the hard disk is in place with the buckle closed). The trained YOLO V5 model is trained multiple times by continuously fine-tuning the parameters and replacing the data set until the loss function is minimized to obtain the final training model. When the model output result reaches 50 times without error between the real judgment and the manual judgment, it is regarded as the end of the model training, and a trained model (that is, a trained deep learning model) is obtained.

[0066] In some embodiments, after the current operation request related to the test task issued by the test management platform in step S2, the following steps are further included:

[0067] S201, the test management platform determines whether the current test case in the test task includes a hard disk plugging and unplugging step. If it includes, it determines whether the robotic arm has completed the initialization operation (that is, whether the robotic arm has entered the working state);

[0068] S202, if it is determined that the robotic arm has not completed the initialization operation, the robotic arm initialization process is carried out. After the process ends, it is determined whether the initialization is successful. If it fails, it is determined whether the number of times of the robotic arm initialization failure is greater than the preset number of times; if it is greater than the preset number of times, the test management platform aborts the test; if it is less than or equal to the preset number of times, the robotic arm initialization process continues; optionally, the preset number of times can be set to 3 times;

[0069] S203, if it is determined that the robotic arm has completed the initialization operation, it is determined whether the current test step is a hot plugging and unplugging step of the hard disk. If so, the test management platform sends a current hard disk hot plugging and unplugging operation request to the management server; the management server sends an operation instruction to the PLC industrial control computer, and the industrial control computer cooperates with the robotic arm to perform the hot plugging and unplugging operation.

[0070] After the current test case in the test task includes the hard disk plugging and unplugging steps, the manipulator that is determined to have not completed the initialization operation is initialized, avoiding the hot plugging operation failure caused by the manipulator problem and ensuring the reliability of each step of the manipulator for hot plugging operation.

[0071] In some embodiments, before step S1, based on the specifications of the hard disk array cabinet connected to the test server, the position serial numbers of each hard disk in the hard disk array cabinet are manually marked, and then, the spatial position coordinates of the manipulator corresponding to each position serial number are calibrated for the manipulator.

[0072] By marking the position serial numbers of each hard disk and then calibrating the spatial position coordinates of the corresponding manipulator based on each position serial number, when it is necessary to perform a hot plugging test on the hard disk, only the position serial number of the hard disk needs to be input, which is convenient for the staff to operate.

[0073] In some embodiments, based on the received operation instruction, the PLC industrial control computer in the manipulator subsystem controls the camera to collect the environmental visual information of the hard disk array cabinet and transmits it to the trained deep learning model in the management server to obtain the judgment result representing the hard disk condition, specifically including:

[0074] S301, input the position serial number of the hard disk to be hot plugged, and determine whether the position serial number is valid. If it is invalid, an error is reported and the test process is ended; if it is valid, continue to the next step:

[0075] S302, the manipulator reaches the spatial coordinate point corresponding to the position serial number according to the pre-set motion trajectory, adjusts the manipulator posture and controls the camera at the end of the manipulator to collect the environmental visual information of the hard disk array cabinet and transmits it back to the management server;

[0076] S303, the management server inputs the environmental visual information into the trained deep learning training model to obtain the judgment result representing the hard disk condition. Specifically, the judgment result includes: the hard disk is not in place, the hard disk is in place and the buckle is open, or the hard disk is in place and the buckle is closed.

[0077] By inputting the environmental visual information of the hard disk array cabinet collected by the camera in real time into the trained deep learning model on the management server for image recognition, accurate hard disk condition information can be obtained, providing a reliable judgment basis for controlling the movement of the robotic arm to help the staff adjust the hot plugging test process in real time.

[0078] In some embodiments, the management server formats the judgment result according to the TCP protocol and transmits it to the test management platform.

[0079] In some embodiments, the PLC industrial control computer in the manipulator subsystem in step S4 controls the manipulator to perform hot plugging and unplugging operations on the hard disks in the hard disk array cabinet connected to the test server based on the received operation instructions and judgment results, including: if the judgment result is that the hard disk is not in place, an error is reported and the test process ends; if the judgment result is that the hard disk is in place and the buckle is open, the industrial control computer cooperates with the manipulator to close the buckle; if the judgment result is that the hard disk is in place and the buckle is closed, the manipulator completes the hot plugging and unplugging operation of the hard disk according to the pre-set movement trajectory and posture.

[0080] In some embodiments, if the judgment result is that the hard disk is in place and the buckle is open, the industrial control computer cooperates with the manipulator to close the buckle, including: determining whether the number of operations of the manipulator to close the buckle is greater than a preset threshold. If it is greater than the preset threshold, an error is reported and the test process ends; if it is not greater than the preset threshold, the manipulator reaches the spatial coordinate point corresponding to the position number according to the pre-set movement trajectory, adjusts the posture of the manipulator, and controls the camera at the end of the manipulator to collect the environmental visual information of the hard disk array cabinet and transmit it back to the management server; the management server inputs the environmental visual information into the trained deep learning training model to obtain the judgment result representing the hard disk condition.

[0081] Exemplarily, the preset threshold is set to 3. Determine whether the number of operations of the control computer to cooperate with the manipulator to close the buckle is greater than 3. If it is greater than 3, an error is reported and the test process ends; if it is not greater than 3, the manipulator reaches the spatial coordinate point corresponding to the position number according to the pre-set movement trajectory, adjusts the posture of the manipulator, and controls the camera at the end of the manipulator to collect the environmental visual information of the hard disk array cabinet and transmit it back to the management server.

[0082] Further perform different operations according to different judgment results, and report errors when the process is incorrect, which can facilitate tracing when a fault occurs during the test process and improve the test efficiency.

[0083] In some embodiments, a ROS system is provided in the PLC industrial computer, and the PLC industrial computer provides an interactive interface between the ROS system and the management server. The ROS system provides gain position compensation for the positioning of the manipulator based on the visual information collected by the camera to ensure that the manipulator will not drift at the fixed coordinates to ensure positioning accuracy, making the operation more stable; and can implement suitable and feasible solutions according to different production environments, different server models, different fixed positions and different hard disk boxes (mainly affecting the plug-in and unplug methods), retain interfaces that are easy to develop, have strong ease of use, are compatible with a variety of user scenarios, and are convenient for testing products of different specifications. The visual information collected by the camera is fed back to the PLC industrial computer as gain compensation, which can avoid the drift of the manipulator when it is fixed, making the operation of the manipulator more stable and more accurate. In addition, by dynamically adjusting the spatial coordinate axis of the manipulator's end joint, the accuracy of the force control sensor, the y-axis offset and other parameters, the subtle differences in the actual plug-in and unplug operations of different testers (such as plug-in and unplug force, plug-in and unplug interval time, etc.) can be reproduced, making the plug-in and unplug hard disk operations more realistic.

[0084] In some embodiments, based on the log service provided by the ROS system, each posture of the robot is traced according to the time point, the controllability of the hard disk hot-swap operation is guaranteed, and the robot can be used and stopped at any time.

[0085] In some embodiments, the robot is a six-axis freedom robot.

[0086] In some embodiments, an nginx server is also deployed on the management server, which is a bridge for data interaction between the host computer and the ROS system. The host computer can be implemented by python code.

[0087] In some embodiments, data is transmitted between the ROS system and the management server via modbusTCP (an Ethernet-based communication protocol), enabling real-time interaction between the two.

[0088] In some embodiments, the management server is also remotely connected to a host computer, which is used to remotely control the posture of the robot. The host computer is implemented by python code, and an API interface is set on the host computer. The setting of the API interface facilitates calling the code in the test process. By modifying the code that controls the running trajectory of the robot in the ROS system and modifying the control program code in the host computer, a variety of solutions can be generated to adapt to hard disk cabinets of different specifications.

[0089] In some embodiments, the management server pre-processes the received environmental visual information before transmitting the received environmental visual information to the cloud server, and transmits the pre-processed environmental visual information to the cloud server.

[0090] Accordingly, if Figure 2As shown, the present application exemplarily further provides a hard disk hot plug test system based on a manipulator, including: a management server, a test management platform, a manipulator subsystem, and a test server; a hard disk array cabinet is connected to the test server to form a storage system; the manipulator subsystem includes a manipulator, a PLC industrial computer, and a camera, the camera is arranged at the end of the manipulator, and a ROS system is set in the PLC industrial computer, and the PLC industrial computer provides an interaction interface between the ROS system and the management server;

[0091] The test management platform is used to send test tasks to the test server and send operation requests related to the test tasks to the management server;

[0092] The management server sends operation instructions to the manipulator in the manipulator subsystem based on the received operation requests, and the management server is also used to receive the environmental visual information transmitted by the PLC industrial computer in the manipulator subsystem;

[0093] The PLC industrial computer in the manipulator subsystem controls the manipulator to perform hot plug operations on the hard disks of the hard disk array cabinet based on the received operation instructions, and controls the camera to collect the environmental visual information of the hard disk array cabinet and transmit it to the management server;

[0094] The test server executes test cases in sequence based on the sent test tasks and feeds back the test results to the test management platform.

[0095] In some embodiments, the hard disk hot plug test system further includes a cloud server. The management server transmits the received historical environmental visual information to the cloud server. The cloud server trains an initial deep learning model based on the historical environmental visual information collected by the camera to obtain a trained deep learning model. This training includes steps such as image calibration and classification. Specifically, the cloud server is used to train the initial deep learning model based on the historical environmental visual information collected by the camera until convergence to obtain a trained deep learning model. The trained deep learning model can output a judgment result of the hard disk status of the hard disk array cabinet based on the environmental visual information of the hard disk array cabinet collected by the camera. The trained deep learning model is deployed on the management server. After the management server receives the environmental visual information transmitted by the PLC industrial computer, it inputs the environmental visual information into the trained deep learning model to obtain a judgment result of the hard disk status of the hard disk array cabinet and transmits the judgment result to the test management platform. The test management platform sends an operation request to the management server based on the judgment result.

[0096] Exemplarily, the initial deep learning model is an image recognition model, specifically the YOLO V5 model. Training the initial deep learning model based on the historical environmental visual information collected by the camera until convergence is specifically as follows: Control the robotic arm to take 150 pictures at each calibration point of the hard disk array cabinet as the data set, and take the first 100 pictures as the pre-training data set (manually annotate and classify the pre-training data set, that is, the hard disk is in place with the buckle open, the hard disk is in place with the buckle closed, and the hard disk is not in place). YOLO V5 performs preprocessing, sampling, feature extraction and other processes on the pictures in the pre-training data set to generate a training model. After the model training is completed, the last 50 pictures are used as the test set for verification. For example, for an input image in the test set, the trained model outputs a set of probability values (such as: the hard disk is in place with the buckle open 20%, the hard disk is in place with the buckle closed 70%, the hard disk is not in place 10%), and the category corresponding to the highest probability is taken as the judgment result (that is, the result is that the hard disk is in place with the buckle closed). The trained YOLO V5 model is trained multiple times by continuously fine-tuning the parameters and replacing the data set until the loss function is minimized to obtain the final training model. When the model output result reaches 50 times without error between the real judgment and the manual judgment, it is regarded as the end of the model training, and a trained model is obtained (that is, a trained deep learning model).

[0097] In some embodiments, the hard disk hot plug test system further includes a host computer, which is remotely connected to the management server. The host computer is used to remotely control the posture of the robotic arm; the host computer is implemented by python code, and an API interface is set on the host computer. The setting of the API interface facilitates the calling of the code in the test process. By modifying the code for controlling the running trajectory of the robotic arm in the ROS system and modifying the control program code in the host computer, various solutions can be generated to adapt to different specifications of hard disk cabinets.

[0098] The above system can implement the hard disk hot plug test method provided in the above embodiments. For the specific system, reference can be made to the specific description of the hard disk hot plug test method in the above embodiments, which will not be elaborated here.

[0099] It can be understood that the circuit structures, names and parameters described in the above embodiments are only examples. Those skilled in the art can also easily combine and adjust the structural features of the above multiple embodiments according to the usage needs, and should not limit the concept of the present application to the specific details of the above examples.

[0100] Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A hard disk hot-swap test method based on a manipulator, characterized in that: include: S1, based on the test tasks issued by the test management platform, the test server executes the test cases in sequence; S2, based on the current operation request related to the test task issued by the test management platform, the management server issues an operation instruction to the manipulator subsystem; S3, based on the received operation instruction, the PLC industrial computer in the manipulator subsystem controls the camera to collect environmental visual information of the hard disk array cabinet connected to the test server and transmits it to the trained deep learning model in the management server to obtain a judgment result representing the hard disk status; S4, the PLC industrial computer in the manipulator subsystem controls the manipulator to perform a hot-swap operation on the hard disk in the hard disk array cabinet connected to the test server based on the received operation instruction and the judgment result; S5, repeat steps S2-S4 and feed back the test results to the test management platform until the test task is completed.

2. The hard disk hot-swap test method according to claim 1, characterized in that: The current operation request related to the test task issued by the test management platform also includes: S201, the test management platform determines whether the current test case in the test task includes a hard disk plugging and unplugging step, and if so, determines whether the manipulator has completed the initialization operation; S202, if it is determined that the manipulator has not completed the initialization operation, the manipulator initialization process is performed, and after the initialization process is completed, it is determined whether the initialization is successful. If it fails, it is determined whether the number of initialization failures of the manipulator is greater than the preset number; if it is greater than the preset number, the test management platform terminates the test; if it is less than or equal to the preset number, the manipulator initialization process continues; S203, if it is determined that the manipulator has completed the initialization operation, then determine whether the current test step is a hard disk hot-swap step. If so, the test management platform initiates a current hard disk hot-swap operation request to the management server; the management server sends an operation instruction to the PLC industrial computer, and the industrial computer cooperates with the manipulator to perform a hot-swap operation.

3. The hard disk hot-swap test method according to claim 1, characterized in that: Before step S1, the position number of each hard disk is marked based on the specification of the hard disk array cabinet connected to the test server, and then the spatial position coordinates of the manipulator corresponding to each position number are calibrated.

4. The hard disk hot-swap test method according to claim 1, characterized in that: Based on the received operation instructions, the PLC industrial computer in the manipulator subsystem controls the camera to collect environmental visual information of the hard disk array cabinet connected to the test server and transmits it to the trained deep learning model in the management server to obtain a judgment result representing the hard disk status, specifically including: S301, input the position number of the hard disk to be hot-swapped, and determine whether the position number is valid. If it is invalid, an error is reported and the test process ends; if it is valid, proceed to the next step: S302, the manipulator arrives at the spatial coordinate point corresponding to the position number according to the pre-set motion trajectory, adjusts the manipulator posture and controls the camera at the end of the manipulator to collect environmental visual information of the hard disk array cabinet and transmits it back to the management server; S303, the management server inputs the environmental visual information into the trained deep learning training model to obtain a judgment result representing the hard disk status.

5. The hard disk hot-swap test method according to claim 1, characterized in that: The judgment results include: the hard disk is not in place, the hard disk is in place and the buckle is open, or the hard disk is in place and the buckle is closed.

6. The hard disk hot-swap test method according to claim 5, characterized in that: The PLC industrial computer in the manipulator subsystem controls the manipulator to perform a hot-swap operation on the hard disk in the hard disk array cabinet connected to the test server based on the received operation instruction and the judgment result, including: If the judgment result is that the hard disk is not in place, an error is reported and the test process ends; if the judgment result is that the hard disk is in place and the buckle is open, the industrial computer cooperates with the manipulator to close the buckle; if the judgment result is that the hard disk is in place and the buckle is closed, the manipulator completes the hard disk hot swap operation according to a pre-set motion trajectory and posture.

7. The hard disk hot-swap test method according to claim 6, characterized in that: If the judgment result is that the hard disk is in place and the buckle is open, the industrial computer cooperates with the manipulator to close the buckle including: Determine whether the number of times the robot closes the buckle is greater than the preset threshold. If so, an error is reported and the test process ends. If not, the robot arrives at the spatial coordinate point corresponding to the position number according to the preset motion trajectory, adjusts the robot posture, and controls the camera at the end of the robot to collect environmental visual information of the hard disk array cabinet and transmit it back to the management server.

8. The hard disk hot-swap test method according to claim 1, characterized in that: The PLC industrial computer is provided with a ROS system, and the ROS system provides gain position compensation for the positioning of the manipulator based on the visual information collected by the camera.

9. A hard disk hot-swap test system based on a manipulator, characterized in that: Used to implement the hard disk hot-swap test method according to any one of claims 1 to 8, the hard disk hot-swap test system comprises: a management server, a test management platform, a manipulator subsystem and a test server; the hard disk array cabinet is connected to the test server to form a storage system; the manipulator subsystem comprises a manipulator, a PLC industrial computer and a camera, and the camera is arranged at the end of the manipulator; The test management platform is used to send test tasks to the test server, and send operation requests related to the test tasks to the management server; The management server sends an operation instruction to the manipulator in the manipulator subsystem based on the received operation request, and the management server is also used to receive environmental visual information transmitted by the PLC industrial computer in the manipulator subsystem; The PLC industrial computer in the manipulator subsystem controls the manipulator to perform hot-swap operations on the hard disks of the hard disk array cabinet based on the received operation instructions, and controls the camera to collect environmental visual information of the hard disk array cabinet and transmit it to the management server; The test server executes the test cases in sequence based on the issued test tasks, and feeds back the test results to the test management platform.

10. The hard disk hot-swap test system based on a manipulator according to claim 9, characterized in that: It also includes a cloud server. The management server transmits the received historical environmental visual information to the cloud server. The cloud server trains the initial deep learning model based on the historical environmental visual information collected by the camera to obtain a trained deep learning model.

Images