Power feedforward control method, system and control device for gravity energy storage system based on digital twinning and edge computing

Abstract

The application relates to a gravity energy storage system power feedforward control method, system and control device based on digital twinning and edge computing. By deploying a digital twinning model in an edge computing controller on the physical layer side of the system, high-frequency local collection of operation data and on-site calculation of control instructions are realized, two-way communication delay is eliminated, and a hardware foundation is provided for sub-second power response. The residual error between actual power and model predicted power is compared in real time, and physical parameters are calibrated online to ensure that the twinning model accurately maps the entity state. When a grid point disturbance or instruction change is detected, based on the calibrated digital twinning model, inverse dynamics are used to calculate feedforward control instructions, which are superimposed with the reference torque output by the PID controller to generate a total torque instruction. The feedforward instruction provides fast transient compensation to overcome mechanical inertia, and the PID controller eliminates residual error, so that the system has both fast dynamic response and steady-state smoothness, and mechanical oscillation is prevented.

Description

Technical Field

[0001] This application relates to the field of motor control technology, and in particular to a power feedforward control method, system and control device for a gravity energy storage system based on digital twin and edge computing. Background Technology

[0002] For gravity energy storage systems, existing control system architectures typically collect local operational data and then centrally upload it to a cloud server. The cloud server then processes the data and centrally issues reference commands. The local control system, based on these commands, initiates PID control to adjust the inverters, motors, or gravity blocks within the gravity energy storage system. However, the data volume during the upload and command issuance processes is massive, resulting in communication delays of tens to hundreds of milliseconds, which is insufficient to meet the low-latency response requirements of modern power systems. Furthermore, traditional PID feedback control relies on error signals; PID adjustment only begins after a deviation in grid frequency has occurred. Due to the high inertia of the mechanical components in gravity energy storage systems, the system's dynamic response lags significantly, and the fixed adjustment parameters make it difficult for gravity energy storage systems to maintain optimal performance. Summary of the Invention

[0003] In view of this, in order to at least partially improve the above problems, this application provides a power feedforward control method, system and control device for gravity energy storage system based on digital twin and edge computing.

[0004] This application first provides a power feedforward control method for a gravity energy storage system based on digital twin and edge computing, applied to a gravity energy storage system. The gravity energy storage system includes an edge computing controller, a PID controller, and a converter, and the edge computing controller and the PID controller are integrated into a gravity energy storage intelligent control device. The edge computing controller integrates a digital twin model. The method includes: The edge computing controller collects real-time operating data of the gravity energy storage system, including motor speed, torque, grid connection point data, and actual power. The digital twin model outputs predicted power based on the actual operating data; The edge computing controller compares the residual data between the actual power and the predicted power, and updates and calibrates the physical parameters of the digital twin model in real time based on the residual data; Based on the actual operating data, determine whether there is a disturbance trend at the grid connection point, or whether the reference power command has changed; If there is a disturbance trend at the grid connection point or a change in the reference power command, the edge computing controller uses the calibrated digital twin model to calculate the feedforward torque based on the principle of inverse dynamics and generates a feedforward control command. The total torque command is generated by superimposing the feedforward control command and the reference torque command output by the PID controller. The total torque command is sent to the converter so that the converter can adjust the power according to the total torque command.

[0005] In one embodiment, the digital twin model is a multiphysics coupled digital twin model, which includes a mutually coupled mechanical dynamics model, an electromagnetic transient model of an electric motor, and a power output model. The mechanical dynamics model described herein is used to describe the mechanical response of the gravity energy storage system and is constructed from differential equations containing nonlinear Coulomb friction terms and smoothed velocity constants. Its expression is as follows: ; Among them, J eq For the equivalent moment of inertia, Let ω be the angular acceleration of the motor. m T is the mechanical angular velocity of the motor. e T is the electromagnetic torque of the motor. g B is the load torque of the gravity block. eq T is the equivalent damping coefficient. c Let ω be the Coulomb friction torque. s T is the smooth velocity constant. s This refers to the elastic torque of the wire rope. The electromagnetic transient model of the motor is used to describe the electromagnetic response of the motor, and the electromagnetic torque of the motor is calculated based on the stator and rotor voltage and current parameters and then input into the mechanical dynamics model. The power output model is used to calculate the predicted power based on the motor's electromagnetic torque, the motor's mechanical angular velocity, and the motor's efficiency.

[0006] In one embodiment, the real-time updating and calibration of the physical parameters of the digital twin model based on the residual data includes: The equivalent damping coefficient and elastic modulus of the wire rope of the digital twin model are obtained as the physical parameters to be estimated by the extended Kalman filter algorithm. Construct an augmented state vector based on the physical parameters to be estimated; Using the extended Kalman filter algorithm, the augmented state vector is iteratively estimated in real time based on the residual data between the actual power and the predicted power to obtain the calibrated equivalent damping coefficient and the elastic modulus of the wire rope at each moment. The physical parameters of the digital twin model are updated in real time based on the calibrated equivalent damping coefficient and the elastic modulus of the wire rope.

[0007] In one embodiment, the real-time iterative estimation of the augmented state vector to obtain the calibrated equivalent damping coefficient and wire rope elastic modulus at each time step includes: Construct the state transition equation, and calculate the prior state estimate and prior covariance matrix for the next time step based on the state transition equation; Construct the actual observation vector for the next moment, which includes the actual angular velocity and actual output power of the motor at the next moment; The actual observation vector is predicted based on the nonlinear observation function to obtain the predicted observation value, and the observation residual, Kalman gain, posterior state estimate and posterior covariance matrix at the next time step are calculated. The updated equivalent damping coefficient and wire rope elastic modulus are extracted from the posterior state estimate.

[0008] In one embodiment, the step of calculating the feedforward torque based on the inverse dynamics principle and generating the feedforward control command includes: The digital twin model pre-plans an ideal rotational speed trajectory for the system; The edge computing controller calculates the feedforward torque that follows the ideal rotational speed trajectory of the system based on the principle of inverse mechanical dynamics and the physical parameters after real-time calibration, and uses the feedforward torque as the feedforward control command.

[0009] In one embodiment, the digital twin model pre-plans an ideal rotational speed trajectory for the system, including: Based on the changed reference power command or the determined grid connection point disturbance trend, the target mechanical angular velocity is calculated using the digital twin model. Obtain the initial mechanical angular velocity at the moment the disturbance occurs or the moment the command changes; A transition time is set, and the ideal rotational speed trajectory of the system is generated using a fifth-order polynomial. The transition time is determined by the edge computing controller through real-time simulation iterations in the digital twin model. The determination steps are as follows: The candidate transition times of different durations are substituted into the fifth-degree polynomial and the digital twin model for trial calculation; Under the constraint that the maximum output electromagnetic torque of the motor is satisfied and the rate of change of the elastic torque of the wire rope does not exceed the set safety threshold, the candidate transition time with the smallest value is selected as the set transition time.

[0010] In one embodiment, after sending the total torque command to the converter, the method further includes: The edge computing controller returns to the step of real-time acquisition of actual operating data of the gravity energy storage system at a set high-frequency sampling rate, so as to obtain the latest actual operating data of the gravity energy storage system after responding to the total torque command in the previous control cycle.

[0011] This application also provides a power feedforward control system for a gravity energy storage system based on digital twins and edge computing, applied to a gravity energy storage system, for implementing the method described in any of the preceding claims. The gravity energy storage system includes an edge computing controller, a PID controller, and a converter. The edge computing controller integrates a digital twin model, including: The data acquisition module is used to collect the actual operating data of the gravity energy storage system in real time through the edge computing controller. The actual operating data includes motor speed, torque, grid connection point data and actual power. A predicted power output module is used to output predicted power based on the actual operating data using the digital twin model; The physical parameter calibration module is used to compare the residual data between the actual power and the predicted power through the edge computing controller, and update and calibrate the physical parameters of the digital twin model in real time based on the residual data; The forward-looking monitoring module is used to determine, based on the actual operating data, whether there is a disturbance trend at the grid connection point or whether the reference power command has changed; The feedforward command calculation module is used to calculate the feedforward torque based on the inverse dynamics principle using the calibrated digital twin model if there is a disturbance trend at the grid connection point or a change in the reference power command, and generate a feedforward control command. The composite control execution module is used to generate a total torque command by superimposing the feedforward control command and the reference torque command output by the PID controller. A power regulation drive module is used to send the total torque command to the converter so that the converter can perform power regulation according to the total torque command.

[0012] This application also provides a gravity energy storage intelligent control device, deployed on the physical layer side of a gravity energy storage system; the gravity energy storage intelligent control device includes: an edge computing controller and a PID controller; The edge computing controller integrates an ARM+FPGA heterogeneous computing unit and a bus communication interface, and directly acquires high-frequency sensor data of the physical layer of the gravity energy storage system through the bus communication interface. The PID controller is communicatively connected to the edge computing controller and is used to calculate a reference torque command based on actual power feedback and send it to the edge computing controller. The edge computing controller includes: Memory used to store executable computer program instructions and multiphysics coupled digital twin models; The processor, coupled to the memory, is configured to, when executing the computer program instructions, implement the power feedforward control method for a gravity energy storage system based on digital twins and edge computing as described above, and superimpose the feedforward control instructions generated by executing the method with the reference torque instructions output by the PID controller to generate a total torque instruction which is sent to the converter of the physical layer of the gravity energy storage system for power regulation.

[0013] This application also provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0014] Beneficial Effects: This application achieves high-frequency local acquisition of operating data and on-site computation of control commands by setting the digital twin model in the edge computing controller at the physical layer of the system. This fundamentally eliminates the bidirectional communication delay caused by data transmission, providing the hardware and architectural foundation for achieving sub-second ultra-fast power response. In the long-term operation of gravity energy storage systems, physical parameters such as wire rope elasticity, mechanical friction damping, and equivalent inertia dynamically drift due to mechanical wear and aging. This application calculates the residual by comparing the actual power with the predicted power of the digital twin model in real time, and adaptively calibrates the physical parameters online based on this residual data. This allows the digital twin model to consistently and faithfully map the real state of the physical entity, effectively avoiding prediction distortion and decreased control accuracy caused by model mismatch in traditional fixed-parameter models. Traditional pure PID feedback control must wait for the system to generate significant errors before passively adjusting, often exhibiting severe response lag when facing the large inertial mechanical loads of gravity energy storage. When this application detects a disturbance trend or command change at the grid connection point, it directly calculates the feedforward control command based on the calibrated digital twin model and the principle of inverse dynamics. This command is then superimposed with the reference torque command output by the PID controller to generate the total torque command. The feedforward control command is used to provide rapid, high-power transient torque compensation in the early stages of disturbance to offset mechanical inertia. The PID controller is used to eliminate residual model errors and unmodeled random disturbances to ensure the steady-state accuracy of the system. Superimposing the two allows the total torque command output by the converter to have both extremely fast dynamic following performance and steady-state smoothness, preventing the system from experiencing severe mechanical oscillations. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating a power feedforward control method for a gravity energy storage system based on digital twins and edge computing, according to one embodiment of this application.

[0016] Figure 2 This is a power response comparison curve of a power feedforward control method for a gravity energy storage system based on digital twin and edge computing according to an embodiment of this application under a step disturbance.

[0017] Figure 3 This is a schematic diagram of the power feedforward control system of a gravity energy storage system based on digital twin and edge computing, according to one embodiment of this application.

[0018] Figure 4 This is a block diagram of a gravity energy storage intelligent control device according to an embodiment of this application.

[0019] Figure 5 This is a block diagram of a readable storage medium according to an embodiment of this application. Detailed Implementation

[0020] The terms "first," "second," and "third," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the term "comprising," and any variations thereof, is intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules not listed, or may optionally include other steps or modules inherent to these processes, methods, products, or apparatuses.

[0021] Please see Figures 1-5 This application provides an embodiment of a power feedforward control method for a gravity energy storage system based on digital twins and edge computing. This method can be applied to gravity energy storage systems, which include an edge computing controller, a PID controller, and a converter. The edge computing controller and the PID controller are integrated into a gravity energy storage intelligent control device. The edge computing controller integrates a digital twin model. The method includes: S1. The edge computing controller collects the actual operating data of the gravity energy storage system in real time. The actual operating data includes motor speed, torque, grid connection point data and actual power. S2. The digital twin model outputs predicted power based on the actual operating data; S3. The edge computing controller compares the residual data between the actual power and the predicted power, and updates and calibrates the physical parameters of the digital twin model in real time based on the residual data; S4. Based on the actual operating data, determine whether there is a disturbance trend at the grid connection point, or whether the reference power command has changed; S5. If there is a disturbance trend at the grid connection point or a change in the reference power command, the edge computing controller uses the calibrated digital twin model to calculate the feedforward torque based on the inverse dynamics principle and generates a feedforward control command. S6. Based on the feedforward control command and the reference torque command output by the PID controller, a total torque command is generated by superimposing them. S7. The total torque command is sent to the converter so that the converter can adjust the power according to the total torque command.

[0022] As described above, by setting the digital twin model in the edge computing controller on the physical layer side of the system, high-frequency local acquisition of operating data and local computation of control commands are achieved. This fundamentally eliminates the two-way communication delay caused by data transmission, providing the hardware and architectural foundation for achieving sub-second ultra-fast power response. During long-term operation, the physical parameters of a gravity energy storage system, such as the elasticity of its wire ropes, mechanical friction damping, and equivalent inertia, dynamically drift due to mechanical wear and aging. This application calculates the residual by comparing the actual power with the predicted power of the digital twin model in real time, and then adaptively calibrates the physical parameters online based on this residual data. This allows the digital twin model to consistently and faithfully map the real state of the physical entity, effectively avoiding prediction distortion and decreased control accuracy caused by model mismatch in traditional fixed-parameter models.

[0023] Traditional pure PID feedback control can only passively adjust after the system generates a significant error, often resulting in severe response lag when facing large inertial mechanical loads with gravity-stored energy. This application, upon detecting disturbance trends or command changes at the grid connection point, directly calculates the feedforward control command based on the calibrated digital twin model and the principle of inverse dynamics. This command is then superimposed with the reference torque command output by the PID controller to generate the total torque command. The feedforward control command provides rapid, high-power transient torque compensation in the early stages of disturbances to counteract mechanical inertia. The PID controller eliminates residual model errors and unmodeled random disturbances, ensuring the system's steady-state accuracy. Superimposing both allows the converter's output total torque command to possess both extremely fast dynamic tracking performance and steady-state smoothness, preventing severe mechanical oscillations in the system.

[0024] In one embodiment, the digital twin model is a multi-physics coupled digital twin model. Since a gravity energy storage system is a complex electromechanical / hydraulic / electromechanical rigid-flexible coupled system involving electrical drive, mechanical transmission, and gravitational potential energy conversion, a single physics model cannot accurately reflect the system's true transient response under grid connection point disturbances. Therefore, the multi-physics coupled digital twin model in this embodiment specifically includes a mutually coupled mechanical dynamics model, a motor electromagnetic transient model, and a power output model.

[0025] The specific construction process is as follows: First, construct the mechanical dynamics model: The mechanical components of a gravity energy storage unit exhibit large inertia, heavy loads, and complex frictional and flexible deformation characteristics. The mechanical dynamics model described herein is used to characterize the mechanical response of the gravity energy storage system and is constructed from differential equations containing nonlinear Coulomb friction terms and a smoothed velocity constant. Its expression is as follows: ; Among them, J eq For the equivalent moment of inertia, Let ω be the angular acceleration of the motor. m T is the mechanical angular velocity of the motor. e T is the electromagnetic torque of the motor. g B is the load torque of the gravity block. eq T is the equivalent damping coefficient. c Let ω be the Coulomb friction torque. s T is the smooth velocity constant. s This refers to the elastic torque of the wire rope. It should be noted that the mechanical friction of gravity energy storage systems exhibits strong nonlinear characteristics during frequent start-stop and low-speed crawling phases. Traditional linear friction models produce abrupt changes (singularities) when the rotational speed crosses zero, which can easily cause the edge computing controller to diverge during ultra-real-time numerical simulations. This embodiment incorporates a nonlinear smoothing term T during its construction. c tanh(ω m / ω s By utilizing the continuous differentiability of the hyperbolic tangent function Tanh(·), a smooth approximation of the nonlinear state of Coulomb friction at low speeds was achieved, which not only ensured the high fidelity of the physical mapping but also improved the stability and convergence speed of the numerical solution of the edge computing controller.

[0026] Secondly, construct the electromagnetic transient model of the motor: The electromagnetic transient model of the motor is used to describe the microsecond-level electromagnetic response of a drive motor (e.g., a permanent magnet synchronous motor). This model is calculated based on the stator and rotor voltage and current parameters acquired in real time by an edge computing controller. Taking the transient equations in the dq rotating coordinate system as an example, they include: ; ; Based on the above voltage and current equations, the key driving force, the motor electromagnetic torque T, is calculated. e : ; Among them, u d， u q For the d-axis and q-axis voltages, i d， i q R represents the d-axis and q-axis currents. s L is the stator resistance. d， L q For d-axis and q-axis inductance, ω e Let Ψ be the electric angular velocity. f For permanent magnet flux linkage, p is the number of pole pairs of the motor, and T is the number of pole pairs of the motor. e This refers to the electromagnetic torque. The transient electromagnetic torque T calculated by this model is... e This will serve as the input torque for the mechanical dynamics model, thereby enabling the dynamic coupling and transmission from the electrical system to the mechanical system.

[0027] Finally, construct the power output model: The power output model is used to calculate the final grid-connected predicted power of the system. The expression is: P out =η e T e ω m ; Among them, P out For the predicted power of the digital twin model, η e T represents the electrical energy to mechanical energy conversion efficiency of the motor. e For transient electromagnetic torque, ω m This refers to the mechanical angular velocity of the motor.

[0028] Combining the above three models, the electromagnetic transient model calculates the electromagnetic torque T based on electrical commands / feedback. e Subsequently, the T e The torque T of the load against gravity is input into the nonlinear mechanical dynamics equations. g Using time-varying damping and nonlinear Coulomb friction, the real-time mechanical angular velocity ω of the motor is calculated. m Finally, extract the torque T from the electrical side. e and the rotational speed ω on the mechanical side m The power output model is used to transform the system's predicted power P into a macroscopic model. outThis tightly coupled transmission chain enables the digital twin model to accurately predict the power output trajectory of the gravity energy storage system under complex disturbances at ultra-real-time speeds (e.g., 10 times faster than physical time), providing high-precision data support for subsequent feedforward compensation calculations.

[0029] In the long-term actual operation of gravity energy storage systems, due to mechanical wear, temperature changes, and fatigue tension of the wire ropes, the system's equivalent damping coefficient B... eq And the elastic modulus E of steel wire rope s Parameter drift can occur. To avoid this, in one embodiment, the edge computing controller updates and calibrates the physical parameters of the digital twin model in real time based on the residual data by comparing the actual power with the predicted power, including: The equivalent damping coefficient and elastic modulus of the wire rope of the digital twin model are obtained as the physical parameters to be estimated by the extended Kalman filter algorithm. Construct an augmented state vector based on the physical parameters to be estimated; Using the extended Kalman filter algorithm, the augmented state vector is iteratively estimated in real time based on the residual data between the actual power and the predicted power to obtain the calibrated equivalent damping coefficient and the elastic modulus of the wire rope at each moment. The physical parameters of the digital twin model are updated in real time based on the calibrated equivalent damping coefficient and the elastic modulus of the wire rope.

[0030] Specifically, traditional digital twin models typically treat physical parameters as fixed constants. To enable real-time observation of these parameters, this embodiment constructs an augmented state vector, linking the equivalent damping coefficient and the elastic modulus of the wire rope to be calibrated with the dynamic mechanical state of the system (i.e., the motor's mechanical angular velocity ω). m Together, for the k-th time, its augmented state vector x k =[ω m (k)B eq (k)E s (k)] T Then, the edge computing controller predicts the prior state estimate for the next time step (k+1) based on the system's state transition equation. and the prior covariance matrix P k+1∣k The state transition equations include the system's mechanical state transition model and parameter evolution model, as detailed below: ; B eq (k+1)=B eq (k)+w B,k ; E s (k+1)=Es (k)+w E,k ; Where, ω m (k) represents the mechanical angular velocity of the motor at the k-th moment, T s (k) represents the elastic torque of the wire rope at the k-th moment, J eq For the equivalent moment of inertia, T e (k) represents the electromagnetic torque of the motor at the k-th moment, T g (k) represents the gravitational load torque at the k-th moment, B eq (k) is the equivalent damping coefficient at time k, T c Let ω be the Coulomb friction torque. s T is the smooth velocity constant. sa For the sampling period, w B,k and w E,k This is process noise.

[0031] In the aforementioned state transition equations, the derivation of the mechanical angular velocity follows dynamic laws that include Coulomb friction and a smoothing constant; while the damping coefficient and elastic modulus are modeled as random walk processes. This specific modeling approach for large inertia gravity equipment ensures that the model does not diverge when predicting low-speed crawling conditions, while allowing the model parameters to smoothly approximate the actual physical wear state in the time domain.

[0032] After completing the prior prediction, the system incorporates the collected actual operational data for error correction and calibration. First, a heterogeneous observation vector y is constructed. (k+1) It also includes the actual angular velocity ω of the motor on the mechanical side. m (k+1) and the actual output power P of the electrical measurement out (k+1), the edge computing controller uses the observation function h(·) to map the corresponding predicted observation value, and calculates the residual r between the actual and predicted values. k+1 : ; Since the observation function h(·) has high nonlinearity, the Jacobian matrix H can be obtained by taking its partial derivative with respect to the state variables. (k+1) Subsequently, the Kalman gain K is calculated by combining the observation noise covariance R. (k+1) : ;

[0033] Using this Kalman gain as a weighting adjustment coefficient, the residual r k+1 This is superimposed on the prior estimate to complete the posterior calibration of the state and simultaneously update the covariance matrix: ; ; Where I is the identity matrix, P is the posterior state estimation vector (i.e., the posterior state estimate). k+1∣k+1 Let be the posterior covariance matrix.

[0034] Finally, the edge calculator uses the updated posterior state vector Extract B eq (k+1) and E s (k+1) is used to achieve real-time calibration of the physical parameters of the digital twin model based on an adaptive observer.

[0035] To address the significant mechanical lag issue in traditional PID feedback control, in one embodiment, based on the actual operating data, it is determined whether there is a grid connection point disturbance trend or whether the reference power command has changed; if there is a grid connection point disturbance trend or the reference power command has changed, the edge computing controller uses the calibrated digital twin model to calculate the feedforward torque based on the inverse dynamics principle and generates a feedforward control command.

[0036] Specifically, traditional systems typically only begin adjustment after detecting a significant error between the actual power and the target power, resulting in a response delay. In this embodiment, the edge computing controller monitors the grid connection point status and host computer commands in real time. First, it calculates the grid frequency deviation at the connection point in real time, i.e., the difference between the current frequency and the power frequency. If the frequency deviation is greater than a preset deviation, it determines that the grid connection point has a disturbance trend; the preset deviation can be set to 0.02Hz. Second, it calculates the change in the reference power command in real time, i.e., the difference between the previous reference power command and the current reference power command. If this change is greater than the product of the system's rated power and the noise prevention triggering ratio coefficient, it determines that the reference power command has changed. For example, the noise prevention triggering ratio coefficient can preferably be set to 1%. If either of these two conditions is met, a feedforward command is calculated and feedforward control is performed.

[0037] If the speed target is directly issued in a step manner after triggering feedforward control, the large-inertia gravity block will experience extremely high acceleration instantaneously. This will not only trigger the converter overcurrent protection but also cause severe mechanical vibration or even breakage of the flexible steel wire rope. Therefore, this embodiment pre-plans a smooth ideal speed trajectory for the system before calculating the feedforward command. Specifically, a fifth-order polynomial can be used to generate the transition trajectory ω. ref (t): ; Where t is the trajectory generation running time, and 0≤t≤T, T is the set transition time, ω0 is the initial mechanical angular velocity at the time of disturbance occurrence or command change, and ω targetFor the target mechanical angular velocity, compared to conventional linear or cubic curves, the trajectory derived from fifth-order polynomial programming not only ensures the continuity of rotational speed and angular acceleration during the transition period, but also guarantees that the jerk is zero at both the starting and target points. This theoretically eliminates the transient step change in driving torque, enabling flexible start-stop of large inertial loads.

[0038] Furthermore, the target mechanical angular velocity ω target The transition time is obtained based on the power output model and the changed reference power command or the determined grid connection point disturbance trend. The edge computing controller internally presets a set of candidate transition times with different time lengths. These are substituted into a fifth-order polynomial to generate candidate trajectories, and then extrapolated and calculated in the digital twin model at a speed more than 10 times faster than physical time. Under the premise of ensuring that the extrapolated motor electromagnetic torque does not exceed the hardware limit and that the drastic rate of change of the wire rope elastic torque is within a safe threshold, the smallest value, T, is automatically selected as the final set transition time. This allows the gravity energy storage system to adaptively release the maximum response potential of the hardware while ensuring absolute physical safety.

[0039] Finally, based on the principle of inverse mechanical dynamics, the latest calibrated equivalent rotational inertia J of the system is obtained. eq Equivalent damping coefficient B eq Load torque T g Together with the angular velocity ω of the ideal trajectory ref and its derivative (angular acceleration) Substitute them into the inverse dynamics equation: ; Among them, T ff This is a feedforward torque command. The term B is used to compensate for the dynamic torque required to overcome acceleration and deceleration of large inertia loads. eq ·ω ref The term T is used to compensate for the current real-time frictional damping torque of the system. g The term is used to overcome the fundamental drag torque generated by gravitational potential energy.

[0040] The feedforward torque command is calculated by the inverse dynamic equation and superimposed on the output of the PID controller. This enables the converter to output sufficient and accurate electromagnetic torque immediately when a disturbance occurs, reducing the regulation lag effect caused by the large inertia mechanical system and realizing rapid, accurate and stable regulation of the output power of gravity energy storage.

[0041] In one embodiment, after sending the total torque command to the converter for power regulation, the method further includes: The edge computing controller returns to the step of executing the real-time acquisition of actual operating data of the gravity energy storage system at a set high-frequency sampling rate, so as to obtain the latest actual operating data of the gravity energy storage system after responding to the total torque command in the previous control cycle; Using the latest actual operating data, the digital twin model prediction, real-time update and calibration of the physical parameters, and generation of the feedforward control command are triggered for the next control cycle. The set high-frequency sampling rate is 1 kHz to 10 kHz.

[0042] To illustrate this application more clearly, the following example, using a 500kW gravity energy storage unit in a power storage station as an example, will provide a detailed explanation of the specific workflow of this application.

[0043] First, during the system startup phase, an edge computing controller integrating an ARM+FPGA heterogeneous computing unit was deployed locally on the physical layer side (such as next to the converter and motor) of the 500kW gravity energy storage unit, and EtherCAT industrial Ethernet communication was established. Subsequently, a multiphysics digital twin (DT) model containing mechanical dynamics and motor transients was built inside the edge computing controller.

[0044] After the system enters steady-state operation, the edge computing controller starts a high-frequency closed-loop cycle: With a high-frequency sampling rate of 1kHz-10kHz, the actual speed, torque, actual output power, and voltage / frequency data of the grid connection point of the motor are collected in real time. The digital twin model performs ultra-real-time simulation at a speed more than 10 times faster than physical time, continuously outputting predicted power. At the same time, the adaptive correction module continuously compares the actual output power with the predicted power to calculate the residual, and uses the extended Kalman filter (EKF) algorithm to continuously correct key physical parameters such as the friction coefficient and equivalent inertia in the digital twin model in the background, ensuring absolute fidelity between the virtual model and the physical entity.

[0045] When no grid disturbance is detected, the system determines the result as "no" and only maintains normal feedback regulation through the local PID controller, and continuously returns to the data acquisition step for high-frequency cycling.

[0046] Disturbance occurrence and feedforward control triggering, such as Figure 2As shown, at time t=1.0s, a sudden external disturbance occurs at the grid connection point of the gravity energy storage system (e.g., a sudden increase in grid load causing a slight fluctuation in the grid connection point frequency). At this time, the edge computing controller detects the disturbance trend at the grid connection point, the system judgment result jumps to "yes", the edge computing controller calls the parameter-calibrated digital twin model, predicts the future state evolution path of the system according to the set frequency regulation target, and calculates the power compensation required to overcome the large inertia of the gravity block based on the principle of inverse mechanical dynamics, and then generates feedforward control commands.

[0047] The generated feedforward control command is superimposed with the feedback command from the local PID controller to form a total torque / power command, which is directly sent to the converter to drive the motor. After the action is completed, the program returns in milliseconds to continue monitoring the physical state after intervention. Figure 2 From the contrast response curves, we can see that: Traditional control: When using traditional pure feedback control (dashed line in the figure), due to the need to wait for the frequency deviation to accumulate and the limitations of communication and mechanical lag, the system reacts slowly after a disturbance occurs at t=1.0s, and does not reach the target power until about t=2.0s, resulting in a significant delay of up to 1 second, accompanied by a slight overshoot.

[0048] In this application, when the power feedforward control method (solid line in the figure) is adopted, due to the digital twin feedforward command, the converter obtains precise inverse dynamic compensation driving force at the instant the disturbance occurs. The system completes the smooth power ramp-up and stabilizes at the target value in less than 0.5 seconds (i.e., at t=1.3s).

[0049] In summary, this embodiment has sub-second rapid response capability. Compared with the significant adjustment lag of traditional control methods, it can smoothly adjust the power output to the target value in less than 0.5 seconds after the disturbance occurs through advanced prediction and feedforward calculation, effectively overcoming the large inertia mechanical lag problem of gravity energy storage system.

[0050] It is understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0051] Based on the same inventive concept, this application also provides a system for implementing the power feedforward control method for a gravity energy storage system based on digital twins and edge computing as described above. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more system embodiments provided below can be found in the limitations of the method described above, and will not be repeated here.

[0052] In one exemplary embodiment, such as Figure 3 As shown, a power feedforward control system for a gravity energy storage system based on digital twin and edge computing is provided. The gravity energy storage system is equipped with an edge computing controller, a PID controller and a converter. The edge computing controller integrates a digital twin model, including: an operation data acquisition module 1, a predicted power output module 2, a physical parameter calibration module 3, a forward monitoring module 4, a feedforward command calculation module 5, a composite control execution module 6, and a power regulation drive module 7.

[0053] The above system corresponds to the aforementioned method embodiments and can implement the corresponding method steps. Its implementation principle and technical effect are similar, and will not be repeated here.

[0054] Please see Figure 4 This application also provides a gravity energy storage intelligent control device, deployed on the physical layer side of a gravity energy storage system; the gravity energy storage intelligent control device includes: an edge computing controller and a PID controller; The edge computing controller integrates an ARM+FPGA heterogeneous computing unit and a bus communication interface, and directly acquires high-frequency sensor data of the physical layer of the gravity energy storage system through the bus communication interface. The PID controller is communicatively connected to the edge computing controller and is used to calculate a reference torque command based on actual power feedback and send it to the edge computing controller. The edge computing controller includes: Memory used to store executable computer program instructions and multiphysics coupled digital twin models; The processor, coupled to the memory, is configured to, when executing the computer program instructions, implement the power feedforward control method for a gravity energy storage system based on digital twins and edge computing as described above, and superimpose the feedforward control instructions generated by executing the method with the reference torque instructions output by the PID controller to generate a total torque instruction which is sent to the converter of the physical layer of the gravity energy storage system for power regulation.

[0055] Please see Figure 5This application also provides a readable storage medium storing a computer application program, which, when executed by a processor, implements the methods of the above-described method embodiments.

[0056] The readable storage medium can be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of computer programs from one location to another. A computer storage medium can be any available medium accessible to a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located on an Application-Specific Integrated Circuit (ASIC). s In an ASIC (Integrated Circuit-Based ASIC), the processor and readable storage medium can reside within the user equipment. Alternatively, the processor and readable storage medium can also exist as discrete components within the communication device.

[0057] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein, and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A power feedforward control method for a gravity energy storage system based on digital twin and edge computing, applied to a gravity energy storage system, characterized in that, The gravity energy storage system includes an edge computing controller, a PID controller, and a converter, and the edge computing controller and the PID controller are integrated into a gravity energy storage intelligent control device. The edge computing controller integrates a digital twin model. The method includes: The edge computing controller collects real-time operating data of the gravity energy storage system, including motor speed, torque, grid connection point data, and actual power. The digital twin model outputs predicted power based on the actual operating data; The edge computing controller compares the residual data between the actual power and the predicted power, and updates and calibrates the physical parameters of the digital twin model in real time based on the residual data; Based on the actual operating data, determine whether there is a disturbance trend at the grid connection point, or whether the reference power command has changed; If there is a disturbance trend at the grid connection point or a change in the reference power command, the edge computing controller uses the calibrated digital twin model to calculate the feedforward torque based on the inverse dynamics principle and generates a feedforward control command. The total torque command is generated by superimposing the feedforward control command and the reference torque command output by the PID controller. The total torque command is sent to the converter so that the converter can adjust the power according to the total torque command.

2. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 1, characterized in that, The digital twin model is a multiphysics coupled digital twin model, which includes a coupled mechanical dynamics model, an electromagnetic transient model of an electric motor, and a power output model. The mechanical dynamics model described herein is used to describe the mechanical response of the gravity energy storage system and is constructed from differential equations containing nonlinear Coulomb friction terms and smoothed velocity constants. Its expression is as follows: ； Among them, J eq For equivalent rotational inertia, Let ω be the angular acceleration of the motor. m T is the mechanical angular velocity of the motor. e T is the electromagnetic torque of the motor. g B is the load torque of the gravity block. eq T is the equivalent damping coefficient. c Let ω be the Coulomb friction torque. s T is the smooth velocity constant. s This refers to the elastic torque of the wire rope. The electromagnetic transient model of the motor is used to describe the electromagnetic response of the motor, and the electromagnetic torque of the motor is calculated based on the stator and rotor voltage and current parameters and then input into the mechanical dynamics model. The power output model is used to calculate the predicted power based on the motor's electromagnetic torque, the motor's mechanical angular velocity, and the motor's efficiency.

3. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 1, characterized in that, The real-time updating and calibration of the physical parameters of the digital twin model based on the residual data includes: The equivalent damping coefficient and elastic modulus of the wire rope of the digital twin model are obtained as the physical parameters to be estimated by the extended Kalman filter algorithm. Construct an augmented state vector based on the physical parameters to be estimated; Using the extended Kalman filter algorithm, the augmented state vector is iteratively estimated in real time based on the residual data between the actual power and the predicted power to obtain the calibrated equivalent damping coefficient and the elastic modulus of the wire rope at each moment. The physical parameters of the digital twin model are updated in real time based on the calibrated equivalent damping coefficient and the elastic modulus of the wire rope.

4. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 3, characterized in that, The real-time iterative estimation of the augmented state vector to obtain the calibrated equivalent damping coefficient and wire rope elastic modulus at each time step includes: Construct the state transition equation, and calculate the prior state estimate and prior covariance matrix for the next time step based on the state transition equation; Construct the actual observation vector for the next moment, which includes the actual angular velocity and actual output power of the motor at the next moment; The actual observation vector is predicted based on the nonlinear observation function to obtain the predicted observation value, and the observation residual, Kalman gain, posterior state estimate and posterior covariance matrix at the next time step are calculated. The updated equivalent damping coefficient and wire rope elastic modulus are extracted from the posterior state estimate.

5. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 1, characterized in that, The step of calculating the feedforward torque based on the inverse dynamics principle and generating the feedforward control command includes: The digital twin model pre-plans an ideal rotational speed trajectory for the system; The edge computing controller calculates the feedforward torque that follows the ideal rotational speed trajectory of the system based on the principle of inverse mechanical dynamics and the physical parameters after real-time calibration, and uses the feedforward torque as the feedforward control command.

6. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 5, characterized in that, The digital twin model pre-plans an ideal rotational speed trajectory for the system, including: Based on the changed reference power command or the determined grid connection point disturbance trend, the target mechanical angular velocity is calculated using the digital twin model. Obtain the initial mechanical angular velocity at the moment the disturbance occurs or the moment the command changes; A transition time is set, and the ideal rotational speed trajectory of the system is generated using a fifth-order polynomial. The transition time is determined by the edge computing controller through real-time simulation iterations in the digital twin model. The determination steps are as follows: The candidate transition times of different durations are substituted into the fifth-degree polynomial and the digital twin model for trial calculation; Under the constraint that the maximum output electromagnetic torque of the motor is satisfied and the rate of change of the elastic torque of the wire rope does not exceed the set safety threshold, the candidate transition time with the smallest value is selected as the set transition time.

7. The power feedforward control method for gravity energy storage systems based on digital twins and edge computing according to claim 1, characterized in that, After sending the total torque command to the converter, the process further includes: The edge computing controller returns to the step of real-time acquisition of actual operating data of the gravity energy storage system at a set high-frequency sampling rate, so as to obtain the latest actual operating data of the gravity energy storage system after responding to the total torque command in the previous control cycle.

8. A power feedforward control system for a gravity energy storage system based on digital twin and edge computing, characterized in that, Applied to a gravity energy storage system, for implementing the method as described in any one of claims 1 to 7, the gravity energy storage system includes an edge computing controller, a PID controller, and a converter, wherein the edge computing controller integrates a digital twin model, including: The data acquisition module is used to collect the actual operating data of the gravity energy storage system in real time through the edge computing controller. The actual operating data includes motor speed, torque, grid connection point data and actual power. A predicted power output module is used to output predicted power based on the actual operating data using the digital twin model; The physical parameter calibration module is used to compare the residual data between the actual power and the predicted power through the edge computing controller, and update and calibrate the physical parameters of the digital twin model in real time based on the residual data; The forward-looking monitoring module is used to determine, based on the actual operating data, whether there is a disturbance trend at the grid connection point or whether the reference power command has changed; The feedforward command calculation module is used to calculate the feedforward torque based on the inverse dynamics principle using the calibrated digital twin model if there is a disturbance trend at the grid connection point or a change in the reference power command, and generate a feedforward control command. The composite control execution module is used to generate a total torque command by superimposing the feedforward control command and the reference torque command output by the PID controller. A power regulation drive module is used to send the total torque command to the converter so that the converter can perform power regulation according to the total torque command.

9. A gravity energy storage intelligent control device, characterized in that, Deployed on the physical layer side of gravity energy storage systems; The gravity energy storage intelligent control device includes: an edge computing controller and a PID controller; The edge computing controller integrates an ARM+FPGA heterogeneous computing unit and a bus communication interface, and directly acquires high-frequency sensor data of the physical layer of the gravity energy storage system through the bus communication interface. The PID controller is communicatively connected to the edge computing controller and is used to calculate a reference torque command based on actual power feedback and send it to the edge computing controller. The edge computing controller includes: Memory used to store executable computer program instructions and multiphysics coupled digital twin models; The processor, coupled to the memory, is configured to, when executing the computer program instructions, implement the power feedforward control method for a gravity energy storage system based on digital twins and edge computing as described in any one of claims 1 to 7, and superimpose the feedforward control instructions generated by executing the method with the reference torque instructions output by the PID controller to generate a total torque instruction which is sent to the converter of the physical layer of the gravity energy storage system for power regulation.

10. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images