A method for dynamic overload detection and protection of servo motors based on thermal accumulation model

Abstract

This invention discloses a dynamic overload detection and protection method for servo motors based on a thermal accumulation model, relating to the field of servo system fault protection application technology. The method includes constructing an equivalent thermal accumulation model, dynamically detecting overload states, executing heat dissipation recovery strategies, implementing multi-level protection mechanisms, adaptive environmental compensation, and historical data learning functions. By discretizing overload curve data and employing nonlinear accelerated calculation strategies, it avoids the protection lag and false triggering problems caused by traditional single-parameter detection, ensuring timely response of the motor under light load, heavy load, and load fluctuation scenarios, reducing the risk of motor burnout. By integrating an ambient temperature sensor and a heat dissipation condition evaluation unit, it dynamically adjusts the thermal accumulation coefficient and heat dissipation attenuation parameters to solve the model deviation problem caused by environmental interference. When heat dissipation deterioration and abnormal temperature feedback are detected, it automatically extends the heat dissipation recovery cycle and corrects the calculation logic, improving the continuous operational reliability of the system.

Description

Technical Field

[0001] This invention relates to the field of servo system fault protection application technology, and in particular to a method for dynamic overload detection and protection of servo motors based on a thermal accumulation model. Background Technology

[0002] Servo motors can control speed and have very accurate positioning. They convert voltage signals into torque and speed to drive the controlled object. The rotor speed of a servo motor is controlled by the input signal and can respond quickly. In automatic control systems, they are used as actuators and have the characteristics of small electromechanical time constant and high linearity. They can convert the received electrical signals into angular displacement and angular velocity output on the motor shaft.

[0003] In the field of servo motor operation control, traditional overload detection methods have the following drawbacks: they cannot quantify the combined effect of load duration on heat accumulation in real time, relying solely on current thresholds and temperature sensors for single-parameter judgments, resulting in insufficient protection sensitivity under light load conditions and protection lag due to thermal inertia under heavy load conditions; they lack dynamic adaptability, and fixed protection thresholds are difficult to match the drastic load fluctuations in industrial robots and CNC machine tools, leading to decreased equipment utilization and accelerated aging of motor insulation; they neglect closed-loop management of the heat dissipation stage, allowing residual heat to accumulate continuously after the motor recovers from overload, increasing the risk of secondary overloads and lacking a precise calibration mechanism for heat dissipation rate; changes in ambient temperature and heat dissipation conditions are not included in the compensation system, leading to increased deviations in heat accumulation calculations, and the deviation between measured temperature and model predictions cannot be self-corrected; historical overload patterns are not effectively explored and utilized, protection thresholds are rigid, and dynamic optimization of warning parameters cannot be achieved through load periodicity.

[0004] Therefore, a dynamic overload detection and protection method for servo motors based on a thermal accumulation model is proposed to solve the above problems. Summary of the Invention

[0005] The main objective of this invention is to provide a method for dynamic overload detection and protection of servo motors based on a thermal accumulation model, so as to solve the problems mentioned in the background above.

[0006] To achieve the above objectives, the technical solution adopted by this invention is as follows: a method for dynamic overload detection and protection of servo motors based on a thermal accumulation model, comprising the following steps:

[0007] S1. Construct an equivalent heat accumulation model: Obtain motor overload curve data, discretize it into a load-time correspondence table, establish a mapping model between load value and equivalent heat generation coefficient, set a total heat capacity threshold, accumulate heat generation value in real time, and trigger protection when the ratio is greater than 1.

[0008] S2. Dynamic overload detection: The load value is obtained in real time through current sampling and encoder conversion. The allowable duration is obtained by querying the corresponding table and the timing is started. When the load continues to exceed the rated value, the cumulative heating amount is accumulated using a nonlinear acceleration model. For load fluctuations, the root mean square of the load value within the window is calculated using the sliding window method as the equivalent load value.

[0009] S3. Implement heat dissipation recovery strategy: calibrate the natural heat dissipation rate of the motor, establish a temperature-time decay curve, and when the load drops below the rated value, change the heat accumulation decay amount according to the preset decay rate, set a safe heat accumulation threshold, and allow overload operation only when the real-time value is lower than the threshold.

[0010] S4. Implement a multi-level protection mechanism: Set early warning thresholds and protection thresholds, and calculate the heat accumulation through closed-loop correction of the measured temperature by the thermistor.

[0011] S5. Adaptive environmental compensation: The integrated temperature sensor dynamically adjusts the heat accumulation coefficient, monitors heat dissipation conditions, and increases the heat dissipation correction coefficient when conditions deteriorate.

[0012] S6. Historical data learning function: Record overload event parameters, analyze patterns using time series algorithms, and optimize early warning and protection thresholds based on machine learning.

[0013] Preferably, the discretization process in step S1 specifically includes:

[0014] Motor overload curve data is obtained through experimental calibration and manufacturer technical documents. The curve is expressed as torque percentage in load units and allowable duration in time units.

[0015] The continuous curve is divided according to a preset load interval, which is set to 5% of the rated torque increment, to generate a discretized load-time correspondence table.

[0016] Each set of data corresponds to the safe duration threshold under the load value. When the load value is 110% of the rated torque, the allowable duration is 10 seconds, and when the load value is 150% of the rated torque, the allowable duration is 3 seconds.

[0017] The discretization process uses a linear interpolation algorithm to compensate for the nonlinear region of the curve, ensuring that the accuracy of the data points is within 1%.

[0018] Preferably, the mapping model between the load value and the equivalent heating coefficient in step S1 is as follows:

[0019] The load value and the equivalent heat generation coefficient are linearly positively correlated; that is, for every 10% increase in the load value, the equivalent heat generation coefficient increases by 0.13 times.

[0020] The coefficient is 1.0 for rated load and 1.2 for 120% overload.

[0021] In the calculation of heat accumulation per unit time, the time increment is set to a millisecond-level sampling period, and the heat accumulation increment is proportional to the load value.

[0022] The mapping model was obtained by fitting the experimental data using the least squares method, with the fitting error controlled within 2% to ensure the model's adaptability under light and heavy load conditions.

[0023] Preferably, in the accelerated model accumulation in step S2:

[0024] The cumulative heat increment is calculated using a nonlinear acceleration model, and the specific formula is as follows:

[0025] ;

[0026] in The cumulative heat increment per unit time This represents the real-time motor load value. This is the rated load value of the motor. As a time increment, this formula is activated only when the load continuously exceeds the rated load by more than 10%. The heat accumulation rate is 1.21 times the baseline value at 110% load and 2.25 times at 150% load to reflect the square-law nonlinear effect of increased load on heat accumulation.

[0027] The activation of the nonlinear acceleration model requires two conditions to be met:

[0028] Load threshold: Real-time load is greater than 110% of rated load;

[0029] Time threshold: Overload state lasting longer than 100ms;

[0030] When load fluctuations cause discontinuous operation, the cumulative duration of continuous overload is used as the criterion. The time threshold is determined by the motor's thermal inertia constant, ensuring that the model is activated only during the effective heat accumulation phase.

[0031] During the accumulation process, the ratio of the accumulated heat to the total heat capacity threshold is compared in real time. The ratio calculation sampling frequency is 100Hz to ensure timeliness.

[0032] Preferably, the sliding window method in step S2 specifically includes:

[0033] Dynamic window duration: The base duration is 5 seconds, which is dynamically adjusted according to the standard deviation σ of the load fluctuation within the window;

[0034] Duration adjustment formula:

[0035] ;

[0036] in The rated load value is σ, which is calculated in real time using the variance of the load value within the window.

[0037] The root mean square of the load samples within the calculation window is used as the equivalent load value. The calculation formula is as follows:

[0038] ;

[0039] in This is the equivalent load value. This represents the number of sampling points within the window. The load value at the i-th sampling point;

[0040] Root mean square (RMS) calculation is applicable in handling load fluctuation scenarios to reduce false triggering;

[0041] When the equivalent load value is included in the thermal accumulation calculation, it is linked with the load-time correspondence table to ensure the stability of the protection action when the load fluctuates.

[0042] Preferably, the cumulative thermal decay in step S3 is specifically:

[0043] ;

[0044] in This is the cumulative thermal decay. The heat dissipation rate constant is For time increments;

[0045] The heat dissipation rate constant was calibrated through no-load experiments, with a baseline value of 0.05 / second set at an ambient temperature of 25℃.

[0046] The attenuation model adopts a piecewise function: linear attenuation is used in the high temperature zone and exponential attenuation is used in the low temperature zone to match the thermal inertia of the motor. The high temperature zone is >40℃ and the low temperature zone is ≤40℃.

[0047] The safe heat accumulation threshold is set at 40% of the total heat capacity threshold and is dynamically adjusted through temperature feedback.

[0048] Preferably, in the multi-level protection mechanism described in step S4:

[0049] The warning threshold is set to 75% of the total heat capacity threshold. When triggered, the motor output power is limited to 80% of the rated value, and the warning signal is activated.

[0050] The protection threshold is set to 97% of the total heat capacity. When triggered, it will force a shutdown and generate an audible and visual alarm.

[0051] The closed-loop correction uses a thermistor to collect the measured temperature of the stator winding. When the deviation between the measured temperature and the calculated heat accumulation exceeds 10%, the model correction module is activated, and the correction coefficient is calculated based on the deviation ratio.

[0052] The correction process uses a sampling frequency of 50Hz to ensure real-time performance.

[0053] Preferably, in the environmental compensation described in step S5:

[0054] For every 10°C increase in ambient temperature, the heat accumulation calculation coefficient increases by 0.15 times;

[0055] Heat dissipation conditions are assessed by monitoring fan speed. A 20% decrease in fan speed is considered a deterioration in heat dissipation, at which point the heat dissipation rate constant drops to 60% of its normal value.

[0056] The compensation coefficients are stored in a lookup table, and the table data is generated based on thermal simulation experiments, covering temperatures up to 60°C.

[0057] The compensation mechanism is linked to the historical data learning function, and the correction coefficient is automatically adjusted when heat dissipation deterioration events occur frequently.

[0058] Preferably, in the historical data learning function of step S6:

[0059] Overload event parameters include three-dimensional data of load value, duration, and ambient temperature, which are stored in the operation log;

[0060] The time series analysis algorithm uses an autoregressive integral moving average model to uncover the periodic patterns of the load.

[0061] The machine learning algorithm uses a support vector machine. The input parameters include historical load peaks, daily average overload frequency, and average heat dissipation conditions. The output is dynamically adjusted warning thresholds and protection thresholds, with the adjustment range being less than 15% of the initial value.

[0062] The training dataset covers 1000 hours of running data, and the model is updated every 24 hours.

[0063] Preferably, it also includes a fault diagnosis linkage strategy:

[0064] When the cumulative protection threshold is triggered 3 times / 24 hours, an automatic motor insulation aging early warning report is generated. The report includes the cumulative heat peak, event time, and environmental conditions.

[0065] Overload events are linked to motor operating conditions and stored to form a thermal stress life map. The map uses load-time as the coordinate axis to visualize the history of thermal accumulation.

[0066] The present invention has the following beneficial effects:

[0067] 1. In this invention, by setting up a thermal accumulation modeling module, an equivalent thermal accumulation model is constructed based on the combined influence of load and time during the overload detection process of the servo motor. The dynamic changes of the load are converted into thermal accumulation in real time, and multi-level protection thresholds are set to improve the accuracy of overload state judgment. By discretizing the overload curve data and using a nonlinear acceleration calculation strategy, the protection lag and false triggering problems caused by traditional single-parameter detection are avoided, ensuring that the motor can respond in a timely manner under light load, heavy load and load fluctuation scenarios, and reducing the risk of motor burnout.

[0068] 2. In this invention, by setting a dynamic compensation module, the natural heat dissipation rate is calibrated in real time during the heat dissipation recovery phase, and a heat accumulation decay mechanism is activated. Combined with temperature feedback closed-loop correction of the heat accumulation calculation value, the heat accumulation coefficient and heat dissipation decay parameter are dynamically adjusted by integrating an ambient temperature sensor and a heat dissipation condition evaluation unit to solve the model deviation problem caused by environmental interference. When heat dissipation deterioration and abnormal temperature feedback are detected, the heat dissipation recovery cycle is automatically extended and the calculation logic is corrected, thereby avoiding the risk of secondary overload from the root and improving the reliability of continuous system operation.

[0069] 3. In this invention, by setting an adaptive learning module, the load spectrum, duration, and environmental parameters of overload events are recorded in historical operation data analysis, and time series algorithms are used to mine the periodic patterns of the load; the warning threshold and protection threshold are dynamically optimized based on machine learning models, so that the protection strategy can adaptively match the needs of different working conditions; by constructing a thermal stress life spectrum to visualize the historical overload distribution, intelligent prediction of motor health status and maintenance decision support are realized, fundamentally improving the adaptability and life management capability of the overload protection system. Attached Figure Description

[0070] Figure 1 This is a flowchart of a dynamic overload detection and protection method for servo motors based on a thermal accumulation model, according to the present invention. Detailed Implementation

[0071] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] Please see Figure 1 This method for dynamic overload detection and protection of servo motors based on a thermal accumulation model includes the following steps:

[0073] S1. Construct an equivalent heat accumulation model: Obtain motor overload curve data, discretize it into a load-time correspondence table, establish a mapping model between load value and equivalent heat generation coefficient, set a total heat capacity threshold, accumulate heat generation value in real time, and trigger protection when the ratio is greater than 1.

[0074] S2. Dynamic overload detection: The load value is obtained in real time through current sampling and encoder conversion. The allowable duration is obtained by querying the corresponding table and the timing is started. When the load continues to exceed the rated value, the cumulative heating amount is accumulated using a nonlinear acceleration model. For load fluctuations, the root mean square of the load value within the window is calculated using the sliding window method as the equivalent load value.

[0075] S3. Implement heat dissipation recovery strategy: calibrate the natural heat dissipation rate of the motor, establish a temperature-time decay curve, and when the load drops below the rated value, change the heat accumulation decay amount according to the preset decay rate, set a safe heat accumulation threshold, and allow overload operation only when the real-time value is lower than the threshold.

[0076] S4. Implement a multi-level protection mechanism: Set early warning thresholds and protection thresholds, and calculate the heat accumulation through closed-loop correction of the measured temperature by the thermistor.

[0077] S5. Adaptive environmental compensation: The integrated temperature sensor dynamically adjusts the heat accumulation coefficient, monitors heat dissipation conditions, and increases the heat dissipation correction coefficient when conditions deteriorate.

[0078] S6. Historical data learning function: Record overload event parameters, analyze patterns using time series algorithms, and optimize early warning and protection thresholds based on machine learning.

[0079] The discretization process in step S1 specifically includes:

[0080] Motor overload curve data is obtained through experimental calibration and manufacturer technical documents. The curve is expressed as torque percentage in load units and allowable duration in time units.

[0081] The continuous curve is divided according to a preset load interval, which is set to 5% of the rated torque increment, to generate a discretized load-time correspondence table.

[0082] Each set of data corresponds to the safe duration threshold under the load value. When the load value is 110% of the rated torque, the allowable duration is 10 seconds, and when the load value is 150% of the rated torque, the allowable duration is 3 seconds.

[0083] The discretization process uses a linear interpolation algorithm to compensate for the nonlinear region of the curve, ensuring that the accuracy of the data points is within 1%.

[0084] The mapping model between the load value and the equivalent heat generation coefficient in step S1 is as follows:

[0085] The load value and the equivalent heat generation coefficient are linearly positively correlated; that is, for every 10% increase in the load value, the equivalent heat generation coefficient increases by 0.13 times.

[0086] The coefficient is 1.0 for rated load and 1.2 for 120% overload.

[0087] In the calculation of heat accumulation per unit time, the time increment is set to a millisecond-level sampling period, and the heat accumulation increment is proportional to the load value.

[0088] The mapping model was obtained by fitting the experimental data using the least squares method, with the fitting error controlled within 2% to ensure the model's adaptability under light and heavy load conditions.

[0089] In the accelerated model accumulation of step S2:

[0090] The cumulative heat increment is calculated using a nonlinear acceleration model, and the specific formula is as follows:

[0091] ;

[0092] in The cumulative heat increment per unit time This represents the real-time motor load value. This is the rated load value of the motor. As a time increment, this formula is activated only when the load continuously exceeds the rated load by more than 10%. The heat accumulation rate is 1.21 times the baseline value at 110% load and 2.25 times at 150% load to reflect the square-law nonlinear effect of increased load on heat accumulation.

[0093] The activation of the nonlinear acceleration model requires two conditions to be met:

[0094] Load threshold: Real-time load is greater than 110% of rated load;

[0095] Time threshold: Overload state lasting longer than 100ms;

[0096] During the accumulation process, the ratio of the accumulated heat to the total heat capacity threshold is compared in real time. The ratio calculation sampling frequency is 100Hz to ensure timeliness.

[0097] The sliding window method in step S2 specifically includes:

[0098] Dynamic window duration: The base duration is 5 seconds, which is dynamically adjusted according to the standard deviation σ of the load fluctuation within the window;

[0099] Duration adjustment formula:

[0100] ;

[0101] in The rated load value is σ, which is calculated in real time using the variance of the load value within the window.

[0102] The root mean square of the load samples within the calculation window is used as the equivalent load value. The calculation formula is as follows:

[0103] ;

[0104] in This is the equivalent load value. This represents the number of sampling points within the window. The load value at the i-th sampling point;

[0105] Root mean square (RMS) calculation is applicable in handling load fluctuation scenarios to reduce false triggering;

[0106] When the equivalent load value is included in the thermal accumulation calculation, it is linked with the load-time correspondence table to ensure the stability of the protection action when the load fluctuates.

[0107] The specific amount of heat accumulation attenuation in step S3 is as follows:

[0108] ;

[0109] in This is the cumulative thermal decay. The heat dissipation rate constant is For time increments;

[0110] The heat dissipation rate constant was calibrated through no-load experiments, with a baseline value of 0.05 / second set at an ambient temperature of 25℃.

[0111] The attenuation model adopts a piecewise function: linear attenuation is used in the high temperature zone and exponential attenuation is used in the low temperature zone to match the thermal inertia of the motor. The high temperature zone is >40℃ and the low temperature zone is ≤40℃.

[0112] The safe heat accumulation threshold is set at 40% of the total heat capacity threshold and is dynamically adjusted through temperature feedback.

[0113] In the multi-level protection mechanism of step S4:

[0114] The warning threshold is set to 75% of the total heat capacity threshold. When triggered, the motor output power is limited to 80% of the rated value, and the warning signal is activated.

[0115] The protection threshold is set to 97% of the total heat capacity. When triggered, it will force a shutdown and generate an audible and visual alarm.

[0116] The closed-loop correction uses a thermistor to collect the measured temperature of the stator winding. When the deviation between the measured temperature and the calculated heat accumulation exceeds 10%, the model correction module is activated, and the correction coefficient is calculated based on the deviation ratio.

[0117] The correction process uses a sampling frequency of 50Hz to ensure real-time performance.

[0118] In step S5, environmental compensation:

[0119] For every 10°C increase in ambient temperature, the heat accumulation calculation coefficient increases by 0.15 times;

[0120] Heat dissipation conditions are assessed by monitoring fan speed. A 20% decrease in fan speed is considered a deterioration in heat dissipation, at which point the heat dissipation rate constant drops to 60% of its normal value.

[0121] The compensation coefficients are stored in a lookup table, and the table data is generated based on thermal simulation experiments, covering temperatures up to 60°C.

[0122] The compensation mechanism is linked to the historical data learning function, and the correction coefficient is automatically adjusted when heat dissipation deterioration events occur frequently.

[0123] In the historical data learning function of step S6:

[0124] Overload event parameters include three-dimensional data of load value, duration, and ambient temperature, which are stored in the operation log;

[0125] The time series analysis algorithm uses an autoregressive integral moving average model to uncover the periodic patterns of the load.

[0126] The machine learning algorithm uses a support vector machine. The input parameters include historical load peaks, daily average overload frequency, and average heat dissipation conditions. The output is dynamically adjusted warning thresholds and protection thresholds, with the adjustment range being less than 15% of the initial value.

[0127] The training dataset covers 1000 hours of running data, and the model is updated every 24 hours.

[0128] It also includes fault diagnosis linkage strategies:

[0129] When the cumulative protection threshold is triggered 3 times / 24 hours, an automatic motor insulation aging early warning report is generated. The report includes the cumulative heat peak, event time, and environmental conditions.

[0130] Overload events are linked to motor operating conditions and stored to form a thermal stress life map. The map uses load-time as the coordinate axis to visualize the history of thermal accumulation.

[0131] Implementation 1: Multi-condition adaptive protection for CNC machine tool spindle motors

[0132] When the spindle servo motor of the machining center is used under heavy cutting conditions, this method is applied. In the heat accumulation model construction stage, the heat dissipation parameters are calibrated through no-load heating experiment: at an ambient temperature of 40℃, it takes 210 seconds for the motor to cool from 120℃ to 80℃. The heat dissipation rate constant a is fitted to be 0.033 / second.

[0133] The dynamic detection module sets two sliding windows for intermittent cutting load characteristics: the short window handles the transient impact of the cutting teeth entering the cutter, and the long window evaluates the continuous load effect. When milling titanium alloys, when the load peak reaches the rated value, the system automatically activates the square law to accelerate the accumulation, so that the heat accumulation rate is increased to the reference value.

[0134] Multi-level protection mechanism linked to the cooling system: When the heat accumulation reaches the warning threshold, the PLC controls the oil cooler to increase the flow rate; when the protection threshold is reached, the tool retraction and shutdown are executed immediately. The environmental compensation module automatically increases the heat accumulation coefficient in the summer workshop through the temperature and humidity sensor built into the control cabinet.

[0135] The load fluctuation frequency of the finishing process is several times that of the normal working condition. Based on this, the adaptive learning module shortens the sliding window duration. After the fault diagnosis system generates a warning of spindle bearing lubrication deterioration after multiple protection shutdowns in a single day, the grease is confirmed to be carbonized and replaced in advance to avoid motor burnout. This implementation reduces the temperature rise of the motor under continuous heavy cutting conditions and extends tool life.

[0136] Implementation 2: Optimized heat dissipation control for multi-motor collaboration in logistics sorting lines

[0137] In the express sorting system, eight conveyor belt servo motors are controlled in a centralized manner. The system synchronously collects the load of each motor through a bus. The heat accumulation model sets different parameters for parallel operation characteristics: the total heat capacity threshold of the inlet motor is set to 4500J due to frequent start and stop; the continuous operation motor at the outlet is set to 5500J.

[0138] During the dynamic detection phase, when package jamming causes a sudden increase in the load on motor 3, the nonlinear acceleration model causes the accumulated heat to reach the protection threshold, triggering the motor to reverse urgently to release the jam. The heat dissipation recovery strategy innovatively introduces neighboring machine collaboration: when motor 5 enters the decay phase due to poor heat dissipation, the scheduling system automatically transfers its load to the adjacent motors 4 and 6 to improve their heat dissipation rate. The environmental compensation module detects the peak temperature period in the workshop during the summer afternoon and automatically lowers the warning threshold of the entire system.

[0139] Analysis of the historical learning module revealed that the peak load on Mondays was higher than the average. Based on this, the LSTM model was trained to activate the enhanced cooling mode in advance. In scenarios where heat dissipation deteriorates, the system automatically lowers the safety threshold and prohibits load distribution operations. After three months of continuous operation, the system has reduced the downtime of the motor group under peak conditions, reduced the overall energy consumption, and the thermal stress life spectrum shows that the aging rate of the motor insulation has slowed down.

[0140] Implementation 3: Overload protection and energy efficiency optimization of the injection molding machine's mold clamping servo motor

[0141] In large-scale injection molding production lines, the servo motor of the mold clamping mechanism adopts this method to achieve dynamic protection and energy efficiency coordinated control. During the system initialization stage, a discretized load-time correspondence table is constructed based on the motor thermal characteristic curve: the maximum allowable duration of the high-pressure mold clamping stage in the mold clamping cycle is set, while no limit is set for the low-pressure mold moving stage; the dynamic detection module monitors the crankshaft angle in real time through a high-precision encoder, and automatically identifies the load state in combination with the feedback from the pressure sensor. When foreign objects in the mold cause an abnormal increase in mold clamping resistance, the load value jumps, and the nonlinear acceleration model immediately increases the heat accumulation rate to several times the benchmark value, so that the system triggers emergency mold opening protection when the heat accumulation reaches the total capacity threshold, thus avoiding deformation of the crank mechanism.

[0142] The heat dissipation recovery strategy is specially designed for the high temperature and high humidity environment of the injection molding workshop: During the mold cooling cycle, the system automatically starts the forced air cooling device to increase the heat dissipation rate constant. The closed-loop correction module automatically adjusts the heat accumulation calculation benchmark when it detects abnormal local temperature rise of the mold through the temperature sensor embedded in the mold plate, eliminating the model deviation caused by uneven heat conduction of the mold. The historical learning module analyzes three months of production data and finds that the peak clamping load is higher when producing polycarbonate material than when producing ABS material. Based on this, a material-load correlation database is established, and the protection threshold parameters are pre-adjusted when switching materials.

[0143] The adaptive compensation system is linked to the workshop environment monitoring network: when the ambient temperature in summer exceeds 35℃, the warning threshold is automatically lowered from the total capacity; when a decrease in cooling tower efficiency is detected, the backup refrigeration unit is immediately activated and the mold opening heat dissipation time is extended. In terms of energy efficiency optimization, the system analyzes the load curve in the mold closing cycle to identify that the high-pressure mold locking stage can be shortened without affecting product quality. The fault prediction mechanism automatically generates a guide rail lubrication inspection work order after multiple abnormal overloads occur. Maintenance personnel check and confirm the wear of the guide rail and replace it in time to avoid unplanned downtime losses. After implementing this solution, the peak temperature of the injection molding machine motor windings is reduced, the quarterly maintenance cost is reduced, and the unit energy consumption is reduced.

[0144] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic overload detection and protection of servo motors based on a thermal accumulation model, characterized in that, Includes the following steps: S1. Construct an equivalent heat accumulation model: Obtain motor overload curve data, discretize it into a load-time correspondence table, establish a mapping model between load value and equivalent heat coefficient, set a total heat capacity threshold, accumulate heat value in real time, and trigger protection when the ratio of heat accumulation to total heat capacity threshold is greater than 1. The discretization process in step S1 specifically includes: Motor overload curve data is obtained through experimental calibration and manufacturer technical documents. The curve is expressed as torque percentage in load units and allowable duration in time units. The continuous curve is divided according to a preset load interval, which is set to 5% of the rated torque increment, to generate a discretized load-time correspondence table. Each set of data corresponds to the safe duration threshold under the load value. When the load value is 110% of the rated torque, the allowable duration is 10 seconds, and when the load value is 150% of the rated torque, the allowable duration is 3 seconds. The discretization process uses a linear interpolation algorithm to compensate for the nonlinear region of the curve, ensuring that the accuracy of the data points is within 1%. S2. Dynamic overload detection: The load value is obtained in real time through current sampling and encoder conversion. The allowable duration is obtained by querying the corresponding table and the timing is started. When the load continues to exceed the rated value, the cumulative heating amount is accumulated using a nonlinear acceleration model. For load fluctuations, the root mean square of the load value within the window is calculated using the sliding window method as the equivalent load value. S3. Implement heat dissipation recovery strategy: calibrate the natural heat dissipation rate of the motor, establish a temperature-time decay curve, and when the load drops below the rated value, change the heat accumulation decay amount according to the preset decay rate, set a safe heat accumulation threshold, and allow overload operation only when the real-time value is lower than the threshold. S4. Implement a multi-level protection mechanism: Set early warning thresholds and protection thresholds, and calculate the heat accumulation through closed-loop correction of the measured temperature by the thermistor. S5. Adaptive environmental compensation: The integrated temperature sensor dynamically adjusts the heat accumulation coefficient, monitors heat dissipation conditions, and increases the heat dissipation correction coefficient when conditions deteriorate. S6. Historical data learning function: Record overload event parameters, analyze patterns using time series algorithms, and optimize early warning and protection thresholds based on machine learning.

2. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, The mapping model between the load value and the equivalent heat generation coefficient in step S1 is as follows: The load value and the equivalent heat generation coefficient are linearly positively correlated; that is, for every 10% increase in the load value, the equivalent heat generation coefficient increases by 0.13 times. The coefficient corresponding to the rated load is 1.0, and the equivalent heat generation coefficient at 120% overload is 1.26; In the calculation of heat accumulation per unit time, the time increment is set to a millisecond-level sampling period, and the heat accumulation increment is proportional to the load value. The mapping model was obtained by fitting the experimental data using the least squares method, with the fitting error controlled within 2% to ensure the model's adaptability under light and heavy load conditions.

3. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, In the accelerated model accumulation in step S2: The cumulative heat increment is calculated using a nonlinear acceleration model, and the specific formula is as follows: Where ΔH is the cumulative heat increment per unit time, L actual L represents the real-time motor load value. rated The rated load value of the motor is given by Δt, which is the time increment. This formula is activated only when the load continuously exceeds the rated load by more than 10%. The heat accumulation rate is 1.21 times the baseline value at 110% load and 2.25 times at 150% load to reflect the square-law nonlinear effect of increased load on heat accumulation. During the accumulation process, the ratio of the accumulated heat to the total heat capacity threshold is compared in real time. The ratio calculation sampling frequency is 100Hz to ensure timeliness.

4. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, The sliding window method in step S2 specifically includes: The window duration is adjustable up to 5 seconds, and the window sliding step is synchronized with the sampling period of the control system. The root mean square of the load samples within the calculation window is used as the equivalent load value. The calculation formula is as follows: Where L eq The equivalent load value is given by n, where n is the number of sampling points within the window, and L is the value of L. i The load value at the i-th sampling point; Root mean square (RMS) calculation is applicable in handling load fluctuation scenarios to reduce false triggering; When the equivalent load value is included in the thermal accumulation calculation, it is linked with the load-time correspondence table to ensure the stability of the protection action when the load fluctuates.

5. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, The specific amount of heat accumulation attenuation in step S3 is as follows: ΔH decay =-a·△t Where ΔH decay denoted as the cumulative thermal decay, a as the heat dissipation rate constant, and Δt as the time increment; The heat dissipation rate constant was calibrated through no-load experiments, with a baseline value of 0.05 / second set at an ambient temperature of 25℃. The attenuation model adopts a piecewise function: linear attenuation is used in the high temperature region and exponential attenuation is used in the low temperature region to match the thermal inertia of the motor. The high temperature region is >40℃ and the low temperature region is ≤40℃. The safe heat accumulation threshold is set at 40% of the total heat capacity threshold and is dynamically adjusted through temperature feedback.

6. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, In the multi-level protection mechanism described in step S4: The warning threshold is set to 75% of the total heat capacity threshold. When triggered, the motor output power is limited to 80% of the rated value, and the warning signal is activated. The protection threshold is set to 97% of the total heat capacity threshold. When triggered, the machine will be forced to shut down and an audible and visual alarm will be generated. The closed-loop correction uses a thermistor to collect the measured temperature of the stator winding. When the deviation between the measured temperature and the calculated heat accumulation exceeds 10%, the model correction module is activated, and the correction coefficient is calculated based on the deviation ratio. The correction process uses a sampling frequency of 50Hz to ensure real-time performance.

7. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, In the environmental compensation described in step S5: For every 10°C increase in ambient temperature, the heat accumulation calculation coefficient increases by 0.15 times; Heat dissipation conditions are assessed by monitoring fan speed. A 20% decrease in fan speed is considered a deterioration in heat dissipation, at which point the heat dissipation rate constant drops to 60% of its normal value. The compensation coefficients are stored in a lookup table, and the table data is generated based on thermal simulation experiments, covering temperatures up to 60°C. The compensation mechanism is linked to the historical data learning function, and the correction coefficient is automatically adjusted when heat dissipation deterioration events occur frequently.

8. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, In the historical data learning function described in step S6: Overload event parameters include three-dimensional data of load value, duration, and ambient temperature, which are stored in the operation log; The time series analysis algorithm uses an autoregressive integral moving average model to uncover the periodic patterns of the load. The machine learning algorithm uses a support vector machine. The input parameters include historical load peaks, daily average overload frequency, and average heat dissipation conditions. The output is a dynamically adjusted warning threshold and a protection threshold, with the adjustment range being less than 15% of the initial value. The training dataset covers 1000 hours of running data, and the model is updated every 24 hours.

9. The method for dynamic overload detection and protection of a servo motor based on a thermal accumulation model according to claim 1, characterized in that, It also includes fault diagnosis linkage strategies: When the cumulative protection threshold is triggered 3 times / 24 hours, an automatic motor insulation aging early warning report is generated. The report includes the cumulative heat peak, event time, and environmental conditions. Overload events are linked to motor operating conditions and stored to form a thermal stress life map. The map uses load-time as the coordinate axis to visualize the history of thermal accumulation.

Images