Intelligent motor remote monitoring method based on industrial bus

Abstract

The application provides an intelligent motor remote monitoring method based on an industrial bus, relates to the technical field of motor control or regulation, and improves the intelligence level of a motor controller by constructing a technical closed loop of "future risk prediction based on digital twinning - coupled and fused decision of multi-domain risks - prospective adaptive adjustment of internal resources - dynamic determinacy broadcast of external capabilities", thereby being capable of enhancing the robustness and reliability of task execution of the motor control system in a high dynamic and high uncertainty industrial environment, effectively improving the collaborative efficiency of the entire distributed control system, and providing a solid technical foundation for realizing higher-order predictive maintenance and autonomous health management in a complex industrial scene.

Description

Technical Field

[0001] This invention relates to the field of electric motor control or regulation technology, specifically to a method for remote monitoring of intelligent motors based on an industrial bus. Background Technology

[0002] In complex industrial scenarios such as high-end manufacturing, robotic collaboration, and new energy, intelligent motors have evolved from traditional power execution units into key network nodes integrating sensing, computing, and communication capabilities. Especially in scenarios with stringent requirements for dynamic performance and system reliability, such as high-end manufacturing and robotic collaboration, individual motor controllers must not only accurately execute local control but also act as intelligent agents, efficiently collaborating with other nodes in the distributed system. Therefore, enhancing the autonomous adjustment capabilities and depth of information interaction of motor controllers in complex network environments has become an important development direction in this field.

[0003] Currently, as industrial bus-based motor remote monitoring technology evolves towards a higher level of intelligence, it still faces several technical challenges:

[0004] 1. Existing technologies generally rely on monitoring current or historical operating data to assess motor status and respond accordingly. For example, the method disclosed in Chinese patent document CN110829937A determines the current state of the motor by collecting its current position and current signals. This type of control paradigm lacks the ability to proactively predict rapidly evolving early potential risks (such as sudden overheating caused by insulation aging), which to some extent limits the safety margin of the system under extreme operating conditions.

[0005] 2. In traditional control strategies, risks in the physical domain of the motor (such as overtemperature and overload) and resource states in the computational domain of the controller (such as processor load and memory usage) are usually assessed separately and independently. This separation of dimensions makes it difficult for the system to identify and quantify complex risks caused by the coupling of factors from multiple domains. For example, minor physical anomalies have a much higher probability of causing system failure when computational resources are under severe strain than when resources are abundant, and existing technologies lack effective means to quantify the coupling gain effects of such risks.

[0006] 3. Existing controllers typically broadcast discrete and static status information, such as "normal," "warning," or "fault" mode identifiers. This information is coarse-grained and cannot convey the dynamic trends of the controller's internal state to external systems. External systems have no way of knowing whether a controller in a "high load mode" is stabilizing or on the verge of deterioration. This ambiguity in communication hinders upper-level systems from making accurate and forward-looking collaborative decisions, necessitating further improvements in the collaborative efficiency and robustness of the entire distributed system.

[0007] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] The purpose of this invention is to provide a remote monitoring method for intelligent motors based on industrial bus, so as to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] The method for remote monitoring of intelligent motors based on industrial bus includes the following steps:

[0011] S1: Real-time operating data acquired by the motor control system for constructing a multi-physics digital twin model of the motor;

[0012] Based on the multiphysics digital twin model, the future physical state parameters of the motor within a preset time window are predicted;

[0013] S2: Integrate the future physical state parameters with the resource state parameters of the motor control system to generate a task criticality index;

[0014] Based on the aforementioned task criticality index, a forward-looking resource scheduling strategy is generated;

[0015] S3: Based on the aforementioned forward-looking resource scheduling strategy, dynamically adjust the resource allocation of computing and communication tasks within the motor control system;

[0016] S4: Based on the dynamically adjusted resource allocation results, generate a controller capability profile that represents the current external service capabilities of the motor control system, and broadcast the controller capability profile to external systems via the industrial bus.

[0017] Compared with existing technologies, the beneficial effects of this invention are: by introducing a multiphysics digital twin model; this model is not satisfied with monitoring the current state, but rather, driven by real-time data, it extrapolates the future physical state evolution of the motor in digital space, thereby quantifying the probability of potential, yet-to-occur, failures and safety margins. This achieves a fundamental shift from "passive response" to "active prediction," providing a crucial time window for all subsequent decisions;

[0018] Through an adaptive weighted fusion mechanism based on risk coupling gain, when multiple risks are superimposed, the generated "task criticality index" can produce a steep nonlinear growth, thereby capturing those complex risks that are easily ignored in a single dimension more sensitively and earlier.

[0019] A dynamic capability profile generation and broadcasting model is constructed. The controller generates and broadcasts a multi-dimensional, dynamic capability profile. The core of this profile is the introduction of a "task execution determinism index," which reveals the "stability" and "predictability" of its future state to the external system by quantifying the time-varying rate of internal comprehensive risk. This enables the collaboration of the entire distributed system to leap from simple synchronization based on discrete states to refined adaptive collaboration based on future stability prediction. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the application process of the present invention;

[0021] Figure 2 This is a schematic diagram illustrating the execution logic of steps S1 and S2 of the present invention;

[0022] Figure 3 A schematic diagram of the execution logic of steps S3 and S4 of the present invention. Detailed Implementation

[0023] 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.

[0024] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various elements, but unless otherwise stated, these elements are not limited by these terms. These terms are used only to distinguish one element from another.

[0025] Example 1:

[0026] Please see Figures 1 to 3 The present invention provides a technical solution:

[0027] The method for remote monitoring of intelligent motors based on industrial bus includes the following steps:

[0028] S1: Real-time operating data acquired by the motor control system for constructing a multi-physics digital twin model of the motor;

[0029] Based on a multiphysics digital twin model, predict the future physical state parameters of the motor within a preset time window;

[0030] S2: Integrate future physical state parameters with the resource state parameters of the motor control system to generate task criticality indicators;

[0031] Based on task criticality indicators, generate forward-looking resource scheduling strategies;

[0032] S3: Based on a forward-looking resource scheduling strategy, dynamically adjust the resource allocation of computing and communication tasks within the motor control system;

[0033] S4: Based on the dynamically adjusted resource allocation results, generate a controller capability profile that represents the current external service capabilities of the motor control system, and broadcast the controller capability profile to external systems via the industrial bus.

[0034] Further explanation: For Figure 1 It should be noted that: "Digital Twin and Future State Prediction" represents step S1; "Multi-Domain Risk Fusion and Strategy Decision" represents step S2; "Dynamic Adaptive Adjustment of Internal Resources" represents step S3; and "Dynamic Capability Profile Generation and Broadcast" represents step S4.

[0035] Real-time operational data specifically includes: acquiring at least one electrical parameter; acquiring non-electrical physical parameters reflecting the internal thermodynamics of the motor; future physical state parameters including future failure probability and system safety margin; and resource state parameters including computation time margin.

[0036] Further explanation: The steps in S2 to integrate and generate the task criticality index are as follows: the task criticality index is generated by adaptively weighting and integrating the future failure probability, system safety margin, and calculation time margin of the motor control system.

[0037] Further explanation: Adaptive weighted fusion includes: calculating the risk coupling factor based on future physical state parameters and resource state parameters.

[0038] Risk coupling factors include physical risk coupling factors, which are calculated based on a combination of future failure probabilities and system safety margins;

[0039] The risk coupling factor also includes a resource risk coupling factor, which is calculated based on a combination of future failure probability and computation time margin.

[0040] The preset basic weights are dynamically adjusted using a risk coupling factor to generate dynamic weights. The future physical state parameters and resource state parameters are weighted and summed using the dynamic weights, and the summation result is nonlinearly normalized to obtain the task criticality index.

[0041] Further explanation: S2 generates a forward-looking resource scheduling strategy, specifically by dividing the numerical range of the task criticality index into at least two threshold ranges; and pre-setting a corresponding resource scheduling mode for each threshold range. The resource scheduling mode includes at least a task priority adjustment strategy and a loading and unloading strategy for functional modules.

[0042] The following is a detailed implementation description of the above content: The core logic of this embodiment is as follows: By synchronously acquiring the electrical parameters of the motor and the non-electrical physical parameters obtained through online identification technology, a high-fidelity multi-physics digital twin model is constructed and calibrated in real time. Secondly, this model is used to predict the potential failure probability of the motor within a future time window and the current safety margin, quantifying the potential risks in the physical domain. This embodiment introduces an adaptive weighted fusion mechanism based on risk coupling gain. This mechanism can non-linearly fuse predicted risk parameters from the physical domain with real-time resource status parameters from the computational domain, generating a task criticality index that more sensitively reflects complex risk scenarios. Finally, based on the comparison results between this task criticality index and multiple preset dynamic thresholds, the system can decisively switch to the optimal resource scheduling mode. This mode specifically defines the execution priority of different software tasks in the central processing unit and the dynamic loading or unloading strategy of specific functional algorithm modules. This allows the task criticality index to produce a "steepened" non-linear growth when multiple risks are superimposed, thereby achieving earlier and more decisive risk avoidance. In this embodiment, all key parameters are calculated and processed by a processor deployed inside the motor controller, as detailed below:

[0043] Electrical parameters: including motor phase currents, whose symbols are as follows. , representing the instantaneous current flowing through the stator windings of the motor; and the motor terminal voltage, whose parameter symbols are . , representing the instantaneous voltage value applied to the motor terminals, these two parameters are the basis for characterizing the electromagnetic state of the motor.

[0044] The "motor phase current" and "motor terminal voltage" are directly measured using a high-precision Hall effect current sensor (Allegro-Microsystems-ACS720 series) and a voltage divider resistor network integrated on the controller hardware circuit. The analog signals output by the sensors are sampled by a 16-bit resolution analog-to-digital converter (ADC) at a sampling frequency of 20kHz, synchronized with the pulse width modulation (PWM) frequency. The sampled raw data sequence is then digitally filtered by a third-order low-pass Butterworth filter with a cutoff frequency set to 5kHz.

[0045] Non-electrical physical parameters: specifically, the equivalent thermal resistance of the stator winding, with the following symbol: Physically, represents the equivalent thermal resistance encountered by the stator windings of a motor as a heat source during heat transfer to the motor housing, measured in Kelvin per watt (K / W). It directly reflects the motor's heat dissipation capacity and health status. This parameter is obtained through an online parameter identification algorithm, specifically based on resistance temperature measurement and recursive least squares (RLS). The "brief pause between two motion commands" in motor control is defined as the non-torque output interval. The inverter module of the controller injects a weak DC test current of constant amplitude into the U and V phase windings of the motor, with the parameter symbol being... The current amplitude is set to 1% of the motor's rated current to ensure that the generated torque is negligible and does not cause a significant temperature rise; during this period, the voltage drop between phases U and V is measured, and its parameter symbol is... Based on Ohm's law, the instantaneous DC resistance of the winding is calculated, and its parameter symbol is... The calculation relationship is equal Divide by Secondly, based on the temperature coefficient of resistance of the winding copper, which is calibrated at the motor factory, its parameter symbol is... And the winding resistance at a reference temperature of 25 degrees Celsius, the parameter symbol of which is Through linear approximation, by The current average temperature of the winding is calculated by reverse calculation, and its parameter symbol is: The relation is expressed as: Equal to reference temperature plus ( minus The difference is then divided by ( Multiply The product of ) . Finally, the controller records the total motor power loss in real time during continuous operating cycles as the input heat flow for the thermal model, and records The case temperature is measured by an NTC thermistor mounted on the case. The temperature difference between the windings and the housing is considered. This input-output data pair is fed into a recursive least squares (RLS) estimator, which identifies online and iteratively the first-order transfer function describing the dynamic relationship between the temperature difference between the windings and the housing and the power loss. The steady-state gain in this function is then determined as the equivalent thermal resistance of the stator windings. .

[0046] Injected in this embodiment The measured value was 0.1 amperes. If it is 0.15 volts, then It is 1.5 ohms. If the motor is at 25 degrees Celsius... The temperature coefficient of resistance of the copper winding is 1.3 ohms. If it is 0.00393, then The calculated value is 25 + (1.5 - 1.3) / (0.00393 × 1.3) ≈ 64.1 degrees Celsius. Over the past 10 seconds, the average power loss was 20 watts, and the measured average temperature difference was 40 degrees Celsius. The initial estimate is 40 / 20 = 2K / W. The RLS algorithm will continuously smooth and correct this value based on new data. Regarding the "multiphysics digital twin model," the following additional information needs to be provided:

[0047] The "multiphysics digital twin model" in this embodiment is a composite model composed of an electrical loss model and a thermal network model. Its construction process is divided into two stages: offline parameter calibration and online parameter identification. Specifically, the electrical loss model is used to calculate the total power loss of the motor, which serves as the heat source input to the thermal network model. The second-order Foster thermal network model is used to simulate the heat transfer path and cumulative effect inside the motor; a second-order Foster network structure is selected, which consists of two sets of parallel "thermal resistance-thermal capacity" (RC) circuits. The first set of RC circuits... Characterizing the heat transfer path from the heat source in the motor windings to the housing; second group of RC circuits Characterizes the heat transfer path from the casing to the external environment. Most thermal network parameters were pre-calibrated using motor design data and laboratory "locked rotor temperature rise experiments" and "no-load temperature rise experiments." Specifically:

[0048] Winding heat capacity and casing heat capacity The initial values ​​are calculated using the material specific heat capacity formula based on the copper mass of the motor windings, the material and mass of the iron core and the casing; the thermal resistance from the windings to the casing is also calculated. After the internal structure of the motor is fixed and remains basically unchanged, a steady-state temperature rise experiment is conducted under controlled heat dissipation conditions to measure the stable temperature difference between the winding and the casing and the motor's heating power, which are then calculated based on Ohm's law of thermal motion.

[0049] Thermal resistance from housing to environment This parameter represents the dynamic change of the external cooling airflow speed of the motor under varying operating conditions; it is identified online; and the recursive least squares method with a forgetting factor (FF-RLS) is employed. The implementation steps of this algorithm are as follows:

[0050] The input is a set of time series data pairs, including: the total power loss of the motor at the current moment. The housing temperature is measured in real time by a temperature sensor mounted on the surface of the motor housing. .

[0051] The total power loss of the motor is calculated by precisely adding copper loss and iron loss. Copper loss is calculated based on Joule's law by obtaining the dq-axis current component in the FOC control algorithm and multiplying it by the winding resistance value after real-time temperature compensation. Specifically, the winding resistance value is calculated in real time by a linear temperature compensation model. The input of this model is the winding temperature estimated in real time by a multiphysics digital twin model. The mathematical relationship is as follows: obtain the temperature difference between the current estimated winding temperature and the preset reference temperature; then, multiply the temperature difference by the known resistance temperature coefficient and add the resulting product to the numerical value to obtain the temperature compensation factor; finally, multiply the temperature compensation factor by the resistance reference value calibrated offline at the reference temperature, and the product is the real-time resistance value represented by the compensated winding resistance value.

[0052] Iron loss is estimated using a pre-calibrated two-dimensional lookup table. The controller takes the current motor angular velocity and stator flux linkage amplitude as input, and obtains the corresponding iron loss value by looking up the table and applying bilinear interpolation.

[0053] The identification algorithm within the controller executes once at fixed time intervals. It utilizes the latest total motor power loss... and The measured values ​​are substituted into the recursive formula of the FF-RLS algorithm to iteratively update the covariance matrix and parameter estimation vector; in this embodiment, the fixed time period is "every 1 second"; the output is the identified thermal resistance from the casing to the environment under the current operating condition. The optimal estimate.

[0054] The controller will identify the latest The values ​​are updated in real time to the parameter matrix of the second-order Foster thermal network model. This completes the construction of a high-fidelity digital twin model that can self-calibrate in real time and reflect the current real-world heat dissipation environment.

[0055] Furthermore, future physical state parameters include future failure probabilities, with the parameter symbol being... represents the probability that the motor's critical temperature point will exceed the safety threshold within the initial preset time window of "100 milliseconds", which is a dimensionless value in the interval [0, 1]; and the system safety margin, whose parameter symbol is . , representing the relative distance between the current critical state parameter and its safety threshold, is a dimensionless value within the interval [0, 1]; in this embodiment, the critical temperature point includes the winding temperature; "future failure probability" and "system safety margin" are predicted by the second-order Foster thermal network model in the multiphysics digital twin model; specifically, utilizing the online identified In conjunction with other offline calibrated thermal resistance and thermal capacity parameters, a second-order Foster thermal network is constructed to accurately describe the heat transfer path from the windings to the housing, and then from the housing to the environment. The network's input is the current total power loss of the motor, and its output is the predicted current average temperature of the windings. and casing temperature The controller uses the current state as initial conditions and transforms the predicted load torque for the next 100 milliseconds into a predicted power loss sequence. This sequence is then used as input to the thermal model. By solving the model's differential equations, the predicted winding temperature trajectory for the next 100 milliseconds is obtained. Based on the uncertainties in the model parameters and load prediction, a zero-mean, zero-standard-deviation parameter is superimposed on this deterministic trajectory. Gaussian random noise, standard deviation The results were obtained through offline experimental calibration; by running 1000 Monte Carlo simulations, the highest temperature in the predicted trajectory exceeded the safety threshold. The number of simulations, divided by the total number of simulations (1000), yields the probability of future failures. The security threshold in this embodiment This is expressed as 155 degrees Celsius.

[0056] The following explanation is provided regarding the acquisition of "load torque": In this embodiment, the AR(2) model extrapolation method based on the second-order autoregressive model is used to generate and calculate the load torque TL. The specific implementation steps are as follows:

[0057] The controller allocates a first-in-first-out (FIFO) queue in memory to store the actual load torque TL values ​​for the most recent N sampling periods. This actual value is obtained by multiplying the q-axis current command value in FOC (Field-Oriented Control) by the torque coefficient; the recursive least squares (RLS) method is used to fit the historical torque data in the buffer queue and identify the three coefficients of the AR(2) model online: constant term c, first-order lag term coefficient a1, and second-order lag term coefficient a2. The model expression is: TL(t) = c + a1 × TL(t-1) + a2 × TL(t-2); where t is the index marker at time t;

[0058] Using the current load torque value TL(t) and the previous torque value TL(t-1) as initial conditions, the AR(2) model with identified coefficients is used to perform iterative calculations to generate a torque prediction sequence for the next 100 milliseconds. In this embodiment, the load torque TL(t+1) is predicted as c + a1 × TL(t) + a2 × TL(t-1), and then TL(t+1) and TL(t) are used to predict TL(t+2), and so on. The obtained torque prediction sequence is combined with the current speed and converted into the total motor power loss for the next 100 milliseconds through the electrical loss model. The prediction sequence is the prediction input for the thermal network model.

[0059] Furthermore, system safety margin The calculation logic is as follows: obtain the winding temperature at the current moment. Safety thresholds corresponding to motor insulation class And the ambient temperature at which the motor is currently operating, as measured by sensors. Calculate the actual temperature rise range, i.e. and The difference; and calculate the total allowable temperature rise range, i.e. and The difference; system safety margin It is determined as the ratio of the actual temperature rise range to the total allowable temperature rise range, and the output value range is limited to the interval [0, 1]; when equal hour, A value of 1 indicates the maximum margin; when equal hour, A value of 0 indicates that the margin has been exhausted. In this embodiment, if... , , Then the system safety margin It equals the quotient of (155-124) divided by (155-25), that is, 31 divided by 130 equals 0.238.

[0060] Resource status parameters, specifically the computation time margin, are denoted by [symbol missing]. This represents the percentage of idle time remaining in the current PWM control cycle after the processor has completed all preset hard real-time calculation tasks, relative to the total control cycle duration. It is a dimensionless value within the range [0, 1]. This parameter is directly provided by the performance monitoring module of the real-time operating system (RTOS) on the controller. At the end of the control cycle, the RTOS reads the CPU's internal high-precision timer to calculate the total idle time of the CPU during that cycle. Dividing this idle time by the total duration of the PWM control cycle yields the result. .

[0061] Risk coupling factors: including physical risk coupling factors, whose parameter symbols are as follows This is used to amplify the impact of multiple risks occurring simultaneously within the physical domain; and a resource risk coupling factor, whose parameter symbol is... This is used to amplify the impact when physical risks and computational resource constraints occur simultaneously. Both are dimensionless values ​​and are greater than or equal to 1.

[0062] For physical risk coupling factors The calculation logic is as follows: calculate the basic physical risk index, which is equal to the future failure probability. Multiply by (1 minus the system safety margin) The difference. Then, =1 plus the basic physical risk index multiplied by the preset physical coupling gain coefficient .

[0063] Resource risk coupling factors The calculation logic is as follows: calculate the basic resource risk index, which is equal to the future failure probability. Multiply by (1 minus the calculation time margin) The difference. Then, It equals 1 plus the basic resource risk index multiplied by the preset resource coupling gain coefficient. If in this embodiment... , , ,and Therefore, the basic physical risk index is 0.8 × (1 - 0.1) = 0.72, and the physical risk coupling factor is... The value is 1 + 0.72 × 2.0 = 2.44. The basic resource risk index is 0.8 × (1 - 0.2) = 0.64, and the resource risk coupling factor is... The result is 1 + 0.64 × 3.0 = 2.92. When multiple risks occur, the risk coupling factor is greater than 1, which amplifies the risk.

[0064] For physical coupling gain coefficient and resource coupling gain coefficient The calibration is performed through the following offline optimization method: establish a simulation platform or a real motor test bench that includes a hardware-in-the-loop (HIL) simulation platform for the motor controller.

[0065] Experienced systems engineers define a set of typical and representative extreme operating conditions and set expected task criticality indicators for each scenario. Output value:

[0066] Scenario A1 (Single High Heat Risk): .expect Output ≈ 0.75.

[0067] Scenario B1 (Single High Computational Risk): .expect Output ≈ 0.65.

[0068] Scenario C1 (Complex High Risk): In this scenario, the expected risk is significantly amplified, and the expected risk is... Output > 0.95.

[0069] Define an objective function whose value is the actual output of the algorithm under all key scenarios. Expectations set by experts The root mean square error (RMSE) between them. Using grid search or the more efficient particle swarm optimization (PSO) algorithm, within the preset range... and Perform an iterative search within the range of values. In each iteration, a set of candidate values ​​is selected. Substitute the values ​​into the algorithm, run all key scenarios on the platform, and calculate the error value of the objective function; after sufficient iteration, find the set of values ​​that minimizes the error of the objective function. This is the final calibration result. This result is embedded into the constant definition of the controller program.

[0070] Furthermore, the parameter symbol for the task criticality index is denoted as... , is a normalized index that comprehensively quantifies the "urgency" or "danger" of the system's operation in the current and future period, and its value range is the value in the interval [0, 1].

[0071] The numerical range of the task criticality index is divided into at least two threshold ranges; the two threshold ranges are set by a strategy threshold parameter consisting of a "low criticality threshold" and a "high criticality threshold".

[0072] Policy threshold parameters: including low threshold, whose parameter symbols are as follows. ; and the high criticality threshold, whose parameter symbol is Both critical thresholds are within the (0,1) interval, used to divide different intervals of the task criticality index to trigger different scheduling strategies; specifically determined through offline experimental calibration, the steps of which are: building a motor test platform, running the motor under different but controllable heat dissipation and load conditions, and artificially injecting computational interference tasks to simulate different computational time margins. During this process, task criticality indicators are continuously recorded. The calculated values ​​and the actual operating status of the motor are considered, including whether overheating protection has occurred or control accuracy has decreased. Through actual experiments, the maximum critical performance index is determined. The value, when it exceeds the maximum task threshold indicator When the value is set to a certain value, the system has a high probability of entering an unsafe state within a short period of time; this value is then labeled as... This embodiment sets The value is 0.8. Another low-value task threshold indicator was found. When the value is below this low-value task threshold indicator, the system can operate stably and efficiently for a long time under any test conditions. This value was calibrated as... In this embodiment, the value is 0.3; it should be noted that "high probability" is adjusted based on actual application requirements, and in this embodiment it is set to 80%;

[0073] The core computation process in this embodiment is executed periodically by the processor within the controller in an independent, high-priority non-hard real-time task, specifically once every 10 milliseconds; the computation process is as follows:

[0074] 1.1) The initial input consists of parameters acquired in real time by sensors and the RTOS, including: motor phase current. Motor terminal voltage Casing temperature Ambient temperature and calculation time margin .

[0075] 1.2) Call the online parameter identification module to calculate and update the equivalent thermal resistance of the stator winding using the input electrical parameters. 1.3) The updated version and real-time measurement and The latest parameters are updated into the multiphysics digital twin model. Subsequently, the model is called to calculate the future failure probability. and system safety margin 1.4) Based on the output and , and the input Calculate the physical risk coupling factor Coupling factors with resource risk 1.5) , , And the calculated and Adaptive weighted fusion is performed to calculate the final task criticality index. .

[0076] 1.6) Condition Judgment 1: Judgment of Task Criticality Indicators Is it greater than or equal to the high criticality threshold? .

[0077] If path one is true: generate a resource scheduling policy for "safety priority mode"; the specific content of this policy is encoded as an instruction, which is a data structure containing a mode identifier and a priority list.

[0078] Condition 2: If Condition 1 is false, then continue to evaluate the task criticality index. Is it less than the low critical threshold? If path two is true: generate a resource scheduling strategy for "energy efficiency priority mode"; if path three is false: generate a resource scheduling strategy for "performance balance mode". In this embodiment, the controller runs the following key tasks, with lower RTOS priority values ​​indicating higher priority: The specific resource scheduling strategies for the three modes are defined as follows:

[0079] 0.1) Set the resource scheduling strategy for "Safety Priority Mode" as follows:

[0080] When the task criticality index Triggered when the threshold exceeds a high-criticality level. In this mode, system survival is the sole objective, and all resources are allocated to security and core control. Details:

[0081] The priority of "Safety Monitoring and Protection" is raised to the highest level 1; the priority of "Motor FOC Core Control" remains at the second highest level 2; the priority of "Digital Twin and Prediction" is set to level 3 to "ensure continuous risk assessment"; the priority of "External Communication" is reduced to level 10 to "report the most critical status only when the CPU is idle"; the priority of "Data Logging" is reduced to level 15 to "the lowest, which can be suspended at any time"; and all unnecessary background diagnostic, data analysis and other functional modules are unloaded or suspended under "Functional Module Adjustment".

[0082] 0.2) The resource scheduling strategy for the "energy efficiency priority mode" is set as follows: when the task criticality index... Less than the low critical threshold Triggered on demand. In this mode, the system has ample safety margin and can execute computationally intensive advanced algorithms to optimize energy efficiency. Details:

[0083] The "Safety Monitoring and Protection" priority is set to level 3, and a Model Predictive Control (MPC) algorithm module is loaded and executed to replace the traditional PID controller. The "Motor FOC Core Control" priority is set to level 2, and a more complex parameter identification algorithm is loaded. "Digital Twin and Prediction" priority is set to level 4 for routine monitoring. "External Communication" priority is set to level 5 for normal bandwidth. "Data Logging" priority is set to level 10 for normal logging.

[0084] 0.3) The resource scheduling strategy of "performance balancing mode" is: when the task criticality index The default mode when the value is between two thresholds. Details:

[0085] The priority level for "Safety Monitoring and Protection" is set to 2, and it executes the standard PID control algorithm. The priority level for "Motor FOC Core Control" is set to 4. The priority level for "Digital Twin and Prediction" is set to 3. The priority level for "External Communication" is set to 5. The priority level for "Data Logging" is set to 10.

[0086] 1.7) The final output is a data structure containing explicit scheduling mode instructions. This data structure is sent to the RTOS scheduler kernel to actually adjust the priorities of each task and load / unload specified functional modules at the start of the next scheduling cycle. Furthermore, this embodiment employs an adaptive weighted fusion method based on risk coupling gain; by introducing a risk coupling factor to dynamically adjust the weights, it can accurately model and quantify this nonlinear risk superposition effect, thereby providing a more timely and accurate risk assessment. Supplementary explanation for "using the risk coupling factor to dynamically adjust the preset basic weights to generate dynamic weights":

[0087] For determining the basic weights: basic weights are set for the three basic risk parameters, including the basic weight for the probability of future failures. Safety margin basic weight and the basic weight for calculating time margin These basic weights reflect the inherent importance of each parameter in the absence of coupling effects; their values ​​are obtained by a group of domain experts (5 motor design engineers and 5 embedded systems engineers) using the Analytic Hierarchy Process (AHP) to pairwise compare and score the relative importance of each risk factor, construct a judgment matrix, and calculate the eigenvector corresponding to its largest eigenvalue; in this embodiment, a set of values ​​is determined as follows: , , In each calculation cycle, the dynamic weights are obtained by multiplying the base weights by the calculated risk coupling factor: the dynamic weights of the future failure probability. equal Multiply by the physical risk coupling factor .

[0088] Dynamic weighting of safety margin equal Multiplied by physical risk coupling factor Calculate the dynamic weights of the time margin. equal Multiply by resource risk coupling factor Calculate the weighted risk composite value, with parameter symbols as follows: Its calculation logic is as follows: Multiply by the probability of future failures , plus Multiply by (1 minus the system safety margin) ), plus Multiply by (1 minus) ).

[0089] The final output is limited to the interval [0, 1], and an improved logical sigmoid function is used, namely the Sigmoid function. After normalization, the final task criticality index is obtained. Its calculation logic is as follows: =1 divided by 1 plus the natural constant e [negative k times ( minus () power). Here, the parameter k controls the steepness of the curve, parameter The center offset of the curve was controlled. These two parameters were determined through offline experimental calibration to ensure accuracy during calibration. and The nearest neighbor is chosen to give the corresponding function the most suitable response sensitivity.

[0090] The following detailed implementation instructions are provided for the above content: When the task criticality index The closer the output value is to 1, the closer the motor control system determines the overall state of the motor and its control unit is to a critical or dangerous state. This is the result of the combined effects of potential future physical failures, currently depleted safety margins, and strained computing resources, which have produced nonlinear coupling effects. A task criticality index value that is closer to 1 is a strong signal that the system must immediately take the highest priority risk avoidance measures, including forcibly switching to a "safety priority mode" at the expense of some performance to ensure the core safety of the system.

[0091] When the task criticality index The closer the output value is to 0, the more secure, stable, and resource-rich the motor control system's assessment of the motor and its control unit's operating status. This indicates a lower probability of future failures, a more ample current safety margin, and more idle time for the processor. A value closer to 0 provides a basis for the system to execute an "energy efficiency priority mode."

[0092] The final output task criticality index The output value is determined by three key input parameters through the adaptive weighted fusion mechanism of this embodiment. The influence of each parameter on the output value and its physical logic rationality analysis are as follows:

[0093] Future failure probability With task criticality index There is a positive correlation between them; when other parameters remain unchanged, The increase will directly lead to a decrease in the weighted risk composite value. The monotonically increasing trend. Meanwhile, The increase will also be through physical risk coupling factors Coupling factors with resource risk The calculation nonlinearly amplifies the dynamic weights of itself and other related risk items, further accelerating the process. The growth of . Ultimately, through the monotonically increasing sigmoid function, it leads to The increase in the probability of overheating failure in the near future, as predicted by the digital twin model, indicates a higher potential danger to the system and suggests that the system's operating state should be considered more "critical".

[0094] System safety margin With task criticality index There is a negative correlation between them; in fusion computing, what directly participates in the weighting is "1 minus the system safety margin". "This item. Therefore, when other parameters remain unchanged, The reduction will result in "1 minus the system safety margin". The addition of items, in turn, makes And the final Monotonically increasing. This indicates that the closer the current state of the system is to its safety boundary, the smaller the system's safety margin and the higher the degree of danger of the system.

[0095] Calculate time margin With task criticality index There is a negative correlation between them; what directly participates in the weighting is "1 minus "This item. When other parameters remain unchanged, The decrease of will result in "1 minus The addition of items thus makes And the final Monotonically increasing. This design reveals that controllers lacking sufficient computational power to execute complex diagnostic, predictive, and control algorithms will experience a significant decrease in their ability to cope with unforeseen circumstances, even if their physical objects are currently in a good state, thus resulting in a higher overall system risk.

[0096] To quantitatively verify the significant advancements of the "Adaptive Weighted Fusion Method Based on Risk Coupling Gain" proposed in this embodiment compared to existing technologies, a set of comparative experiments was designed. The experiments were conducted on a high-precision motor durability test bench, simulating various typical operating conditions by precisely controlling the load, heat dissipation conditions, and controller background computation tasks. This experiment aims to verify the core innovation of this embodiment: the risk coupling effect. Therefore, multiple scenarios ranging from single risk to complex risk were set up. The experiment will compare the outputs of the two methods: Method 1 (this embodiment), using adaptive weighted fusion to calculate the "task criticality index". Method 2 (Comparative Technique) uses the traditional linear weighted fusion calculation to obtain the "linear task criticality index," which does not include risk coupling factors. By comparing the output differences of the two methods in different scenarios, the improvement in early warning sensitivity and accuracy of this embodiment in complex risk scenarios can be clearly demonstrated. The comparative experiments are shown in the table below:

[0097]

[0098] The linear task criticality index was calculated using a comparative technique. This index is calculated by directly using preset base weights to weight and sum the future failure probability, (1 minus system safety margin), and (1 minus computation time margin), and then normalizing it using the same Sigmoid function as in this embodiment. This method represents a conventional technical solution that does not consider risk coupling effects.

[0099] Based on benchmark verification under a single risk scenario, comparing scenarios one, two, and three: In scenario one (normal stable operation), all input risk parameters are at low levels, and the average value of the "task criticality index" represented by the output indicators of both methods is consistent and low. This proves that this embodiment will not generate false alarms under safe operating conditions, which is consistent with the behavior of existing technologies;

[0100] In Scenario 2 (single heat risk accumulation) and Scenario 3 (single computing resource shortage), the average values ​​of the "task criticality index" represented by the output indicators of this embodiment are 0.49 and 0.42, respectively, both higher than the average values ​​of 0.40 and 0.36 of the corresponding output indicators of the comparative technologies. In the case of a single risk, the risk coupling factor of this embodiment will have a slight amplification effect, reflecting a higher risk sensitivity;

[0101] Based on the demonstration of core advantages in a complex risk scenario, a comparison is made between Scenario 4 and Scenarios 2 and 3: Scenario 4 simulates the simultaneous occurrence of thermal risk and computing resource risk. The output index of the comparative technology (Experiment D-1 group: 0.56) is equal to the simple summation of its risk increments in Scenarios 2 and 3 (0.39-0.18+0.35≈0.56), which reflects the essence of linear fusion, namely, the simple accumulation of risks;

[0102] The output metric of this embodiment (Experiment D-1 group: 0.82) is significantly higher than its output in a single risk scenario and also significantly higher than the output of the comparative technology. Specifically, compared to the comparative technology's 0.56, the output metric value of this embodiment is improved by 46.4%. Improvement ratio = (Indicator of this embodiment - Indicator of comparative technology) / Indicator of comparative technology = (0.82 - 0.56) / 0.56 ≈ 0.464. This data proves the effectiveness of the "risk coupling gain" design of this embodiment. When multiple risks occur concurrently, the real danger faced by the system is amplified synergistically rather than simply added together. This embodiment successfully captured and quantified this nonlinear effect by calculating and applying the physical risk coupling factor (1.420) and the resource risk coupling factor (2.000), enabling the task criticality index to "leap" to a higher level. In practical applications, an index value of 0.82 will trigger the highest level of "safety priority mode," while the comparative technology's 0.56 still keeps the system in "performance equilibrium mode," thus missing the opportunity for risk avoidance.

[0103] Comparing the data from Group 1 and Group 2 for each scenario: In all scenarios, the data from the two independent experiments (Group 1 and Group 2) showed a high degree of consistency. In the critical scenario four, the output metrics of the two experiments were approximately 0.82 and 0.84, respectively. This demonstrates that the algorithm of this embodiment possesses high determinism and stability, and its output results are reliable and reproducible, meeting the high reliability requirements of the industrial control field.

[0104] Further explanation: The steps for dynamically adjusting resource allocation include: according to the preset resource scheduling mode corresponding to the resource scheduling strategy, calling the kernel service of the real-time operating system to modify the execution priority of at least one computing task and one communication task.

[0105] The steps for generating a controller capability profile include: performing calculations using a dynamic capability profile generation model; the inputs to the dynamic capability profile generation model include the current resource scheduling mode determined by the resource scheduling strategy, and resource status parameters.

[0106] Further explanation: The controller capability profile includes a task execution determinism index; the task execution determinism index is calculated based on the time change rate of the task criticality index, which characterizes the overall risk status within the controller.

[0107] The steps for broadcasting controller capability profiles include: encapsulating the controller capability profile data into process data objects (PDOs) of the industrial bus protocol and sending them periodically.

[0108] The following is a detailed implementation description of the above content: The core logic of this embodiment is to accurately translate the forward-looking resource scheduling strategy generated by the upper-layer decision-making module into a deterministic adjustment of the priority of internal computing and communication tasks of the controller by calling the kernel service of the real-time operating system (RTOS), ensuring the priority execution right of critical tasks in resource competition. This embodiment constructs a dynamic capability profile generation model. This model not only relies on the current resource scheduling mode, but also integrates the root causes of mode switching, namely the underlying physical risks and computing resource status parameters, to dynamically and in real time calculate and generate a multi-dimensional controller capability profile. In addition to the conventional expected response time and available data bandwidth, this profile also introduces a task execution determinism index. This index characterizes the stability of the controller's future state by quantifying the time change rate of the "task criticality index" corresponding to the internal comprehensive risk index. The dynamic capability profile is then formatted and broadcast to external systems through the industrial bus, enabling the entire distributed system to upgrade from simple state synchronization to collaborative adaptation based on future stability prediction, resulting in improved system-level robustness and collaborative efficiency.

[0109] In this embodiment, the controller capability profile and its constituent parameters are calculated and encapsulated by a processor deployed inside the motor controller, as detailed below:

[0110] Current resource scheduling mode: its parameter symbols are as follows This is an enumerated type of status identifier used to explicitly indicate the current macro-level resource scheduling strategy of the controller. Its value range includes at least: "Safety Priority Mode", "Performance Balancing Mode", and "Energy Efficiency Priority Mode"; the current resource scheduling mode. The value is directly determined by the "proactive resource scheduling strategy" generated in the preceding step S2;

[0111] Expected instruction response time: its parameter symbol is It is a value in milliseconds (ms), representing the maximum expected time from when the controller receives a high-level motion command from the external PLC system until the command is fully executed and reflected in the motor output; in this embodiment, the high-level motion command includes a position closed-loop control command; the expected command response time Specifically, an "algorithm-benchmark response time mapping table" is pre-stored in the controller firmware. This table is calibrated through offline benchmark testing and records the core computation time required for different control algorithms to complete one control cycle under the processor's nominal load. The parameter symbol for the core computation time is... The "different control algorithms" in this embodiment include PID and MPC;

[0112] Current resource scheduling model Once determined, the corresponding core computation time is obtained based on the "Algorithm-Baseline Response Time Mapping Table" loaded under this mode. Subsequently, the expected instruction response time. The calculation method is as follows: Multiply by the dynamic response time modulation factor; this dynamic response time modulation factor equals 1 plus (1 minus the current calculation time margin). The difference is then multiplied by the preset response time sensitivity coefficient. The response time sensitivity coefficient Offline stress testing was used to calibrate and reflect the impact of CPU load on instruction execution latency; this embodiment obtains the PID algorithm by looking up a table. It takes 0.8ms. Currently... A value of 0.2 indicates that the CPU is busy. The calibration is set to 0.5. Therefore, the dynamic response time modulation factor is 1 + (1 - 0.2) × 0.5 = 1.4. The final calculated... The time is 0.8ms × 1.4 = 1.12ms.

[0113] Available data processing bandwidth: its parameter symbol is This is a value measured in kilobits per second, representing the effective data transmission rate that the controller can ensure for non-core communication tasks, including "remote monitoring data reporting," under the current resource allocation strategy; available data processing bandwidth. The calculation logic is as follows: A "task priority-baseline bandwidth mapping table" is pre-stored in the controller firmware. This table is calibrated through offline benchmark testing. It records the theoretical maximum bandwidth that a communication task can obtain when assigned different priorities, given a fixed system bus bandwidth. The parameter symbols are as follows: If the current resource scheduling mode Once determined, the system obtains the target bandwidth based on the "task priority-baseline bandwidth mapping table" set for the communication task under this resource scheduling mode. Subsequently, the available data processing bandwidth will be... Multiply by the current computation time margin This characterizes the time margin for computation even when the communication task has a high priority, if the CPU has absolutely no idle time. When the bandwidth approaches zero, data cannot be effectively processed or transmitted, and therefore the available bandwidth also approaches zero. In this embodiment, under the "security priority mode," the communication task priority is reduced to the lowest level, as shown in the table lookup. It is 20kbps. If at this time If it is 0.22, then the available data processing bandwidth is... The value is 20kbps × 0.22 = 4.4kbps.

[0114] For obtaining the "Algorithm-Baseline Response Time Mapping Table" and the "Task Priority-Baseline Bandwidth Mapping Table," taking the acquisition of the "Algorithm-Baseline Response Time Mapping Table" as an example: A benchmark testing platform is built. The hardware core of this platform is an Infineon AURIX-TC397 series microcontroller, running at a 300MHz clock frequency, equipped with 4MB of program flash memory and 1.7MB of SRAM. The software environment running on this platform is ETASRTA-OS based on the OSEK / VDX standard, configured with a fixed-priority preemptive scheduling kernel; the calibration process is executed. Each control algorithm to be calibrated (including basic PID control algorithm, PID algorithm with feedforward compensation, and model predictive control (MPC) algorithm) is encapsulated into an independent test task with the highest priority. This test task is called 10,000 times in a loop through the test host program. Before and after each call, the number of clock cycles consumed by the task execution is accurately recorded by reading the 64-bit CPU clock cycle counter (Cycle-Counter-Register) of the TC397 core. To ensure that the measurement reflects pure computation time, all interrupts were disabled during testing, and data was cached. Data processing and table creation were then performed. For the 10,000 clock cycle samples collected for each algorithm, statistical filtering was performed, removing the highest and lowest 5% of data points to eliminate the impact of occasional system jitter. Then, the arithmetic mean of the remaining 90% of samples was calculated. Dividing this average clock cycle count by the processor's clock frequency of 300MHz yielded the core computation time of that algorithm. The unit is milliseconds.

[0115] Finally, all algorithms and their corresponding benchmark response times are organized into a two-column array (algorithm identifier, ...). The lookup table for the value is stored in the controller's read-only program storage area.

[0116] Furthermore, the task execution determinism index: its parameter symbol is... is a normalized value with a range of (0,1]. It represents the controller's expected response time to its declared commands. and available data processing bandwidth This parameter is a measure of the sustainability and stability of capability indicators; it is generated by a prediction model based on the time derivative of the internal risk state, and its theoretical basis stems from the analysis of the trend of system state changes in control system stability theory; the specific determination logic is as follows: the controller caches the task criticality index of at least the previous calculation cycle. Then, calculate the current task criticality index. The rate of change of time relative to the previous moment, its parameter sign is: The calculation method is the current task criticality index. minus ; and divide this difference by the time interval between the two calculations; this is the rate of change over time. It reflects the rate at which the system's risk status deteriorates or improves; and is an index of certainty in task execution. The mapping is performed using a negative exponential function, specifically expressed as follows: Equal to the negative of the natural constant e Multiply The absolute value of (the power of) is given by itself. This is a preset deterministic decay coefficient, with its physical unit being seconds (s). It is calibrated through offline experiments and used to adjust the sensitivity of the exponent to the rate of change of risk. This function ensures that: when the risk is stably characterized... When it approaches 0, the task execution determinism index Approaching 1; when risks change drastically, the task execution certainty index... Approaching 0.

[0117] The current task criticality index in this embodiment It is 0.82, while 10ms ago The value is 0.62, and the time interval is 0.01 seconds. Therefore... The value is (0.82-0.62) / 0.01=20. If... If calibrated to 0.1, then the task execution determinism index is... The calculation is e^(-0.1×|20|)=e^(-2)≈0.135. This low task execution determinism index warns external systems that although the controller is currently working, its state is unstable and it may enter a worse mode or fail at any time.

[0118] Controller capability profile: its parameter symbols are It is a structured dataset used to encapsulate and broadcast all the aforementioned capability parameters. Its data structure is defined as a record or structure containing four fields: .

[0119] The core calculation processes for steps S3 and S4 above are executed periodically by a medium-priority task within the controller responsible for system management; in this embodiment, this is executed every 20 milliseconds. The calculation process is as follows:

[0120] 2.1) The initial input is the prospective resource scheduling strategy generated in the preceding step S2, and the real-time updated internal state parameters, including the computation time margin. Current and historical task criticality indicators .

[0121] 2.2) Parse the input resource scheduling policy and determine the current resource scheduling mode. According to the current resource scheduling mode Query the preset "mode-task priority mapping table".

[0122] In one specific embodiment, the motor control system incorporates a real-time operating system (RTOS) responsible for scheduling and managing multiple concurrently executing software tasks within the controller. A resource scheduling mechanism defines a set of critical tasks managed by the RTOS within the motor controller. This task set includes at least:

[0123] 1. Safety monitoring task: The function of this task is to perform high-frequency inspections of the critical status of the controller hardware, such as detecting serious electrical faults such as overcurrent and overvoltage, so as to execute the highest priority hardware protection actions in the event of an emergency.

[0124] 2. Motor control loop task: The function of this task is to execute the core motor current loop, speed loop and other closed-loop control algorithms, such as the field-oriented control (FOC) algorithm. Its real-time and deterministic nature is directly related to the dynamic performance and stability of the motor.

[0125] 3. Thermal Model Prediction Task: The function of this task is to run a preset multiphysics digital twin model of the motor to calculate and predict the physical state parameters of the motor in future time windows, which is the data basis for the forward-looking decision-making of this invention.

[0126] 4. Resource scheduling decision task: The function of this task is to integrate the output of the thermal model prediction task and the system's own resource state parameters, calculate the task criticality index, and generate the final forward-looking resource scheduling strategy based on the index.

[0127] 5. Remote communication task: The function of this task is to be responsible for the data exchange between the controller and external devices (in this embodiment, the host PLC or monitoring system), including but not limited to broadcasting the controller capability profile of this invention.

[0128] 6. Data logging task: The function of this task is to store key data or event logs during system operation into non-volatile storage media for subsequent offline analysis or fault tracing.

[0129] Based on the task set defined above, this embodiment constructs a "mode-task priority mapping table" to accurately and deterministically map the macro-resource scheduling mode output by the upper-layer decision module to the execution priority configuration of each task in the underlying RTOS. In a preferred embodiment, the RTOS priority adopts the convention that the smaller the value, the higher the priority, where 0 is the highest allocatable execution priority; the mapping table is shown in the following table:

[0130]

[0131] In the "Safety Priority Mode," the priority configuration aims to maximize system security and decision-making response speed. In this mode, safety monitoring tasks (priority 0) and motor control loop tasks (priority 1) maintain their inherent highest priority to ensure basic hardware safety and motor drive. Simultaneously, the execution priority of thermal model prediction tasks (priority 2) and resource scheduling decision tasks (priority 3) is increased. This ensures the system can update risk state assessments and react with the highest frequency and lowest latency, forming a rapid risk avoidance loop. Correspondingly, the priority of remote communication tasks (priority 15) and data logging tasks (priority 25) is decreased, thereby concentrating processor time slice resources on core tasks directly related to security and control.

[0132] The "Performance Balanced Mode" prioritizes tasks to achieve balanced performance across all controller functions, representing the standard operating state of the system under normal conditions. In this mode, task priorities are distributed in a reasonable gradient. Core control tasks (priority 1) receive priority, while prediction (priority 4) and decision-making (priority 5) tasks run periodically with moderate priorities. Simultaneously, remote communication tasks (priority 8) are given higher priority to ensure smooth and timely information exchange between the controller and external systems.

[0133] For the "Energy Efficiency Priority Mode," the priority configuration aims to provide resource allocation for executing energy efficiency optimization algorithms or strategies while ensuring basic safety and performance. In this mode, the priority of motor control loop tasks can be appropriately reduced (Priority 2) to reserve execution time for other computationally more complex energy efficiency optimization algorithm tasks. At the same time, thermal model prediction tasks (Priority 3) and resource scheduling decision tasks (Priority 4) still maintain high priority, because accurate perception of the system state is a prerequisite for implementing advanced energy efficiency control strategies.

[0134] When the dynamic resource allocation adjustment represented by step S3 of this embodiment is triggered, the resource scheduling decision task within the controller determines the current resource scheduling mode. Subsequently, the task queries the "mode-task priority mapping table" using the enumeration value or index corresponding to the mode to obtain the configuration set corresponding to the mode, which contains the new priorities of all managed tasks. Finally, through a loop or iterative process, the standard API functions provided by the RTOS are called one by one to apply the queried new priority values ​​to the corresponding Task Control Block (TCB), thereby completing the dynamic and deterministic adjustment of the entire system's task scheduling strategy within microseconds.

[0135] Furthermore, the kernel API function of the real-time operating system (RTOS) is invoked to modify the priority parameters of the core computing tasks related to security as defined in the table to the highest level, while modifying the priority parameters of the non-core communication tasks to a lower level. This operation ensures that when the next RTOS clock tick arrives, the scheduler will immediately execute task preemption according to the new priority order;

[0136] 2.3) Call The computational model, input current and Calculate the expected instruction response time Call The computational model, input current and The available data processing bandwidth is calculated.

[0137] Call the "Prediction Model Based on the Time Derivative of Internal Risk State" and input the current... Calculate the task execution determinism index .

[0138] 2.4) The resource scheduling mode determined in 2.2) and 2.3) calculated , , The four parameters are encapsulated into a complete controller capability profile according to a predefined data structure. Data frame. Profil the controller's capabilities. Data frames are written to a dedicated shared memory area or message queue for communication tasks to read;

[0139] When a communication task is executed by the scheduler, it reads the latest controller capability profile from this shared memory region. The data frame is mapped to a specified byte offset position in the process data object (PHP-Data-Object, PDO) according to the industrial bus protocol used; in the next bus communication cycle, the PDO containing the latest capability profile will be automatically and periodically broadcast to the industrial bus network.

[0140] 2.5) The final output is a standardized data message broadcast on the industrial bus, containing a dynamic capability profile of the controller, which can be received and parsed by any PLC, HMI, or other motor controller node in the network. This embodiment uses a dynamic calculation model to enable the capability profile to reflect the controller's "true" capability and "future" stability in real time and accurately, providing the upper-level system with unprecedented depth of decision-making information.

[0141] Sensitivity coefficient for response time and deterministic attenuation coefficient All parameters were determined through offline performance calibration experiments. The specific method involved building a hardware-in-the-loop (HIL) test platform for the controller. Scripts were used to automatically change the processor's computational load and simulate motor risk states. Simultaneously, a high-precision oscilloscope and bus analyzer were used to accurately measure the controller's actual response time and data throughput. By performing nonlinear regression fitting on a large amount of input-output data, the optimal coefficient estimates were obtained and embedded into the controller firmware.

[0142] In a preferred embodiment of this invention, the response time sensitivity coefficient The value was set to 0.5. This value was determined through the following offline stress test experiments: The core control algorithm task was run on a benchmark platform. A computationally adjustable perturbation task was run in parallel, simulating different CPU loads by varying its loop count, thus allowing for computational time margins. It decreased linearly from 0.9 to 0.1. In each At each value point, the actual response time of the core control algorithm task was measured repeatedly one thousand times, and the average value was taken. Finally, the measured actual response time was compared with the core computation time. The growth rate is equal to (1 minus the current calculation time margin). The slope of the fitted line obtained by performing linear regression on the value of ) is determined as the response time sensitivity coefficient. The value of 0.5 is determined when the best fit between the performance degradation model and the measured data is achieved.

[0143] Furthermore, this embodiment has the following technical logic paths, specifically:

[0144] Technical logic path one: "Predicting overheating risk" → triggering "safety priority mode" → S3 performs resource adjustment → thermal management task gets the highest priority → motor temperature is actively controlled → avoiding hard protection shutdown.

[0145] When overheating risk is predicted: This corresponds to the input prerequisites of the S3 process. The preceding steps (S1, S2) predict a high "future failure probability" using a digital twin model. "And calculate the high "task criticality index" After that, a scheduling strategy of "safety priority mode" was generated, which is the basis for the execution of S3.

[0146] The core action of S3 is to call the kernel API function of the real-time operating system (RTOS) to modify the priority parameters of the core computing tasks related to security as defined in the table to the highest level, while modifying the priority parameters of non-core communication tasks to a lower level. In this embodiment, the core computing tasks include thermal management and protection monitoring tasks, achieving the technical effect of "the controller actively allocating more resources to the thermal management algorithm." In a preemptive RTOS, setting the task priority to the highest level signifies unconditionally preempting CPU time slices whenever needed; this is the most reliable way to "allocate more resources" in the field of real-time computing. The above "Technical Logic Path One" has the following technical effects:

[0147] The motor's thermal performance has been optimized to avoid faults: it has been given the highest priority "thermal management and protection monitoring task," enabling it to operate more promptly and frequently. By executing more refined control operations, it can perform optimized dynamic current derating based on real-time temperature and model predictions, or more precisely control the cooling fan speed, thereby actively stabilizing the motor temperature below the safe threshold and fundamentally avoiding overheating faults.

[0148] This solution avoids unplanned downtime caused by triggering traditional hard protection: Traditional hard protection (including thermal relay or temperature sensor interruption) is a passive, precipitous protection that inevitably leads to shutdown once triggered. This solution, through S3's proactive and forward-looking resource scheduling, allows thermal management tasks to intervene well before the temperature reaches the hard protection threshold. The motor will therefore experience reduced output performance, but it will not stop, thus ensuring production continuity.

[0149] Technical logic path two: predicting stable load → triggering "energy efficiency priority mode" → S3 performs resource adjustment → high-energy-efficiency complex algorithm tasks get high priority → smoother motor output and lower power consumption → dual improvement in energy efficiency and quality.

[0150] When a stable load is predicted: This corresponds to another input for S3. At this point, the "task criticality metric"... "At a low level, the system decision-making enters the 'energy efficiency priority mode'."

[0151] "The controller actively loads and executes complex algorithms with higher energy efficiency": The S3 mechanism is a general framework, specifically tailored to the current resource scheduling mode. The system queries the preset "Mode-Task Priority Mapping Table." In "Energy Efficiency Priority Mode," this table prioritizes tasks pre-embedded with "Model Predictive Control (MPC)" while lowering the priority of simple PID control algorithm tasks or putting them into a dormant state. This mode-based dynamic adjustment of task priorities is a concrete implementation of "Active Loading and Execution." The above "Technical Logic Path Two" has the following technical effects:

[0152] "Reduces its own operating power consumption": MPC's complex algorithm can predict the future state of the motor and perform more global and optimized voltage and current vector control, thereby significantly reducing copper loss and iron loss while meeting the same torque output, and directly achieving energy saving and consumption reduction.

[0153] Technical Logic Path 3: Downstream motor performs step S3, resulting in a decrease in capacity due to safety risks → Downstream motor performs step S4, generating and broadcasting a profile containing this "capacity decrease" information → Upstream PLC receives and parses the profile → PLC proactively adjusts its own strategy to adapt to downstream changes → Avoiding system conflicts and ensuring overall line efficiency.

[0154] "When the upstream master controller (PLC) receives the capability profile": The final output of S4 is "a standardized data message broadcast on the industrial bus containing the controller's dynamic capability profile," with the PLC as its default receiver. "Knowing that the selected motor's capability has decreased for safety reasons": When the controller enters "safety priority mode," S4 dynamically calculates new capability parameters. The profile received by the PLC will clearly show the current resource scheduling mode. The field changes to "Security Priority Mode". This example demonstrates: Expected command response time. Increased from 0.8ms to 1.12ms; available data processing bandwidth The bandwidth dropped sharply from 100kbps to 4.4kbps. These three quantitative indicators together constitute a description of the "capacity decline" state.

[0155] The above-mentioned "Technical Logic Path Three" has the technical effect of "proactively adjusting the overall production cycle time or process parameters." Specifically, when high-frequency coordinated movement with the motor is required, it is achieved by reading... It has been changed to 1.12ms, and then it actively adjusts its own waiting or synchronization timing from 0.8ms to 1.2ms to adapt to the changes in downstream motors and avoid mechanical interference or product scrapping caused by timing mismatch.

[0156] The introduced "task execution determinism index" To achieve the following technical effects;

[0157] Technical Logic Path Four: Multiple motors broadcast their respective dynamic capability profiles → The PLC not only knows "what it can do now," but also uses the task execution determinism index Knowing whether the future state is stable → the PLC assigns critical tasks to the most stable node → the system avoids the risk of assigning tasks to nodes that are about to fail → the system's robustness is improved.

[0158] "The main controller performs real-time, intelligent task redistribution and load balancing based on the capability profiles broadcast by all motor nodes in the network." "Intelligent" is reflected in the deterministic index of task execution. In its application. Task execution certainty index. Based on "task criticality index" The "time rate of change" of the node is calculated. High value, and Its task execution certainty index It will approach 1. This indicates that although the node is heavily loaded, its state is controllable and stable. The other node's... When the value is relatively low, but The characteristics are rising rapidly. "This indicates that it is much larger than the certainty index of its task execution" It will approach 0. The node's state is unstable and is sliding towards a dangerous edge;

[0159] The technical effect of the fourth technical logic path is to assign critical tasks to nodes with stronger "capability profiles"; "stronger" refers to prioritizing nodes with a higher certainty index for task execution. The highest node. The PLC will assign high-precision, long-cycle machining tasks to tasks with a deterministic execution index. Nodes approaching 1 are selected to ensure their state does not change abruptly during task execution. This relates to the task execution determinism index. For nodes approaching zero, the PLC will proactively isolate them or assign only monitoring tasks that are allowed to fail, thereby achieving intelligent load balancing based on future stability predictions.

[0160] The following detailed implementation instructions are provided for the above content: When the task execution determinism index The closer a value is to 1, the more stable the overall risk state within the controller; this state signifies the criticality index representing the risk. It changes little over a continuous time period, i.e., its rate of change over time. The closer the value is to 0, the more likely the system is in either a low-risk, stable operating state or a high-risk, stable overload state. In this state, the controller's externally broadcast capability profile, including expected command response time and available data processing bandwidth, has higher reliability and predictability.

[0161] When the task execution determinism index The closer the value is to 0, the more unstable the overall risk state within the controller is, or the more drastic the dynamic changes it is undergoing; this state characterizes the task's criticality index. The greater the increase or decrease, the greater its rate of change over time. The larger the absolute value, the more reliable and predictive the current capability profile broadcast by the controller becomes in this state. Therefore, the lower the reliability and predictability of this profile, the more unreliable or uncertain the controller node will be perceived as when it is received by external systems.

[0162] Index affecting task execution certainty The only direct input parameter is the time rate of change of the task criticality index. The time rate of change of the task criticality index The absolute value of the result, and the task execution determinism index of the final output. There is a nonlinear, monotonically decreasing negative correlation between them; the predictability of a control system depends on the rate of its state change.

[0163] When the rate of change over time When the absolute value of the index increases from 0, it indicates that the system state begins to change and its uncertainty increases. Therefore, the task execution determinism index... Consequently, it decreases. Using a negative exponential function for mapping inherently results in the independent variable... When the value is small, the dependent variable is the task performance determinism index. The decrease is rapid; however, when the independent variable is large, the decrease in the dependent variable tends to be gradual and approach 0. This makes the index sensitive to changes in the system state from "stable" to "small fluctuations," enabling it to capture early signs of instability in a timely manner. At the same time, for systems that are already in a state of violent fluctuation, it stably outputs a low-determinism signal close to 0.

[0164] To quantitatively verify the beneficial effects of the method in this embodiment on improving the collaborative efficiency and robustness of distributed control systems, the following comparative experiment was designed. The experiment was conducted in an industrial bus network environment containing one master controller (PLC) and multiple motor controller nodes. The comparative technical solution is the "existing technology," meaning that the motor controller can only broadcast its current operating mode to the PLC and cannot provide a task execution deterministic index. The specific experimental data comparing controller status broadcasting and system decision-making under different operating conditions are shown in the table below:

[0165]

[0166] System task assignment success rate: A metric for measuring system-level performance, defined as the percentage of control commands issued by the main controller (PLC) to the entire network that are successfully received and completed on time by the target motor controller nodes per unit time. A decrease in this metric is caused by command response timeouts or node offline due to faults.

[0167] Comparing Scenario 1 and Scenario 2: In Scenario 2 (stable overload), the controller enters "safety priority mode" due to high load. With existing technology, the PLC can only observe the "safety priority mode" signal and conservatively reduces task allocation to that node, causing the system task allocation success rate to drop from nearly 100% to 85.2%. However, in this embodiment, the PLC receives not only the mode signal but also a task execution determinism index of over 0.9. This informs the PLC that although the node has a high load, its state is completely stable. Therefore, the PLC continues to allocate non-critical and simple tasks to it, avoiding unnecessary load shifting and maintaining a high system task allocation success rate of over 98%. In the stable overload scenario, this embodiment improves the system task allocation success rate by (98.35% - 85.2%) / 85.2% ≈ 15.4% compared to existing technology. This data strongly demonstrates that this embodiment can accurately distinguish between "high load" and "instability," improving the system's resource utilization under extreme operating conditions.

[0168] Comparing Scenario 2 and Scenario 3: In both scenarios, the controller broadcasts a "safety priority mode." With existing technology, the PLC cannot distinguish between these two fundamentally different "safety priority" states, and therefore adopts the same conservative load reduction strategy as in Scenario 2. However, Scenario 3 involves a sudden, rapidly deteriorating overheating process. This conservative strategy is insufficient to avoid the impending failure, causing a large number of tasks to fail due to the node's impending overheating and shutdown, resulting in a sharp drop in the system's task allocation success rate to 44.5%. In contrast, in this embodiment, in Scenario 3, the PLC receives a task execution determinism index of 0.23. This signal plays a crucial early warning role; the PLC immediately identifies the node as a "high-risk unstable node" and executes the highest priority avoidance strategy, namely, immediately ceasing to allocate any new tasks to it and attempting to migrate existing tasks. This proactive avoidance action allows the system to successfully avoid instruction loss and execution failure caused by the sudden shutdown of a single node, maintaining the system's task allocation success rate above 99%. The unique task execution determinism index in this embodiment provides the upper-level control system with a new information dimension for predicting the stability of future states, thereby enhancing the robustness of the entire distributed control system and the continuity of task execution.

[0169] To increase the certainty index of task execution The quantized output is converted into standardized operating strategies executable by the upper-level main controller (PLC). This embodiment defines the following application range division. This standard is based on the analysis of a large amount of experimental data and combined with expert experience in system stability in the field of industrial automation. Its core idea is the task execution determinism index. The higher the certainty index, the higher the PLC's trust in the node, and the more aggressive the task allocation strategy; conversely, the lower the certainty index, the lower the task execution certainty index. The lower the value, the more decisive the PLC's avoidance and intervention strategies. The specific application range divisions are shown in the table below:

[0170]

[0171] In a preferred embodiment, the intelligent motor controller node and the remote PLC master node are physically independent and separate hardware entities. Their only connection is established via a deterministic industrial fieldbus. The industrial bus uses protocols such as Ether-CAT, Profinet-IRT, or CANopen that support periodic, deterministic data exchange. This physical separation and connection via the bus constitute the physical basis for "remote" monitoring. For example, the remote PLC may be installed in a central control cabinet, while the intelligent motor controller is directly integrated into the motor equipment on the production line; the spatial distance between the two can be tens of meters or even further. The "controller capability profile" generated in step S4... "To efficiently and deterministically transmit data to a remote PLC, this embodiment of the invention employs a standardized data encapsulation mechanism."

[0172] The "controller capability profile" is solidified as a structured data object. In embodiments based on Ether-CAT or CANopen buses, this data object is mapped to a Process-Data-Object (PDO). "Tx-PDO" is defined as the transmission process data object, and its internal data structure is as follows:

[0173] Byte 0: Current resource scheduling mode It is mapped to an 8-bit unsigned integer (UINT8).

[0174] Bytes 1-2: Expected instruction response time It is mapped to a 16-bit unsigned integer (UINT16) in microseconds.

[0175] Bytes 3-4: Available data processing bandwidth Mapped to a 16-bit unsigned integer (UINT16), in units of KB / s.

[0176] Bytes 5-8: Task execution determinism index Mapped to a 32-bit floating-point number (FLOAT), or a 32-bit integer (DINT) that has been magnified 1000 times and then rounded down to save bandwidth.

[0177] The intelligent motor controller is configured to autonomously broadcast the encapsulated PDO data frames to the industrial bus at a fixed, preset interval of 10 milliseconds, without requiring a master station request. This "publish-subscribe" model ensures that the remote PLC can continuously and nearly in real-time receive the latest capability profiles of each controller node, forming the data stream foundation for remote monitoring.

[0178] The remote PLC master station acts as the monitoring terminal. Its internal application includes a dedicated logic module for receiving, parsing, and utilizing the "Controller Capability Profile" (PDO) broadcast from each motor controller node. The decision logic of this remote terminal includes at least the following:

[0179] The PLC maintains a real-time updated status table in its memory for each motor controller node on the network. The contents of this table are a complete mirror of the latest "controller capability profile" broadcast by each node.

[0180] Before issuing new, computationally intensive, or time-sensitive motion control commands, the PLC's task scheduler will first query the status image of the target node.

[0181] If the node's expected command response time and available data processing bandwidth If all requirements of the instruction are met, the task will be issued normally.

[0182] If the conditions are not met, the PLC will automatically reroute the task to another node in the network with more capacity, or place it in a waiting queue, thereby achieving dynamic and forward-looking load balancing at the system level.

[0183] The PLC's diagnostic and alarm logic module continuously monitors the "soft state" indicators in the profile.

[0184] Current resource scheduling mode of the node When switching from "Performance Balanced" to "Safety First", the PLC's remote monitoring interface will immediately display a "Warning" message, indicating that "Node X has entered resource contraction mode, please pay attention." Simultaneously, the PLC's scheduling logic will automatically reduce the task allocation priority for that node.

[0185] When the node's task execution determinism index When the value falls below a preset "threshold", the PLC will trigger a higher-level "unstable state" alarm and can execute preset risk avoidance strategies, such as safely suspending the production process it controls or migrating its tasks to a redundant backup node.

[0186] The computational logic involved in this application can be constructed using algorithms such as regression analysis in machine learning, establishing a mathematical model by analyzing the inherent trends and interrelationships of the collected parameters. This process can be implemented using professional computational tools (such as Python's Scikit-learn library or the R language environment). To ensure the effectiveness and accuracy of the model, techniques such as cross-validation will be used to evaluate model performance, and iterative optimization will be performed based on continuous feedback to ensure that the computational content truly reflects the inherent laws of objective data. In all computational processes, to eliminate the influence of different physical dimensions and ensure that data is compared and analyzed on the same scale, the input parameters in each formula are dimensionless. The dimensionless techniques used include, but are not limited to, min-max-normalization or Z-score standardization.

[0187] It should be emphasized that the foregoing embodiments are merely illustrative of preferred implementations of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art can make various modifications, equivalent substitutions, or improvements based on the technical solutions, spirit, and principles disclosed in the present invention. Any such changes that do not depart from the scope of the technical solutions of the present invention should fall within the protection scope of the claims of the present invention. Furthermore, the technical features of the various embodiments in the specification can be arbitrarily combined without creating technical contradictions. All possible, non-contradictory combinations should be considered within the scope of this specification; this application also provides a computer-readable storage medium storing computer program instructions thereon.

Claims

1. A method for remote monitoring of intelligent motors based on industrial bus, characterized in that, The specific steps include: S1: Real-time operating data acquired by the motor control system for constructing a multi-physics digital twin model of the motor; Based on the multiphysics digital twin model, the future physical state parameters of the motor within a preset time window are predicted; the future physical state parameters include the future failure probability and the system safety margin. S2: Integrate the future physical state parameters with the resource state parameters of the motor control system to generate a task criticality index; the resource state parameters include computation time margin; The task criticality index is generated by adaptively weighting and fusing the future failure probability, system safety margin, and computation time margin of the motor control system. Based on the aforementioned task criticality index, a forward-looking resource scheduling strategy is generated; S3: Based on the aforementioned forward-looking resource scheduling strategy, dynamically adjust the resource allocation of computing and communication tasks within the motor control system; S4: Based on the dynamically adjusted resource allocation results, generate a controller capability profile that characterizes the current external service capabilities of the motor control system, and broadcast the controller capability profile to external systems via the industrial bus; The controller capability profile includes a task execution determinism index; The task execution determinism index is calculated by using the time change rate of the task criticality index, which characterizes the comprehensive risk state within the controller.

2. The method for remote monitoring of intelligent motors based on industrial bus according to claim 1, characterized in that: The real-time operating data specifically includes: acquiring at least one electrical parameter; and acquiring non-electrical physical parameters that reflect the internal thermodynamics of the motor.

3. The method for remote monitoring of intelligent motors based on industrial bus according to claim 2, characterized in that: The adaptive weighted fusion includes: calculating a risk coupling factor based on the future physical state parameters and resource state parameters; The risk coupling factor includes a physical risk coupling factor, which is calculated based on a combination of future failure probability and system safety margin. The risk coupling factor also includes a resource risk coupling factor, which is calculated based on a combination of future failure probability and computation time margin. The risk coupling factor is used to dynamically adjust the preset base weights to generate dynamic weights; The future physical state parameters and resource state parameters are weighted and summed using the dynamic weights, and the summation result is then subjected to nonlinear normalization to obtain the task criticality index.

4. The method for remote monitoring of intelligent motors based on industrial bus according to claim 3, characterized in that: S2 generates a forward-looking resource scheduling strategy, specifically as follows: The numerical range of the task criticality index is divided into at least two threshold ranges; A corresponding resource scheduling mode is preset for each threshold range, and the resource scheduling mode includes at least a task priority adjustment strategy.

5. The method for remote monitoring of intelligent motors based on industrial bus according to claim 4, characterized in that: For "dividing the numerical range of the task criticality index into at least two threshold ranges", the two threshold ranges are formed by a strategy threshold parameter consisting of "low criticality threshold" and "high criticality threshold". Condition 1: Determine whether the task criticality index is greater than or equal to the high criticality threshold; If path one is true: generate a resource scheduling strategy of "safety priority mode"; Condition 2: If condition 1 is false, then continue to check whether the task criticality index is less than the low criticality threshold. If path 2 is true: generate a resource scheduling strategy of "energy efficiency priority mode"; If path 3 is false: generate a resource scheduling strategy for "performance balancing mode".

6. The method for remote monitoring of intelligent motors based on industrial bus according to claim 5, characterized in that: The larger the output value of the task criticality index, the closer the motor control system determines the overall state of the motor and its control unit to the critical state, and the more unstable the operating state.

7. The method for remote monitoring of intelligent motors based on industrial bus according to claim 6, characterized in that: Dynamically adjusting the resource allocation of computing and communication tasks within the motor control system includes: according to the preset resource scheduling mode corresponding to the resource scheduling strategy, calling the kernel service of the real-time operating system to modify the execution priority of at least one computing task and one communication task. The steps for generating a controller capability profile include: performing calculations using a dynamic capability profile generation model; The inputs to the dynamic capability profile generation model include the current resource scheduling mode and resource status parameters.

8. The method for remote monitoring of intelligent motors based on industrial bus according to claim 7, characterized in that: The step of broadcasting the controller capability profile to an external system via an industrial bus includes: encapsulating the data of the controller capability profile into a process data object of the industrial bus protocol and sending it periodically.

9. The method for remote monitoring of intelligent motors based on industrial bus according to claim 8, characterized in that: The higher the determinism index of the task execution, the more stable the overall risk state inside the controller corresponding to the motor control system.

Images