An elevator energy consumption control system based on multi-objective optimization

Abstract

This invention discloses an elevator energy consumption control system based on multi-objective optimization, comprising: a parameterization module for generating an adaptive non-uniform B-spline control point set and corresponding non-uniform node vectors; an elevator discrete nonlinear dynamics module for constructing an elevator discrete nonlinear dynamics model; an index function construction module for constructing an optimal control performance index function for the elevator in the finite-time domain; an iteration module for constructing a linear approximation system; an improved Newton-GMRES module for obtaining the optimal correction amount for the adaptive non-uniform B-spline control point set; an update module for generating an updated adaptive non-uniform B-spline trajectory; a motor drive signal generation module for driving the elevator along the updated adaptive non-uniform B-spline trajectory; and a loop module for repeatedly executing the steps from the index function construction module to the motor drive signal generation module. This invention significantly improves the speed and accuracy of trajectory generation and adaptive adjustment.

Description

Technical Field

[0001] This invention relates to the field of elevator technology, and in particular to an elevator energy consumption control system based on multi-objective optimization. Background Technology

[0002] With the rapid development of high-rise buildings in cities, elevators, as key equipment for vertical transportation, have seen operational efficiency, energy consumption optimization, and passenger comfort become important research directions in the field of elevator control. Traditional elevator trajectory planning methods typically use a preset S-shaped speed curve and achieve acceleration, stabilization, and deceleration through table lookup or simple logic control. To improve energy efficiency and passenger experience, some high-end elevator systems have introduced model-based predictive control methods or optimization algorithms for trajectory generation and adjustment.

[0003] Existing technologies still have several prominent shortcomings in trajectory planning and control: Elevator trajectory optimization based on nonlinear model predictive control requires solving high-dimensional nonlinear optimization problems in real time within each control cycle. These problems involve a large number of matrix operations, especially the inversion of the Hessian matrix, which has an extremely high computational load. This makes it difficult to achieve millisecond-level real-time closed-loop control on hardware platforms with limited computing power, such as embedded elevator controllers, thus affecting the feasibility of industrial applications. Most existing trajectory generation methods use fixed curve shapes, making it difficult to adaptively adjust according to real-time changes in elevator load. Under light or heavy loads, this can easily lead to energy redundancy, wasted capacity, or decreased passenger comfort. Under sudden load changes, traditional methods cannot guarantee trajectory continuity and consistency with physical constraints. Summary of the Invention

[0004] One objective of this invention is to propose an elevator energy consumption control system based on multi-objective optimization. This invention significantly improves the speed and accuracy of trajectory generation and adaptive adjustment, ensuring the industrial applicability of the elevator system under variable load conditions.

[0005] An elevator energy consumption control system based on multi-objective optimization according to an embodiment of the present invention includes:

[0006] The parameterization module collects real-time sensor information on elevator operation under varying loads to construct a real-time elevator operating state vector, and initializes an adaptive non-uniform B-spline trajectory parameterization model to generate an adaptive non-uniform B-spline control point set and corresponding non-uniform node vectors.

[0007] The elevator discrete nonlinear dynamics module is used to construct a discrete nonlinear dynamics model of the elevator.

[0008] The index function construction module constructs the performance index function of the elevator finite time domain nonlinear optimal control, and combines it with the elevator discrete nonlinear dynamics model to transform the elevator nonlinear optimal control problem into an elevator nonlinear parameter optimization problem with an adaptive non-uniform B-spline control point set as the decision variable.

[0009] The iterative module performs Newton linearization iteration on the elevator nonlinear parameter optimization problem to construct a linear approximation system.

[0010] An improved Newton-GMRES module is applied to the linear approximation system. The improved Newton-GMRES algorithm is used to iteratively solve the linear approximation system in the Krylov subspace to obtain the optimal correction amount for the adaptive non-uniform B-spline control point set.

[0011] The update module adds the optimal correction to the current set of adaptive non-uniform B-spline control points to generate the updated adaptive non-uniform B-spline trajectory.

[0012] The motor drive signal generation module calculates the target electromagnetic torque command of the permanent magnet synchronous motor based on the updated adaptive non-uniform B-spline trajectory, and converts it into a space vector pulse width modulation motor drive signal to drive the elevator to run along the updated adaptive non-uniform B-spline trajectory.

[0013] The loop module repeatedly executes the steps from the index function construction module to the motor drive signal generation module until the elevator car reaches the target floor corresponding to the elevator target floor data and stops running, thus completing the elevator energy consumption control.

[0014] Optionally, the parameterization module includes:

[0015] Collect real-time sensor information on elevator operation under varying loads;

[0016] Construct a real-time elevator operating status vector based on real-time sensor information of elevator operation under variable load;

[0017] Based on the real-time operating state vector of the elevator, an adaptive non-uniform B-spline trajectory parameterization model is initialized, and a B-spline expression of the elevator trajectory function is established.

[0018] Construct an initial set of adaptive non-uniform B-spline control points;

[0019] Set an adaptive non-uniform node vector, where each node in the adaptive non-uniform node vector is a node parameter value;

[0020] The initial adaptive non-uniform B-spline control point set is combined with the adaptive non-uniform node vector to form an initial adaptive non-uniform B-spline trajectory parameterization model.

[0021] Optionally, the elevator discrete nonlinear dynamics module includes:

[0022] By combining the elevator car mass parameters, the set of elastic parameters of the elevator traction steel wire rope, the parameters of the equivalent damping coefficient of the elevator traction steel wire rope, the set of electromagnetic torque equation parameters of the permanent magnet synchronous motor, and the real-time operating state vector of the elevator, a discrete nonlinear dynamic model of the elevator is constructed.

[0023] Calculate the time-varying stiffness of the wire rope as it changes with displacement, based on the cross-sectional area and elastic modulus of the wire rope.

[0024] The longitudinal dynamic equilibrium equation of the multi-mass-spring-damped system is constructed. The longitudinal dynamic equilibrium equation of the multi-mass-spring-damped system is a linear superposition of the product of the equivalent mass and the car acceleration, the product of the equivalent damping coefficient of the traction wire rope and the car velocity, and the product of the time-varying stiffness of the wire rope and the car displacement. The result of the linear superposition is equal to the difference between the thrust of the permanent magnet synchronous motor and the gravity term and the nonlinear friction term.

[0025] Based on the three-phase current data of the permanent magnet synchronous motor and the parameter set of the electromagnetic torque equation of the permanent magnet synchronous motor, the electromagnetic torque of the motor is calculated, and the electromagnetic torque of the motor is divided by the radius of the traction wheel to obtain the thrust of the permanent magnet synchronous motor.

[0026] A nonlinear friction term is constructed from the Coulomb friction force and the peak static friction force.

[0027] Optionally, the indicator function construction module includes:

[0028] Based on the discrete nonlinear dynamics model of the elevator and the adaptive non-uniform B-spline trajectory parameterization model, the optimal control performance index function of the elevator in the finite time domain is determined.

[0029] The energy consumption performance index of the elevator is obtained by integrating the heat loss of the motor stator resistance in the prediction time domain.

[0030] The elevator ride comfort performance index is obtained by integrating the square of the car acceleration in the prediction time domain.

[0031] The elevator displacement end-of-time error performance index is obtained by weighting the square of the deviation between the actual elevator displacement and the target displacement at the end of the prediction time domain, and the elevator speed end-of-time error performance index is obtained by weighting the square of the deviation between the actual elevator speed and zero speed at the end of the prediction time domain.

[0032] The elevator energy consumption performance index, elevator ride comfort performance index, elevator displacement end error performance index, and elevator speed end error performance index are weighted and summed to form the elevator finite time domain nonlinear optimal control performance index function.

[0033] Minimizing the performance index function of the finite-time nonlinear optimal control of an elevator is used to obtain the optimal control trajectory that balances energy consumption and comfort while maintaining accurate end-point state.

[0034] Optionally, the iterative module includes:

[0035] The finite-time-domain nonlinear optimal control performance index function of the elevator is expressed as a multivariable nonlinear function with an adaptive non-uniform B-spline control point set as the input variable.

[0036] With the goal of minimizing the performance index function of the finite-time nonlinear optimal control of the elevator, the KKT optimality condition for the elevator nonlinear parameter optimization problem is established. The first necessary condition of the KKT optimality condition is that the gradient vector of the performance index function, which consists of all partial derivatives of the finite-time nonlinear optimal control performance index function with respect to the set of adaptive non-uniform B-spline control points, is equal to zero.

[0037] At the current set of adaptive non-uniform B-spline control points, perform Newton linearization to construct a linear approximation system for the elevator nonlinear parameter optimization problem;

[0038] For linear approximation systems, calculate the Newton residual vector.

[0039] Optionally, the improved Newton-GMRES module includes:

[0040] The initial solution of the improved Newton-GMRES iteration is set to a vector of all zeros. The initial residual vector is calculated to be equal to the right-hand side of the linear approximation system. The second norm of the initial residual is calculated, and the initial residual vector is normalized using this norm to obtain the first basis vector of the Krylov subspace.

[0041] In the process of constructing the Krylov subspace, for any current basis vector, the product of the Hessian matrix and the basis vector at the current set of adaptive non-uniform B-spline control points is approximately calculated using the finite difference method, resulting in a matrix-vector product.

[0042] The matrix-vector product obtained by the finite difference method is orthogonalized according to the Krylov orthogonalization process, and the orthogonal basis vector set of the Krylov subspace is generated sequentially. The orthogonal basis vector set of the Krylov subspace is used to construct the orthogonal basis matrix of the Krylov subspace and the corresponding upper Hessenberg matrix. The least squares problem is solved in the subspace to approximately satisfy the linear approximation system and obtain the optimal correction amount of the adaptive non-uniform B-spline control point set.

[0043] Calculate the current residual vector during the Newton-GMRES iteration process, determine whether the L2 norm of the residual vector is less than the currently set GMRES convergence threshold, and stop the Krylov subspace expansion when the residual vector meets the convergence condition.

[0044] Optionally, the update module includes:

[0045] The optimal correction amount of the adaptive non-uniform B-spline control point set obtained by solving the Krylov subspace in the Newton-GMRES algorithm is added point by point to the current adaptive non-uniform B-spline control point set to generate the updated adaptive non-uniform B-spline control point set.

[0046] Based on the updated set of adaptive non-uniform B-spline control points and the given adaptive non-uniform node vector, the updated adaptive non-uniform B-spline trajectory is generated using the B-spline curve reconstruction method.

[0047] Optionally, the motor drive signal generation module includes:

[0048] Based on the updated adaptive non-uniform B-spline trajectory function, the target acceleration characteristics required for elevator operation are calculated, and the target thrust of the permanent magnet synchronous motor is determined.

[0049] Multiply the target thrust of the permanent magnet synchronous motor by the radius of the traction wheel to obtain the target electromagnetic torque command of the permanent magnet synchronous motor;

[0050] Divide the target electromagnetic torque command of the permanent magnet synchronous motor by the product of the number of pole pairs of the motor and the excitation flux of the permanent magnet, and multiply by a scaling factor to obtain the target q-axis current command of the permanent magnet synchronous motor. Set the d-axis current command to zero to obtain the target d-axis and q-axis current commands of the permanent magnet synchronous motor.

[0051] Based on the target d-axis and q-axis current commands of the permanent magnet synchronous motor, the target d-axis voltage command and the target q-axis voltage command are obtained;

[0052] The target d-axis voltage command and the target q-axis voltage command are sequentially subjected to inverse Park transformation and inverse Clarke transformation to transform the voltage command in the two-phase stationary coordinate system into the three-phase stator target voltage;

[0053] Based on the three-phase target voltage, the corresponding permanent magnet synchronous motor drive signal is generated using the space vector pulse width modulation method.

[0054] The motor drive signal is output to the elevator drive system, which drives the elevator traction mechanism to run according to the updated adaptive non-uniform B-spline trajectory.

[0055] The beneficial effects of this invention are:

[0056] This invention reconstructs the elevator trajectory parameter optimization problem into a multivariate nonlinear optimization problem with adaptive non-uniform B-spline control points as decision variables. It introduces a differentiable improved Newton-GMRES solution framework, uses finite difference approximation to replace the explicit storage and computation of Hessian matrices, and combines it with Krylov subspace iteration and residual orthogonalization mechanisms to effectively avoid the high time and space complexity of traditional inversion operations. This enables online real-time closed-loop optimization of high-order nonlinear, time-varying parameter systems on ordinary microcontrollers, significantly improving the speed and accuracy of trajectory generation and adaptive adjustment, and ensuring the industrial practicality of elevator systems under variable load conditions.

[0057] This invention constructs an adaptive non-uniform B-spline trajectory model, combining the flexible dynamic characteristics of the wire rope's elastic modulus, cross-sectional area, and time-varying rope length. It systematically introduces a multi-mass-spring-damped dynamic model, achieving deep coupling between the trajectory and the physical system. Control points and node vectors are automatically adjusted according to real-time conditions. Higher local degrees of freedom can be allocated to specific trajectory intervals during load abrupt changes and high-dynamic segments. This allows the trajectory to meet the constraints of ultimate acceleration, jerk, and physical safety while improving ride comfort and leveling accuracy, overcoming the technical shortcomings of existing fixed curves in dealing with complex dynamics and nonlinear disturbances. Attached Figure Description

[0058] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0059] Figure 1 This is a structural block diagram of an elevator energy consumption control system based on multi-objective optimization proposed in this invention;

[0060] Figure 2 This is a flowchart of the adaptive non-uniform B-spline control point optimization iteration process of the improved Newton-GMRES algorithm in an elevator energy consumption control system based on multi-objective optimization proposed in this invention. Detailed Implementation

[0061] Example 1: Reference Figures 1-2 An elevator energy consumption control system based on multi-objective optimization includes:

[0062] The parameterization module collects real-time sensor information on elevator operation under varying loads to construct a real-time elevator operating state vector, and initializes an adaptive non-uniform B-spline trajectory parameterization model to generate an adaptive non-uniform B-spline control point set and corresponding non-uniform node vectors.

[0063] In this embodiment, the parameterization module includes:

[0064] Collect real-time sensor information on elevator operation under varying loads;

[0065] In Example 1, the real-time sensing information for elevator operation under varying loads includes: real-time elevator load data acquired by a weighing sensor, representing the effective load mass of the current car; real-time elevator position data and real-time elevator speed data acquired by a rotary encoder, where the real-time elevator position data represents the current position of the car and the real-time elevator speed data represents the current running speed of the car; real-time three-phase current data of the permanent magnet synchronous motor acquired by a current sensor, representing the current three-phase current of the drive system; and elevator target floor data input by the user control command, representing the displacement to the target floor.

[0066] Construct a real-time elevator operating status vector based on real-time sensor information of elevator operation under variable load;

[0067] In Example 1, the elevator real-time operating state vector is composed of elevator real-time position data, elevator real-time speed data, elevator real-time load data, permanent magnet synchronous motor real-time three-phase current data, and elevator target floor data in sequence, which is used to describe the operating state of the elevator system at the current moment.

[0068] Based on the real-time operating state vector of the elevator, an adaptive non-uniform B-spline trajectory parameterization model is initialized, and a B-spline expression of the elevator trajectory function is established.

[0069] In Example 1, the start and end parameters of the elevator trajectory function are determined based on the real-time elevator operating state vector. The start of the elevator trajectory function is jointly determined by the real-time elevator position data and the real-time elevator speed data, and the end of the elevator trajectory function is jointly set by the elevator target floor data and the expected arrival speed. Based on the start and end parameters, combined with the real-time elevator load data and other components in the real-time elevator operating state vector, a set of initial control points is generated by setting the initial values ​​of each control point using parameter mapping rules. According to the preset non-uniform node vector and node distribution requirements, the initial control point set and the non-uniform node vector are jointly calculated using a B-spline curve generation algorithm to obtain an adaptive non-uniform B-spline trajectory parameterization model with the initial control point set as control points and the non-uniform node vector as node parameters. The B-spline expression of the elevator trajectory function is established, and the B-spline expression is used to describe the smooth running trajectory of the elevator from the current state to the target floor.

[0070] Construct an initial set of adaptive non-uniform B-spline control points;

[0071] In Example 1, each control point in the initial adaptive non-uniform B-spline control point set is a geometric control point of the elevator trajectory within the corresponding parameter segment, used to control the speed change trend of the elevator during that time period.

[0072] Set an adaptive non-uniform node vector, where each node in the adaptive non-uniform node vector is a node parameter value;

[0073] The node spacing distribution satisfies the following: during the elevator start-up and braking periods, the node spacing is small and the node density is high; during the elevator steady-speed period, the node spacing is large and the node density is low. This node density distribution allows the B-spline curve to have higher local degrees of freedom during the start-up and braking periods to finely control trajectory changes, while reducing the number of control points during the steady-speed period to improve computational efficiency.

[0074] The initial adaptive non-uniform B-spline control point set is combined with the adaptive non-uniform node vector to form an initial adaptive non-uniform B-spline trajectory parameterization model.

[0075] The elevator discrete nonlinear dynamics module is used to construct a discrete nonlinear dynamics model of the elevator.

[0076] In this embodiment, the elevator discrete nonlinear dynamics module includes:

[0077] By combining the elevator car mass parameters, the set of elastic parameters of the elevator traction steel wire rope, the parameters of the equivalent damping coefficient of the elevator traction steel wire rope, the set of electromagnetic torque equation parameters of the permanent magnet synchronous motor, and the real-time operating state vector of the elevator, a discrete nonlinear dynamic model of the elevator is constructed.

[0078] In Example 1, the elevator car mass parameters describe the static mass of the car; the set of elastic parameters for the elevator traction wire rope includes the elastic modulus of the wire rope, the cross-sectional area of ​​the wire rope, and the reference rope length; the equivalent damping coefficient parameter of the elevator traction wire rope represents the damping characteristics of the wire rope; the set of electromagnetic torque equation parameters for the permanent magnet synchronous motor includes the d-axis inductance, the q-axis inductance, the excitation flux linkage, the stator resistance, the number of pole pairs, and the radius of the traction sheave; and the real-time operating state vector of the elevator reflects the elevator's position, speed, load, and target information at the current moment.

[0079] Calculate the time-varying stiffness of the wire rope as it changes with displacement, based on the cross-sectional area and elastic modulus of the wire rope.

[0080] In Example 1, the time-varying stiffness of the wire rope is calculated as the product of the wire rope's elastic modulus and its cross-sectional area, divided by the sum of the reference rope length and the car displacement. This is used to describe the flexibility of the elevator traction wire rope under different displacements. The wire rope's elastic modulus represents the material's elastic properties, the wire rope's cross-sectional area represents the size of the load-bearing section, the reference rope length represents the initial undeformed length of the wire rope, and the car displacement represents the car's current position in the hoistway.

[0081] The longitudinal dynamic equilibrium equation of the multi-mass-spring-damped system is constructed. The longitudinal dynamic equilibrium equation of the multi-mass-spring-damped system is a linear superposition of the product of the equivalent mass and the car acceleration, the product of the equivalent damping coefficient of the traction wire rope and the car velocity, and the product of the time-varying stiffness of the wire rope and the car displacement. The result of the linear superposition is equal to the difference between the thrust of the permanent magnet synchronous motor and the gravity term and the nonlinear friction term.

[0082] In Example 1, the equivalent mass is the sum of the car mass and the current real-time load, the car acceleration is the rate of change of velocity over time, the car speed is the current speed, the car displacement is the current position, the thrust of the permanent magnet synchronous motor is used to drive the elevator motion, the gravity term is the product of the equivalent mass and the gravitational acceleration, and the nonlinear friction term is used to compensate for the frictional losses of the elevator guide system and the wire rope.

[0083] ;

[0084] in, This represents the equivalent mass of the elevator system, which includes the sum of the car mass and the effective load mass at the current moment. Let represent the derivative of the elevator's speed with respect to time, and let represent the elevator's acceleration. This represents the equivalent damping coefficient of the traction wire rope. Indicates the elevator at time The speed of the car, Indicates the time-varying stiffness of a steel wire rope. Indicates the elevator at time The displacement of the car, Indicates the permanent magnet synchronous motor at time The thrust generated Represents the gravity term. This represents the nonlinear friction term.

[0085] Based on the three-phase current data of the permanent magnet synchronous motor and the parameter set of the electromagnetic torque equation of the permanent magnet synchronous motor, the electromagnetic torque of the motor is calculated, and the electromagnetic torque of the motor is divided by the radius of the traction wheel to obtain the thrust of the permanent magnet synchronous motor.

[0086] In Example 1, the three-phase current data of the permanent magnet synchronous motor is subjected to coordinate transformation. The three-phase current data is converted into d-axis current components and q-axis current components by Clark transformation and Park transformation respectively. The d-axis current component is used to represent the current in the same direction as the excitation flux linkage, and the q-axis current component is used to represent the current perpendicular to the excitation flux linkage direction. The electromagnetic torque of the permanent magnet synchronous motor at the current moment is calculated. The value of the electromagnetic torque is equal to the sum of the number of pole pairs of the motor multiplied by the product of the excitation flux linkage and the q-axis current component, and the sum of the difference between the motor's d-axis inductance and q-axis inductance multiplied by the product of the d-axis current component and the q-axis current component, and then multiplied by the scaling factor. The calculated electromagnetic torque of the permanent magnet synchronous motor is divided by the traction wheel radius in the parameter set of the electromagnetic torque equation of the permanent magnet synchronous motor, and the electromagnetic torque in Newton-meters is converted into the thrust of the permanent magnet synchronous motor in Newtons.

[0087] A nonlinear friction term is constructed from the Coulomb friction force and the peak static friction force.

[0088] In Example 1, the nonlinear friction term is obtained by multiplying the difference between the Coulomb friction force and the peak static friction force by an exponential decay function, adding the Coulomb friction force to the result, multiplying the result by the hyperbolic tangent function of the car speed, and then adding the product of the viscous friction coefficient and the car speed. The Coulomb friction force is the minimum friction force when the system is moving, the peak static friction force is the maximum static friction force at the moment of system startup, the exponential decay function reflects the influence of speed change on friction force, the hyperbolic tangent function of the car speed achieves a continuous and smooth transition of the friction term near zero speed, and the viscous friction coefficient is the friction component that increases linearly with speed.

[0089] ;

[0090] in, Coulomb friction, This represents the peak value of static friction. Stribeck's characteristic car speed, The exponential decay coefficient is... As a smoothing factor, It is the coefficient of viscous friction.

[0091] The index function construction module constructs the performance index function of the elevator finite time domain nonlinear optimal control, and combines it with the elevator discrete nonlinear dynamics model to transform the elevator nonlinear optimal control problem into an elevator nonlinear parameter optimization problem with an adaptive non-uniform B-spline control point set as the decision variable.

[0092] In this embodiment, the index function construction module includes:

[0093] Based on the discrete nonlinear dynamics model of the elevator and the adaptive non-uniform B-spline trajectory parameterization model, the optimal control performance index function of the elevator in the finite time domain is determined.

[0094] In Example 1, the finite-time-domain nonlinear optimal control performance index function of the elevator consists of a process performance term and an elevator terminal state penalty term. The process performance term includes the elevator energy consumption performance index and the elevator ride comfort performance index, while the elevator terminal state penalty term includes the elevator displacement end error performance index and the elevator speed end error performance index.

[0095] The energy consumption performance index of the elevator is obtained by integrating the heat loss of the motor stator resistance in the prediction time domain.

[0096] The elevator energy consumption performance index is obtained by integrating the weighted sum of the stator resistance of the motor and the squares of the d-axis current component and the q-axis current component.

[0097] The elevator ride comfort performance index is obtained by integrating the square of the car acceleration in the prediction time domain.

[0098] Integrating the square of the second derivative of the car speed with respect to time yields an elevator ride comfort performance index, which measures the smoothness of the elevator's trajectory and ride comfort.

[0099] The elevator displacement end-of-time error performance index is obtained by weighting the square of the deviation between the actual elevator displacement and the target displacement at the end of the prediction time domain, and the elevator speed end-of-time error performance index is obtained by weighting the square of the deviation between the actual elevator speed and zero speed at the end of the prediction time domain.

[0100] The elevator energy consumption performance index, elevator ride comfort performance index, elevator displacement end error performance index, and elevator speed end error performance index are weighted and summed to form the elevator finite time domain nonlinear optimal control performance index function.

[0101] Minimizing the performance index function of the finite-time nonlinear optimal control of an elevator is used to obtain the optimal control trajectory that balances energy consumption and comfort while maintaining accurate end-point state.

[0102] The finite-time-domain nonlinear optimal control performance index function of the elevator is used as the objective function for the elevator nonlinear parameter optimization problem.

[0103] The iterative module performs Newton linearization iteration on the elevator nonlinear parameter optimization problem to construct a linear approximation system.

[0104] In this embodiment, the iteration module includes:

[0105] The finite-time-domain nonlinear optimal control performance index function of the elevator is expressed as a multivariable nonlinear function with an adaptive non-uniform B-spline control point set as the input variable.

[0106] The adaptive non-uniform B-spline control point set represents an ordered combination of all B-spline control points in the current control cycle. Each control point is a specific physical location of the elevator trajectory in the parameter space. The adaptive non-uniform B-spline control point set controls the geometry of the elevator's running trajectory.

[0107] With the goal of minimizing the performance index function of the finite-time nonlinear optimal control of the elevator, the KKT optimality condition for the elevator nonlinear parameter optimization problem is established. The first necessary condition of the KKT optimality condition is that the gradient vector of the performance index function, which consists of all partial derivatives of the finite-time nonlinear optimal control performance index function with respect to the set of adaptive non-uniform B-spline control points, is equal to zero.

[0108] The gradient vector of the performance index function is used to measure the direction and magnitude of the change in the performance index corresponding to the current set of adaptive non-uniform B-spline control points.

[0109] At the current set of adaptive non-uniform B-spline control points, perform Newton linearization to construct a linear approximation system for the elevator nonlinear parameter optimization problem;

[0110] In Example 1, the linear approximation system takes the current set of control points as the reference point and is jointly constructed by the corresponding Hessian matrix and the gradient vector of the current performance index function. The Hessian matrix is ​​a symmetric matrix composed of all second-order partial derivatives of the elevator finite-time nonlinear optimal control performance index function with respect to the adaptive non-uniform B-spline control point set.

[0111] ;

[0112] in, This represents the Hessian matrix at the current set of control points. This represents a column vector representing the control point update values ​​in the current iteration step. This is the gradient vector of the performance index function at the current control point.

[0113] For linear approximation systems, calculate the Newton residual vector.

[0114] The Newton residual vector is a vector composed of the partial derivatives of the performance index function with respect to each control point of the current set of adaptive non-uniform B-spline control points. The Newton residual vector is used to measure the degree of deviation between the current trajectory control point configuration and the optimal performance solution.

[0115] An improved Newton-GMRES module is applied to the linear approximation system. The improved Newton-GMRES algorithm is used to iteratively solve the linear approximation system in the Krylov subspace to obtain the optimal correction amount for the adaptive non-uniform B-spline control point set.

[0116] In this embodiment, the Newton-GMRES module is improved, including:

[0117] The initial solution of the improved Newton-GMRES iteration is set to a vector of all zeros. The initial residual vector is calculated to be equal to the right-hand side of the linear approximation system. The second norm of the initial residual is calculated, and the initial residual vector is normalized using this norm to obtain the first basis vector of the Krylov subspace.

[0118] In the process of constructing the Krylov subspace, for any current basis vector, the product of the Hessian matrix and the basis vector at the current set of adaptive non-uniform B-spline control points is approximately calculated using the finite difference method, resulting in a matrix-vector product.

[0119] ;

[0120] in, This indicates that in the Newton-GMRES algorithm, the... In the nth iteration, the Krylov subspace is... The resulting vector after performing a matrix-vector product on each basis vector. Indicates the first In the next iteration, the adaptive non-uniform B-spline control point set The Hessian matrix at point A is a matrix composed of the second-order partial derivatives of the finite-time-domain nonlinear optimal control performance index function of the elevator with respect to the set of control points. Indicates the first In the Newton-GMRES iteration, during the Krylov subspace construction process... orthogonal basis vectors This represents the finite difference perturbation step size, used to approximate the product of the Hessian matrix and the basis vectors, with a range of values. to .

[0121] The matrix-vector product obtained by the finite difference method is orthogonalized according to the Krylov orthogonalization process, and the orthogonal basis vector set of the Krylov subspace is generated sequentially. The orthogonal basis vector set of the Krylov subspace is used to construct the orthogonal basis matrix of the Krylov subspace and the corresponding upper Hessenberg matrix. The least squares problem is solved in the subspace to approximately satisfy the linear approximation system and obtain the optimal correction amount of the adaptive non-uniform B-spline control point set.

[0122] In Example 1, the matrix-vector product obtained by the finite difference method is orthogonalized according to the Krylov orthogonalization process, so that each newly generated basis vector is orthogonal to the existing Krylov subspace basis vectors, and these are accumulated sequentially to form the set of orthogonal basis vectors of the Krylov subspace. All orthogonal basis vectors are arranged in the order of generation to form the Krylov subspace orthogonal basis matrix. During the orthogonalization process, the projection coefficients of each step are recorded simultaneously, and the corresponding upper Hessenberg matrix is ​​constructed. Based on the Krylov subspace orthogonal basis matrix and the upper Hessenberg matrix, the linear approximation system in the Krylov subspace is transformed into a least squares problem with a dimension much smaller than that of the original system. By solving for the weight coefficient vector with the smallest residual norm in the least squares sense, the weight coefficient vector of the optimal solution in the Krylov subspace is obtained. The optimal correction amount of the adaptive non-uniform B-spline control point set is obtained by multiplying the Krylov subspace orthogonal basis matrix with the weight coefficient vector of the optimal solution.

[0123] Calculate the current residual vector during the Newton-GMRES iteration process, determine whether the L2 norm of the residual vector is less than the currently set GMRES convergence threshold, and stop the Krylov subspace expansion when the residual vector meets the convergence condition.

[0124] The current residual vector is the difference between the right-hand side of the linear approximation system and the coefficient matrix multiplied by the correction amount of the current adaptive non-uniform B-spline control point set. The dimension of the GMRES convergence threshold is the elevator finite-time nonlinear optimal control performance index function unit divided by meters.

[0125] The update module adds the optimal correction to the current set of adaptive non-uniform B-spline control points to generate the updated adaptive non-uniform B-spline trajectory.

[0126] In this embodiment, the update module includes:

[0127] The optimal correction amount of the adaptive non-uniform B-spline control point set obtained by solving the Krylov subspace in the Newton-GMRES algorithm is added point by point to the current adaptive non-uniform B-spline control point set to generate the updated adaptive non-uniform B-spline control point set.

[0128] The optimal correction is a column vector with the same dimension as the current set of control points, representing the correction increment for each control point in the current optimization cycle. The current set of adaptive non-uniform B-spline control points is added to the optimal correction according to their corresponding components to obtain a new set of adaptive non-uniform B-spline control points.

[0129] Based on the updated set of adaptive non-uniform B-spline control points and the given adaptive non-uniform node vector, the updated adaptive non-uniform B-spline trajectory is generated using the B-spline curve reconstruction method.

[0130] According to the definition of B-spline basis functions, all updated adaptive non-uniform B-spline control points are weighted and summed with their corresponding B-spline basis functions to obtain the updated adaptive non-uniform B-spline trajectory function. Based on the updated adaptive non-uniform B-spline trajectory function, the position, velocity, acceleration, and jerk characteristics of the trajectory in the prediction time domain are recalculated. The updated adaptive non-uniform B-spline trajectory function is used as the trajectory position function. The trajectory position function is differentiated by the first derivative with respect to the time parameter to obtain the trajectory velocity function. The trajectory velocity function is further differentiated by the first derivative with respect to the time parameter to obtain the trajectory acceleration function. The trajectory acceleration function is further differentiated by the first derivative with respect to the time parameter to obtain the trajectory jerk function.

[0131] The motor drive signal generation module calculates the target electromagnetic torque command of the permanent magnet synchronous motor based on the updated adaptive non-uniform B-spline trajectory, and converts it into a space vector pulse width modulation motor drive signal to drive the elevator to run along the updated adaptive non-uniform B-spline trajectory.

[0132] In this embodiment, the motor drive signal generation module includes:

[0133] Based on the updated adaptive non-uniform B-spline trajectory function, the target acceleration characteristics required for elevator operation are calculated, and the target thrust of the permanent magnet synchronous motor is determined.

[0134] In Example 1, the trajectory position function is differentiated by the first order with respect to the time parameter to obtain the trajectory velocity function. Then, the trajectory velocity function is differentiated by the first order to obtain the trajectory acceleration function. The product of the elevator's equivalent mass and the trajectory acceleration function, plus the product of the traction wire rope's equivalent damping coefficient and the trajectory velocity function, the product of the wire rope's time-varying stiffness and the trajectory position function, the nonlinear friction term, and the gravity term, are added together to obtain the target thrust of the permanent magnet synchronous motor at the current moment.

[0135] Multiply the target thrust of the permanent magnet synchronous motor by the radius of the traction wheel to obtain the target electromagnetic torque command of the permanent magnet synchronous motor;

[0136] The target electromagnetic torque command of a permanent magnet synchronous motor describes the electromagnetic torque output required by the motor at the current moment to achieve the trajectory control target.

[0137] Divide the target electromagnetic torque command of the permanent magnet synchronous motor by the product of the number of pole pairs of the motor and the excitation flux of the permanent magnet, and multiply by a scaling factor to obtain the target q-axis current command of the permanent magnet synchronous motor. Set the d-axis current command to zero to obtain the target d-axis and q-axis current commands of the permanent magnet synchronous motor.

[0138] Based on the target d-axis and q-axis current commands of the permanent magnet synchronous motor, the target d-axis voltage command and the target q-axis voltage command are obtained;

[0139] In Example 1, by combining the motor stator resistance, motor d-axis inductance, motor q-axis inductance, permanent magnet excitation flux linkage, and motor electric angular velocity, and substituting them into the permanent magnet synchronous motor stator voltage equation, the target d-axis voltage command and the target q-axis voltage command are obtained. The target d-axis voltage command and the target q-axis voltage command are used to describe the magnitude of the voltage required to control the motor stator winding.

[0140] The target d-axis voltage command and the target q-axis voltage command are sequentially subjected to inverse Park transformation and inverse Clarke transformation to transform the voltage command in the two-phase stationary coordinate system into the three-phase stator target voltage;

[0141] The target voltages for the three-phase stator are target phase a voltage, target phase b voltage, and target phase c voltage, respectively.

[0142] Based on the three-phase target voltage, the corresponding permanent magnet synchronous motor drive signal is generated using the space vector pulse width modulation method.

[0143] In Example 1, based on the target phase a voltage, target phase b voltage, and target phase c voltage, the three-phase target voltages are combined into an equivalent space vector using the space vector pulse width modulation method. By calculating the projection relationship between the space vector and the six main switch vectors of the inverter in each sector, the two sets of main switch vectors and the zero vector to be activated in the current control cycle are determined. The duty time of the main switch vector and the zero vector are calculated according to the amplitude and direction of the target space vector. The duty time of each main switch vector and the zero vector is normalized to the duty cycle in the switching cycle. The obtained switching duty cycle sequence of each bridge arm is used as the drive signal of the permanent magnet synchronous motor and output to the inverter. This realizes the application of voltage pulses consistent with the target three-phase voltage to the three-phase windings of the permanent magnet synchronous motor, thereby driving the permanent magnet synchronous motor to generate precise electromagnetic torque according to the updated adaptive non-uniform B-spline trajectory.

[0144] The three-phase target voltage space vector synthesis process refers to transforming the voltage quantity in the three-phase stationary coordinate system into a space vector in the rotating coordinate system. The duty time of the main switch vector and the zero vector is related to the distribution of the space vector in different sectors and needs to be calculated based on the real-time target space vector. The final generated drive signal is the pulse width modulation waveform sequence of each bridge arm switch of the inverter.

[0145] The motor drive signal is output to the elevator drive system, which drives the elevator traction mechanism to run according to the updated adaptive non-uniform B-spline trajectory.

[0146] The loop module repeatedly executes the steps from the index function construction module to the motor drive signal generation module until the elevator car reaches the target floor corresponding to the elevator target floor data and stops running, thus completing the elevator energy consumption control.

[0147] Example 2: On a multi-story elevator test prototype platform in a residential area, the research team conducted a 20-day system comparison experiment on the traditional S-curve trajectory planning method and the trajectory smoothing planning method based on the Newton-GMRES algorithm and adaptive non-uniform B-spline curve proposed in this invention.

[0148] The scenario was set as follows: the elevator's rated load was 1000kg, the maximum operating speed was 2.5 m / s, the traction sheave radius was 0.32m, the empty car mass was 850kg, the steel wire rope's elastic modulus was 205GPa, the cross-sectional area was 3.14cm², and the reference rope length was 31m. The experiment used high-precision weighing sensors and position encoders, and synchronously collected the motor's three-phase current and the car's acceleration signals. To simulate real high-frequency passenger entry and exit and load changes, the team set up 60 different load conditions, including empty, half-load, full-load, and sudden load changes (passengers suddenly jumping during operation).

[0149] In one experiment, the elevator started at floor 1 and targeted floor 18. The initial load was 250 kg. When the elevator reached the 10th floor, the system detected a sudden load change (an additional 350 kg was added). In the later part of the operation, the load was unloaded again (200 kg was removed). The system needed to smoothly complete the entire trajectory without interrupting control or affecting passenger comfort.

[0150] The traditional method uses a five-segment S-shaped speed curve for planning, and assigns values ​​by looking up values ​​from a preset form. It cannot dynamically correct the trajectory. The acceleration during operation is preset to 0.8 m / s², and the jerk is set to 1.5 m / s³. Passengers reported slight shaking in some areas, the leveling error of the terminal floor was 2.8 cm, the running time was 34.7 seconds, the energy consumption meter measured the total power consumption to be 2.13 kWh, and the maximum three-phase current peak was 32.1 A.

[0151] Under the same operating conditions, the method of this invention collects the current state data of the car in real time through a weighing sensor, a position encoder, and a current sensor, and the system automatically constructs the following real-time state vector:

[0152] The starting position is 0.00m, the starting speed is 0.00m / s, the real-time load is 250kg, and the target layer distance is 52.8m.

[0153] The wire rope stiffness is updated in real time using the formula k=EA / (L0+x(t)), with an initial stiffness of 20.77kN / m.

[0154] After entering the acceleration phase, the Newton-GMRES algorithm optimizes the B-spline control points in real time. At startup, the density of trajectory nodes is increased to ensure that there are no sudden changes in the acceleration curve at the beginning of the trajectory. The actual maximum acceleration is 0.75 m / s², and the maximum jerk is 1.19 m / s³, which is lower than that of traditional methods.

[0155] When the load changes abruptly at the 10th layer, the state vector is updated in real time, and the B-spline control points are locally adaptively optimized. The nodes are adjusted only during the period when the load changes, resulting in a smooth and continuous trajectory with no fluctuations. The maximum jump value of the trajectory acceleration curve is only 0.26 m / s².

[0156] The system continuously calculates the target thrust and target current command. The maximum three-phase current peak value is 27.6A. The energy consumption meter measured the total power consumption as 1.83 kWh, which is 14% more energy-efficient than the traditional method.

[0157] Throughout the entire operation, the Newton-GMRES optimization took an average of 0.94ms per cycle (maximum 1.13ms), which fully meets the controller's 1ms control cycle requirement.

[0158] The leveling error at the terminal level was 0.9cm, which is lower than that of traditional methods. Passengers' subjective ratings showed no obvious discomfort throughout the journey and no significant shaking at the abrupt change point.

[0159] In another set of extreme operating condition simulations, the experiment simulated a car being maliciously jolted mid-journey, causing the load to suddenly increase by 150 kg within 0.1 seconds. Traditional methods exhibit overshoot and minor oscillations after this sudden change, increasing the leveling error to 4.7 cm, energy consumption to 2.36 kWh, and the motor current peak to 34.9 A.

[0160] The method of this invention still maintains the continuity of the global trajectory, with the acceleration at the abrupt change point fluctuating by only 0.12 m / s², the leveling error of the terminal layer controlled within 1.1 cm, the energy consumption of 1.95 kWh, the maximum three-phase current of 28.3 A, the system dynamic response recovery time of less than 0.2 seconds, and the subjective score shows no obvious discomfort.

[0161] Table 1. Summary of Experimental Data (Partial Scenarios):

[0162]

[0163] All comparative experiments show that the trajectory obtained by the method of this invention can automatically correct the jerk and smoothly compensate for the changes in flexible modes during each load change, without numerical divergence or computational instability. The B-spline node density distribution automatically adapts to load changes and key dynamic constraints, realizing full-process physical modeling of the wire rope stiffness, damping, and friction models. The improved Newton-GMRES solution does not require large-scale matrix inversion, the calculation cycle is stable within 1ms, and it is stable and reliable in long-term operation.

[0164] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-objective optimization based elevator energy consumption control system, characterized in that, The method comprises the following steps: A parameterization module collects real-time sensor information of elevator variable load operation to construct an elevator real-time operation state vector, initializes an adaptive non-uniform B-spline trajectory parameterization model, generates an adaptive non-uniform B-spline control point set and a corresponding non-uniform node vector, and constructs an elevator discrete nonlinear dynamics model. An index function construction module constructs an elevator finite time domain nonlinear optimal control performance index function, and combines the elevator discrete nonlinear dynamics model to convert the elevator nonlinear optimal control problem into an elevator nonlinear parameter optimization problem with the adaptive non-uniform B-spline control point set as the decision variable. An iteration module performs Newton linearization iteration on the elevator nonlinear parameter optimization problem to construct a linear approximation system. An improved Newton-GMRES module applies the improved Newton-GMRES algorithm to the Krylov subspace to iteratively solve the linear approximation system on the linear approximation system to obtain the optimal correction of the adaptive non-uniform B-spline control point set. An update module adds the optimal correction to the current adaptive non-uniform B-spline control point set to generate an updated adaptive non-uniform B-spline trajectory. A motor drive signal generation module calculates a permanent magnet synchronous motor target electromagnetic torque instruction according to the updated adaptive non-uniform B-spline trajectory, and converts it into a space vector pulse width modulation motor drive signal to drive the elevator to run along the updated adaptive non-uniform B-spline trajectory. A cycle module repeatedly executes the steps of the index function construction module to the motor drive signal generation module until the elevator car reaches the target floor corresponding to the elevator target floor data and stops running, and completes the elevator energy consumption control. The parameterization module comprises:

2. The elevator energy consumption control system based on multi-objective optimization according to claim 1, characterized in that, Collecting real-time sensor information of elevator variable load operation; Constructing an elevator real-time operation state vector based on the real-time sensor information of elevator variable load operation; Based on the elevator real-time operation state vector, initializing the adaptive non-uniform B-spline trajectory parameterization model, establishing the B-spline expression of the elevator trajectory function; Constructing an initial adaptive non-uniform B-spline control point set; Setting an adaptive non-uniform node vector, each node in the adaptive non-uniform node vector is a node parameter value; Combining the initial adaptive non-uniform B-spline control point set and the adaptive non-uniform node vector to form an initial adaptive non-uniform B-spline trajectory parameterization model. The elevator discrete nonlinear dynamics module comprises:

3. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, Combining the elevator car mass parameters, elevator hoisting steel wire rope elastic parameter set, elevator hoisting steel wire rope equivalent damping coefficient parameters, permanent magnet synchronous motor electromagnetic torque equation parameter set and elevator real-time operation state vector to construct an elevator discrete nonlinear dynamics model; According to the steel wire rope cross-sectional area and the steel wire rope elastic modulus, the steel wire rope time-varying stiffness varying with displacement is calculated; ​ The longitudinal dynamics equilibrium equation of the multi-mass-spring-damper system is linearly superimposed by the product of the equivalent mass and the acceleration of the car, the product of the equivalent damping coefficient of the traction steel wire rope and the velocity of the car, and the product of the time-varying stiffness of the steel wire rope and the displacement of the car, and the linear superimposition result is equal to the difference between the thrust of the permanent magnet synchronous motor and the gravity term and the nonlinear friction term; According to the three-phase current data of the permanent magnet synchronous motor and the parameter set of the electromagnetic torque equation of the permanent magnet synchronous motor, the electromagnetic torque of the motor is calculated, and the electromagnetic torque of the motor is divided by the radius of the traction sheave to obtain the thrust of the permanent magnet synchronous motor; The nonlinear friction term is constructed by the coulomb friction and the static friction peak value.

4. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, The index function construction module comprises: Based on the discrete nonlinear dynamics model of the elevator and the adaptive non-uniform B-spline trajectory parameterization model, the elevator finite time domain nonlinear optimal control performance index function is determined; The elevator energy consumption performance index is obtained by integrating the heat loss of the motor stator resistance in the prediction time domain; The elevator ride comfort performance index is obtained by integrating the square of the car acceleration in the prediction time domain; The elevator displacement end error performance index is obtained by weighting the square of the deviation between the actual displacement of the elevator at the end of the prediction time domain and the target displacement, and the elevator speed end error performance index is obtained by weighting the square of the deviation between the actual speed of the elevator at the end of the prediction time domain and zero speed; The elevator energy consumption performance index, the elevator ride comfort performance index, the elevator displacement end error performance index and the elevator speed end error performance index are weighted and summed to form the elevator finite time domain nonlinear optimal control performance index function; The minimization of the elevator finite time domain nonlinear optimal control performance index function is used to obtain the optimal control trajectory with balanced energy consumption and comfort and accurate end state.

5. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, The iteration module comprises: The elevator finite time domain nonlinear optimal control performance index function is expressed as a multivariable nonlinear function form with the adaptive non-uniform B-spline control point set as the input variable; The KKT optimality condition of the elevator nonlinear parameter optimization problem is established to minimize the elevator finite time domain nonlinear optimal control performance index function, and the first order necessary condition of the KKT optimality condition is that the performance index function gradient vector composed of all partial derivatives of the elevator finite time domain nonlinear optimal control performance index function with respect to the adaptive non-uniform B-spline control point set is equal to zero; The Newton linearization operation is performed at the current adaptive non-uniform B-spline control point set to construct the linear approximation system of the elevator nonlinear parameter optimization problem; The Newton residual vector is calculated for the linear approximation system.

6. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, The improved Newton-GMRES module comprises: The initial solution of the improved Newton-GMRES iteration is set to be a zero vector, the initial residual vector is calculated to be equal to the right end item of the linear approximation system, the two norm of the initial residual is calculated, and the initial residual vector is normalized by the norm to obtain the first basis vector of the Krylov subspace; In the process of constructing the Krylov subspace, the finite difference method is used to approximate the product of the Hessian matrix at the current adaptive non-uniform B-spline control point set and the basis vector for any current basis vector, to obtain a matrix-vector product; The matrix-vector product obtained by the finite difference method is orthogonalized according to the Krylov orthogonalization process, and the orthogonal basis vector set of the Krylov subspace is generated in turn, which is used to construct the Krylov subspace orthogonal basis matrix and the corresponding upper Hessenberg matrix, and the least squares problem is solved in the subspace to approximately satisfy the linear approximation system, to obtain the optimal correction of the adaptive non-uniform B-spline control point set; The current residual vector in the Newton-GMRES iteration process is calculated, and it is judged whether the two-norm of the residual vector is less than the current set GMRES convergence threshold, and when the residual vector meets the convergence condition, the Krylov subspace expansion is stopped.

7. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, The update module comprises: The optimal correction of the adaptive non-uniform B-spline control point set obtained by solving the Krylov subspace in the Newton-GMRES algorithm is added to the current adaptive non-uniform B-spline control point set point by point to generate an updated adaptive non-uniform B-spline control point set; Based on the updated adaptive non-uniform B-spline control point set and the established adaptive non-uniform node vector, the B-spline curve reconstruction method is used to calculate and generate an updated adaptive non-uniform B-spline trajectory.

8. The multi-objective optimization based elevator energy consumption control system according to claim 1, wherein, The motor drive signal generation module comprises: According to the updated adaptive non-uniform B-spline trajectory function, the target acceleration characteristic required for elevator operation is calculated, and the permanent magnet synchronous motor target thrust is determined; The permanent magnet synchronous motor target thrust is multiplied by the traction sheave radius to obtain the permanent magnet synchronous motor target electromagnetic torque instruction; The permanent magnet synchronous motor target electromagnetic torque instruction is divided by the product of the motor pole pair number and the permanent magnet excitation flux linkage and multiplied by a proportional coefficient to obtain the permanent magnet synchronous motor target q-axis current instruction, and the d-axis current instruction is set to zero to obtain the permanent magnet synchronous motor target d-axis and q-axis current instructions; According to the permanent magnet synchronous motor target d-axis and q-axis current instructions, the target d-axis voltage instruction and the target q-axis voltage instruction are obtained; The target d-axis voltage instruction and the target q-axis voltage instruction are sequentially subjected to inverse Park transformation and inverse Clarke transformation to transform the two-phase static coordinate system voltage instruction into a three-phase stator target voltage; Based on the three-phase target voltage, the corresponding permanent magnet synchronous motor drive signal is generated by using the space vector pulse width modulation method; The motor drive signal is output to the elevator drive system to drive the elevator traction mechanism to run according to the updated adaptive non-uniform B-spline trajectory.

Images