Aurora power load control system for home elevator

Abstract

The invention relates to the technical field of power load control, and discloses an aurora power load control system for a home elevator, and the system comprises a system state sensing unit which is configured to collect the home total electrical load power, the elevator calling signal of the home elevator and the target floor information in real time; the collaborative decision-making and optimization controller is in communication connection with the system state sensing unit and is used for executing core decision-making and optimization calculation; the elevator driving instruction reconstruction unit is connected with the collaborative decision and optimization controller and is used for receiving the optimization instruction and generating a corresponding elevator driving signal; and a load margin management pool. According to the system, the system state sensing unit collects the total household electrical load in real time, accurate sensing and prospective regulation and control of the total household power grid load are achieved on the basis of the prediction model and dynamic margin management, and the system can actively evaluate the residual capacity of the power grid and dynamically adjust the elevator operation curve before the elevator is started.

Description

Technical Field

[0001] This invention relates to the field of power load control technology, specifically to an aurora power load control system for home elevators. Background Technology

[0002] With the acceleration of urbanization and the continuous improvement of residents' living standards, home elevators, as key equipment for improving the quality of life and facilitating vertical travel for family members, especially the elderly and children, are becoming increasingly popular in villas, duplexes, and old residential communities.

[0003] In existing technologies, home elevators typically employ a simple on-demand start-up mode. When a user presses the call button, the elevator drive system and auxiliary systems such as lighting and ventilation immediately start and operate at rated power. While this mode offers simple control logic, the capacity of the home power grid is usually limited, especially during peak electricity consumption periods. When high-power appliances such as air conditioners, water heaters, induction cookers, and electric vehicle charging stations are running simultaneously, the total household load may approach or even exceed the capacity limit of the incoming electricity meter. If the home elevator suddenly starts at this time, its instantaneous surge current and operating power can easily cause the main switch to overload and trip, resulting in a power outage for the entire household. This not only affects the normal use of the elevator but also interrupts the operation of other household appliances, causing significant inconvenience and safety hazards. This invention designs an Aurora Power Load Control System for home elevators to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to provide an aurora power load control system for home elevators, which solves the problem of limited capacity of home power grids in the prior art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] An aurora power load control system for home elevators includes:

[0007] The system status sensing unit is configured to collect the total household power load P_total(t), the elevator call signal, and the target floor information in real time.

[0008] The collaborative decision-making and optimization controller is communicatively connected to the system state perception unit and is used to perform core decision-making and optimization calculations.

[0009] An elevator drive command reconfiguration unit, connected to the collaborative decision and optimization controller, is used to receive optimization commands and generate corresponding elevator drive signals.

[0010] The load margin management pool, as a virtual functional module, is integrated into the collaborative decision and optimization controller to manage the dynamic real-time available load margin P_pool_max(t).

[0011] The collaborative decision-making and optimization controller is configured to perform the following operations:

[0012] Step S1: Based on historical load data, use the recursive least squares method with a forgetting factor to predict the non-elevator base load P_base_pred(t+Δt) in the future time period Δt online;

[0013] Step S2, based on the household power grid safety threshold P_grid_max, the predicted base load P_base_pred(t+Δt), and the buffer power P_buffer dynamically adjusted based on the historical load fluctuation standard deviation σ(t), the formula is as follows:

[0014] ,

[0015] k is the safety factor, ranging from 1.5 to 3.0. Calculate the real-time available load margin:

[0016] ;

[0017] Step S3: Upon receiving an elevator call signal, using the real-time available load margin as a constraint, perform multi-objective optimization on the power-time curve of the target elevator's operation process to generate optimized elevator operation instructions;

[0018] In step S4, the elevator drive command reconfiguration unit controls the elevator to execute the optimized running curve by dynamically adjusting the inverter output of the elevator host according to the optimized command.

[0019] Preferably, the online prediction in step S1 uses an autoregressive moving average model to characterize the non-elevator foundation load:

[0020]

[0021] u(t) is an external influencing factor vector, including time and weather temperature; e(t) is white noise; A, B, and C are polynomials to be identified; the model parameters θ(t) are updated in real time using the recursive least squares method with a forgetting factor λ to predict P_base(t+Δt), where 0.95≤λ≤0.99.

[0022] Preferably, the objective function J of the multi-objective optimization problem in step S3 is:

[0023]

[0024] Where T_trip is the actual operating time, T_rated is the rated operating time, E_trip is the actual energy consumption, and E_rated is the rated energy consumption; α, β, γ are weighting coefficients that satisfy α+β+γ=1, used to balance operating efficiency, power smoothness, and energy economy.

[0025] Preferably, it also includes a distributed home appliance collaborative communication unit, which is connected to the collaborative decision and optimization controller;

[0026] When the real-time available load margin is insufficient to support the normal operation of the elevator, the collaborative decision and optimization controller sends a power adjustment request to at least one controllable home appliance in the home network through the distributed home appliance collaborative communication unit.

[0027] The power adjustment request is generated based on a dynamic contractual game model and includes a power reduction amount ΔP. cut Duration ΔT and corresponding excitation parameter I.

[0028] Preferably, the dynamic contract game model achieves power resource allocation through a two-way auction mechanism;

[0029] The collaborative decision-making and optimization controller, acting as the auctioneer, will determine the required total power reduction ΔP. cut Decomposed into n power reduction contracts, where the i-th contract is represented as a triple (ΔP) cuti， ΔT i I i ),in:

[0030] ΔP cuti ΔT represents the power reduction requested from the i-th controllable appliance, in kilowatts; i Indicates the duration, in seconds, required for the i-th controllable appliance to perform power reduction; i This represents the incentive compensation value provided to the i-th controllable home appliance in exchange for the power reduction, and the incentive compensation value may be virtual points, electricity price discount, or actual monetary amount;

[0031] Each controllable home appliance, acting as a bidder, independently evaluates and responds to the contract based on its own operating status and user settings. The collaborative decision-making and optimization controller selects a contract set S from all responding contracts, which must simultaneously satisfy the following two constraints:

[0032] Condition 1, Total power reduction requirement constraint: ;

[0033] Condition 2: Minimize the total incentive cost of the system. .

[0034] Preferably, in step S4, the elevator drive command reconfiguration unit tracks the optimized running curve through an adaptive sliding mode controller;

[0035] The sliding surface of the adaptive sliding mode controller is designed as follows:

[0036]

[0037] Where e=v optimal -v actual For speed tracking error, v optimal v is the ideal speed value defined in the optimized elevator operation curve. actual The actual running speed value fed back by the elevator host is given by , e' is the first derivative of the speed tracking error e with respect to time, i.e., the acceleration error; h is the sliding surface gain coefficient, which is a constant greater than zero, used to configure the dynamic characteristics of error convergence.

[0038] The control law u of the adaptive sliding mode controller is designed as follows:

[0039]

[0040] Where u is the output signal of the controller, used to drive the frequency converter of the elevator main unit; u eq is the equivalent control term, used to maintain the system state on the sliding surface under ideal, disturbance-free conditions; its value is calculated online based on the system model. K is the switching gain, an adaptively adjusted positive number used to compensate for uncertainties in the system model and external disturbances. sat is the saturation function, used to replace the sign function to eliminate the inherent chattering phenomenon of sliding mode control. Φ is the boundary layer thickness, used to define the linear range of the saturation function.

[0041] Preferably, the load margin management pool adopts a hierarchical management strategy, dividing the available load margin into multiple priority sub-pools;

[0042] This includes a basic operation sub-pool, an optimized operation sub-pool, and an emergency coordination sub-pool;

[0043] The collaborative decision-making and optimization controller dynamically adjusts the elevator operation strategy based on the sub-pool level where the current available margin is located.

[0044] Preferably, the system further includes a user behavior learning module, which establishes a user elevator usage habit model through a deep reinforcement learning algorithm;

[0045] The model predicts the probability of elevator usage in future periods based on historical usage data, and adjusts the configuration parameters of the load margin management pool accordingly.

[0046] Preferably, the deep reinforcement learning algorithm adopts a deep Q-network architecture;

[0047] The state space includes time, date type, weather conditions, and recent electricity consumption patterns;

[0048] The action space includes the pre-allocation strategy for the load margin management pool;

[0049] The reward function is designed as follows:

[0050]

[0051] Where U_satisfaction is the user satisfaction assessment, E_cost is the energy cost, P_penalty is the grid over-limit penalty, and w1, w2, and w3 are weighting coefficients.

[0052] Preferably, the collaborative decision-making and optimization controller further includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the system's control functions.

[0053] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0054] 1. This invention, through a system status sensing unit, collects the total household electricity load in real time and, based on a predictive model and dynamic margin management, achieves accurate sensing and proactive control of the total load on the household power grid. The system can proactively assess the remaining capacity of the power grid before the elevator starts and dynamically adjust the elevator's operating curve, effectively avoiding the tripping of the main power switch due to excessive instantaneous elevator starting power, eliminating the risk of power outages during peak electricity consumption periods, and ensuring the continuity and safety of household electricity use.

[0055] 2. This invention introduces a multi-objective optimization algorithm and a distributed home appliance coordination mechanism. Under the premise of ensuring power grid safety, the system intelligently coordinates elevator operation and home appliance power consumption. When the available load margin is insufficient, the system can actively and flexibly interact with other high-power home appliances and temporarily adjust their power through incentive contracts, thereby creating the necessary power space for elevator operation.

[0056] 3. This invention collects the total household electricity load in real time through a system status sensing unit, and based on a prediction model and load margin management, proactively assesses the remaining capacity of the power grid before the elevator starts and dynamically adjusts the elevator operation curve. This fundamentally avoids the problem of the main household switch tripping due to the instantaneous starting power of the elevator being superimposed on the operating load of other household appliances, ensuring the continuity and safety of household electricity. Attached Figure Description

[0057] Figure 1 This is a block diagram of the overall system structure of the present invention;

[0058] Figure 2This is a flowchart of the collaborative decision-making and optimization controller operation of the present invention;

[0059] Figure 3 This is a flowchart of the multi-objective optimization solution of the present invention;

[0060] Figure 4 This is a flowchart of the distributed home appliance collaborative communication process of the present invention;

[0061] Figure 5 This is a flowchart of the adaptive sliding mode controller of the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1;

[0064] Please see Figures 1-5 In this embodiment of the invention, an Aurora Power Load Control System for Home Elevators includes: a system state sensing unit configured to collect in real time the total household power load P_total(t), the elevator call signal, and the target floor information; a collaborative decision-making and optimization controller, communicatively connected to the system state sensing unit, for performing core decision-making and optimization calculations; an elevator drive command reconfiguration unit, connected to the collaborative decision-making and optimization controller, for receiving optimization commands and generating corresponding elevator drive signals; and a load margin management pool, integrated as a virtual functional module within the collaborative decision-making and optimization controller, for managing the dynamic real-time available load margin P_pool_max(t).

[0065] The collaborative decision-making and optimization controller is configured to perform the following operations:

[0066] Step S1: Based on historical load data, use the recursive least squares method with a forgetting factor to predict the non-elevator base load P_base_pred(t+Δt) in the future time period Δt online;

[0067] Step S2, based on the household power grid safety threshold P_grid_max, the predicted base load P_base_pred(t+Δt), and the buffer power P_buffer dynamically adjusted based on the historical load fluctuation standard deviation σ(t), the formula is as follows:

[0068] ,

[0069] k is the safety factor, ranging from 1.5 to 3.0. Calculate the real-time available load margin:

[0070] ;

[0071] Step S3: Upon receiving an elevator call signal, with the real-time available load margin as a constraint, perform multi-objective optimization on the power-time curve of the target elevator's operation process to generate optimized elevator operation instructions;

[0072] In step S4, the elevator drive command reconfiguration unit controls the elevator to execute the optimized running curve by dynamically adjusting the inverter output of the elevator host according to the optimized command.

[0073] The online prediction in step S1 uses an autoregressive moving average model to characterize the non-elevator foundation load:

[0074]

[0075] u(t) is an external influencing factor vector, including time and weather temperature; e(t) is white noise; A, B, and C are polynomials to be identified; the model parameters θ(t) are updated in real time using the recursive least squares method with a forgetting factor λ to predict P_base(t+Δt), where 0.95≤λ≤0.99.

[0076] The objective function J of the multi-objective optimization problem in step S3 is:

[0077]

[0078] Where T_trip is the actual operating time, T_rated is the rated operating time, E_trip is the actual energy consumption, and E_rated is the rated energy consumption; α, β, γ are weighting coefficients that satisfy α+β+γ=1, used to balance operating efficiency, power smoothness, and energy economy.

[0079] It also includes a distributed home appliance collaborative communication unit, which connects to the collaborative decision and optimization controller;

[0080] When the real-time available load margin is insufficient to support the normal operation of the elevator, the collaborative decision-making and optimization controller sends a power adjustment request to at least one controllable home appliance in the home network through the distributed home appliance collaborative communication unit.

[0081] Power adjustment requests are generated based on a dynamic contractual game model and include the power reduction amount ΔP. cut Duration ΔT and corresponding excitation parameter I.

[0082] The dynamic contract game model achieves power resource allocation through a two-way auction mechanism;

[0083] The collaborative decision-making and optimization controller, acting as the auctioneer, determines the required total power reduction ΔP. cut Decomposed into n power reduction contracts, where the i-th contract is represented as a triple (ΔP) cuti， ΔT i I i ),in:

[0084] ΔP cuti ΔT represents the power reduction requested from the i-th controllable appliance, in kilowatts; i Indicates the duration, in seconds, required for the i-th controllable appliance to perform power reduction; i This represents the incentive compensation value provided to the i-th controllable appliance in exchange for the power reduction. The incentive compensation value can be virtual credits, electricity price discounts, or actual monetary amounts.

[0085] Each controllable home appliance, acting as a bidder, independently evaluates and responds to the contract based on its own operating status and user settings. The collaborative decision-making and optimization controller selects a contract set S from all responding contracts, which must simultaneously satisfy the following two constraints:

[0086] Condition 1, Total power reduction requirement constraint: ;

[0087] Condition 2: Minimize the total incentive cost of the system. .

[0088] In step S4, the elevator drive command reconfiguration unit tracks the optimized running curve through an adaptive sliding mode controller;

[0089] The sliding surface design of the adaptive sliding mode controller is as follows:

[0090]

[0091] Where e=v optimal -v actual For speed tracking error, v optimal To optimize the ideal speed value defined in the elevator operating curve, v actual denoted as , where is the actual operating speed value fed back by the elevator host; e' is the first derivative of the speed tracking error e with respect to time, i.e., the acceleration error; h is the sliding surface gain coefficient, a constant greater than zero, used to configure the dynamic characteristics of error convergence.

[0092] The control law u of the adaptive sliding mode controller is designed as follows:

[0093]

[0094] Where u is the output signal of the controller, used to drive the frequency converter of the elevator main unit; u eqis the equivalent control term, used to maintain the system state on the sliding surface under ideal, disturbance-free conditions; its value is calculated online based on the system model. K is the switching gain, an adaptively adjusted positive number used to compensate for uncertainties in the system model and external disturbances. sat is the saturation function, used to replace the sign function to eliminate the inherent chattering phenomenon of sliding mode control. Φ is the boundary layer thickness, used to define the linear range of the saturation function.

[0095] The load margin management pool adopts a hierarchical management strategy, dividing the available load margin into multiple priority sub-pools;

[0096] This includes a basic operation sub-pool, an optimized operation sub-pool, and an emergency coordination sub-pool;

[0097] The collaborative decision-making and optimization controller dynamically adjusts the elevator operation strategy based on the sub-pool level where the current available margin is located.

[0098] The system also includes a user behavior learning module, which uses deep reinforcement learning algorithms to build a model of users' elevator usage habits.

[0099] The model predicts the probability of elevator usage in future periods based on historical usage data, and adjusts the configuration parameters of the load margin management pool accordingly.

[0100] The working principle of this invention is as follows: The system state perception unit first collects the total household electricity load P_total(t), elevator call signals, and target floor information in real time. These data serve as the basic inputs for system decision-making. The collaborative decision-making and optimization controller communicates with the system state perception unit and performs core decision-making and optimization calculations. Specifically, based on historical load data, the controller uses a recursive least squares method with a forgetting factor to predict the non-elevator base load P_base_pred(t+Δt) in the future time period Δt online. This prediction process is characterized using an autoregressive moving average model, in which the external influencing factor vector u(t) (including time and weather temperature) and white noise e(t) are incorporated into the model, and the model parameters are updated in real time through a forgetting factor λ (ranging from 0.95 to 0.99) to improve prediction accuracy and adaptability.

[0101] Subsequently, the collaborative decision-making and optimization controller calculates the real-time available load margin P_pool_max(t) based on the home grid safety threshold P_grid_max, the predicted base load P_base_pred(t+Δt), and the buffer power P_buffer dynamically adjusted based on the historical load fluctuation standard deviation σ(t). This calculation is achieved through the formula P_pool_max(t) = P_grid_max - P_base_pred(t+Δt) - P_buffer, ensuring that the system operates within the grid safety threshold and reserving buffer space to cope with load fluctuations. When an elevator call signal is received, the controller performs multi-objective optimization on the power-time curve of the target elevator operation process, constrained by the real-time available load margin. The optimization objective function J is used to balance operating efficiency, power smoothness, and energy economy. The optimized elevator operation command is sent to the elevator drive command reconfiguration unit.

[0102] The elevator drive command reconfiguration unit is connected to the collaborative decision-making and optimization controller, receives optimization commands, and tracks the optimized running curve through an adaptive sliding mode controller. The control law u is designed as u=u eq +K·sat(s / Φ), where u eq The equivalent control term is calculated online based on the system model to maintain the sliding surface; K is the adaptive switching gain used to compensate for system uncertainties and disturbances; sat is the saturation function, and Φ is the boundary layer thickness, which together eliminate chattering in sliding mode control. By dynamically adjusting the inverter output of the elevator host, the system controls the elevator to execute the optimized operating curve, ensuring smooth, efficient operation that meets load constraints.

[0103] In addition, the system includes a distributed home appliance collaborative communication unit. When the real-time available load margin is insufficient to support the normal operation of the elevator, the collaborative decision-making and optimization controller sends a power adjustment request to at least one controllable home appliance in the home network through this unit. The load margin management pool, as a virtual functional module, adopts a hierarchical management strategy, dividing the available load margin into a basic operation sub-pool, an optimized operation sub-pool, and an emergency coordination sub-pool. The controller dynamically adjusts the elevator operation strategy according to the current margin level, further enhancing the system's flexibility.

[0104] Example 2;

[0105] Please see Figures 1-5 In this embodiment of the invention, the deep reinforcement learning algorithm adopts a deep Q-network architecture;

[0106] The state space includes time, date type, weather conditions, and recent electricity consumption patterns;

[0107] The action space includes the pre-allocation strategy for the load margin management pool;

[0108] The reward function is designed as follows:

[0109]

[0110] Where U_satisfaction is the user satisfaction assessment, E_cost is the energy cost, P_penalty is the grid over-limit penalty, and w1, w2, and w3 are weighting coefficients.

[0111] The collaborative decision-making and optimization controller further includes a memory, a processor, and a computer program stored in the memory and executable on the processor, which implements the control functions of the system when the processor executes the program.

[0112] The working principle of this invention is as follows: The system further integrates a user behavior learning module, which establishes a user elevator usage habit model through deep reinforcement learning algorithms to optimize the configuration of the load margin management pool and the overall system performance. This module adopts a deep Q-network (DQN) architecture, and its state space includes multi-dimensional variables such as time, date type, weather conditions, and recent electricity consumption patterns. These variables collectively capture the spatiotemporal characteristics of user behavior and environmental influencing factors. The action space is defined as the pre-allocation strategy of the load margin management pool, including adjustments to the margin allocation of different priority sub-pools (such as the basic operation sub-pool, the optimized operation sub-pool, and the emergency coordination sub-pool).

[0113] The user behavior learning module is trained using historical usage data (such as elevator call frequency, operating hours, and user preferences) and continuously optimizes the policy network using deep reinforcement learning algorithms. During training, the module selects actions (i.e., pre-allocated strategies) based on the current state (such as electricity consumption patterns during specific time periods), observes environmental feedback after execution (such as changes in user satisfaction and load exceedance situations), and updates the Q-value function to maximize cumulative rewards. Through iterative learning, the module can predict the probability of elevator usage in future periods, such as identifying changes in elevator demand during peak electricity consumption periods or under specific weather conditions. Based on these predictions, the collaborative decision-making and optimization controller pre-adjusts the configuration parameters of the load margin management pool, such as allocating more margin to the optimized operation sub-pool in advance or triggering an emergency coordination mechanism, thereby responding quickly when an elevator call signal occurs and reducing optimization calculation latency.

[0114] Furthermore, this learning module works collaboratively with other units in the system. The prediction results are input into the collaborative decision-making and optimization controller to refine the prediction model for non-elevator base loads. This includes adjusting the weights of external factors in the autoregressive moving average model and optimizing the weight coefficients α, β, and γ in the multi-objective function to adapt to changes in user habits. For instance, during peak elevator usage periods, the system can prioritize adjusting the operating efficiency weight α to ensure elevator response speed; while during periods of higher energy costs, it emphasizes the energy economy weight γ. This adaptive learning mechanism enables the system to dynamically adapt to user behavior patterns, improving the accuracy and efficiency of load management. Simultaneously, by pre-adjusting, it reduces the computational load of real-time decision-making, enhancing the system's real-time performance and reliability.

[0115] Example 3;

[0116] Please see Figures 1-5 A specific embodiment is provided, in which the home power grid safety threshold P_grid_max is set to 10 kW; the prediction time interval Δt is set to 30 seconds; the forgetting factor λ is set to 0.98; the safety factor k is set to 2.0; the multi-objective optimization weight coefficients are set to α=0.5, β=0.3, γ=0.2 respectively; the boundary layer thickness Φ is set to 0.1; and the reward function weight coefficients are set to w1=0.6, w2=0.25, w 3= 0.15.

[0117] When the system starts working, the system status sensing unit collects the current total household power load in real time, with a monitored value of 7.2 kilowatts. At the same time, it detects the elevator call signal and the target floors as the 1st to 3rd floors.

[0118] The user behavior learning module first initiates its prediction function. Based on a deep Q-network architecture, this module takes into account the current time (14:30), a weekday date, sunny weather conditions, and a recent power consumption pattern of medium load. After calculation, the module outputs a prediction: the probability of elevator use within the next 30 minutes is 0.75. Based on this prediction, the module pre-adjusts the configuration parameters of the load margin management pool, setting the basic operation sub-pool threshold to 1.0 kW, the optimized operation sub-pool threshold to 2.0 kW, and the emergency coordination sub-pool threshold to 3.0 kW.

[0119] The collaborative decision-making and optimization controller then performs load forecasting calculations. A recursive least squares method with a forgetting factor is used for online forecasting based on an autoregressive moving average model. The model inputs an external influencing factor vector, including timestamps and temperature data, and outputs a non-elevator base load forecast value P_base_pred of 7.1 kW for the next 30 seconds.

[0120] Next, the controller calculates the buffer power P_buffer. The system obtains the historical load fluctuation standard deviation σ(t) as 0.3 kW, and combined with the safety factor k=2.0, calculates P_buffer=0.6 kW. Then, calculate the real-time available load margin P_pool_max(t)=10-7.1-0.6=2.3 kW.

[0121] Upon receiving a call signal, the controller initiates a multi-objective optimization solution with a real-time available load margin of 2.3 kW as a constraint. The optimization objective function J comprehensively considers operating time deviation, power change rate, and energy consumption differences. After calculation, an optimized elevator operating curve is generated, which adjusts the rated operating time from 10 seconds to 12 seconds and the rated energy consumption from 0.25 kWh to 0.22 kWh.

[0122] Since the current available load margin is within the optimized operation sub-pool range, the system directly executes the optimized operation strategy. The elevator drive command reconfiguration unit receives the optimization command and accurately tracks the optimization curve through the adaptive sliding mode controller. The controller monitors the speed tracking error in real time, and when the detected error value exceeds the threshold of 0.05 m / s, it automatically adjusts the control law output and dynamically adjusts the elevator host's operating status through the frequency converter.

[0123] In special circumstances, when the system detects that the real-time available load margin has dropped to 0.8 kW, which is below the basic operating sub-pool threshold, the collaborative decision-making and optimization controller immediately activates the distributed home appliance collaborative mechanism. The controller generates a total power reduction demand ΔP. cut =1.5 kW, sending power adjustment requests to controllable home appliances in the home network through a two-way auction mechanism.

[0124] In practice, the controller decomposes the total demand into three power reduction contracts: Contract 1 requires the air conditioner to reduce its power by 0.8 kW for 120 seconds, with an incentive parameter of 0.5 yuan; Contract 2 requires the water heater to reduce its power by 0.5 kW for 90 seconds, with an incentive parameter of 0.3 yuan; Contract 3 requires the electric vehicle charging station to reduce its power by 0.7 kW for 180 seconds, with an incentive parameter of 0.6 yuan.

[0125] After a game-theoretic decision-making process, the system ultimately selected a combination of Contract 1 and Contract 2, resulting in a total power reduction of 1.3 kilowatts and a total incentive cost of 0.8 yuan, thus meeting the minimum power requirement. During this period, the elevator completed the transportation task in basic operating mode.

[0126] During system operation, the user behavior learning module continuously collects operational data, including actual running time of 12.3 seconds, actual energy consumption of 0.21 kWh, and user satisfaction score of 0.85. Based on this data, the module updates the parameters of the deep Q-network to optimize subsequent prediction accuracy and decision quality.

[0127] Working Principle: Based on real-time power load monitoring and dynamic optimization control, the system's state sensing unit collects the total household power load and elevator operation demand. The collaborative decision-making and optimization controller uses a recursive least squares method with a forgetting factor to predict the non-elevator base load, and calculates the real-time available load margin by combining the power grid safety threshold and dynamic buffer power. Upon receiving an elevator call signal, the system uses this margin as a constraint to perform multi-objective optimization of the elevator operation process, generating operating instructions that balance operating efficiency, power smoothing, and energy economy. The adaptive sliding mode controller accurately tracks the optimization curve to achieve safe, smooth, and energy-efficient elevator operation.

[0128] By constructing a user elevator usage habit model through deep reinforcement learning, its deep Q-network predicts the future elevator usage probability and pre-adjusts the configuration parameters of the load margin management pool based on state features such as time, date, weather, and electricity consumption patterns. This module comprehensively optimizes user satisfaction, energy costs, and grid over-limit penalties through a reward function, enabling the system to adaptively respond to changes in user behavior and pre-optimize load allocation strategies, thereby improving the overall intelligence level of the system.

[0129] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An aurora power load control system for home elevators, characterized in that, include: The system status sensing unit is configured to collect the total household power load P_total(t), the elevator call signal, and the target floor information in real time. The collaborative decision-making and optimization controller is communicatively connected to the system state perception unit and is used to perform core decision-making and optimization calculations. An elevator drive command reconfiguration unit, connected to the collaborative decision and optimization controller, is used to receive optimization commands and generate corresponding elevator drive signals. The load margin management pool, as a virtual functional module, is integrated into the collaborative decision and optimization controller to manage the dynamic real-time available load margin P_pool_max(t). The collaborative decision-making and optimization controller is configured to perform the following operations: Step S1: Based on historical load data, use the recursive least squares method with a forgetting factor to predict the non-elevator base load P_base_pred(t+Δt) in the future time period Δt online; Step S2, based on the household power grid safety threshold P_grid_max, the predicted base load P_base_pred(t+Δt), and the buffer power P_buffer dynamically adjusted based on the historical load fluctuation standard deviation σ(t), the formula is as follows: ， k is the safety factor, ranging from 1.5 to 3.

0. Calculate the real-time available load margin: ； Step S3: Upon receiving an elevator call signal, using the real-time available load margin as a constraint, perform multi-objective optimization on the power-time curve of the target elevator's operation process to generate optimized elevator operation instructions; In step S4, the elevator drive command reconfiguration unit controls the elevator to execute the optimized running curve by dynamically adjusting the inverter output of the elevator host according to the optimized command.

2. The aurora power load control system for home elevators according to claim 1, characterized in that, The online prediction in step S1 uses an autoregressive moving average model to characterize the non-elevator foundation load: Among them, Z -1 The shift operator, in time series models, represents a time delay. ; u(t) is an external influencing factor vector, including time and weather temperature; e(t) is white noise; A, B, and C are polynomials to be identified; the model parameters θ(t) are updated in real time using the recursive least squares method with a forgetting factor λ to predict P_base(t+Δt), where 0.95≤λ≤0.

99.

3. The aurora power load control system for a home elevator according to claim 1, characterized in that, The objective function J of the multi-objective optimization problem in step S3 is: Where dP_elevator / dt represents the rate of change of elevator power with respect to time, T_trip is the actual running time, T_rated is the rated running time, E_trip is the actual energy consumption, and E_rated is the rated energy consumption; α, β, γ are weighting coefficients that satisfy α+β+γ=1, used to balance operating efficiency, power smoothness, and energy economy.

4. The aurora power load control system for a home elevator according to claim 3, characterized in that, It also includes a distributed home appliance collaborative communication unit, which is connected to the collaborative decision and optimization controller; When the real-time available load margin is insufficient to support the normal operation of the elevator, the collaborative decision and optimization controller sends a power adjustment request to at least one controllable home appliance in the home network through the distributed home appliance collaborative communication unit. The power adjustment request is generated based on a dynamic contractual game model and includes a power reduction amount ΔP. cut Duration ΔT and corresponding excitation parameter I.

5. The aurora power load control system for a home elevator according to claim 4, characterized in that, The dynamic contract game model achieves power resource allocation through a two-way auction mechanism; The collaborative decision-making and optimization controller, acting as the auctioneer, will determine the required total power reduction ΔP. cut Decomposed into n power reduction contracts, where the i-th contract is represented as a triple (ΔP) cuti， ΔT i I i ),in: ΔP cuti ΔT represents the power reduction requested from the i-th controllable appliance, in kilowatts; i Indicates the duration, in seconds, required for the i-th controllable appliance to perform power reduction; i This represents the incentive compensation value provided to the i-th controllable home appliance in exchange for the power reduction, and the incentive compensation value may be virtual points, electricity price discount, or actual monetary amount; Each controllable home appliance, acting as a bidder, independently evaluates and responds to the contract based on its own operating status and user settings. The collaborative decision-making and optimization controller selects a contract set S from all responding contracts, which must simultaneously satisfy the following two constraints: Condition 1, Total power reduction requirement constraint: ; Condition 2: Minimize the total incentive cost of the system. .

6. The aurora power load control system for a home elevator according to claim 1, characterized in that, In step S4, the elevator drive command reconstruction unit tracks the optimized running curve through an adaptive sliding mode controller; The sliding surface of the adaptive sliding mode controller is designed as follows: Where e=v optimal -v actual For speed tracking error, v optimal v is the ideal speed value defined in the optimized elevator operation curve. actual The actual running speed value fed back by the elevator host is given by , e' is the first derivative of the speed tracking error e with respect to time, i.e., the acceleration error; h is the sliding surface gain coefficient, which is a constant greater than zero, used to configure the dynamic characteristics of error convergence. The control law u of the adaptive sliding mode controller is designed as follows: Where u is the output signal of the controller, used to drive the frequency converter of the elevator main unit; u eq The equivalent control term is used to maintain the system state on the sliding surface under ideal, disturbance-free conditions. Its value is calculated online based on the system model. K is the switching gain, which is an adaptively adjusted positive number used to compensate for uncertainties in the system model and external disturbances. sat is the saturation function, used to replace the sign function to eliminate the inherent chattering phenomenon of sliding mode control; Φ is the boundary layer thickness, used to define the linear range of the saturation function.

7. The aurora power load control system for a home elevator according to claim 1, characterized in that, The load margin management pool adopts a hierarchical management strategy, dividing the available load margin into multiple priority sub-pools; This includes a basic operation sub-pool, an optimized operation sub-pool, and an emergency coordination sub-pool; The collaborative decision-making and optimization controller dynamically adjusts the elevator operation strategy based on the sub-pool level where the current available margin is located.

8. The aurora power load control system for a home elevator according to claim 1, characterized in that: The system also includes a user behavior learning module, which establishes a user elevator usage habit model through deep reinforcement learning algorithms. The model predicts the probability of elevator usage in future periods based on historical usage data, and adjusts the configuration parameters of the load margin management pool accordingly.

9. The aurora power load control system for a home elevator according to claim 1, characterized in that, The deep reinforcement learning algorithm employs a deep Q-network architecture; The state space includes time, date type, weather conditions, and recent electricity consumption patterns; The action space includes the pre-allocation strategy for the load margin management pool; The reward function is designed as follows: Where U_satisfaction is the user satisfaction assessment, E_cost is the energy cost, P_penalty is the grid over-limit penalty, and w1, w2, and w3 are weighting coefficients used to balance the importance of user satisfaction, energy cost, and grid over-limit penalty in the reward function, and satisfy the normalization condition.

10. The aurora power load control system for a home elevator according to claim 1, characterized in that, The collaborative decision-making and optimization controller further includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the control functions of the system as described in any one of claims 1-9.

Images