Compact shelving power supply control system

Abstract

The invention relates to the technical field of automatic control, and discloses a compact shelving power supply control system comprising a power supply assembly; each compact shelf assembly comprises a driving module and at least one state sensor; and the central control assembly is configured to respond to a target instruction representing the target state of the compact shelving and execute the following operations: S1, updating a discrete time state space model representing the unified behavior of all compact shelving assemblies in real time according to actual state information fed back by the state sensor; and S2, based on the updated discrete time state space model and the representation compact shelving target state. An online identification mechanism is introduced, a system model can be automatically updated according to real-time data, and changes of physical characteristics such as loads and friction are automatically compensated. According to the design, high-precision and high-stability control can be kept for a long time without manual intervention, and the environmental adaptability and robustness of the system are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of automatic control technology, in particular to a compact shelving power supply control system. Background Art

[0002] Currently, compact shelving systems, a core component of modern archive management and book storage, significantly increase storage capacity per unit space by arranging shelves densely and moving them on fixed tracks. To facilitate user access, modern compact shelving systems generally utilize motor-driven and automated controls to automatically open and close the shelves. Typically, these control systems include a central controller, drive modules, and sensors for position and obstacle detection, managing the movement of the compact shelving using pre-set control logic.

[0003] However, in practical applications, existing control methods often expose their inherent limitations when faced with the dynamically changing physical characteristics of compact shelving systems. Most of these control systems rely on a set of control parameters that are fixed during the installation and commissioning phase, such as fixed acceleration curves, motor output power, or PID (proportional-integral-differential) gains. However, the physical properties of compact shelving systems are not static. Their total mass changes frequently as archival materials are stored and retrieved, and the friction coefficients of the rails and transmission mechanisms drift due to long-term use, wear, and changes in ambient temperature and humidity. Control strategies using fixed parameters are unable to proactively adapt to these changes, often leading to a gradual deterioration of control effectiveness, which may manifest as jitter and impact during operation, reduced positioning accuracy, or the use of excessive safety margins to ensure versatility, sacrificing operational efficiency and energy efficiency.

[0004] In addition, existing technologies also have shortcomings in their strategies for ensuring operational safety and improving operational efficiency. Traditional safety assurance mechanisms are usually passively triggered, for example, emergency braking after detecting an obstacle through infrared radiation or collision sensors. Although this reactive control logic can avoid serious collisions, its "after-the-fact" nature cannot achieve forward-looking risk avoidance, and frequent emergency stops and starts not only affect the user experience, but also have an impact on the mechanical structure and stored items. In the scenario of coordinated movement of multiple columns of compact shelving, how to plan a path that can ensure that a safe distance is always maintained between the shelves, and make the overall movement process the smoothest, most time-consuming, and least energy-consuming, is a technical problem that has not been fully solved in existing technologies.

[0005] For the coordinated control of the entire compact shelving array, existing technologies often use relatively simple independent control or preset timing logic. This approach treats each column of compact shelving as an independent control object, ignoring the dynamic coupling and mutual influence that may occur when they move on the track. Therefore, when complex operations such as multi-frame linkage or sequential movement are required, it is difficult to achieve truly smooth and coordinated group movement, and the overall intelligence and automation level of the system are therefore limited. Therefore, the industry urgently needs an advanced control method that can actively adapt to system changes, proactively plan safe and efficient paths, and achieve deep coordination of multiple frames. Summary of the Invention

[0006] The purpose of the present invention is to provide a compact shelving power supply control system to solve the technical problems in existing compact shelving control technology, which cannot adapt to dynamic changes of the system due to fixed control parameters and is difficult to balance forward-looking safety planning and overall operating efficiency.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a compact shelving power supply control system, comprising: Power supply components; A plurality of compact shelving components, each of which comprises a drive module and at least one status sensor; and a central control component configured to, in response to a target instruction representing a target state of the compact shelving system, perform the following operations: S1, updating a discrete-time state-space model representing a unified behavior of all compact shelving components in real time according to actual state information fed back by the state sensor; S2. Based on the updated discrete-time state-space model and the target state of the compact shelving system, predict and generate an optimal control sequence by solving a constrained optimization problem; S3. Apply the optimal control sequence to drive the compact shelving assembly to move toward the target state representing the compact shelving.

[0008] Preferably, the discrete-time state-space model is represented by the following formula: x(k+1)=A(k)x(k)+B(k)u(k); Where x(k) represents the global state vector; u(k) is the global control input vector; A(k) and B(k) are time-varying system matrices.

[0009] Preferably, the central control component updates the time-varying system matrices A(k) and B(k) by adopting a recursive update rule represented by the following formula: in, represents the estimation vector containing the matrix parameters; K(k) is the gain matrix; e(k) is the prediction error vector.

[0010] Preferably, the constrained optimization problem solved by the central control component aims to minimize a cost function J, which is represented by the following formula: Among them, x pred represents the predicted state vector; x ref is the reference state vector; x pred -x ref is the tracking error; Q is the state weight matrix; u is the control input vector; R is the input weight matrix.

[0011] Preferably, the constraint condition of the constrained optimization problem includes a minimum safety distance constraint represented by the following formula: |p j -p j+1 |≥d min ; Among them, p j and p j+1 Represents the predicted position of any adjacent compact shelving components; d min A preset safety distance.

[0012] Preferably, at least one of the state sensors includes a displacement sensor and a force sensor; and the constraint condition further includes an external force limit set based on a signal detected by the force sensor.

[0013] Preferably, the central control component adopts a rolling time domain strategy when applying the optimal control sequence, that is, only the first control vector of the sequence is applied in each control cycle, and the update of the discrete-time state space model and the generation of the optimal control sequence are re-executed in the next cycle.

[0014] Preferably, the global state vector x(k) includes position information and speed information of all compact shelving components detected by the state sensor.

[0015] Preferably, the central control component is further configured to automatically use the target position of the next preset compact shelving component as the new target instruction after a compact shelving component moves to its target state to trigger a new round of control operations.

[0016] The present invention also provides a compact shelving power supply control method, comprising the following steps: 1) In response to target instructions, the discrete-time state-space model that characterizes the unified behavior of all compact shelving components is updated in real time; 2) predicting and generating an optimal control sequence based on the updated discrete-time state-space model and the target state corresponding to the target instruction by solving an optimization problem aimed at minimizing a multi-objective cost function and satisfying a minimum safety distance constraint; 3) Applying the first control vector of the optimal control sequence in a rolling time domain manner to drive the compact shelving assembly, and returning to step 1) in the next control cycle.

[0017] In summary, the present invention includes at least one of the following beneficial technical effects: 1. This invention significantly enhances the environmental adaptability and long-term robustness of the compact shelving control system by introducing an online system identification mechanism. Unlike traditional control methods that rely on fixed parameters, this invention continuously and automatically updates its internal dynamic system model based on actual status information fed back by sensors in real time. This inventive concept enables the control system to autonomously learn and compensate for changes in the system's physical characteristics caused by factors such as file load changes, track friction and wear, and temperature effects, thereby maintaining high-precision control performance without the need for manual intervention or recalibration, ensuring the long-term stability and reliability of the system.

[0018] 2. The present invention has achieved significant optimization between safety and operational efficiency by adopting a control strategy based on model prediction. The core of this method is that before performing any physical movement, the system will make forward-looking plans for the movement trajectories of all compact shelving units in the future by solving a constrained optimization problem. In this planning process, key safety constraints such as the minimum safe spacing between adjacent shelves are strictly met as hard boundary conditions, fundamentally eliminating the risk of collision. At the same time, by optimizing a cost function aimed at minimizing trajectory errors and controlling energy consumption, it is ensured that the generated movement path is not only absolutely safe, but also achieves a better level in terms of operational smoothness and energy economy.

[0019] 3. The present invention achieves coordination and smoothness in the overall operation of the system by treating the entire compact shelving array as a unified, multivariable coupled system for collaborative control, and combining it with a rolling time domain strategy for closed-loop feedback. This method does not control each compact shelving in isolation, but rather centrally plans the output of all drive modules, ensuring the inherent connection and harmony of the movement of each unit. The rolling time domain execution method ensures that the control instructions are re-optimized based on the latest system status at each time step, so that various unforeseen disturbances can be quickly responded to and suppressed. This system-level collaborative planning combined with high-frequency feedback correction allows the compact shelving to move coherently and respond quickly when performing complex sequential opening or dense gathering operations, significantly improving the overall performance of the system and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is one of the flow charts of the method of the present invention; Figure 2 This is the second flow chart of the method of the present invention. DETAILED DESCRIPTION

[0021] The following is combined with Figure 1 -Attached Figure 2 , the present invention is described in further detail.

[0022] The present invention provides a compact shelving power supply control system, comprising: Power supply components; A plurality of compact shelving components, each of which comprises a drive module and at least one status sensor; and a central control component configured to, in response to a target instruction representing a target state of the compact shelving system, perform the following operations: S1, updating a discrete-time state-space model representing a unified behavior of all compact shelving components in real time according to actual state information fed back by the state sensor; S2. Based on the updated discrete-time state-space model and the target state of the compact shelving system, predict and generate an optimal control sequence by solving a constrained optimization problem; S3. Apply the optimal control sequence to drive the compact shelving assembly to move toward the target state representing the compact shelving.

[0023] S1: Real-time update of dynamic system models In this embodiment, step S1 is described in detail as follows. Its core purpose is to establish and continuously maintain a mathematical model that can accurately characterize the dynamic characteristics of the entire compact shelving array, thereby providing a high-fidelity prediction basis for the subsequent predictive control step S2.

[0024] To achieve coordinated control of multiple rows of compact shelving systems, this paper first abstracts the physically separate N rows of compact shelving systems into a unified, multivariable, coupled, multiple-input, multiple-output (MIMO) system at the control level. This systematic modeling approach is the fundamental prerequisite for achieving overall optimal control, enabling the controller to comprehensively consider the interactions among all compact shelving systems, rather than controlling them in isolation.

[0025] Specifically, the central control component builds a discrete-time state-space model to describe the dynamic behavior of the MIMO system. This model framework is the basis for all subsequent operations.

[0026] First, define a global state vector x(k) that can fully describe the motion state of the system at any discrete moment k. This vector is preferably composed of the position and velocity information of all compact shelving components, and its dimension is 2N: x(k)=[p1(k),v1(k),p2(k),v2(k),…,p N (k),v N (k)] T ; Among them, p N (k) and v N (k) are the position and velocity of the Nth row of compact shelving at time k; T is the transposed sign. These state quantities are measured in real time or calculated by differential calculation using state sensors installed on each compact shelving component, preferably high-precision displacement sensors (such as photoelectric encoders).

[0027] Accordingly, a global control input vector u(k) is defined, which represents the control signal issued by the central control component at time k and applied to each column of the compact rack drive module (such as a DC motor). Its dimension is N: u(k) = [u1(k),u2(k),…,u N (k)] T ; Where k represents the discrete time index; N is the total number of compact shelving components; u N (k) is the control input applied to the Nth column of the compact rack; T is the transposition symbol.

[0028] Based on the above definition, the dynamic evolution of the system can be described by the following linear time-varying state space equation: x(k+1)=A(k)x(k)+B(k)u(k); A(k) is the time-varying state transition matrix, whose internal elements reflect the physical properties of the system itself, such as the inertia and damping of each compact rack, as well as the coupling effects between racks. B(k) is the time-varying input matrix, whose elements reflect the efficiency of the control input signal in driving the system state. Designing these two matrices to be time-varying (i.e., varying with time step k) is a key technical feature of this invention, enabling the model to capture and adapt to changes in the system's physical characteristics.

[0029] In practical applications, the physical properties of compact shelving systems are not static. For example, accessing files significantly changes the load mass and inertia, and long-term operation can cause changes in the rail friction coefficient. These factors can cause a fixed, offline model to deviate from the actual system conditions, a phenomenon known as model mismatch. To address this technical issue, the present invention introduces an online system identification mechanism.

[0030] In this embodiment, the central control component embeds an online identification module. The core function of this module is to update and correct the matrices A(k) and B(k) in the state-space model using the latest measured data during each control cycle. Preferably, this module uses the Recursive Least Squares with Forgetting Factor (FF-RLS) algorithm because of its advantages of moderate computational complexity, fast convergence, and effective tracking of time-varying parameters.

[0031] The execution process of the FF-RLS algorithm is as follows: In each control cycle k, the module first obtains the current actual state measurement value x(k) from the sensor, and calls the actual state x(k-1) at the previous moment k-1 and the control input u(k-1) applied at that time from the memory.

[0032] Then, the algorithm uses the following recursive formula to calculate the parameter vector containing all the parameters to be identified in matrices A and B: To update: Calculate the prediction error (Prediction Error) e(k): This error represents the deviation between the state predicted based on the old model and the actual measured state.

[0033] Among them, x(k) represents the actual state vector; is the one-step predicted state vector.

[0034] Calculate the gain matrix (Gain Matrix) K(k): The gain matrix determines the weight of the prediction error on the parameter correction.

[0035] Among them, P(k-1) is the parameter error covariance matrix; φ(k-1) is the regression vector; λ is the forgetting factor.

[0036] Parameter Estimation Update This is the core step of the algorithm, which uses prediction error and gain to correct parameters.

[0037] in, represents the parameter matrix before updating; K(k) is the gain matrix; e T (k) is the transposed prediction error vector.

[0038] Update the covariance matrix (Covariance Matrix Update) P(k): prepare for the next round of gain calculation.

[0039] Here, P(k) is the covariance matrix of the parameters; I is the identity matrix; and λ is a key forgetting factor, preferably ranging from (0.9 to 1]. The introduction of the forgetting factor enables the algorithm to give higher weight to newly collected data when updating parameters, while gradually "forgetting" older historical data. This mechanism is crucial for tracking time-varying systems, ensuring that the model can promptly respond to sudden changes in system characteristics caused by load changes, for example.

[0040] In this embodiment, force sensors are preferably installed on each compact shelving component. The external disturbance force information detected by the force sensor can be integrated into the online identification process, for example, as a known disturbance input into the system model, or used to identify an additional disturbance model. This can further improve the fidelity of the dynamic system model, allowing it to not only describe the system's own motion but also reflect the effects of the external environment on the system.

[0041] By executing the above steps, at the beginning of each control cycle, the central control component can obtain a set of model matrices that best represent the real physical characteristics of the current system. and This real-time updated, high-precision dynamic system model lays a solid and reliable foundation for accurate trajectory prediction and optimal control planning in the subsequent step S2.

[0042] S2: Predictive Generation of Optimal Control Sequences In this embodiment, step S2 is described in detail below. Its core purpose is to utilize the high-fidelity dynamic system model updated in real time in step S1 to proactively plan an optimal path that can safely and efficiently guide the entire compact shelving array from its current state to its target state. This step is a key manifestation of the invention's intelligence, transforming traditional passive feedback control into an active, predictive optimization plan.

[0043] When the central control component receives the target instruction from the outside (such as the user interface or the upper management system), the instruction is first parsed into an accurate, mathematical target state vector x ref This vector clearly defines the desired position and desired speed (usually zero) that each column of the compact rack should reach when the task is completed. ref It will serve as the final guide for the optimization problem in this step.

[0044] The core of this step is that the predictive control module within the central control component solves a finite-time constrained optimization problem. This method is often called Model Predictive Control (MPC) in control theory. Its basic idea is to use the current system state x(k) and the latest model obtained in step S1 at each control time k to calculate the optimal solution. and The system behavior is predicted in the future (i.e., the prediction horizon), and a series of optimal control inputs (i.e., the control horizon) are calculated through optimization algorithms to achieve the preset control objectives.

[0045] Specifically, the constrained optimization problem consists of an objective function to be minimized and a series of constraints that must be satisfied.

[0046] In this embodiment, the objective function (or cost function J) is preferably designed as a quadratic function, which aims to comprehensively balance the trajectory tracking accuracy and control energy consumption of the system. A typical mathematical expression of the cost function J is as follows: Among them, x pred represents the predicted state vector; x ref is the reference state vector; x pred -x ref is the tracking error; Q is the state weight matrix; u is the control input vector; R is the input weight matrix.

[0047] By adjusting the relative sizes of the weight matrices Q and R, those skilled in the art can flexibly balance the rapid response performance of the system with operational stability and economy.

[0048] In the process of minimizing the above cost function, the solution process must strictly comply with a series of constraints. These constraints are the fundamental guarantee for ensuring the physical feasibility and safety of the system operation. In this embodiment, these constraints include at least: System dynamic constraints: This is the basis for the connection between the optimization problem and the physical world. All predictions of future states x pred All must strictly follow the dynamic system model updated in real time by step S1.

[0049] Among them, x pred (k+i+1|k) represents the predicted value of the state at time k for the next time k+i+1; k represents the current real time step; i is a counter in the prediction horizon, starting at i=0; k+i+1 represents the future time point whose state we want to predict.

[0050] Control input constraints: Considering the limitations of physical hardware, the control input u(k+i|k) applied to each driver module must be limited to its executable range.

[0051] u min ≤u j (k+i|k)≤u max ; Among them, u min and u max Respectively represent the minimum and maximum output capabilities of the driver module.

[0052] State constraint: This is the core constraint to ensure the safe operation of the system.

[0053] Minimum safety distance constraint: In order to fundamentally eliminate the risk of collision between compact shelving units, the position information p j , the optimizer is required to p Ensure that the distance between any two adjacent rows of compact shelving j and j+1 is always no less than a preset minimum safety distance d min .

[0054] External force limit constraints: When the system is equipped with force sensors, force feedback-based safety constraints can be introduced. For example, the compressive or tensile forces generated by the motor drive can be limited to a certain threshold to protect the stored items or the shelving structure itself. This constraint can be expressed as a linear or nonlinear inequality constraint on state variables or input variables.

[0055] The central control component uses its computing power to numerically solve the quadratic programming (QP) problem defined above with equality and inequality constraints in each control cycle. The result of the solution is an optimal control sequence that satisfies all safety and physical constraints while minimizing the overall performance cost. This sequence contains N future times from the current moment c The ideal control action for each time step represents the optimal planning solution for the system to solve the problem of "how to move". The solution will then be passed to step S3 for execution.

[0056] S3: Application of control sequences and rolling horizon execution In this embodiment, step S3 is described in detail as follows. Its core purpose is to accurately and robustly convert the theoretically optimal control sequence generated by the optimized planning in step S2 into the actual action of the compact shelving components in the physical world, and constitute the key execution link of the closed-loop control strategy of the present invention.

[0057] In step S2, the predictive control module of the central control component has calculated a set of optimal control sequences for the future Its form is: This sequence represents the time from the current moment k to the next N c Within a control cycle, a series of control actions is considered "optimal." A straightforward approach is to sequentially apply all control commands in this sequence to the system. However, to achieve greater robustness and adaptability to unforeseen disturbances, the present invention preferably employs a strategy called Receding Horizon Control to apply this sequence.

[0058] The essence of the rolling horizon control strategy is to "focus on the present, look to the future, and make continuous corrections." Specifically, in the current control cycle k, the central control component does not execute the entire optimal control sequence.

[0059] Instead, the central control component extracts only the first control vector u from the sequence * (k|k). This vector represents the optimal control input applied to all compact shelving drive modules at the current time k.

[0060] Then, this single control vector u * (k|k) is sent to the drive module of each compact shelving component. The drive module precisely adjusts its output torque or speed based on the value of the instruction, completing the physical drive within the current control cycle and causing the compact shelving to produce a slight displacement.

[0061] After applying u * After (k|k), the optimal control sequence The remaining part, that is, u * (k+1|k),…,u * (k+N c -1|k), will be completely discarded. This "discarding" action is crucial, as it reflects the present invention's profound understanding of system uncertainty. * During the brief period of (k|k) execution, the system may have encountered minor disturbances that the model failed to fully capture (e.g., a momentary obstruction on the track, a slight voltage fluctuation, etc.), or the system's parameters may have undergone subtle changes that the online identification module has not yet fully identified. Continuing to execute the old plan based on past information can lead to accumulated deviations.

[0062] Therefore, when the time step enters the next control cycle k+1, the control process of the present invention will return to step S1. The central control component will first use the state sensor to measure the new actual state x(k+1) of the system at this moment.

[0063] Based on this true state x(k+1) that includes the impact of the latest disturbance, the system will restart the entire "learning-prediction-planning" chain: Return to S1: Use x(k+1) and the u just executed * (k|k), execute the online identification algorithm again to identify the dynamic system model and Carry out a new round of updates and get and

[0064] Return to S2: Based on this updated model and the new current state x(k+1), solve the constrained optimization problem again to generate a new, future-oriented optimal control sequence

[0065] Then, at time (k+1), the logic of step S3 is repeated again: only the new sequence is applied The first element u in * (k+1|k+1), and then discard the rest.

[0066] Through this repetitive cycle of "measure-update-optimize-execute-remeasure," the present invention cleverly transforms an open-loop optimization problem into a closed-loop feedback control structure. Its fundamental advantage lies in the fact that every actual control action applied to the system is based on the optimal decision made based on the latest actual system state and the latest system model. This continuous feedback and replanning mechanism makes the entire control system highly adaptable and robust to model uncertainties and external environmental disturbances, ensuring that the compact shelving array can move smoothly and precisely along the optimal trajectory toward the target state until final convergence.

[0067] This cycle continues until the central control component detects that the current state x(k) is consistent with the target state x ref When the error between them is less than a preset convergence threshold, the task is considered completed and the control loop is terminated.

[0068] The present invention also provides a compact shelving power supply control method, comprising the following steps: 1) In response to target instructions, the discrete-time state-space model that characterizes the unified behavior of all compact shelving components is updated in real time; 2) predicting and generating an optimal control sequence based on the updated discrete-time state-space model and the target state corresponding to the target instruction by solving an optimization problem aimed at minimizing a multi-objective cost function and satisfying a minimum safety distance constraint; 3) Applying the first control vector of the optimal control sequence in a rolling time domain manner to drive the compact shelving assembly, and returning to step 1) in the next control cycle.

[0069] The method of this embodiment can be used to execute the above system embodiment. Its principles and technical effects are similar and will not be described in detail here.

[0070] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A power supply control system for a compact shelving system, characterized in that: include: Power supply components; A plurality of compact shelving components, each of which comprises a drive module and at least one status sensor; and a central control component configured to, in response to a target instruction representing a target state of the compact shelving unit, perform the following operations: S1. updating a discrete-time state-space model representing the unified behavior of all compact shelving components in real time based on actual state information fed back by the state sensor; S2. Based on the updated discrete-time state-space model and the target state of the compact shelving system, predict and generate an optimal control sequence by solving a constrained optimization problem; S3. Apply the optimal control sequence to drive the compact shelving assembly to move toward the target state representing the compact shelving.

2. A compact shelving power supply control system according to claim 1, characterized in that: The discrete-time state-space model is represented by: x(k+1)=A(k)x(k)+B(k)u(k); Where x(k) represents the global state vector; u(k) is the global control input vector; A(k) and B(k) are time-varying system matrices.

3. A compact shelving power supply control system according to claim 2, characterized in that: The central control component updates the time-varying system matrices A(k) and B(k) by employing a recursive update rule represented by the following equation: in, represents the estimation vector containing the matrix parameters; K(k) is the gain matrix; e(k) is the prediction error vector.

4. A compact shelving power supply control system according to claim 1, characterized in that: The constrained optimization problem solved by the central control component aims to minimize a cost function J, which is represented by the following formula: Among them, x pred represents the predicted state vector; x ref is the reference state vector; x pred -x ref is the tracking error; Q is the state weight matrix; u is the control input vector; R is the input weight matrix.

5. The power supply control system for a compact shelving system according to claim 1, characterized in that: The constraints of the constrained optimization problem include the minimum safe distance constraint expressed as follows: |p j -p j+1 |≥d min ; Among them, p j and p j+1 Represents the predicted position of any adjacent compact shelving components; d min A preset safety distance.

6. A compact shelving power supply control system according to claim 5, characterized in that: At least one of the state sensors includes a displacement sensor and a force sensor; and the constraint condition further includes an external force limit set based on a signal detected by the force sensor.

7. The power supply control system for a compact shelving system according to claim 1, characterized in that: When applying the optimal control sequence, the central control component adopts a rolling horizon strategy, that is, only the first control vector of the sequence is applied in each control cycle, and the update of the discrete-time state space model and the generation of the optimal control sequence are re-executed in the next cycle.

8. The power supply control system for a compact shelving system according to claim 2, characterized in that: The global state vector x(k) includes the position information and speed information of all compact shelving components detected by the state sensor.

9. The power supply control system for a compact shelving system according to claim 1, characterized in that: The central control component is also configured to automatically use the target position of the next preset compact shelving component as the new target instruction after a compact shelving component moves to its target state to trigger a new round of control operations.

10. A method for controlling power supply of a compact shelving system, applied to a system comprising a plurality of compact shelving components, characterized in that: The following steps are involved: 1) In response to target instructions, the discrete-time state-space model that characterizes the unified behavior of all compact shelving components is updated in real time; 2) predicting and generating an optimal control sequence based on the updated discrete-time state-space model and the target state corresponding to the target instruction by solving an optimization problem aimed at minimizing a multi-objective cost function and satisfying a minimum safety distance constraint; 3) Applying the first control vector of the optimal control sequence in a rolling time domain manner to drive the compact shelving assembly, and returning to step 1) in the next control cycle.

Images