A fault early warning and remote diagnosis method for an industrial robot electric control system

Abstract

This invention relates to the field of industrial robot fault diagnosis technology, and discloses a fault early warning and remote diagnosis method for the electrical control system of an industrial robot. This method obtains mechanical load parameters through a potential load model, an inertial load model, and a basic dynamic model; obtains electrical state parameters through an electrical efficiency model; generates correction factors through a correction model based on ambient temperature; calculates dynamic current thresholds through a threshold model based on the correction factors and a basic current threshold; and finally outputs health scores and fault early warning information through a diagnostic model by combining vibration acceleration and servo loop following error. This invention integrates multi-source parameters from mechanical, electrical, and environmental sources to achieve dynamic threshold adjustment and multi-dimensional health assessment, effectively distinguishing between normal operating conditions and abnormal states, significantly improving the accuracy of fault early warnings, reducing false alarm rates, and providing a reliable remote diagnosis solution for industrial robots.

Description

Technical Field

[0001] This invention belongs to the field of industrial robot fault diagnosis technology, and in particular relates to a fault early warning and remote diagnosis method for an industrial robot electrical control system. Background Technology

[0002] As a core component of high-end manufacturing equipment, the reliability of the electrical control system of industrial robots directly determines the operating efficiency and stability of the entire production line. In actual industrial production, robots typically operate under complex conditions for extended periods. Key components in their electrical control systems, such as servo drives, motors, and power modules, are continuously subjected to the combined effects of load changes, motion state switching, and environmental factors, significantly increasing the risk of system failure. Especially under high load, high speed, or extreme temperature environments, the operating states of various components in the electrical control system exhibit obvious dynamic changes, making it difficult for traditional single-parameter monitoring methods to accurately capture abnormal signs under these complex conditions.

[0003] Currently, the industrial robot field mainly employs two maintenance strategies: periodic preventative maintenance and reactive fault repair. The former involves maintenance based on fixed cycles, lacking accurate assessment of the equipment's actual operating status, easily leading to either "over-maintenance" or "under-maintenance." The latter involves emergency repairs after a failure occurs, inevitably causing unplanned downtime and resulting in significant production losses for enterprises. Neither of these methods can meet the stringent requirements of modern intelligent manufacturing for equipment reliability and operational efficiency.

[0004] In recent years, with the popularization of predictive and health management concepts, some state-based monitoring methods have begun to be applied. However, these existing technical solutions still have significant limitations: First, the monitoring parameters are too singular, relying only on individual parameters such as current or vibration for judgment, lacking a comprehensive consideration of load characteristics, motion state, and environmental factors, resulting in a high false alarm rate; second, the alarm threshold settings are rigid and cannot adapt to the dynamic changes in different working conditions of the robot, often leading to misjudgments under special working conditions such as heavy load, high speed, or high temperature; finally, existing methods are insufficient in identifying early fault characteristics, often only detecting them when the fault has developed to a considerable extent, missing the best opportunity for early warning. Summary of the Invention

[0005] The purpose of this invention is to provide a fault early warning and remote diagnosis method for the electrical control system of an industrial robot, in order to solve the above-mentioned problems.

[0006] This invention is implemented as follows: a fault early warning and remote diagnosis method for an industrial robot electrical control system, comprising the following steps: obtaining a potential load coefficient based on load mass and joint angle using a potential load model; obtaining an inertial load coefficient based on load rotational inertia and acceleration using an inertial load model; obtaining a basic dynamic coefficient based on joint velocity, potential load coefficient, and inertial load coefficient using a basic dynamic model; obtaining an electrical efficiency coefficient based on bus voltage and power factor using an electrical efficiency model; obtaining a correction factor based on the basic dynamic coefficient and electrical efficiency coefficient using a correction model; obtaining a dynamic current threshold based on ambient temperature, correction factor, and basic current threshold using a threshold model; and outputting diagnostic information based on vibration acceleration, servo loop following error, and dynamic current threshold using a diagnostic model.

[0007] A further technical solution, specifically the steps for outputting diagnostic information through a diagnostic model based on vibration acceleration, servo loop following error, and dynamic current threshold, are as follows: The current vibration acceleration and servo loop following error are compared with the maximum allowable vibration acceleration and the maximum allowable following error, respectively, to obtain the vibration acceleration index and the servo loop following error index; the vibration acceleration, servo loop following error index, and dynamic current threshold are then imported into the formula... Get a health score ,in, The value range is 0-1. This is a normal state; continue monitoring. 0.70 To monitor the status, increase the monitoring frequency to 0.5. In a state of alert, planned inspections are conducted. If the machine is in maintenance condition, stop it immediately for maintenance. For dynamic current threshold, This is the actual current. The vibration acceleration index, The servo loop following error index. , and All are weighting coefficients.

[0008] A further technical solution, the specific steps for obtaining the dynamic current threshold based on ambient temperature, correction factor, and base current threshold through a threshold model, are as follows: The difference between the current ambient temperature and the lowest operating temperature is compared with the difference between the highest and lowest operating temperatures to obtain the ambient temperature index; the ambient temperature index, correction factor, and base current threshold are then imported... Formula acquisition ,in, Based on the base current threshold, As a correction factor, The ambient temperature index. This is the temperature influence coefficient.

[0009] A further technical solution is that the modified model is: ,in, As a correction factor, Basic dynamic coefficients, The electrical efficiency coefficient, This is the safety margin coefficient.

[0010] A further technical solution involves obtaining the electrical efficiency coefficient based on bus voltage and power factor using an electrical efficiency model. The specific steps are as follows: Ratio the current bus voltage and power factor to the rated bus voltage and rated power factor, respectively, to obtain the bus voltage index and power factor index; then, import the bus voltage index and power factor index into the formula... Obtain the electrical efficiency coefficient ,in, This refers to the bus voltage index. The power factor index, It is the basic efficiency coefficient.

[0011] A further technical solution involves obtaining the basic dynamic coefficients based on joint velocity, potential load coefficient, and inertial load coefficient through a basic dynamic model. The specific steps are as follows: The current joint velocity is compared to the maximum joint velocity to obtain the joint velocity index; the joint velocity index, potential load coefficient, and inertial load coefficient are then imported into the formula. Obtain the basic dynamic coefficients ,in, The potential loading coefficient, The inertial load factor, The joint velocity index, , and All are weighting coefficients.

[0012] A further technical solution involves obtaining the inertial load coefficient based on the load's moment of inertia and acceleration using an inertial load model. The specific steps are as follows: The load's moment of inertia and acceleration are compared to the maximum moment of inertia and maximum angular acceleration, respectively, to obtain the load's moment of inertia index and acceleration index; the load's moment of inertia index and acceleration index are multiplied together to obtain the inertial load coefficient. .

[0013] A further technical solution, the specific steps for obtaining the potential load coefficient based on load mass and joint angle through a potential load model are as follows: Based on the joint angle, obtain the lever arm length at the current angle through a lever arm simplification function; calculate the ratios of the load mass and the lever arm length at the current angle with the maximum allowable load mass and the maximum lever arm length, respectively, to obtain the load mass index and the joint angle index; import the load mass index and the joint angle index into the formula. Obtain the potential load coefficient ,in, The load quality index, This is the joint angle index.

[0014] A further technical solution is that the simplified function of the lever arm is: ,in, The reference lever arm length, The lever arm variation coefficient, This refers to the joint angle.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0016] 1. By integrating mechanical dynamic parameters (load mass, moment of inertia, joint angle, velocity and acceleration), electrical status parameters (bus voltage, power factor) and environmental parameters (temperature), a multi-dimensional fault assessment system is constructed, which effectively overcomes the limitations of traditional single-parameter monitoring and significantly reduces false alarms and missed alarms caused by fluctuations in normal operating conditions.

[0017] 2. Based on real-time load, efficiency, and temperature data, the current protection threshold is dynamically calculated and corrected through a series of models, enabling the diagnostic benchmark to adapt to the actual operating state of the robot and solving the problem of poor adaptability of fixed thresholds under complex working conditions such as heavy load, high speed, or extreme temperature.

[0018] 3. A health scoring mechanism is introduced, which integrates multiple parameters such as vibration, servo error and current for weighted evaluation, to achieve early identification and graded warning of equipment performance degradation trends (normal, observation, warning, maintenance), providing a valuable time window for planned maintenance and avoiding unplanned downtime. Attached Figure Description

[0019] Figure 1 The flowchart illustrates a fault early warning and remote diagnosis method for an industrial robot electrical control system provided by this invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0021] In existing technologies, the maintenance strategies for industrial robot electrical control systems mainly rely on periodic preventative maintenance and reactive fault repair. Periodic maintenance, based on fixed cycles, cannot accurately determine the actual operating status of the equipment, easily leading to over-maintenance or under-maintenance. Reactive repair involves emergency repairs after a fault occurs, causing unplanned downtime. With the popularization of predictive and health management concepts, condition-based monitoring methods have begun to be applied, but existing technologies suffer from problems such as single monitoring parameters, rigid threshold settings, and a lack of early warning capabilities. For example, single current monitoring methods struggle to distinguish between load surges and equipment failures, and fixed alarm thresholds cannot adapt to dynamic operating conditions, resulting in high false alarm rates and poor adaptability.

[0022] To address these issues, researchers noted that the fault mechanism involves the coupling of multiple parameters, including mechanical load, electrical efficiency, and environmental factors. Analysis revealed that real-time changes in gravity and inertial load directly affect current demand, while bus voltage fluctuations and power factor changes reflect electrical system efficiency. Furthermore, changes in ambient temperature can cause component performance drift. Based on this, the idea of ​​constructing dynamic thresholds by integrating mechanical dynamics parameters, electrical efficiency parameters, and environmental parameters was developed, and a multi-source information fusion diagnostic model was established to achieve early anomaly identification.

[0023] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0024] like Figure 1 The diagram illustrates a fault warning and remote diagnosis method for an industrial robot electrical control system, provided by an embodiment of the present invention. The method includes the following steps: obtaining a potential load coefficient using a potential load model based on load mass and joint angle; obtaining an inertial load coefficient using an inertial load model based on load rotational inertia and acceleration; obtaining a basic dynamic coefficient using a basic dynamic model based on joint velocity, potential load coefficient, and inertial load coefficient; obtaining an electrical efficiency coefficient using an electrical efficiency model based on bus voltage and power factor; obtaining a correction factor using a correction model based on the basic dynamic coefficient and electrical efficiency coefficient; obtaining a dynamic current threshold using a threshold model based on ambient temperature, correction factor, and basic current threshold; and outputting diagnostic information using a diagnostic model based on vibration acceleration, servo loop following error, and dynamic current threshold.

[0025] The potential load model is a mathematical model that calculates the impact of gravity on joint torque by considering load mass and joint angle. Specifically, it can be implemented using a simplified lever arm function combined with mass ratio processing, and is used to quantify the pressure of gravity load on the drive system. The inertial load model is a mathematical model that calculates dynamic motion inertial torque by considering rotational inertia and acceleration. Specifically, it can be implemented by multiplying the inertia ratio by the acceleration exponent, and is used to characterize instantaneous load changes caused by motion state transitions. The basic dynamics model is an algorithm that weights and fuses potential and inertial loads based on joint velocities. Specifically, it can be implemented using a linear combination of the velocity exponent and load coefficient, and is used to reflect the comprehensive load state of the mechanical system. The electrical efficiency model is an algorithm that evaluates power conversion efficiency based on bus voltage and power factor. Specifically, it can be implemented by multiplying the voltage exponent and power factor exponent combined with a basic efficiency coefficient, and is used to monitor power supply quality and energy loss. The power factor can be obtained through actuators, sensors, or computational models. The correction model refers to a function that couples mechanical load parameters with electrical efficiency parameters. Specifically, it can be implemented by combining the product of dynamic coefficients and efficiency coefficients with a safety margin coefficient to generate a comprehensive system state correction factor. The threshold model is an algorithm that dynamically adjusts the base current threshold based on ambient temperature. Specifically, it can be implemented by multiplying the temperature index with a correction factor and then adding the result to the base threshold to generate a dynamic protection threshold that adapts to changing operating conditions. The diagnostic model is a comprehensive evaluation algorithm that integrates vibration characteristics, servo errors, and dynamic thresholds. Specifically, it can be implemented by calculating a health score using multi-parameter exponential weighting to classify and identify abnormal states.

[0026] Specifically, this method first calculates the influence of load mass and joint angle on gravitational torque using a potential load model, generating a potential load coefficient. Simultaneously, it calculates the influence of moment of inertia and acceleration on inertial torque using an inertial load model, generating an inertial load coefficient. The joint velocity and these two coefficients are input into a basic dynamics model and weighted to obtain a basic dynamics coefficient reflecting the comprehensive load state of the mechanical system. On the electrical side, the bus voltage and power factor are normalized using an electrical efficiency model, and an electrical efficiency coefficient is generated by combining it with the basic efficiency coefficient. The basic dynamics coefficient from the mechanical side and the efficiency coefficient from the electrical side are input into a correction model, and a correction factor is generated by combining it with a safety margin coefficient. This correction factor, along with the ambient temperature index, acts on the basic current threshold, generating a dynamically adjusted protection threshold through a threshold model. Finally, vibration acceleration, servo following error, and dynamic threshold are input into a diagnostic model, and a health score is calculated through multi-parameter fusion. Different levels of warning information are output based on the score range.

[0027] Compared to existing technologies, traditional methods monitor only a single current parameter and use a fixed threshold, failing to distinguish between normal load fluctuations and actual faults. This solution quantifies mechanical load changes in real time using potential load coefficients and inertial load coefficients, combined with electrical efficiency coefficients to reflect power supply quality, enabling the dynamic current threshold to automatically adjust according to operating conditions. Simultaneously, vibration acceleration and servo error are introduced as auxiliary diagnostic parameters to effectively differentiate between anomalies caused by instantaneous overload and mechanical wear. This multi-dimensional parameter fusion mechanism overcomes the limitations of single-parameter monitoring, the dynamic threshold model solves the problem of poor adaptability of fixed thresholds, and the multi-parameter weighted scoring mechanism enables the identification of early, subtle anomalies.

[0028] Through the above technical solutions, this application can reduce the false alarm rate caused by sudden load changes or environmental variations, and improve the adaptability of the diagnostic system to complex operating conditions. A dynamic threshold adjustment mechanism avoids erroneous shutdowns under normal operating conditions, while simultaneously capturing early fault characteristics in minute current fluctuations. The multi-source information fusion diagnostic model can identify coupled faults of abnormal mechanical vibration, decreased servo control accuracy, and electrical parameter deviations, achieving a leap from single fault detection to composite fault early warning.

[0029] This application further proposes that the specific steps for outputting diagnostic information through a diagnostic model based on vibration acceleration, servo loop following error, and dynamic current threshold are as follows:

[0030] The current vibration acceleration and servo loop following error are compared with the maximum permissible vibration acceleration and the maximum permissible following error, respectively, to obtain the vibration acceleration index and the servo loop following error index.

[0031] Vibration acceleration, servo loop following error exponent, and dynamic current threshold are imported into the formula. Get a health score .

[0032] The value range is 0-1. This is a normal state; continue monitoring. 0.70 To monitor the status, increase the monitoring frequency to 0.5. In a state of alert, planned inspections are conducted. If the system is in maintenance mode, it should be shut down immediately for maintenance. These thresholds can be optimized based on historical fault data.

[0033] in, The dynamic current threshold is a current limit that is dynamically adjusted according to operating conditions. Specifically, it can be generated by real-time correction of the base threshold based on temperature and load parameters to adapt to changes in electrical load under different operating conditions. This is the actual current. The vibration acceleration index is the ratio of the current vibration acceleration to the maximum permissible vibration acceleration. Specifically, it can be obtained by collecting real-time vibration data from an accelerometer and then normalizing the data. It is used to quantify how close the mechanical vibration intensity is to the safety threshold. The servo loop following error index is the ratio of the current following error to the maximum allowable following error. It can be calculated from the position error signal output by the servo controller and is used to characterize the deviation in motion control accuracy. , and All are weighting coefficients. , and All are greater than or equal to 0 and less than or equal to 1, and The system uses machine learning or expert experience to assign values. The health score is a composite index that integrates mechanical vibration, motion error, and electrical parameters. Specifically, it can be calculated using a weighted summation algorithm to achieve a quantitative assessment of equipment status.

[0034] Specifically, vibration acceleration and following error data are collected by an accelerometer and a servo controller, respectively. The raw data are then compared with preset maximum allowable values ​​to obtain the vibration acceleration index and servo loop following error index within the range of 0-1. The dynamic current threshold is adjusted according to real-time operating conditions, and the ratio of it to the actual current is limited to within 1 using a min function to avoid sudden changes in the score due to instantaneous current fluctuations. The weighting coefficients can be configured according to the equipment type; for example, the servo error weight is increased in precision assembly scenarios, and the current weight is increased in heavy-duty handling scenarios. The health score is divided into four levels. When the score is in the observation state, the monitoring frequency can be increased from once per minute to once every ten seconds; when the warning state is entered, the system automatically generates an inspection work order and pushes it to the maintenance terminal.

[0035] Compared to existing technologies, traditional methods only monitor whether the current exceeds a fixed threshold. This solution, however, uses a multi-dimensional fusion of vibration, servo error, and dynamic current thresholds to distinguish between normal load fluctuations and abnormal fault characteristics. For example, when moving heavy objects causes an increase in current but vibration and servo error are normal, traditional methods may falsely trigger alarms. This solution, however, can still determine a normal state by balancing the current ratio with vibration and error components in the health score. Furthermore, existing technologies lack early warning mechanisms, while this solution initiates a tiered response when the score falls below 0.85, detecting performance degradation trends earlier than traditional binary judgment.

[0036] Through the above technical solutions, this application can effectively reduce the false alarm rate caused by single parameter monitoring, such as avoiding accidental shutdown when the current instantaneously exceeds the limit but the mechanical condition is normal; improve the adaptability of working conditions by integrating dynamic thresholds and multiple parameters, such as automatically relaxing the current threshold while strengthening vibration monitoring in high temperature environments; and achieve early warning through a four-level health scoring mechanism, such as triggering inspection and maintenance in advance when the servo error gradually increases but does not exceed the limit.

[0037] This application further proposes that the specific steps for obtaining the dynamic current threshold based on ambient temperature, correction factor, and base current threshold through a threshold model are as follows:

[0038] The ambient temperature index is obtained by comparing the difference between the current ambient temperature and the lowest operating temperature with the difference between the highest operating temperature and the lowest operating temperature.

[0039] Import ambient temperature index, correction factor, and base current threshold. Formula acquisition .

[0040] in, The base current threshold is a preset reference current protection value. Specifically, the initial value can be set according to the motor's rated parameters and historical operating data. This threshold serves as a reference value for dynamic adjustment. The correction factor is an adjustment coefficient that combines mechanical dynamics and electrical efficiency. Specifically, it can be achieved by multiplying the basic dynamic coefficient and the electrical efficiency coefficient by the safety margin coefficient. This coefficient reflects the overall load-bearing capacity of the system under specific operating conditions. The ambient temperature index is a standardized temperature influence parameter obtained by processing the temperature difference ratio. Specifically, it can be achieved by collecting ambient temperature data with a temperature sensor and then normalizing it by combining the preset minimum and maximum operating temperatures. This parameter quantifies the actual impact of the current temperature on the system's heat dissipation capacity. This is the temperature influence coefficient. If the value is less than 1, the temperature influence coefficient refers to the control parameter of the temperature compensation intensity. The temperature influence coefficient is assigned a value based on expert experience.

[0041] Specifically, this technical solution monitors ambient temperature data in real time using a temperature sensor and calculates an ambient temperature index. This index maps actual temperature changes to a standardized range of 0-1. When the ambient temperature approaches the maximum operating temperature, the index approaches 1, indicating a decrease in system heat dissipation capacity and necessitating an appropriate increase in the allowable current threshold. Substituting the ambient temperature index, correction factor, and base current threshold into the calculation formula, the dynamic current threshold exhibits a linear growth trend with increasing temperature. The temperature influence coefficient serves as an adjustment factor, controlling the intensity of temperature compensation and preventing over-compensation that could lead to missed faults. For example, under high-temperature conditions, this model can increase the current threshold to 1.2-1.5 times the base value, accommodating normal changes such as increased conductor resistance due to temperature, while ensuring the threshold adjustment remains within a safe range through a safety margin coefficient.

[0042] Compared to existing technologies, traditional methods using a fixed current threshold are prone to false alarms in high-temperature environments due to normal temperature effects such as increased conductor resistance and reduced heat dissipation efficiency, while in low-temperature conditions, the threshold may be too high, leading to missed early faults. This solution establishes a dynamic correlation model between temperature and threshold, enabling the current protection value to adaptively adjust with ambient temperature. This avoids false alarms in high-temperature conditions while maintaining threshold sensitivity during abnormal temperature events. Compared to the conservative approach of simply widening the threshold safety range, this solution achieves precise control of the compensation intensity through a temperature influence coefficient, improving diagnostic accuracy while ensuring safety.

[0043] Through the above technical solution, this application effectively solves the problem of inaccurate current threshold setting caused by ambient temperature fluctuations. In high-temperature environments, it avoids false alarms caused by normal load fluctuations by appropriately increasing the threshold, and maintains fault detection sensitivity even when temperatures rise abnormally. Furthermore, this solution ensures the rationality and consistency of threshold adjustments under different temperature conditions through standardized processing of the temperature index and a constraint mechanism for the temperature influence coefficient, significantly reducing the interference of temperature factors on the fault diagnosis system.

[0044] This application further proposes that the modified model is as follows:

[0045] in, The value range is 0-1. As a correction factor, The basic dynamic coefficients are the load characteristic parameters of a mechanical system calculated by combining joint velocity, potential load coefficient, and inertial load coefficient. Specifically, they can be implemented using a weighted summation model to reflect the dynamic load changes of the mechanical system under different motion states. The electrical efficiency coefficient is a real-time efficiency parameter of an electrical system calculated based on bus voltage and power factor. Specifically, it can be achieved by the normalized product of voltage and power factor, and is used to characterize the operating efficiency of the electrical system. For safety margin coefficient, A value less than 1 indicates that the safety margin coefficient is a conservative adjustment parameter used to reserve a safety margin. The safety margin coefficient can be assigned a value based on expert experience, for example, between 0.85 and 0.95, to reduce the threshold setting value and enhance the system's fault tolerance.

[0046] Specifically, the correction model generates a dynamic correction factor by multiplying the basic dynamic coefficient, the electrical efficiency coefficient, and the safety margin coefficient. The basic dynamic coefficient captures the dynamic changes in mechanical load, such as inertial load fluctuations under high-speed motion or heavy-load conditions; the electrical efficiency coefficient reflects the real-time performance of the electrical system, such as efficiency losses caused by voltage dips or power factor declines; and the safety margin coefficient narrows the range of the correction factor through multiplication, providing a conservative adjustment for threshold calculation. After coupling these three factors, the correction factor can dynamically balance mechanical load demands, electrical performance fluctuations, and safety margins, thereby achieving adaptive adjustment of the current threshold through the subsequent threshold model. For example, when electrical efficiency decreases in a high-temperature environment, the correction factor automatically decreases, and the dynamic current threshold is adjusted accordingly to avoid overload misjudgments caused by electrical performance degradation.

[0047] Compared to existing technologies, traditional methods typically adjust the current threshold using a fixed threshold or a single parameter, such as linear compensation based solely on temperature, which cannot simultaneously adapt to sudden changes in mechanical load and fluctuations in electrical parameters. This solution achieves multi-dimensional dynamic threshold correction by integrating mechanical dynamic parameters, electrical efficiency parameters, and safety margin design, significantly improving the adaptability and reliability of the threshold setting under operating conditions.

[0048] Through the above technical solution, this application solves the problem of false alarms or missed alarms caused by fixed thresholds. It can dynamically adjust the current threshold according to changes in mechanical load, fluctuations in electrical performance and environmental factors. For example, it can automatically increase the threshold under heavy load and high speed conditions to avoid false alarms triggered by normal loads, and at the same time, it can reduce the threshold when electrical efficiency decreases to enhance the sensitivity of abnormal detection, thereby improving the accuracy and timeliness of fault warning.

[0049] This application further proposes that the specific steps for obtaining the electrical efficiency coefficient based on the bus voltage and power factor through the electrical efficiency model are as follows:

[0050] The current bus voltage and power factor are compared with the rated bus voltage and rated power factor, respectively, to obtain the bus voltage index and power factor index.

[0051] The bus voltage index and power factor index are imported into the formula. Obtain the electrical efficiency coefficient .

[0052] in, The value range is 0-1. The bus voltage index refers to the ratio of the current bus voltage to the rated value. Specifically, it can be achieved by collecting the bus voltage in real time using a voltage sensor and then performing a ratio calculation with the rated voltage stored in the control system. It is used to characterize the impact of power supply voltage fluctuations on electrical efficiency. The power factor index refers to the ratio of the current power factor to the rated value. Specifically, it can be achieved by measuring the real-time power factor with a power analyzer and calculating the ratio with the rated power factor. It is used to reflect the reactive power loss status of the motor drive system. Based on the basic efficiency coefficient, A value less than 1 indicates that the basic efficiency coefficient refers to the baseline efficiency parameter of the equipment under rated operating conditions. Specifically, it can be obtained by fitting the equipment's factory test data or historical operating data, or it can be assigned by expert experience to provide a unified efficiency evaluation benchmark for different models of equipment.

[0053] Specifically, by standardizing bus voltage and power factor separately, physical quantities with different dimensions are transformed into dimensionless exponents, eliminating the influence of parameter dimensional differences on the model. The two exponents are then multiplied and introduced into a base efficiency coefficient to form the electrical efficiency coefficient, which comprehensively reflects the coupled effect of voltage fluctuations and power factor changes on the electrical system efficiency. For example, when the bus voltage is lower than the rated value, the bus voltage exponent decreases, leading to a decrease in the electrical efficiency coefficient; when the power factor deviates from the rated value due to changes in motor load, the power factor exponent changes synchronously, further affecting the electrical efficiency coefficient. This dynamic evaluation mechanism can accurately capture the real-time efficiency status of the electrical system, providing precise input for subsequent correction factor calculations, thereby avoiding misjudgments caused by traditional methods neglecting dynamic changes in electrical efficiency.

[0054] Compared to existing technologies, traditional fault early warning methods typically only monitor whether the absolute value of the current exceeds a fixed threshold, without considering the impact of voltage fluctuations and power factor changes on system efficiency. For example, during voltage sags or power factor deterioration, even if the actual current does not exceed the fixed threshold, a decrease in system efficiency may already indicate a potential fault risk. This solution, by establishing an electrical efficiency model and incorporating key electrical parameters such as voltage and power factor into the efficiency evaluation system, can identify abnormal states caused by electrical performance degradation earlier, overcoming the technical deficiency that single current monitoring cannot distinguish between normal load fluctuations and true efficiency degradation.

[0055] Through the above technical solutions, this application can dynamically assess changes in the overall efficiency of electrical systems and accurately distinguish between current fluctuations under normal operating conditions and abnormal states caused by actual efficiency decline. For example, when voltage fluctuations lead to efficiency reduction, even if the actual current does not exceed the traditional fixed threshold, the decrease in the electrical efficiency coefficient can still trigger an early warning mechanism, thereby detecting potential faults caused by deteriorating power supply quality in advance. Simultaneously, by introducing the power factor index, it can identify increased reactive power losses caused by motor winding aging or capacitor failure, avoiding the problem of misjudging normal heavy-load conditions as faults due to a decrease in the power factor, significantly improving the accuracy of early warning and adaptability to operating conditions.

[0056] This application further proposes that the specific steps for obtaining the basic dynamic coefficients based on the joint velocity, potential load coefficient, and inertial load coefficient through the basic dynamic model are as follows:

[0057] The ratio of the current joint velocity to the maximum joint velocity is used to obtain the joint velocity index;

[0058] The joint velocity index, potential load coefficient, and inertial load coefficient are imported into the formula. Obtain the basic dynamic coefficients .

[0059] in, The value range is 0-1. Potential load factor refers to the quantitative value of the load effect caused by gravity. Specifically, it can be obtained by measuring the product of the load mass and the lever arm length corresponding to the joint angle. This parameter is used to reflect the impact of static load on the system. The inertial load factor is a quantitative value of the dynamic load effect caused by motion acceleration. Specifically, it can be achieved by detecting the acceleration component in the motor drive current. This parameter is used to characterize the impact of dynamic load on the system. The joint velocity index is the ratio of the current joint movement speed to the maximum speed allowed by the system. Specifically, it can be achieved by acquiring encoder data in real time and calculating the speed percentage. This parameter is used to quantify the degree of influence of the motion state on the mechanical load. , and All are weighting coefficients. , and All are greater than or equal to 0 and less than or equal to 1, and The weighting coefficient refers to the proportion of contribution of different load types to the overall dynamic state. Specifically, it can be dynamically adjusted through expert experience or machine learning algorithms. This parameter is used to balance the interaction between various load factors under different working conditions.

[0060] Specifically, by collecting joint motion velocities in real time and standardizing them with a preset maximum speed, a velocity index reflecting the current motion intensity is generated. The standardized velocity index is then linearly weighted and combined with pre-calculated potential load coefficients and inertial load coefficients. The potential load coefficient reflects the static load under gravity, the inertial load coefficient characterizes the dynamic load caused by motion acceleration, and the velocity index reflects the superposition effect of motion state on system load. By setting adjustable weighting coefficient combinations, the system can dynamically adjust the contribution ratio of each parameter according to actual operating conditions. For example, the weight of the potential load coefficient is increased under low-speed, high-load conditions, and the influence of the inertial load coefficient is enhanced under high-speed, direction-changing conditions. The resulting comprehensive dynamic coefficient accurately reflects the current composite load state of the mechanical system, providing precise input parameters for subsequent diagnostic threshold adjustments.

[0061] Compared to existing technologies, traditional methods typically monitor only a single current parameter or a fixed threshold to determine the load state, failing to distinguish the different effects of gravity loads and inertial loads. This solution, however, establishes a multi-parameter fusion model that can simultaneously capture the three-dimensional characteristics of static load, dynamic load, and motion state. For example, it accurately identifies the potential load-dominated condition when moving heavy objects at low speeds, and effectively detects changes in inertial load during high-speed, unloaded movement. This multi-dimensional parameter fusion mechanism overcomes the shortcomings of single-parameter monitoring, which is prone to misjudgment. For instance, when the system is simultaneously subjected to gravity and acceleration loads, traditional methods may misjudge the combined load as an abnormal current, while this solution accurately identifies normal operating conditions by decomposing each load component.

[0062] Through the above technical solution, this application can effectively distinguish the impact of different load types on the system, avoiding misjudging composite loads under normal operating conditions as abnormal states. For example, when the robot performs high-speed picking actions, the system can automatically increase the weight of the inertial load coefficient, accurately identify current fluctuations caused by acceleration as normal operating conditions, thereby avoiding false alarm triggering. At the same time, the dynamic weight adjustment mechanism enables the system to adapt to changes in load characteristics at different process stages. For example, when potential load dominates in the welding process, the system automatically adjusts the parameter weights, and enhances the monitoring sensitivity of inertial load in the handling process, thereby achieving accurate load state assessment.

[0063] This application further proposes that the specific steps for obtaining the inertial load coefficient based on the load rotational inertia and acceleration through the inertial load model are as follows:

[0064] The load moment of inertia and acceleration are compared with the maximum moment of inertia and the maximum angular acceleration, respectively, to obtain the load moment of inertia index and acceleration index.

[0065] Multiply the load's moment of inertia exponent and acceleration exponent to obtain the inertial load coefficient. , The value range is 0-1.

[0066] Among them, the load rotational inertia refers to the inertial measure of the robot's end effector and its load rotating around the joint axis. It can be calculated using 3D modeling software combined with mass distribution parameters, and is used to characterize the inertial resistance of the load on the joint drive system. Acceleration refers to the angular acceleration during joint movement, which can be obtained through differential calculation of servo motor encoder data, and is used to reflect the dynamic impact of changes in motion state on the inertial load. Ratio processing refers to the normalization calculation between the measured physical quantity and the system's maximum allowable value, which can be implemented using a linear proportional conversion algorithm, used to eliminate dimensional differences and quantify the deviation of actual working conditions from system limits. The inertial load coefficient is a dimensionless parameter that comprehensively reflects the coupling effect of rotational inertia and acceleration, specifically implemented using multiplication logic, and is used to accurately characterize the composite effect of inertial torque.

[0067] Specifically, by dividing the load's moment of inertia by the system's maximum permissible moment of inertia, an inertia exponent within the range of 0-1 is obtained. This exponent reflects the proportion of the current load inertia to the system's maximum load-bearing capacity. Simultaneously, dividing the measured angular acceleration by the system's maximum set angular acceleration yields an acceleration exponent within the range of 0-1. This exponent characterizes the relative relationship between the actual motion state and the ultimate acceleration capability. Multiplying these two exponents accurately replicates the product relationship between inertial torque, moment of inertia, and angular acceleration in classical mechanics, enabling the inertial load coefficient to dynamically reflect the actual load conditions at different motion stages. When the load's moment of inertia approaches the system's limit or a high-speed start-stop operation is performed, this coefficient will approach 1, triggering the dynamic adjustment mechanism of the subsequent threshold model.

[0068] Compared to existing technologies, traditional methods typically monitor only current or a single mechanical parameter, failing to accurately distinguish between inertial load variations under normal operating conditions and abnormal fault signals. This solution overcomes the limitations of single-parameter monitoring through dual-parameter coupled modeling and solves the problem of fusing parameters with different dimensions through normalization processing. Existing technologies using fixed threshold settings struggle to adapt to normal load fluctuations during high-speed motion, while the dynamic coefficients provided in this solution offer a precise input reference for subsequent threshold adjustments.

[0069] Through the above technical solution, this application effectively solves the problem of false alarms caused by inaccurate inertial load assessment, and can accurately distinguish between normal acceleration processes and abnormal overload states. When the robot performs high-speed sorting operations, this method can accurately calculate the reasonable load range during the acceleration phase, avoiding misjudging rapid start-stop required by the process as a motor overload fault. When handling heavy workpieces, the system can automatically identify the reasonable current demand of large inertial loads, preventing fixed thresholds from triggering shutdown protection prematurely.

[0070] This application further proposes that the specific steps for obtaining the potential load coefficient based on the potential load model using the load mass and joint angle are as follows:

[0071] Based on the joint angle, the lever arm length at the current angle is obtained through a lever arm simplification function;

[0072] The load mass and lever arm length at the current angle are compared with the maximum allowable load mass and the maximum lever arm length, respectively, to obtain the load mass index and the joint angle index.

[0073] Import the load mass index and joint angle index into the formula. Obtain the potential load coefficient .

[0074] in, The value range is 0-1. The load quality index is the ratio of the actual load quality to the maximum allowable load quality. Specifically, it can be obtained by dividing the load measured by the sensor by a preset threshold. It is used to reflect the proportion of the load relative to the design limit. The joint angle index is the ratio of the current lever arm length to the maximum lever arm length. It is calculated by substituting the joint angle obtained from an angle sensor into the lever arm function and is used to characterize the amplification effect of lever arm changes on torque under different postures. The potential load model is a mathematical model that calculates the potential load coefficient using load mass and joint angles. It can be implemented using a simplified lever arm function combined with the mass-to-angle ratio, and is used to quantify the impact of load states on the system. The simplified lever arm function is a functional relationship that converts joint angles into lever arm lengths; it can be implemented using a nonlinear expression containing a sine function to simulate the motion characteristics of the robotic arm.

[0075] Specifically, by collecting joint angle parameters in real time and calling a simplified lever arm function containing a sine function, the equivalent lever arm length under the current robotic arm configuration is dynamically calculated. Simultaneously, the actual load mass is normalized to the system's maximum allowable load mass to obtain a load mass index. Further, the calculated lever arm length is compared with the system's maximum lever arm length to generate a joint angle index. The minimum value of the two indices is then used as a constraint to ensure that the potential load coefficient does not exceed 1, thus outputting a standardized evaluation result. This process, by coupling the dynamic relationship between load mass and joint posture, solves the evaluation bias problem caused by traditional methods relying solely on static mass parameters. For example, when the robotic arm is at a specific angle, even if the load mass has not reached its limit, an excessively long equivalent lever arm may still trigger an increase in the coefficient through a multiplicative effect, achieving early warning.

[0076] Compared to existing technologies, traditional methods typically use fixed lever arm lengths or linear proportional models to assess potential loads, failing to accurately reflect the nonlinear impact of joint angle changes on torque. This solution, however, introduces a simplified lever arm model incorporating a sine function, better reflecting the actual kinematic characteristics of the robotic arm. Furthermore, combined with dual-parameter normalization, it effectively distinguishes between normal load variations and abnormal over-limit states. For example, in material handling operations, when the robotic arm extends to a large angle, existing technologies may underestimate the actual load torque by neglecting the lever arm growth effect, leading to missed detections. This solution, through dynamically calculating the coupling effect of lever arm length and mass index, can accurately identify potential risks in such conditions.

[0077] Through the above technical solution, this application can dynamically correct the potential load assessment results based on the real-time motion posture of the robotic arm, avoiding torque calculation errors caused by changes in joint angles. By normalizing the mass and lever arm parameters and applying product constraints, the combined influence of load mass and mechanical configuration can be accurately identified, reducing false alarms caused by single-parameter judgments or static models. For example, in high-speed motion or complex trajectory operations, this method can distinguish between normal load fluctuations required by the process and abnormal equipment conditions, improving the accuracy and timeliness of fault warnings.

[0078] This application further proposes that the simplified function of the lever arm is:

[0079] in, The reference lever arm length, The lever arm variation coefficient, Less than 1, This refers to the joint angle.

[0080] Among them, the length of the reference lever arm This refers to the effective lever arm length when the robotic arm is in a specific reference position. Specifically, it can be achieved by measuring the lever arm length when the robotic arm is in a horizontal position, used to establish a correspondence with the physical structure. The lever arm variation coefficient is a proportional coefficient used to adjust the magnitude of lever arm variation with angle. It can be determined through expert experience, experimental calibration, or simulation optimization, used to balance model simplification requirements with computational accuracy. The joint angle refers to the rotation angle of the robotic arm joint relative to the reference position, specifically measured using an encoder or angle sensor, used to characterize the actual pose of the robotic arm. Absolute value function. This refers to the operation of taking the absolute value of the result of the sine function calculation. It can be implemented through the mathematical operation module to eliminate the influence of the joint rotation direction on the calculation of the lever arm length.

[0081] Specifically, this function establishes a fundamental mechanical relationship based on the length of a reference lever arm, combined with... The trigonometric function terms dynamically reflect the nonlinear effect of joint angle changes on the lever arm length. When the joint angle... When the sine function term changes, the length of the lever arm is automatically adjusted. For example, when... When it is 90 degrees, When the maximum value of 1 is reached, the lever arm length expands to... ;when When it is 0 degrees or 180 degrees, The value is 0, and the lever arm length reverts to the baseline value. The introduction of the lever arm variation coefficient allows the model to adapt to the characteristics of different mechanical structures by adjusting the coefficient value. For example, for cases where the direction of gravity is perpendicular to the joint axis, the coefficient can be increased to enhance the effect of angle; for cases where the direction of gravity is parallel to the joint axis, the coefficient can be decreased to reduce angle sensitivity. Absolute value calculations ensure that a symmetrical change in lever arm length occurs when the joint rotates in either the forward or reverse direction, avoiding load calculation deviations due to incorrect direction judgment.

[0082] In some specific implementations, the reference lever arm length can be set as the vertical distance from the end of the robotic arm to the joint axis when the arm is in a horizontally extended state; the lever arm variation coefficient can be determined through load experiments, for example, by measuring the actual torque at multiple typical angles and obtaining the optimal value through fitting calculations; joint angle The data can be collected in real time by an optical encoder installed at the joint and transmitted to the controller for function calculation.

[0083] Compared to existing technologies, traditional methods typically simplify the lever arm length to a fixed value or only linearly relate it to the angle, leading to significant errors in potential load assessment under non-horizontal poses. For example, when the robotic arm vertically lifts a load, traditional models may underestimate the lever arm length, resulting in an underestimation of the load torque and an excessively low dynamic current threshold, failing to effectively identify abnormal loads. This proposed solution accurately characterizes the nonlinear relationship between the angle and the lever arm length using trigonometric functions, combined with experimentally calibrated coefficients, enabling the model to reflect actual physical laws while maintaining computational efficiency.

[0084] Through the above technical solution, this application solves the problem of load assessment error caused by the oversimplification of lever arm length calculation in traditional potential load models, and significantly improves the calculation accuracy of potential load coefficient under different joint poses. For example, under the condition of large-angle swing of the robotic arm, this function can accurately capture the dynamic characteristics of lever arm length changing with angle, thereby providing reliable mechanical parameters for subsequent dynamic current threshold generation and avoiding false alarms or missed alarms caused by load assessment deviations.

[0085] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for fault early warning and remote diagnosis of an industrial robot electrical control system, characterized in that, Includes the following steps: Based on the load mass and joint angle, the potential load coefficient is obtained through the potential load model. Based on the load's rotational inertia and acceleration, the inertial load coefficient is obtained through an inertial load model; Based on joint velocity, potential load coefficient, and inertial load coefficient, the basic dynamic coefficients are obtained through the basic dynamic model. The electrical efficiency coefficient is obtained based on the bus voltage and power factor through an electrical efficiency model. Based on the fundamental dynamic coefficients and electrical efficiency coefficients, correction factors are obtained through a modified model; Based on ambient temperature, correction factor and base current threshold, dynamic current threshold is obtained through threshold model; Based on vibration acceleration, servo loop following error, and dynamic current threshold, diagnostic information is output through a diagnostic model. The specific steps for obtaining the dynamic current threshold based on ambient temperature, correction factor, and base current threshold through a threshold model are as follows: The ambient temperature index is obtained by comparing the difference between the current ambient temperature and the lowest operating temperature with the difference between the highest operating temperature and the lowest operating temperature. Import ambient temperature index, correction factor, and base current threshold. Formula acquisition ,in, Based on the base current threshold, As a correction factor, The ambient temperature index. This is the temperature influence coefficient; The corrected model is as follows: ,in, As a correction factor, Basic dynamic coefficients, The electrical efficiency coefficient, This is the safety margin coefficient.

2. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 1, characterized in that, The specific steps for outputting diagnostic information through the diagnostic model based on vibration acceleration, servo loop following error, and dynamic current threshold are as follows: The current vibration acceleration and servo loop following error are compared with the maximum permissible vibration acceleration and the maximum permissible following error, respectively, to obtain the vibration acceleration index and the servo loop following error index. Vibration acceleration, servo loop following error exponent, and dynamic current threshold are imported into the formula. Get a health score ,in, The value range is 0-1. This is a normal state; continue monitoring. 0.70 To monitor the status, increase the monitoring frequency to 0.

5. In a state of alert, planned inspections are conducted. If the machine is in maintenance condition, stop it immediately for maintenance. For dynamic current threshold, This is the actual current. The vibration acceleration index, The servo loop following error index. , and All are weighting coefficients.

3. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 1, characterized in that, The specific steps for obtaining the electrical efficiency coefficient based on bus voltage and power factor through an electrical efficiency model are as follows: The current bus voltage and power factor are compared with the rated bus voltage and rated power factor, respectively, to obtain the bus voltage index and power factor index. The bus voltage index and power factor index are imported into the formula. Obtain the electrical efficiency coefficient ,in, This refers to the bus voltage index. The power factor index, It is the basic efficiency coefficient.

4. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 3, characterized in that, The specific steps for obtaining the basic dynamic coefficients based on joint velocity, potential load coefficient, and inertial load coefficient through the basic dynamic model are as follows: The ratio of the current joint velocity to the maximum joint velocity is used to obtain the joint velocity index; The joint velocity index, potential load coefficient, and inertial load coefficient are imported into the formula. Obtain the basic dynamic coefficients ,in, The potential loading coefficient, The inertial load factor, The joint velocity index, , and All are weighting coefficients.

5. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 4, characterized in that, The specific steps for obtaining the inertial load coefficient based on the load's rotational inertia and acceleration through an inertial load model are as follows: The load moment of inertia and acceleration are compared with the maximum moment of inertia and the maximum angular acceleration, respectively, to obtain the load moment of inertia index and acceleration index. Multiply the load's moment of inertia exponent and acceleration exponent to obtain the inertial load coefficient. .

6. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 4, characterized in that, The specific steps for obtaining the potential load coefficient based on the load mass and joint angle through the potential load model are as follows: Based on the joint angle, the lever arm length at the current angle is obtained through a lever arm simplification function; The load mass and lever arm length at the current angle are compared with the maximum allowable load mass and the maximum lever arm length, respectively, to obtain the load mass index and the joint angle index. Import the load mass index and joint angle index into the formula. Obtain the potential load coefficient ,in, The load quality index, This is the joint angle index.

7. The fault early warning and remote diagnosis method for the electrical control system of an industrial robot according to claim 6, characterized in that, The simplified function of the lever arm is: ,in, The reference lever arm length, The lever arm variation coefficient, This refers to the joint angle.

Images