Mechanical arm photovoltaic module feeding and discharging method and system

Abstract

This invention relates to the technical field of loading and unloading photovoltaic modules with robotic arms, specifically to a method and system for loading and unloading photovoltaic modules with robotic arms. The method includes the following steps: acquiring torque signals at the end effector or joints of the robotic arm using a torque sensor to obtain ideal motion response data; acquiring torque signals at the end effector or joints of the robotic arm in real time using a torque sensor to obtain actual torque data; aligning the actual torque data with the ideal motion response data in time and subtracting them point-by-point to obtain torque difference signals; identifying fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion commands; determining whether the photovoltaic module has undergone planar rotation relative to the support portion of the robotic arm; and if planar rotation of the photovoltaic module is determined, triggering adjustments to the robotic arm's motion parameters. This allows for timely adjustment of the robotic arm's motion parameters.

Description

Technical Field

[0001] This invention relates to the technical field of loading and unloading photovoltaic modules with robotic arms, and specifically to a method and system for loading and unloading photovoltaic modules with robotic arms. Background Technology

[0002] In the production process of photovoltaic modules, heavy-duty robotic arms are often used to remove freshly laminated double-glass modules from the lamination table and precisely place them into the slots of the cooling buffer rack. Double-glass modules consist of two layers of tempered glass bonded together with an intermediate encapsulation film. They have high overall rigidity but their edges are sensitive to impact. After lamination, a small amount of encapsulation film often overflows from the module edges. This overflow adhesive is in a high-viscosity, semi-fluid state before it has fully cooled, with a temperature typically between 50 and 70 degrees Celsius. If the outer suction cups of the robotic arm's end-effector press against these areas of residual adhesive, the contact interface changes from traditional dry friction between rubber and glass to a hybrid friction layer of rubber and semi-fluid glass. This state results in a significantly reduced shear resistance under tangential force.

[0003] In existing technologies, robotic arm end effectors typically employ distributed multi-suction cup arrays to distribute the load and ensure stable vacuum adsorption. However, when the vacuum circuit exhibits a normal negative pressure value and remains sealed, the adsorption detection system uses negative pressure stability as the criterion for successful gripping, making it difficult to detect the deterioration of interfacial friction properties. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned shortcomings by proposing a robotic arm method and system for loading and unloading photovoltaic modules.

[0005] The present invention adopts the following technical solution:

[0006] A method for loading and unloading photovoltaic modules using a robotic arm, the method comprising the following steps:

[0007] With the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, the robotic arm moves according to the preset transport trajectory, and the torque sensor collects the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array.

[0008] When the robotic arm actually grasps the photovoltaic module and performs the transport action, the torque sensor is used to collect the torque signal at the end of the robotic arm or at the joint in real time to obtain the actual torque data.

[0009] The actual torque data and the ideal motion response data are time-aligned and subtracted point by point to obtain the torque difference signal;

[0010] The torque difference signal is analyzed to identify fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command;

[0011] Based on the fluctuation characteristics, determine whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm;

[0012] If the photovoltaic module is detected to be rotating in a plane, the motion parameters of the robotic arm will be adjusted.

[0013] This technical solution enables real-time and accurate determination of whether photovoltaic modules are rotating in a plane by analyzing the fluctuation characteristics of torque difference signals. This allows for timely adjustment of the robotic arm's motion parameters, effectively preventing collisions between the modules and the cooling rack. It also solves the problem in existing adsorption detection systems where it is difficult to detect slippage caused by deterioration of interface friction properties.

[0014] This application also discloses a robotic arm photovoltaic module loading and unloading system, applied to the above-mentioned robotic arm photovoltaic module loading and unloading method, the system comprising:

[0015] The acquisition module, with the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, makes the robotic arm move according to the preset transport trajectory, and uses a torque sensor to collect the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array.

[0016] The data acquisition module uses a torque sensor to collect torque signals at the end of the robotic arm or at the joints in real time when the robotic arm actually grasps the photovoltaic module and performs the handling action, so as to obtain the actual torque data.

[0017] The calculation module aligns the actual torque data with the ideal motion response data in time and subtracts them point by point to obtain the torque difference signal;

[0018] The analysis module analyzes the torque difference signal and identifies fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command.

[0019] The judgment module determines whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm based on the fluctuation characteristics.

[0020] If the processing module determines that the photovoltaic module is rotating in a plane, it triggers the adjustment of the motion parameters of the robotic arm.

[0021] This technical solution provides a system for implementing the aforementioned robotic arm method for loading and unloading photovoltaic modules. Through its modular design, the system can efficiently and accurately perform loading and unloading operations on photovoltaic modules, and promptly detect and address planar rotation issues of the modules, thereby improving production efficiency and safety.

[0022] This application, by monitoring the fluctuation characteristics of torque signals in real time, can sensitively detect minute slippage caused by insufficient tangential friction between the suction cup and the module. Even when the vacuum circuit and adsorption detection display are normal, it can accurately determine the relative attitude drift of the module. This real-time slippage detection mechanism based on torque signals overcomes the limitation of existing adsorption detection systems in detecting the degradation of interfacial friction properties, and avoids the risk of rigid collision between the module's corners and the cooling rack guide groove due to slight rotation of the module causing its edges to deviate from the expected geometric position. Therefore, this application significantly improves the safety, stability, and production efficiency of the photovoltaic module loading and unloading process.

[0023] To further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and illustration only and are not intended to limit the present invention. Attached Figure Description

[0024] Figure 1 This is a flowchart of a robotic arm method for loading and unloading photovoltaic modules according to the present invention.

[0025] Figure 2 This is a schematic diagram of the structure of a robotic arm photovoltaic module loading and unloading system according to the present invention. Detailed Implementation

[0026] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Furthermore, the accompanying drawings of the present invention are for simple illustrative purposes only and are not depictions of actual dimensions; this is stated in advance. The following embodiments will further describe the relevant technical content of the present invention in detail, but the disclosed content is not intended to limit the scope of protection of the present invention.

[0027] This embodiment provides a method and system for loading and unloading photovoltaic modules with a robotic arm, combined with... Figure 1 and Figure 2 As shown.

[0028] refer to Figure 1 A method for loading and unloading photovoltaic modules with a robotic arm, the method comprising the following steps:

[0029] With the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, the robotic arm moves according to the preset transport trajectory, and the torque sensor collects the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array.

[0030] When the robotic arm actually grasps the photovoltaic module and performs the transport action, the torque sensor is used to collect the torque signal at the end of the robotic arm or at the joint in real time to obtain the actual torque data.

[0031] The actual torque data and the ideal motion response data are time-aligned and subtracted point by point to obtain the torque difference signal;

[0032] The torque difference signal is analyzed to identify fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command;

[0033] Based on the fluctuation characteristics, determine whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm;

[0034] If the photovoltaic module is detected to be rotating in a plane, the motion parameters of the robotic arm will be adjusted.

[0035] In this context, a robotic arm refers to a multi-jointed, programmable automated device used to perform operations such as grasping and handling. The support unit is the end effector of the robotic arm used to grasp and support photovoltaic modules; in this application, it specifically refers to a suction cup structure arranged in an array to ensure stable adhesion to the photovoltaic modules. A torque sensor is a device capable of measuring rotational force or torque; in this application, it is used to collect torque signals at the end effector or joints of the robotic arm to reflect the force state of the robotic arm during movement. Ideal motion response data refers to the torque signal data generated when the robotic arm moves along a preset handling trajectory under ideal conditions where there is no relative sliding connection between the robotic arm and the photovoltaic module. Actual torque data refers to the torque signal data collected in real time by the torque sensor when the robotic arm actually grasps the photovoltaic module and performs handling actions. The torque difference signal is the signal obtained by time alignment and point-by-point subtraction between the actual torque data and the ideal motion response data; it reflects the deviation between the actual motion and the ideal motion. Fluctuation characteristics refer to the periodic changes in the torque difference signal with a specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion commands. These fluctuation characteristics are key evidence for determining whether photovoltaic modules have undergone planar rotation. Planar rotation refers to the rotation of the photovoltaic module relative to the robotic arm's support portion on a horizontal plane after it has been grasped by the robotic arm. Motion parameter adjustment refers to modifying parameters such as the robotic arm's speed, acceleration, and trajectory based on the judgment results to correct module attitude deviations or avoid potential collisions.

[0036] The core of the robotic arm photovoltaic module loading and unloading method proposed in this application lies in achieving real-time and accurate judgment of the planar rotation of the photovoltaic module through refined analysis of torque signals.

[0037] Specifically, with the robotic arm's support section maintaining a non-slip connection with the photovoltaic module, the robotic arm moves along a preset transport trajectory. Torque sensors collect torque signals at the robotic arm's end effector or joints to obtain ideal motion response data. The support section of the robotic arm is a suction cup structure located at the end effector, arranged in an array. For example, a standard photovoltaic module can be pre-securely fixed to the suction cup structure of the robotic arm, ensuring no relative slippage between them. Then, the robotic arm performs multiple no-load or standard-load movements according to the transport trajectory used in actual production (e.g., from the laminating table to the cooling rack). During this process, torque sensors installed at the robotic arm's joints or end effector continuously collect torque signals in various axes. These collected torque signals, after averaging and filtering, become the ideal motion response data. This data represents the torque performance that the robotic arm should exhibit when executing a specific transport trajectory under ideal, slip-free conditions.

[0038] When the robotic arm actually grasps and transports photovoltaic modules, torque sensors are used to collect torque signals at the end effector or joints of the robotic arm in real time to obtain actual torque data. For example, when the robotic arm grasps a newly laminated photovoltaic module from the lamination table, the torque sensor continuously monitors the torque changes of the robotic arm during the transport process. This real-time collected torque data contains all the force information of the robotic arm in the actual working environment, including the weight of the module, inertial forces, and additional torque caused by possible slippage.

[0039] The torque difference signal is obtained by aligning the actual torque data with the ideal motion response data in time and subtracting them point by point. For example, since there may be slight time deviations between the actual handling process and the preset ideal trajectory, it is necessary to first synchronize the two sets of torque data to ensure that the data at each time point corresponds accurately. Time alignment can be achieved through cross-correlation algorithms or dynamic time warping algorithms in signal processing. After alignment, the value of the actual torque data at each time point is subtracted from the value of the ideal motion response data at the corresponding time point to obtain a torque difference signal. This difference signal can effectively filter out the normal torque generated by the robot arm's own movement and highlight the torque changes caused by abnormal conditions such as component slippage.

[0040] Analyzing the torque difference signal identifies fluctuations with specific frequencies and amplitudes that exhibit a phase delay relationship with the robotic arm's motion commands. For example, when a photovoltaic module rotates in a plane, its rotational motion often lags behind the robotic arm's tangential or torsional motion commands due to the module's inertia, resulting in specific periodic fluctuations in the torque difference signal. These fluctuations may have specific frequency components related to the robotic arm's motion frequency, and their amplitude increases with the degree of slippage. These fluctuations with specific frequencies and amplitudes can be extracted from the torque difference signal using spectral analysis methods such as Fourier transform and wavelet analysis. Furthermore, analyzing the phase relationship between these fluctuation features and the robotic arm's motion commands further confirms whether these fluctuations are caused by the relative rotation of the module.

[0041] Based on the wave characteristics, it can be determined whether the photovoltaic module has undergone planar rotation relative to the support of the robotic arm. For example, a threshold can be preset; when the frequency, amplitude, or phase delay relationship of the identified wave characteristics meets specific conditions and exceeds the preset threshold, it is determined that the photovoltaic module has undergone planar rotation. This threshold can be trained and optimized through a large amount of experimental data to ensure the accuracy and robustness of the judgment.

[0042] If the system detects that the photovoltaic module is rotating in a plane, it triggers an adjustment of the robotic arm's motion parameters. For example, once the system detects that the photovoltaic module is rotating in a plane, the robotic arm's control system will immediately receive a trigger signal. At this time, the robotic arm's motion parameters can be automatically adjusted according to the degree and direction of rotation, such as reducing the handling speed, decreasing the acceleration, or fine-tuning the robotic arm's posture to correct the module's deviation and prevent it from colliding with the cooling rack guide structure during subsequent handling.

[0043] The proposed robotic arm method for loading and unloading photovoltaic modules utilizes torque sensors to collect and analyze torque signals at the end effector or joints of the robotic arm. This enables real-time and accurate identification of planar rotation of the photovoltaic module relative to the robotic arm's support structure. Its working principle lies in the fact that, under ideal, slip-free conditions, the robotic arm's motion response torque exhibits a predictable pattern—the ideal motion response data. When the photovoltaic module undergoes planar rotation during actual handling, due to the module's inertia and changes in friction between the suction cup and the module, additional torques, differing from the ideal state, are generated at the end effector or joints of the robotic arm. These additional torques are reflected in the actual torque data. By aligning the actual torque data with the ideal motion response data over time and subtracting them point-by-point, a torque difference signal can be obtained. This signal effectively filters out the conventional torques generated by the robotic arm's own movement, thus highlighting the abnormal torques caused by module slippage.

[0044] Furthermore, the torque difference signal is analyzed to identify fluctuation characteristics with specific frequencies and amplitudes that have a phase delay relationship with the robotic arm's motion commands. These fluctuation characteristics are typical manifestations of planar rotation of the photovoltaic module. For example, when the robotic arm accelerates or decelerates, if the module slips, its inertia will cause the actual motion of the module to lag behind the robotic arm's commanded motion. This lag will manifest as periodic fluctuations in the torque signal that have a phase delay relationship with the robotic arm's motion commands. By analyzing the frequency and amplitude of these fluctuation characteristics, the degree and nature of the module slippage can be quantified. Once these fluctuation characteristics are identified and it is determined that the photovoltaic module has undergone planar rotation, the system will immediately trigger adjustments to the robotic arm's motion parameters. For example, the robotic arm's handling speed can be reduced, acceleration decreased, or the robotic arm's posture can be fine-tuned to correct the module's deviation, thereby preventing collisions when the module is placed into the cooling rack. The entire process forms a closed-loop control system, realizing real-time monitoring and proactive intervention of potential slippage risks during the handling of photovoltaic modules, significantly improving the stability and safety of the loading and unloading process.

[0045] This application further proposes a method for loading and unloading photovoltaic modules with a robotic arm, which includes the following steps:

[0046] After the position is locked as displayed by the encoder of the robotic arm motor, the motion information fed back by the servo driver is obtained;

[0047] Based on motion information, determine whether the robotic arm is swaying;

[0048] If the robotic arm is shaking, do not update the zero point or reduce the frequency of zero point updates;

[0049] Adjust the threshold and response strategy for slippage determination based on the reliability of the zero-point update.

[0050] Specifically, after the robotic arm motor encoder displays a locked position, it means that each joint motor of the robotic arm has reached its target position and remains stable. At this time, the position data fed back by the motor encoder should be in a stable state. In this state, motion information fed back by the servo driver is acquired. This motion information may include, but is not limited to, parameters such as motor current, voltage, speed fluctuations, and position errors. These parameters can reflect the subtle dynamic behavior of the robotic arm in a seemingly static state. Determining whether the robotic arm is swaying based on the motion information can be understood as analyzing the motion information fed back by the servo driver, such as monitoring the periodic fluctuations, frequency characteristics, or amplitude changes of the motor current or position error, to identify whether the robotic arm is in an unexpected vibration or unstable state. For example, when periodic fluctuations of the motor current or position error exceeding a preset threshold are detected within a certain frequency range, it can be determined that the robotic arm is swaying.

[0051] In practical applications, if the robotic arm wobbles, not updating the zero point or reducing the frequency of zero point updates aims to avoid zero-point calibration when the robotic arm is unstable. An unstable robotic arm state leads to inaccurate zero-point data, which in turn affects the accuracy of subsequent torque difference signals. Reducing the frequency of zero-point updates decreases the chance of introducing erroneous zero-point data during wobbling. Furthermore, the threshold for slip detection and the response strategy are adjusted based on the reliability of the zero-point update. The reliability of the zero-point update can be comprehensively evaluated based on factors such as the degree and duration of robotic arm wobbling and the frequency of zero-point updates. For example, when the reliability of the zero-point update is low, the threshold for slip detection can be appropriately increased to avoid misjudging torque fluctuations caused by robotic arm wobbling as planar rotation of the photovoltaic module. Simultaneously, the response strategy can be adjusted to a more conservative approach, such as adding an additional confirmation step or delaying the adjustment before triggering motion parameter adjustments.

[0052] The solution proposed in this application acquires motion information fed back by the servo driver after the position is locked by the encoder of the robotic arm motor, and determines whether the robotic arm is swaying, thereby enabling real-time monitoring of the robotic arm's stability. When swaying is detected, the error introduced by zero-point calibration under unstable conditions is effectively avoided by not updating the zero point or reducing the frequency of zero-point updates, ensuring the accuracy of the subsequent torque difference signal. Therefore, based on the reliability of the zero-point update, the threshold and response strategy for slip judgment are dynamically adjusted, allowing the slip judgment system to adapt to the actual working state of the robotic arm, reducing the risk of misjudgment caused by the robotic arm's own swaying, and thus improving the accuracy and robustness of photovoltaic module planar rotation detection.

[0053] In some preferred embodiments, it is assumed that after the robotic arm grasps the photovoltaic module, slight motor resonance or external environmental vibration causes minor, periodic wobbling of the robotic arm even after the motor encoder indicates position lock. In this case, the motor current or position error data fed back by the servo driver will show specific fluctuation characteristics. The system will determine that the robotic arm is wobbling based on this motion information. To avoid these wobblings affecting the accuracy of slip detection, the system will temporarily suspend zero-point updates or reduce the frequency of zero-point updates from once per second to once every five seconds. Simultaneously, due to the reduced reliability of zero-point updates, the system will dynamically increase the slip detection threshold, for example, increasing the torque difference signal threshold from 0.5 Nm to 0.8 Nm, and adjusting the response strategy so that the robotic arm's motion parameters are only triggered when the torque difference signal exceeds 0.8 Nm three times consecutively, rather than triggering it all at once. In this way, even with slight wobbling of the robotic arm, misjudging the photovoltaic module's planar rotation can be effectively avoided, thereby improving the stability and reliability of the entire loading and unloading process.

[0054] This application further proposes steps for adjusting the threshold and response strategy for slip determination based on the reliability of the zero-point update, including:

[0055] Obtain the adhesive overflow status parameters, module type information, and ambient temperature of the photovoltaic modules;

[0056] Based on the photovoltaic module's glue overflow state parameters, module type information, ambient temperature, and the reliability of zero-point updates, the threshold and response strategy for slippage determination are dynamically adjusted.

[0057] Based on the adjusted slip determination threshold and response strategy, determine whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm;

[0058] If it is determined that the photovoltaic module is rotating in a plane, the motion parameters of the robotic arm are adjusted.

[0059] Specifically, the adhesive overflow parameters of photovoltaic modules refer to the physical properties of the adhesive (e.g., EVA adhesive) remaining on the edge or backsheet of the photovoltaic module, such as its viscosity, distribution uniformity, and degree of curing. These parameters directly affect the actual contact area and friction coefficient between the suction cup structure and the photovoltaic module. Module type information can be understood as the inherent attributes of the photovoltaic module, such as its specifications, dimensions, weight, and surface material. Different types of modules may have different sensitivities to slippage during handling. Ambient temperature refers to the real-time temperature of the robotic arm's working area, which affects the adhesive viscosity and the physical properties of the module materials, thus influencing the likelihood and characteristics of slippage. Dynamically adjusting the slippage judgment threshold and response strategy refers to adjusting the critical value used to determine whether the photovoltaic module has undergone planar rotation and the subsequent countermeasures based on the various parameters obtained above, either in real-time or periodically. For example, when the adhesive overflow viscosity is high, a higher torque difference signal may be needed to determine slippage; when the module type is heavy, a more sensitive response strategy may be required.

[0060] This application's solution, by incorporating photovoltaic module overflow state parameters, module type information, and ambient temperature, can more comprehensively assess the actual contact and friction conditions between the robotic arm's support and the photovoltaic module. Traditional solutions rely solely on the reliability of zero-point updates to adjust slip judgment, failing to fully consider the complex influence of the photovoltaic module's own physical characteristics and its environment on slip behavior. For example, when there is uneven overflow of adhesive in the photovoltaic module, the contact between the suction cup structure and the module may be weak. In this case, even if the zero-point update is reliable, a stricter slip judgment threshold is required. Furthermore, the viscosity of the adhesive may increase at low temperatures and decrease at high temperatures, all of which affect the probability and characteristics of slip. By incorporating these key parameters, a more refined and accurate slip model can be constructed, allowing the slip judgment threshold and response strategy to adaptively adjust according to actual working conditions, thereby avoiding misjudgments or omissions due to insufficient consideration of a single factor.

[0061] In some preferred embodiments, the specific implementation is as follows:

[0062] First, after the robotic arm grasps the photovoltaic module, sensors integrated on the support unit acquire parameters related to the excess adhesive, such as average excess adhesive temperature and average excess adhesive viscosity. Simultaneously, the system obtains information on the type of photovoltaic module to be handled from the production management system and uses environmental sensors to monitor the ambient temperature of the work area in real time.

[0063] Suppose that at a certain moment, the system detects that the average viscosity of the photovoltaic module's adhesive overflow is low, the ambient temperature is high, and the module type is a large-size, heavy-duty module. In this case, even if the reliability of the zero-point update is high, the system will dynamically lower the torque difference threshold for slip detection based on this comprehensive information, and may choose a more aggressive response strategy (e.g., immediately decelerating or stopping upon detecting even slight signs of slippage) to improve the sensitivity of slip detection and ensure the safe handling of heavy-duty modules under high-temperature, low-viscosity conditions.

[0064] Conversely, if the detected adhesive viscosity is high, the ambient temperature is moderate, and the modules are of standard size, the system may appropriately increase the slippage detection threshold to avoid frequent adjustments due to oversensitivity, thereby ensuring handling efficiency. Through this dynamic adjustment mechanism, the robotic arm system can intelligently optimize the slippage detection logic according to changes in actual working conditions, thus achieving more precise and reliable photovoltaic module loading and unloading operations.

[0065] Specifically, the steps for obtaining the adhesive overflow state parameters of photovoltaic modules include:

[0066] Multiple miniature infrared temperature sensors and multiple miniature spectral sensors are distributed on the support part of the robotic arm;

[0067] After the robotic arm grasps the photovoltaic module, during the robotic arm's handling process, multiple miniature infrared temperature sensors are used to collect real-time overflow temperature data at different locations on the edge of the photovoltaic module, and multiple miniature spectral sensors are used to collect real-time overflow spectral data at different locations on the edge of the photovoltaic module.

[0068] Based on the overflow temperature data and overflow spectrum data, calculate the spatial distribution differences of the overflow temperature data and the spatial distribution differences of the overflow spectrum data.

[0069] The uniformity of the glue distribution is determined based on the spatial distribution differences of the glue temperature data and the spatial distribution differences of the glue spectrum data.

[0070] If it is determined that the glue overflow distribution is uneven, the average glue overflow temperature and average glue overflow viscosity in the contact area between the robotic arm's load-bearing part and the photovoltaic module are calculated based on the glue overflow temperature data and glue overflow spectrum data, and the average glue overflow temperature and average glue overflow viscosity are used as glue overflow state parameters of the photovoltaic module.

[0071] Among them, a miniature infrared temperature sensor is configured to non-contactly measure the surface temperature of adhesive overflow at the edge of the photovoltaic module. This can be achieved using technologies such as thermopile sensors or microbolometers to ensure measurement accuracy and response speed. A miniature spectral sensor is configured to analyze the spectral characteristics of the overflow adhesive to identify its chemical composition or physical state. This can be achieved using technologies based on filter arrays or miniature spectrometers to meet the needs of miniaturization and integration. The overflow temperature and spectral data are key information for assessing the overflow state, reflecting the physical and chemical characteristics of the adhesive overflow at the edge of the photovoltaic module. Spatial distribution difference refers to the difference between the overflow temperature and spectral data collected at different locations on the edge of the photovoltaic module. This difference reflects the uniformity of the overflow distribution at the module edge. The uniformity of the overflow distribution directly affects the contact stability and frictional characteristics between the robotic arm's support and the photovoltaic module. Uneven overflow distribution may lead to insufficient or excessive local adhesion, thereby increasing the risk of slippage. When the adhesive overflow distribution is uneven, a comprehensive adhesive overflow state parameter can be obtained by calculating the average adhesive overflow temperature and average adhesive overflow viscosity in the contact area between the robotic arm's support and the photovoltaic module. This parameter is used to more accurately assess the overall adhesion characteristics. The average adhesive overflow viscosity can be estimated by combining adhesive overflow spectral data with a preset viscosity-spectral model. Ultimately, the adhesive overflow state parameter, as a comprehensive parameter, characterizes the properties of adhesive overflow at the edge of the photovoltaic module, including its temperature, viscosity, and distribution uniformity. This provides an important basis for subsequently dynamically adjusting the threshold and response strategy for slip determination.

[0072] This application's solution achieves refined sensing of the adhesive overflow status at the edges of photovoltaic modules by distributing multiple miniature infrared temperature sensors and multiple miniature spectral sensors on the support section of the robotic arm. During the robotic arm's grasping and handling of the photovoltaic modules, these sensors can collect temperature and spectral data of the overflow adhesive in real time and at multiple points. By calculating the spatial distribution differences of this data, the uniformity of the adhesive overflow distribution at the module edges can be accurately determined. If the overflow distribution is uneven, the average overflow temperature and average overflow viscosity in the contact area between the robotic arm's support section and the photovoltaic module are further calculated. These detailed overflow status parameters, including distribution uniformity, average temperature, and average viscosity, provide more comprehensive and accurate input for subsequent dynamic adjustment of the slippage determination threshold and response strategy based on the reliability of zero-point updates, thereby improving the accuracy and reliability of slippage determination.

[0073] If it is determined that the photovoltaic module has undergone planar rotation, the steps for adjusting the motion parameters of the robotic arm include:

[0074] Obtain the component type information of photovoltaic modules;

[0075] Based on the photovoltaic module type information, query the preset module collision sensitivity data;

[0076] Based on the degree and direction of planar rotation, calculate the potential collision point location and contact angle of the photovoltaic module's corners relative to the cooling rack guide structure;

[0077] Assess the collision risk level of potential collision points based on component collision sensitivity data, potential collision point locations, and contact angles;

[0078] Select a strategy to adjust the robotic arm's motion parameters based on the collision risk level of potential collision points;

[0079] Adjust the motion parameters of the robotic arm according to the robotic arm motion parameter adjustment strategy.

[0080] Specifically, obtaining photovoltaic (PV) module type information refers to acquiring physical characteristic data such as the specific model, size, and weight of the PV modules currently being handled. Querying pre-defined module collision sensitivity data based on the PV module type information involves retrieving collision sensitivity data related to that module type from a pre-stored database. This data may include damage thresholds and vulnerable areas of the module under different impact forces or contact angles, aiming to provide a basis for subsequent collision risk assessment. In practical applications, calculating the potential collision point position and contact angle of the PV module's corner relative to the cooling rack guide structure based on the degree and direction of planar rotation involves predicting the corner position where the PV module might contact the cooling rack guide structure during continued movement, based on the detected degree and direction of planar rotation, combined with the current pose of the robotic arm, the geometric model of the PV module, and the geometric model of the cooling rack guide structure, and calculating the contact angle at that potential contact point. This aims to accurately identify the probability of a collision and its geometric conditions. Furthermore, assessing the collision risk level of potential collision points involves comprehensively considering the module collision sensitivity data, the calculated potential collision point position, and the contact angle. For example, if a potential collision point is located in a vulnerable area of ​​a component and the contact angle is unfavorable for cushioning, the collision risk level will be assessed as high, with the aim of quantifying the severity of the collision. Selecting a robotic arm motion parameter adjustment strategy involves choosing an appropriate strategy from a pre-defined strategy library based on the assessed collision risk level of the potential collision point. For example, for high-risk collision points, more conservative strategies such as deceleration, path alteration, or fine-tuning of the posture might be chosen. Therefore, adjusting the robotic arm's motion parameters according to the selected adjustment strategy means modifying parameters such as the robotic arm's speed, acceleration, path, or end effector posture in real time to mitigate potential collision risks.

[0081] This application's solution addresses the issue of potential collisions arising from simple adjustments to motion parameters by introducing a mechanism for assessing potential collision risks. Specifically, when it's determined that the photovoltaic module is undergoing planar rotation and adjustments to the robotic arm's motion parameters are necessary, the module type information is first obtained, and corresponding collision sensitivity data is retrieved, providing a foundation for subsequent risk assessment. Subsequently, based on the degree and direction of planar rotation, the potential collision point location and contact angle between the photovoltaic module's corners and the cooling rack guide structure are precisely calculated, thereby identifying specific risk areas and collision geometry. Based on this information, combined with the module's collision sensitivity data, the collision risk level of potential collision points is assessed, enabling the robotic arm to quantify potential hazards. Finally, based on the assessed risk level, a corresponding robotic arm motion parameter adjustment strategy is selected and executed. For example, if the risk level is high, a more significant path correction or deceleration operation may be implemented, effectively avoiding collisions between the photovoltaic module and the cooling rack guide structure while adjusting the robotic arm's motion parameters, ensuring the safety of the handling process.

[0082] In some preferred embodiments, it is assumed that a robotic arm is moving a large photovoltaic module onto a cooling rack. During the movement, a torque sensor detects a slight planar rotation of the photovoltaic module. At this point, the system first obtains the module type information, for example, "monocrystalline silicon module, 2000mm x 1000mm". Based on this information, the system determines that the corner areas of this type of module are more sensitive to lateral impacts, and its collision sensitivity data indicates that microcracks are easily formed under specific impact forces. Next, based on the detected degree of planar rotation (e.g., 2 degrees clockwise) and direction, and combined with the CAD model of the cooling rack guide structure, the system calculates that the lower right corner of the photovoltaic module may come into contact with a guide rod of the cooling rack within the next 0.5 seconds. The potential collision point is located 5mm inward from the module corner, with a contact angle of 45 degrees. Based on the module collision sensitivity data, the location of the potential collision point, and the contact angle, the system assesses the collision risk level of this potential collision point as "medium-high". Given this risk level, the system adopted an adjustment strategy: while maintaining the overall handling path unchanged, the posture of the robotic arm's end effector was fine-tuned, shifting it 2mm to the left horizontally, while simultaneously reducing the robotic arm's descent speed by 20%. Based on this adjustment strategy, the robotic arm's motion parameters were modified in real time, effectively avoiding potential collisions between the photovoltaic modules and the cooling rack guide rods, thus ensuring the safe placement of the photovoltaic modules.

[0083] This application further proposes a step for selecting a strategy for adjusting the motion parameters of a robotic arm, which includes:

[0084] Identify all potential collision points;

[0085] Calculate the collision risk level for each potential collision point;

[0086] Based on the collision risk level of each potential collision point, determine the collision point with the highest risk level;

[0087] Select a strategy to adjust the robotic arm's motion parameters based on the collision point with the highest risk level;

[0088] Based on the robotic arm motion parameters, adjust the strategy and reassess the collision risk level of all potential collision points;

[0089] If there are still other potential collision points with a collision risk higher than or equal to the preset risk threshold, the robot arm motion parameter adjustment strategy is iterated until the collision risk of all potential collision points is lower than the preset risk threshold.

[0090] Specifically, identifying all potential collision points refers to predicting and determining all possible contact points between the corners of the photovoltaic modules and the cooling rack guide structure during the robotic arm's handling of photovoltaic modules. This is achieved by combining the geometric model of the photovoltaic modules, the pose data of the robotic arm, and the geometric information of the cooling rack guide structure. Calculating the collision risk level of each potential collision point can be understood as quantifying the probability of a collision and the potential damage based on the location of the potential collision point, the collision sensitivity data of the photovoltaic modules, and the motion state of the robotic arm. For example, the collision risk level can be a numerical value that comprehensively considers factors such as distance, relative velocity, and material properties. Furthermore, determining the collision point with the highest risk level means selecting the point with the highest current collision risk level from all potential collision points and prioritizing it for the current adjustment strategy. Selecting a robotic arm motion parameter adjustment strategy based on the highest risk level collision point means developing a corresponding robotic arm motion adjustment plan for that highest risk point, such as fine-tuning the robotic arm's pose, speed, or acceleration to reduce the collision risk at that point. The process of reassessing the collision risk level of all potential collision points based on the robotic arm motion parameter adjustment strategy involves conducting a comprehensive risk assessment of all potential collision points again after selecting and assuming the adjustment strategy has been implemented. This verifies the effectiveness of the adjustment strategy and identifies any new risks that may be introduced by the adjustment. If other potential collision points still exist with a collision risk higher than or equal to a preset risk threshold, the robotic arm motion parameter adjustment strategy is iterated until the collision risk of all potential collision points is lower than the preset risk threshold. This means that if an adjustment fails to bring all potential collision points to a safety standard, the system will repeat the above process to continuously optimize the adjustment strategy until all potential risks are effectively avoided.

[0091] The proposed solution effectively addresses the aforementioned issues by introducing an iterative adjustment mechanism. Specifically, after initially selecting a strategy for adjusting the robotic arm's motion parameters, it is not executed immediately. Instead, a comprehensive reassessment of all potential collision points is conducted. This reassessment mechanism promptly identifies other potential collision risks that may be triggered by the initial adjustment strategy. By identifying the highest-risk collision points and selecting targeted adjustment strategies, it ensures that each adjustment focuses on the most pressing risks. More importantly, when the reassessment reveals other potential collision points with collision risks higher than or equal to a preset risk threshold, the system triggers iterative adjustments until the collision risks of all potential collision points are below the preset risk threshold. This iterative process ensures the comprehensiveness and thoroughness of the adjustment strategy, avoiding the problem of neglecting overall safety while focusing on local optimization.

[0092] In some preferred embodiments, it is assumed that the photovoltaic module grasped by the robotic arm, due to slight planar rotation, presents multiple potential collision points as it approaches the cooling rack guide structure. First, the system identifies all these potential collision points and calculates their respective collision risk levels. For example, the front left corner of the photovoltaic module might be assessed as having the highest collision risk level. In this case, the robotic arm's motion parameter adjustment strategy might be selected to perform a slight counter-clockwise rotation to move the front left corner away from the guide structure. However, after performing this adjustment, the system immediately reassesses the collision risk levels of all potential collision points. If it is found that the rear right corner of the photovoltaic module now has a collision risk higher than a preset risk threshold due to this rotation, the system will not stop but will re-identify the rear right corner as the current highest-risk collision point and select a new adjustment strategy, such as a slight translation of the robotic arm, to simultaneously avoid the risk at the rear right corner. This iterative process continues until the collision risk of all potential collision points is below the preset risk threshold, thereby ensuring the absolute safety of the photovoltaic module during handling.

[0093] This application further proposes steps for reassessing the collision risk level of all potential collision points based on a strategy adjusted according to the robotic arm's motion parameters, including:

[0094] Multiple distance sensors are integrated into the support section of the robotic arm;

[0095] After the robotic arm executes the robotic arm motion parameter adjustment strategy, multiple distance sensors are used to measure the distance between the corners of the photovoltaic module and the cooling rack guide structure in real time to obtain distance data;

[0096] Based on distance data, the geometric model of the photovoltaic module, and the pose data of the robotic arm, the potential collision point position and contact angle of the photovoltaic module's corner relative to the cooling rack guide structure are recalculated.

[0097] Based on component collision sensitivity data, recalculated potential collision point locations and contact angles, the collision risk level of all potential collision points is reassessed.

[0098] Specifically, the aforementioned distance sensors can be understood as sensors capable of acquiring real-time distance information between objects, such as laser rangefinders, ultrasonic rangefinders, or visual rangefinder systems. These sensors are integrated into the support structure of the robotic arm to directly and in real-time perceive the spatial relationship between the photovoltaic modules and the cooling rack guide structure. The geometric model of the photovoltaic module refers to a mathematical or digital model describing its shape, size, and structural characteristics, used to provide precise geometric information when calculating potential collision points. The robotic arm's pose data refers to its position and orientation information in space, typically provided by the robotic arm's own encoder or an external positioning system, used to determine the absolute or relative position of the photovoltaic modules in space. In practical applications, recalculating the potential collision point position and contact angle means, after adjusting the robotic arm's motion parameters, using the real-time acquired distance data, combined with the photovoltaic module's geometric model and the robotic arm's pose data, accurately determining the specific location and angle at which the photovoltaic module's corners may contact the cooling rack guide structure. This aims to provide accurate input for subsequent collision risk assessment.

[0099] This application's solution integrates multiple distance sensors on the carrier section of the robotic arm, enabling real-time acquisition of actual distance data between the corners of the photovoltaic module and the cooling rack guide structure after the robotic arm executes its motion parameter adjustment strategy. It is precisely because of the introduction of this real-time distance data, combined with the geometric model of the photovoltaic module and the pose data of the robotic arm, that the potential collision point position and contact angle of the photovoltaic module's corners relative to the cooling rack guide structure can be accurately recalculated. This calculation method based on actual measurement data effectively compensates for potential errors that may exist when relying solely on theoretical models, thus providing a more accurate and reliable basis for subsequent collision risk assessment. In this way, it can be ensured that after adjusting the robotic arm's motion parameters, the risk assessment of all potential collision points is based on the latest and most realistic spatial relationships, significantly improving the accuracy of the assessment.

[0100] In some preferred embodiments, after the robotic arm is fine-tuned according to the adjustment strategy, for example, by slightly raising or shifting to the side to avoid a high-risk collision point, multiple distance sensors integrated on the robotic arm's support, such as laser rangefinders, immediately activate and measure in real time the distances between the four corners of the photovoltaic module and the edges of the cooling rack guide structure. This distance data is transmitted to the control system, which, combined with a pre-stored 3D geometric model of the photovoltaic module and the robotic arm's current precise pose data, reconstructs the precise position of the photovoltaic module in space and calculates all potential points where its corners might contact the cooling rack guide structure, along with their contact angles. Subsequently, these recalculated potential collision point positions and contact angles, along with the module's collision sensitivity data, are input to the risk assessment module to reassess the collision risk level of each potential collision point. If any points with a collision risk exceeding a preset threshold are found, the system iteratively adjusts the strategy again until all potential collision risks are effectively eliminated.

[0101] Specifically, the steps for recalculating the potential collision point location and contact angle of the photovoltaic module corners relative to the cooling rack guide structure include:

[0102] On the support part of the robotic arm, multiple laser rangefinders and multiple ultrasonic rangefinders are distributed at intervals along the circumferential and radial directions of the suction cup structure array.

[0103] After the robotic arm picks up the photovoltaic module, during the robotic arm's handling process, the laser rangefinder and the ultrasonic rangefinder simultaneously measure the distance at different positions on the edge of the photovoltaic module.

[0104] The validity of the distance data collected by the laser rangefinder is determined to obtain the effective distance data of the laser rangefinder.

[0105] The validity of the distance data collected by the ultrasonic ranging sensor is determined to obtain the effective distance data of the ultrasonic ranging sensor.

[0106] If the measurement of a laser rangefinder or ultrasonic rangefinder at a certain measurement location is abnormal, the effective distance data of another type of sensor at the corresponding measurement location will be used to supplement the measurement and obtain the supplemented distance information.

[0107] If both the laser rangefinder and the ultrasonic rangefinder at a certain measurement location show abnormalities, then the reconstructed distance information is obtained by interpolating the effective distance data of the adjacent sensors at the corresponding measurement location.

[0108] Based on the supplemented or reconstructed distance information, combined with the geometric model of the photovoltaic module and the pose data of the robotic arm, the potential collision point position and contact angle of the photovoltaic module's corners relative to the cooling rack guide structure are recalculated.

[0109] The robotic arm's support structure features multiple laser ranging sensors and multiple ultrasonic ranging sensors spaced circumferentially and radially along the suction cup array. This design aims to achieve multi-point, multi-type sensor collaborative measurement of the photovoltaic module's edge distance data, thereby improving measurement coverage and data redundancy. Laser ranging sensors typically offer high measurement accuracy and good directionality, making them suitable for precise measurements; while ultrasonic ranging sensors are insensitive to the target object's material and lighting conditions, making them suitable for auxiliary measurements in complex environments. Combining these two sensor types complements their respective advantages and disadvantages.

[0110] Furthermore, after the robotic arm grasps the photovoltaic module, during the robotic arm's handling process, laser and ultrasonic ranging sensors simultaneously measure the distance to different locations on the edge of the photovoltaic module. This synchronous measurement ensures that distance data from different locations are acquired at the same time, providing temporal consistency for subsequent data fusion and processing. The validity of the distance data collected by the laser and ultrasonic ranging sensors is assessed separately to identify and eliminate abnormal data caused by environmental interference, sensor malfunction, or poor measurement conditions, thereby ensuring that the distance data used in subsequent processing is reliable.

[0111] Specifically, if the measurement from a laser rangefinder or ultrasonic rangefinder at a certain measurement location is abnormal, the effective distance data from another type of sensor at the same measurement location is used to supplement the measurement, resulting in supplemented distance information. For example, if a laser rangefinder cannot obtain effective data due to strong reflection, and the ultrasonic rangefinder at the same location has valid data, it can be used to supplement the distance information for that location. If both the laser rangefinder and ultrasonic rangefinder at a certain measurement location are abnormal, interpolation is performed based on the effective distance data from adjacent sensors at the corresponding measurement location to obtain reconstructed distance information. This interpolation method can utilize reliable surrounding measurement data to estimate the distance to missing or abnormal locations, further improving the completeness of the distance information.

[0112] This application's solution simultaneously deploys laser and ultrasonic ranging sensors on the support of a robotic arm, distributing them circumferentially and radially along a suction cup array. This enables multi-source, multi-point synchronous acquisition of distance data at different locations on the edge of photovoltaic modules. Laser ranging sensors typically offer high accuracy and good directionality, but may be affected by factors such as illumination and reflectivity. Ultrasonic ranging sensors, on the other hand, are insensitive to material properties and illumination, but have relatively lower accuracy and are susceptible to atmospheric interference. By separately evaluating the validity of the distance data acquired by the two types of sensors, abnormal data in each measurement can be identified and eliminated. When a single type of sensor malfunctions at a measurement location, valid data from the other type of sensor can be used to supplement the measurement, thereby compensating for the limitations of a single sensor and improving the reliability of the distance data at that location. When both sensors malfunction at a location, interpolation reconstruction using valid distance data from adjacent locations further ensures the integrity of the distance information. Therefore, the supplemented or reconstructed distance information has higher accuracy and robustness, providing reliable input for recalculating the potential collision point position and contact angle by combining the geometric model of the photovoltaic module and the pose data of the robotic arm, thereby significantly improving the accuracy of collision risk assessment.

[0113] In some preferred embodiments, it is assumed that eight laser ranging sensors and eight ultrasonic ranging sensors are uniformly distributed circumferentially and radially along the suction cup structure array on the robotic arm's support. When the robotic arm grasps a photovoltaic module and begins to move it, all sensors work synchronously to measure the distance to the module's edge in real time. For example, at a certain moment, the laser ranging sensor located at the upper left corner of the module returns abnormal data due to excessive reflection from the module's surface, but the ultrasonic ranging sensor at the same location measures normal data and is judged as valid. In this case, the system will use the valid data from the ultrasonic ranging sensor to supplement the distance information at the upper left corner. As another example, at the lower right corner of the module, both the laser ranging sensor and the ultrasonic ranging sensor return abnormal data due to the special structure of the module's edge or environmental interference. In this case, the system will reconstruct the distance information at the lower right corner based on the distance data measured by the adjacent sensors at the lower right corner (e.g., the valid sensors on the right and bottom sides), using linear interpolation or a more complex surface fitting algorithm. Ultimately, this supplemented or reconstructed distance information, along with the known geometric model of the photovoltaic module and the real-time pose data of the robotic arm, is input into the calculation module to accurately recalculate the potential collision point positions and contact angles between the corners of the photovoltaic module and the cooling rack guide structure, thereby providing a reliable basis for subsequent motion parameter adjustments.

[0114] In some embodiments of this application described above, the step of determining the validity of distance data acquired by the laser rangefinder specifically includes:

[0115] Analyze the intensity of the echo signal from the laser rangefinder sensor;

[0116] Analyze the signal-to-noise ratio of the echo signal from the laser rangefinder sensor;

[0117] Analyze the waveform characteristics of the echo signal from the laser rangefinder sensor;

[0118] The validity of the distance data collected by the laser rangefinder is determined based on the intensity, signal-to-noise ratio, and waveform characteristics of the echo signal.

[0119] Specifically, analyzing the intensity of the echo signal from the laser rangefinder aims to assess the energy level of the received laser signal. Generally, a higher echo signal intensity indicates a better reflection effect between the laser beam and the target object, resulting in higher reliability of the measurement results. Conversely, if the echo signal intensity is too low, it may mean that the laser beam is partially absorbed, scattered, or blocked. In this case, the distance data may contain significant errors and should be considered invalid.

[0120] Analyzing the signal-to-noise ratio (SNR) of the laser rangefinder sensor echo signal quantifies the clarity of the signal relative to background noise. A high SNR indicates a clear signal, minimal noise interference, and high measurement accuracy. Conversely, a low SNR indicates severe noise contamination, potentially leading to deviations or drift in distance measurements; therefore, such data needs to be discarded or corrected.

[0121] Furthermore, analyzing the waveform characteristics of the laser rangefinder sensor's echo signal can reveal whether distortion or multipath effects occur during the propagation and reflection of the laser pulse. For example, an ideal echo signal typically has a single, sharp peak. If the waveform exhibits broadening, multiple peaks, or irregular shapes, it may indicate an anomaly when the laser beam encounters complex surfaces, transparent media, or multiple reflective surfaces, making distance calculations based on this waveform unreliable.

[0122] Therefore, by comprehensively considering the intensity, signal-to-noise ratio, and waveform characteristics of the laser rangefinder sensor's echo signal, a comprehensive evaluation mechanism can be established to determine whether the collected distance data is true and valid. Only when all these parameters meet the preset validity criteria will the distance data be adopted for subsequent calculations and analysis.

[0123] This application's solution addresses the potential unreliability of laser ranging data in complex environments by performing multi-dimensional validity assessments on the distance data acquired by the laser ranging sensor. In the aforementioned robotic arm photovoltaic module loading and unloading method, both the laser ranging sensor and the ultrasonic ranging sensor simultaneously measure distances at different locations on the edge of the photovoltaic module. To ensure the accuracy of subsequent supplementation or reconstruction of abnormal measurement data, the reliability of the original valid distance data must first be ensured. By analyzing the intensity of the echo signal, weak signal measurements caused by signal attenuation or obstruction can be eliminated; by analyzing the signal-to-noise ratio, measurement data severely affected by environmental noise can be filtered out; and by analyzing waveform characteristics, distorted signals caused by multipath effects or target characteristics can be identified. It is precisely through the comprehensive evaluation of these key parameters that the laser ranging sensor can provide high-quality valid distance data, thus laying a solid foundation for subsequent distance information supplementation, reconstruction, and the recalculation of potential collision point positions and contact angles.

[0124] The above technical solution enables multi-dimensional and refined validity assessment of distance data acquired by laser rangefinders. By comprehensively analyzing the intensity, signal-to-noise ratio, and waveform characteristics of the echo signal, abnormal measurement data caused by environmental interference, target characteristics, or sensor malfunctions can be effectively identified and eliminated, ensuring that subsequent distance information supplementation or reconstruction is based on highly reliable original data. This significantly improves the accuracy and reliability of distance measurement between the corners of photovoltaic modules and the cooling rack guide structure, providing a solid data foundation for accurately calculating the location and contact angle of potential collision points, thereby enhancing the accuracy and safety of robotic arm motion parameter adjustments.

[0125] refer to Figure 2 This application proposes a robotic arm photovoltaic module loading and unloading system, applied to a robotic arm photovoltaic module loading and unloading method. The system includes:

[0126] The acquisition module, with the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, makes the robotic arm move according to the preset transport trajectory, and uses a torque sensor to collect the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array.

[0127] The data acquisition module uses a torque sensor to collect torque signals at the end of the robotic arm or at the joints in real time when the robotic arm actually grasps the photovoltaic module and performs the handling action, so as to obtain the actual torque data.

[0128] The calculation module aligns the actual torque data with the ideal motion response data in time and subtracts them point by point to obtain the torque difference signal;

[0129] The analysis module analyzes the torque difference signal and identifies fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command.

[0130] The judgment module determines whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm based on the fluctuation characteristics.

[0131] If the processing module determines that the photovoltaic module is rotating in a plane, it triggers the adjustment of the motion parameters of the robotic arm.

[0132] Through modular design, the system integrates the detection of possible planar rotation of photovoltaic modules during handling with the function of adjusting the motion parameters of the robotic arm, forming a closed-loop intelligent control system, thereby effectively avoiding module collisions and improving production efficiency and safety.

[0133] To better understand the system proposed in this application, some key terms and system components are first explained. A robotic arm refers to a multi-jointed, programmable automated device used to perform operations such as grasping and handling. The support unit is the end effector of the robotic arm used to grasp and support photovoltaic modules; in this application, it specifically refers to a suction cup structure arranged in an array to ensure stable adhesion to the photovoltaic modules. A torque sensor is a device capable of measuring rotational force or torque; in this application, it is used to collect torque signals at the end effector or joints of the robotic arm to reflect the force state of the robotic arm during movement. Ideal motion response data refers to the torque signal data generated when the robotic arm moves along a preset handling trajectory under ideal conditions where there is no relative sliding connection between the robotic arm and the photovoltaic module. Actual torque data refers to the torque signal data collected in real time by the torque sensor when the robotic arm actually grasps the photovoltaic module and performs the handling action. The torque difference signal is the signal obtained by time alignment and point-by-point subtraction between the actual torque data and the ideal motion response data; it reflects the deviation between the actual motion and the ideal motion. Fluctuation characteristics refer to the periodic changes in the torque difference signal with a specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command. These fluctuation characteristics are key evidence for determining whether the photovoltaic module has undergone planar rotation. Planar rotation refers to the rotation of the photovoltaic module relative to the robotic arm's support portion on a horizontal plane after it has been grasped by the robotic arm. Motion parameter adjustment refers to modifying parameters such as the robotic arm's speed, acceleration, and trajectory based on the judgment results to correct the module's posture deviation or avoid potential collisions. The acquisition module, data collection module, calculation module, analysis module, judgment module, and processing module are the functional units of this system. They can be hardware circuits, software programs, or a combination of both, each undertaking specific data processing and control tasks, jointly realizing the detection and response to the planar rotation of the photovoltaic module.

[0134] The core of the robotic arm photovoltaic module loading and unloading system proposed in this application lies in its modular design, which integrates the detection of possible planar rotation of photovoltaic modules during handling with the function of adjusting the motion parameters of the robotic arm.

[0135] Specifically, the acquisition module in the system is responsible for ensuring that the robotic arm moves along a preset transport trajectory while maintaining a non-slip connection between the support part of the robotic arm and the photovoltaic module. It uses torque sensors to collect torque signals at the end effector or joints of the robotic arm to obtain ideal motion response data. This acquisition module can be a standalone hardware unit, such as a dedicated data acquisition and processing circuit board, which integrates a microcontroller and memory to execute the preset motion trajectory and record torque data. Alternatively, the acquisition module can be a software function module within the robotic arm controller, driving the robotic arm to perform ideal motion through programmed instructions and transmitting the torque sensor data to the main control unit via a bus interface for storage and processing.

[0136] The acquisition module is responsible for acquiring real-time torque signals at the end effector or joints of the robotic arm using a torque sensor when the robotic arm actually grasps the photovoltaic modules and performs the handling action, thus obtaining the actual torque data. This acquisition module can be a torque sensor integrated into the end effector or joints of the robotic arm and its corresponding data transmission interface, such as transmitting real-time torque data to the central processing unit via a CAN bus or EtherCAT protocol. In some embodiments, the acquisition module can also be a separate signal conditioning unit, responsible for amplifying, filtering, and converting the analog signal output by the torque sensor to digital, and then sending the digitized data to the subsequent processing unit.

[0137] The calculation module is responsible for time-aligning and point-by-point subtraction of the actual torque data with the ideal motion response data to obtain the torque difference signal. This calculation module can be a high-performance industrial PC or embedded processor, running specialized data processing algorithms, such as cross-correlation algorithms or dynamic time warping algorithms, for time alignment and point-by-point subtraction. As an alternative, the calculation module can also be a digital signal processor inside the robotic arm controller, utilizing its powerful floating-point arithmetic capabilities to perform data alignment and difference calculation in real time.

[0138] The analysis module is responsible for analyzing the torque difference signal and identifying fluctuation features with specific frequencies and amplitudes that have a phase delay relationship with the robotic arm's motion commands. This analysis module can be an industrial computer equipped with spectrum analysis software, such as using algorithms like Fast Fourier Transform or wavelet analysis to extract frequency and amplitude features from the torque difference signal. In some preferred embodiments, the analysis module can also be a hardware accelerator based on a Field-Programmable Gate Array (FPGA), capable of processing large amounts of torque data in parallel to achieve highly efficient real-time fluctuation feature identification.

[0139] The judgment module is responsible for determining whether the photovoltaic module has undergone planar rotation relative to the support of the robotic arm based on the fluctuation characteristics. This judgment module can be a microcontroller running a decision logic algorithm, for example, by comparing the identified fluctuation characteristics (frequency, amplitude, phase delay) with preset thresholds or a pattern library to make a judgment. As another implementation, the judgment module can also be an inference engine based on a machine learning model, which, through training on a large amount of historical data, can more intelligently determine the occurrence of planar rotation.

[0140] The processing module is responsible for determining if the photovoltaic module has undergone planar rotation and triggering adjustments to the robotic arm's motion parameters. This processing module can be an interface unit that communicates directly with the robotic arm controller, for example, by sending adjustment commands via industrial Ethernet or a dedicated I / O port. In some implementations, the processing module can also be a motion planner within the robotic arm controller, dynamically modifying the robotic arm's trajectory, speed, acceleration, and other parameters based on the output of the determination module to correct module attitude deviations or avoid potential collisions.

[0141] The aforementioned acquisition, data collection, calculation, analysis, judgment, and processing modules work together to achieve real-time detection and response to planar rotation during photovoltaic module handling. The robotic arm's support structure is a suction cup structure located at its end, arranged in an array. The implementation methods of these modules can be flexibly selected and combined according to specific application scenarios and performance requirements to achieve optimal system performance and cost-effectiveness.

[0142] The robotic arm photovoltaic module loading and unloading system proposed in this application works by leveraging the collaborative efforts of various functional modules to achieve real-time and accurate judgment and intervention of the planar rotation of photovoltaic modules. Specifically, the acquisition module first establishes an ideal motion torque benchmark, i.e., ideal motion response data, which represents the torque performance of the robotic arm under a specific handling trajectory in a slip-free state. When the robotic arm actually grasps the photovoltaic module and performs the handling action, the acquisition module acquires the actual torque data in real time. Subsequently, the calculation module aligns the actual torque data with the ideal motion response data in time and subtracts them point by point to obtain a torque difference signal. This difference signal can effectively filter out the normal torque generated by the robotic arm's own movement and highlight the torque changes caused by abnormal situations such as module slippage. Furthermore, the analysis module performs in-depth analysis of the torque difference signal to identify fluctuation characteristics with specific frequencies and amplitudes that have a phase delay relationship with the robotic arm's motion commands. These fluctuation characteristics are typical manifestations of planar rotation of the photovoltaic module. For example, when the robotic arm accelerates or decelerates, if the module slips, its inertia will cause the actual motion of the module to lag behind the robotic arm's commanded motion. This lag will manifest as periodic fluctuations in the torque signal that have a phase delay relationship with the robotic arm's motion command. Based on these identified fluctuation characteristics, the judgment module determines whether the photovoltaic module has undergone planar rotation relative to the robotic arm's support. Once planar rotation is determined, the processing module immediately triggers adjustments to the robotic arm's motion parameters, such as reducing the robotic arm's handling speed, decreasing acceleration, or fine-tuning the robotic arm's posture to correct the module's deviation, thereby preventing collisions when the module is placed into the cooling rack. The entire system forms a closed-loop control loop, realizing real-time monitoring and proactive intervention of potential slippage risks during photovoltaic module handling, significantly improving the stability and safety of the loading and unloading process.

[0143] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A method for loading and unloading photovoltaic modules with a robotic arm, characterized in that, The method includes the following steps: With the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, the robotic arm moves according to the preset transport trajectory, and the torque sensor collects the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array. When the robotic arm actually grasps the photovoltaic module and performs the handling action, the torque sensor is used to collect the torque signal at the end of the robotic arm or at the joint in real time to obtain the actual torque data. The actual torque data is time-aligned with the ideal motion response data and subtracted point by point to obtain the torque difference signal; The torque difference signal is analyzed to identify fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command; Based on the fluctuation characteristics, determine whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm; If a planar rotation of the photovoltaic module is detected, the motion parameters of the robotic arm are adjusted accordingly. The method also includes: After the position is locked as displayed by the encoder of the robotic arm motor, the motion information fed back by the servo driver is obtained; Based on motion information, determine whether the robotic arm is swaying; If the robotic arm is shaking, do not update the zero point or reduce the frequency of zero point updates; Obtain the adhesive overflow status parameters, module type information, and ambient temperature of the photovoltaic modules; Based on the photovoltaic module's glue overflow state parameters, module type information, ambient temperature, and the reliability of zero-point updates, the threshold and response strategy for slippage determination are dynamically adjusted. Based on the adjusted slip determination threshold and response strategy, determine whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm; If it is determined that the photovoltaic module is undergoing planar rotation, adjust the motion parameters of the robotic arm; If it is determined that the photovoltaic module has undergone planar rotation, the steps for adjusting the motion parameters of the robotic arm include: Obtain the component type information of photovoltaic modules; Based on the photovoltaic module type information, query the preset module collision sensitivity data; Based on the degree and direction of planar rotation, calculate the potential collision point location and contact angle of the photovoltaic module's corners relative to the cooling rack guide structure; Assess the collision risk level of potential collision points based on component collision sensitivity data, potential collision point locations, and contact angles; Select a strategy to adjust the robotic arm's motion parameters based on the collision risk level of potential collision points; Adjust the motion parameters of the robotic arm according to the robotic arm motion parameter adjustment strategy.

2. The robotic arm method for loading and unloading photovoltaic modules as described in claim 1, characterized in that, The steps to obtain the adhesive overflow state parameters of photovoltaic modules include: Multiple miniature infrared temperature sensors and multiple miniature spectral sensors are distributed on the support part of the robotic arm; After the robotic arm grasps the photovoltaic module, during the robotic arm's handling process, multiple miniature infrared temperature sensors are used to collect real-time overflow temperature data at different locations on the edge of the photovoltaic module, and multiple miniature spectral sensors are used to collect real-time overflow spectral data at different locations on the edge of the photovoltaic module. Based on the overflow temperature data and overflow spectrum data, calculate the spatial distribution differences of the overflow temperature data and the spatial distribution differences of the overflow spectrum data. The uniformity of the glue distribution is determined based on the spatial distribution differences of the glue temperature data and the spatial distribution differences of the glue spectrum data. If it is determined that the glue overflow distribution is uneven, the average glue overflow temperature and average glue overflow viscosity in the contact area between the robotic arm's load-bearing part and the photovoltaic module are calculated based on the glue overflow temperature data and glue overflow spectrum data, and the average glue overflow temperature and average glue overflow viscosity are used as glue overflow state parameters of the photovoltaic module.

3. The robotic arm method for loading and unloading photovoltaic modules as described in claim 1, characterized in that, The steps for selecting a strategy to adjust the motion parameters of a robotic arm include: Identify all potential collision points; Calculate the collision risk level for each potential collision point; Based on the collision risk level of each potential collision point, determine the collision point with the highest risk level; Select a strategy to adjust the robotic arm's motion parameters based on the collision point with the highest risk level; Based on the robotic arm motion parameters, adjust the strategy and reassess the collision risk level of all potential collision points; If there are still other potential collision points with a collision risk higher than or equal to the preset risk threshold, the robot arm motion parameter adjustment strategy is iterated until the collision risk of all potential collision points is lower than the preset risk threshold.

4. The robotic arm method for loading and unloading photovoltaic modules as described in claim 3, characterized in that, The steps for reassessing the collision risk level of all potential collision points based on the strategy for adjusting the robotic arm's motion parameters include: Multiple distance sensors are integrated into the support section of the robotic arm; After the robotic arm executes the robotic arm motion parameter adjustment strategy, multiple distance sensors are used to measure the distance between the corners of the photovoltaic module and the cooling rack guide structure in real time to obtain distance data; Based on distance data, the geometric model of the photovoltaic module, and the pose data of the robotic arm, the potential collision point position and contact angle of the photovoltaic module's corner relative to the cooling rack guide structure are recalculated. Based on component collision sensitivity data, recalculated potential collision point locations and contact angles, the collision risk level of all potential collision points is reassessed.

5. The robotic arm method for loading and unloading photovoltaic modules as described in claim 4, characterized in that, The steps for recalculating the potential collision point location and contact angle of the photovoltaic module corners relative to the cooling rack guide structure include: On the support part of the robotic arm, multiple laser rangefinders and multiple ultrasonic rangefinders are distributed at intervals along the circumferential and radial directions of the suction cup structure array. After the robotic arm picks up the photovoltaic module, during the robotic arm's handling process, the laser rangefinder and the ultrasonic rangefinder simultaneously measure the distance at different positions on the edge of the photovoltaic module. The validity of the distance data collected by the laser rangefinder is determined to obtain the effective distance data of the laser rangefinder. The validity of the distance data collected by the ultrasonic ranging sensor is determined to obtain the effective distance data of the ultrasonic ranging sensor. If the measurement of a laser rangefinder or ultrasonic rangefinder at a certain measurement location is abnormal, the effective distance data of another type of sensor at the corresponding measurement location will be used to supplement the measurement and obtain the supplemented distance information. If both the laser rangefinder and the ultrasonic rangefinder at a certain measurement location show abnormalities, then the reconstructed distance information is obtained by interpolating the effective distance data of the adjacent sensors at the corresponding measurement location. Based on the supplemented or reconstructed distance information, combined with the geometric model of the photovoltaic module and the pose data of the robotic arm, the potential collision point position and contact angle of the photovoltaic module's corners relative to the cooling rack guide structure are recalculated.

6. The method for loading and unloading photovoltaic modules with a robotic arm as described in claim 5, characterized in that, The steps for determining the validity of distance data acquired by a laser rangefinder include: Analyze the intensity of the echo signal from the laser rangefinder sensor; Analyze the signal-to-noise ratio of the echo signal from the laser rangefinder sensor; Analyze the waveform characteristics of the echo signal from the laser rangefinder sensor; The validity of the distance data collected by the laser rangefinder is determined based on the intensity, signal-to-noise ratio, and waveform characteristics of the echo signal.

7. A robotic arm photovoltaic module loading and unloading system, applied to the robotic arm photovoltaic module loading and unloading method as described in claim 1, characterized in that, The system includes: The acquisition module, with the support part of the robotic arm and the photovoltaic module in a state of no relative sliding connection, makes the robotic arm move according to the preset transport trajectory, and uses a torque sensor to collect the torque signal at the end of the robotic arm or the joint to obtain ideal motion response data. The support part of the robotic arm is a suction cup structure set at the end of the robotic arm, and the suction cup structure is arranged in an array. The data acquisition module uses a torque sensor to collect torque signals at the end of the robotic arm or at the joints in real time when the robotic arm actually grasps the photovoltaic module and performs the handling action, so as to obtain the actual torque data. The calculation module aligns the actual torque data with the ideal motion response data in time and subtracts them point by point to obtain the torque difference signal; The analysis module analyzes the torque difference signal and identifies fluctuation characteristics with specific frequency and amplitude that have a phase delay relationship with the robotic arm's motion command. The judgment module determines whether the photovoltaic module has undergone planar rotation relative to the support part of the robotic arm based on the fluctuation characteristics. If the processing module determines that the photovoltaic module is rotating in a plane, it triggers the adjustment of the motion parameters of the robotic arm.

Images