Elevator control system and method based on dynamic passenger flow prediction and life coordination control

Abstract

The application discloses a kind of based on dynamic passenger flow prediction and life collaborative control's elevator control system, including passenger flow prediction module, energy decomposition module, life optimization module and closed-loop control module;Passenger flow prediction module is used to predict elevator passenger flow density in real time according to elevator load dynamic characteristics, and export elevator passenger flow prediction result;Energy decomposition module is used to dynamically adjust empirical mode decomposition parameter according to elevator passenger flow prediction result, adjust frequency domain demarcation point, realize energy storage energy distribution;Life optimization module is used to establish the coupling aging model of energy storage element, the weight of optimization target is adjusted in real time dynamically, realize the collaborative optimization of multi-objective life;Closed-loop control module is used to carry out collaborative control to each module and hybrid energy storage system according to passenger flow prediction result, energy storage energy distribution and life optimization information. Energy decomposition is carried out, passenger flow prediction and energy distribution control are combined, closed-loop control is formed, energy recovery efficiency is improved and energy storage system equipment life is balanced.

Description

Technical Field

[0001] This invention relates to the field of elevator control technology, and more specifically, to an elevator hybrid energy storage control system and method based on dynamic passenger flow prediction and lifespan coordinated control. Background Technology

[0002] With the acceleration of urbanization and the increasingly urgent need for energy structure transformation, elevator energy-saving control technology has become a focus. Traditional elevator systems consume a lot of energy and are inefficient in high-frequency usage scenarios such as office buildings and hospitals. However, a hybrid energy storage system using supercapacitors and batteries can serve as an effective solution for recovering and regenerating energy, smoothing load fluctuations, and improving system efficiency.

[0003] However, existing technologies have the following problems: hybrid energy storage systems based on fixed filtering frequencies cannot adapt to the drastic fluctuations in elevator loads; batteries and supercapacitors have uneven lifespans due to differences in their operating characteristics; and there is a lack of effective closed-loop control strategies for elevator scenarios, resulting in a disconnect between prediction and control. Summary of the Invention

[0004] The purpose of this invention is to overcome the defects and shortcomings of the prior art and provide an elevator control system and method based on dynamic passenger flow prediction and lifespan collaborative control. This system dynamically adjusts the frequency domain boundary point to achieve adaptive energy distribution; establishes a coupled aging model of energy storage elements to collaboratively optimize multi-objective lifespan; and combines passenger flow prediction with energy distribution control to form a closed-loop control, thereby improving energy recovery efficiency and balancing equipment lifespan.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An elevator control system based on dynamic passenger flow prediction and lifespan coordinated control includes a passenger flow prediction module, an energy decomposition module, a lifespan optimization module, and a closed-loop control module.

[0007] The passenger flow prediction module is used to predict elevator passenger flow density in real time based on the dynamic characteristics of elevator load, and output the elevator passenger flow prediction results.

[0008] The energy decomposition module is used to dynamically adjust the empirical mode decomposition parameters and adjust the frequency domain boundary point based on the elevator passenger flow prediction results to realize the energy storage distribution.

[0009] The lifetime optimization module is used to establish a coupled aging model of energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives.

[0010] The closed-loop control module is used to coordinate the control of each module and the hybrid energy storage system based on passenger flow forecast results, energy storage allocation and lifespan optimization information.

[0011] Furthermore, the passenger flow prediction module is used to predict elevator passenger density in real time based on the dynamic characteristics of elevator load, and output the elevator passenger flow prediction results, specifically:

[0012] The passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM. By collecting data from inside the elevator car, the call buttons on each floor, and the turnstiles, it constructs a spatiotemporal dataset to achieve a refined prediction of the short-term passenger flow density.

[0013] The passenger flow prediction module consists of an input layer, a hidden layer, and an output layer. The hidden layer adopts a structure that combines one-dimensional convolution and long short-term memory networks. The input data includes passenger flow density, building type coding, and time period information. The network is trained and optimized by using historical data and adding Gaussian noise for data augmentation. After the network is trained, it outputs the predicted passenger flow density sequence by inputting the feature data of the current moment.

[0014] Furthermore, the energy decomposition module is used to dynamically adjust the empirical mode decomposition parameters and frequency domain boundary points based on elevator passenger flow prediction results to achieve energy storage allocation, specifically:

[0015] The energy decomposition module dynamically adjusts the window length and wavelet entropy threshold of the empirical mode decomposition based on the predicted passenger flow density, realizes the adaptive calculation of the frequency domain boundary point, and divides the frequency band into high-frequency and low-frequency parts through the boundary frequency mapping function.

[0016] In the energy decomposition module, the window length W of the empirical mode decomposition algorithm is dynamically adjusted according to the predicted passenger flow density D, and the calculation formula is as follows:

[0017] W = W0 - k·D;

[0018] In the formula, W0 is the base window length; k is the adjustment coefficient; D is the predicted passenger flow density; when the passenger flow density is high, the window becomes shorter to capture high-frequency components; when the passenger flow density is low, the window becomes longer to capture low-frequency components.

[0019] The wavelet entropy threshold Th is obtained by fitting historical data, and the calculation formula is as follows:

[0020] Th = ab·D;

[0021] In the formula, a and b are fitting coefficients.

[0022] Furthermore, the lifetime optimization module is used to establish a coupled aging model of energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of multi-objective lifetimes, specifically:

[0023] The lifespan optimization module establishes a coupled model of the battery cycle life decay factor λ_bat and the aging increment ΔESR of the supercapacitor's equivalent series resistance (ESR). The weights of the three optimization objectives are dynamically adjusted in real time through an improved particle swarm optimization algorithm to achieve collaborative optimization of multi-objective lifespan.

[0024] The weights of the three optimization objectives are calculated using the following formulas:

[0025] α = α0 - c1·λ_bat;

[0026] β = β0;

[0027] γ = γ0 + c2·Δ_ESR;

[0028] In the formula, α is used to minimize grid impact, β is used to maximize energy recovery efficiency, γ is used to minimize the aging rate of energy storage system, α0, β0, and γ0 are the basic weight values ​​of each optimization objective, and c1 and c2 are adjustment coefficients.

[0029] The formula for calculating the supercapacitor power allocation ratio Ratio_SC is:

[0030] Ratio_SC=R_base-k1·Δ_ESR+k2·λ_bat;

[0031] In the formula, R_base is the basic allocation ratio, and k1 and k2 are fitting coefficients.

[0032] Furthermore, the closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the work of the passenger flow prediction module, energy decomposition module, and lifespan optimization module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller through the CAN bus to realize real-time data acquisition and control command issuance.

[0033] Furthermore, the closed-loop control module uses a bidirectional DC / DC converter to connect the hybrid energy storage system and the elevator frequency converter, achieving efficient energy conversion through PWM complementary mode.

[0034] Furthermore, during the operation of the elevator control system, the closed-loop control module monitors passenger flow prediction errors, energy storage system aging rate, and grid fluctuation rate in real time. When an anomaly is detected, the elevator control system automatically adjusts and optimizes its strategy. When the prediction error is large, the elevator control system increases the length of the empirical mode decomposition analysis window. When the grid fluctuation rate is high, the elevator control system increases the weight of maximizing energy recovery efficiency to optimize energy recovery efficiency.

[0035] An elevator control method based on dynamic passenger flow prediction and lifespan coordinated control, applied to the aforementioned elevator control system based on dynamic passenger flow prediction and lifespan coordinated control, includes the following steps:

[0036] Passenger flow forecasting: Based on the dynamic characteristics of elevator load, predict elevator passenger flow density in real time and output the elevator passenger flow forecast results;

[0037] Energy decomposition: Based on elevator passenger flow forecast results, the empirical mode decomposition parameters are dynamically adjusted, and the frequency domain boundary point is adjusted to achieve energy storage distribution;

[0038] Lifetime optimization: Establish a coupled aging model for energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives;

[0039] Closed-loop control: Based on passenger flow forecasts, energy storage allocation, and lifespan optimization information, the system performs coordinated control of each module and the hybrid energy storage system.

[0040] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described above.

[0041] A storage medium, a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described above.

[0042] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0043] 1. This invention employs a spatiotemporal convolutional LSTM dynamic passenger flow prediction module, which integrates one-dimensional convolution and long short-term memory networks to achieve refined prediction of future short-term passenger flow density. It adapts to the dynamic characteristics of elevator load, solves the problem that traditional fixed filtering frequencies are difficult to cope with sudden passenger flow, and improves the response speed and energy efficiency of elevator control systems.

[0044] 2. This invention uses an adaptive frequency domain energy decomposition module to dynamically adjust the Empirical Mode Decomposition (EMD) window length and wavelet entropy threshold based on the predicted passenger flow density, thereby achieving adaptive calculation of the frequency domain boundary point. This overcomes the limitations of fixed filtering frequencies, solves the problem of high-frequency component response lag, fully leverages the advantages of supercapacitor's fast response, and improves the accuracy and efficiency of energy allocation.

[0045] 3. This invention establishes a coupling model between the battery cycle life decay factor and the aging increment of the equivalent series resistance of the supercapacitor. By improving the particle swarm optimization algorithm, the weights of multiple objectives are adjusted in real time to achieve life synergistic optimization. It comprehensively considers the coupled aging effect of energy storage components, solves the problem of large differences in life decay, and extends the overall life of the energy storage system.

[0046] 4. This invention integrates passenger flow forecasting with energy distribution control to form a complete closed-loop control. It uses the forecast results to guide the adjustment of power distribution strategies in real time, solving the energy efficiency loss caused by the disconnect between traditional "forecasting" and "control", and improving overall operating efficiency.

[0047] 5. This invention adaptively adjusts the boundary frequency and window length to achieve refined management of energy in different frequency bands, enabling rapid response to sudden passenger flow, smoothing load fluctuations, improving the dynamic response capability of the elevator control system, and meeting actual operational needs. Attached Figure Description

[0048] Figure 1 This is a topology diagram of an elevator control system based on dynamic passenger flow prediction and lifespan coordinated control.

[0049] Figure 2 This is a schematic diagram of the passenger flow forecasting process.

[0050] Figure 3 This is a schematic diagram of the closed-loop control process.

[0051] Figure 4 This is a flowchart of an elevator control method based on dynamic passenger flow prediction and lifespan coordinated control. Detailed Implementation

[0052] The elevator control system and method based on dynamic passenger flow prediction and lifespan coordinated control of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0053] Please see Figure 1 This invention discloses an elevator control system based on dynamic passenger flow prediction and lifespan coordinated control, including a passenger flow prediction module, an energy decomposition module, a lifespan optimization module and a closed-loop control module.

[0054] The passenger flow prediction module is used to predict elevator passenger flow density in real time based on the dynamic characteristics of elevator load, and output the elevator passenger flow prediction results.

[0055] The energy decomposition module is used to dynamically adjust the empirical mode decomposition parameters and adjust the frequency domain boundary point based on the elevator passenger flow prediction results to realize the energy storage distribution.

[0056] The lifetime optimization module is used to establish a coupled aging model of energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives.

[0057] The closed-loop control module is used to coordinate the control of each module and the hybrid energy storage system based on passenger flow forecast results, energy storage allocation and lifespan optimization information.

[0058] Please see Figure 2Specifically, the passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM. By collecting data from inside the elevator car, call buttons on each floor, and turnstiles, it constructs a spatiotemporal dataset to achieve refined prediction of future short-term passenger flow density.

[0059] The passenger flow prediction module consists of an input layer, a hidden layer, and an output layer. The hidden layer adopts a structure that combines one-dimensional convolution and long short-term memory networks. The input data includes passenger flow density, building type coding, and time period information. The network is trained and optimized by using historical data and adding Gaussian noise for data augmentation. After the network is trained, it outputs the predicted passenger flow density sequence by inputting the feature data of the current moment.

[0060] Specifically, the energy decomposition module dynamically adjusts the window length and wavelet entropy threshold of the empirical mode decomposition based on the predicted passenger flow density, realizes the adaptive calculation of the frequency domain boundary point, and divides the frequency band into high-frequency and low-frequency parts through the boundary frequency mapping function.

[0061] In the energy decomposition module, the window length W of the empirical mode decomposition algorithm is dynamically adjusted according to the predicted passenger flow density D, and the calculation formula is as follows:

[0062] W = W0 - k·D;

[0063] In the formula, W0 is the base window length; k is the adjustment coefficient; D is the predicted passenger flow density; when the passenger flow density is high, the window becomes shorter to capture high-frequency components; when the passenger flow density is low, the window becomes longer to capture low-frequency components.

[0064] The wavelet entropy threshold Th is obtained by fitting historical data, and the calculation formula is as follows:

[0065] Th = ab·D;

[0066] In the formula, a and b are fitting coefficients.

[0067] Specifically, the life optimization module establishes a coupled model of the battery cycle life decay factor λ_bat and the aging increment ΔESR of the supercapacitor's equivalent series resistance (ESR). Through an improved particle swarm optimization algorithm, the weights of the three optimization objectives are dynamically adjusted in real time to achieve collaborative optimization of multi-objective lifespan.

[0068] The weights of the three optimization objectives are calculated using the following formulas:

[0069] α = α0 - c1·λ_bat;

[0070] β = β0;

[0071] γ = γ0 + c2·Δ_ESR;

[0072] In the formula, α is used to minimize grid impact, β is used to maximize energy recovery efficiency, γ is used to minimize the aging rate of energy storage system, α0, β0, and γ0 are the basic weight values ​​of each optimization objective, and c1 and c2 are adjustment coefficients.

[0073] The formula for calculating the supercapacitor power allocation ratio Ratio_SC is:

[0074] Ratio_SC=R_base-k1·Δ_ESR+k2·λ_bat;

[0075] In the formula, R_base is the basic allocation ratio, and k1 and k2 are fitting coefficients.

[0076] Please see Figure 3 Specifically, the closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the operation of the passenger flow prediction module, energy decomposition module, and lifespan optimization module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller via a CAN bus to achieve real-time data acquisition and control command issuance. The closed-loop control module uses a bidirectional DC / DC converter to connect the hybrid energy storage system and the elevator frequency converter, achieving efficient energy conversion through a PWM complementary mode.

[0077] During the operation of the elevator control system, the closed-loop control module monitors passenger flow prediction errors, energy storage system aging rate, and power grid fluctuation rate in real time. When an anomaly is detected, the elevator control system automatically adjusts and optimizes its strategy. When the prediction error is large, the elevator control system increases the length of the empirical mode decomposition analysis window. When the power grid fluctuation rate is high, the elevator control system increases the weight of maximizing energy recovery efficiency to optimize energy recovery efficiency.

[0078] Example 1

[0079] This embodiment provides an elevator control system based on dynamic passenger flow prediction and lifespan collaborative control, including a passenger flow prediction module, an energy decomposition module, a lifespan optimization module, and a closed-loop control module.

[0080] The passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM to construct a spatiotemporal dataset by collecting data from inside the elevator car, call buttons on each floor, and turnstiles, thereby enabling refined prediction of short-term passenger flow density.

[0081] Specifically, the passenger flow prediction module comprises an input layer, a hidden layer, and an output layer. The hidden layer employs a structure that fuses one-dimensional convolutional and long short-term memory networks. Input data includes passenger flow density, building type encoding, and time period information. Data augmentation is performed using historical data and adding Gaussian noise to train and optimize the network (Conv-LSTM model). After training, the network outputs a predicted passenger flow density sequence by inputting the feature data of the current moment.

[0082] The energy decomposition module dynamically adjusts the window length W and wavelet entropy threshold Th of the Empirical Mode Decomposition (EMD) based on the predicted passenger flow density D. As a specific implementation, the window length W of the EMD algorithm can be dynamically adjusted according to the following formula:

[0083] W = 300 - 1.2 × D (unit: milliseconds)

[0084] In this embodiment, the base window length W0 is set to 300ms, and the adjustment coefficient k is set to 1.2. When the passenger flow density D is high, the window W is shortened to capture high-frequency components; when the passenger flow density D is low, the window W is lengthened to capture low-frequency components.

[0085] The wavelet entropy threshold Th is obtained by fitting historical data. One specific calculation formula is: Th = 0.85 - 0.005 × D, that is, fitting coefficients a = 0.85, b = 0.005.

[0086] The lifespan optimization module establishes a coupled model of the battery cycle life degradation factor λ_bat and the aging increment Δ_ESR of the supercapacitor's equivalent series resistance (ESR). An improved particle swarm optimization algorithm is used to optimize the weights of the three objective functions in real time. As a specific implementation method, the weight calculation formula is as follows:

[0087] α=0.5-0.01×λ_bat, that is, α0=0.5, c1=0.01.

[0088] β = 0.3, that is, β0 = 0.3

[0089] γ=0.2+0.005×Δ_ESR, that is, γ0=0.2, c2=0.005,

[0090] The supercapacitor allocation ratio Ratio_SC can be calculated using one specific formula as follows:

[0091] Ratio_SC=0.6-0.1×Δ_ESR+0.05×λ_bat, that is, R_base=0.6, k1=0.1, k2=0.05.

[0092] The closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the work of each module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller via a CAN bus to achieve real-time data acquisition and control command issuance. The closed-loop control module uses a bidirectional DC / DC converter to connect the hybrid energy storage system and the elevator frequency converter, achieving efficient energy conversion through a PWM complementary mode.

[0093] During the operation of the elevator control system, the closed-loop control module monitors the system's operating status in real time, including passenger flow prediction errors, energy storage system aging rate, and grid fluctuations. When an anomaly is detected, the elevator control system automatically adjusts its optimization strategy. For example, when the prediction error is large, the elevator control system increases the EMD analysis window length; when the grid fluctuation is too high, the elevator control system increases the weight of β to optimize energy recovery efficiency.

[0094] Example 2

[0095] This embodiment focuses on application scenarios that extend elevator life, such as high-end office buildings and hospitals. It provides an elevator control system based on dynamic passenger flow prediction and lifespan collaborative control, including a passenger flow prediction module, an energy decomposition module, a lifespan optimization module, and a closed-loop control module.

[0096] The passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM to construct a spatiotemporal dataset by collecting data from inside the elevator car, call buttons on each floor, and turnstiles, thereby enabling refined prediction of short-term passenger flow density.

[0097] Specifically, the passenger flow prediction module comprises an input layer, a hidden layer, and an output layer. The hidden layer employs a structure that fuses one-dimensional convolutional and long short-term memory networks. Input data includes passenger flow density, building type encoding, and time period information. Data augmentation is performed using historical data and adding Gaussian noise to train and optimize the network (Conv-LSTM model). After training, the network outputs a predicted passenger flow density sequence by inputting the feature data of the current moment.

[0098] The energy decomposition module dynamically adjusts the window length W and wavelet entropy threshold Th of the Empirical Mode Decomposition (EMD) based on the predicted passenger flow density D. As a specific implementation, the window length W of the EMD algorithm can be dynamically adjusted according to the following formula:

[0099] W = 280 - 1.0 × D (unit: milliseconds)

[0100] In this embodiment, the base window length W0 is set to 280ms, and the adjustment coefficient k is set to 1.0. When the passenger flow density D is large, the window W is shortened to capture high-frequency components; when the passenger flow density D is small, the window W is lengthened to capture low-frequency components.

[0101] The wavelet entropy threshold Th is obtained by fitting historical data. One specific calculation formula is: Th = 0.82 - 0.004 × D, that is, fitting coefficients a = 0.82, b = 0.004.

[0102] The lifespan optimization module establishes a coupled model of the battery cycle life degradation factor λ_bat and the aging increment Δ_ESR of the supercapacitor's equivalent series resistance (ESR). An improved particle swarm optimization algorithm is used to optimize the weights of the three objective functions in real time. As a specific implementation method, the weight calculation formula is as follows:

[0103] α=0.4-0.008×λ_bat (i.e. α0=0.4, c1=0.008)

[0104] β = 0.25 (i.e., β0 = 0.25)

[0105] γ=0.35+0.008×Δ_ESR (i.e. γ0=0.35, c2=0.008)

[0106] The supercapacitor allocation ratio Ratio_SC can be calculated using one specific formula as follows:

[0107] Ratio_SC=0.5-0.15×Δ_ESR+0.03×λ_bat, that is, R_base=0.5, k1=0.15, k2=0.03.

[0108] The closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the work of each module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller via a CAN bus to achieve real-time data acquisition and control command issuance. The closed-loop control module uses a bidirectional DC / DC converter to connect the energy storage system and the elevator frequency converter, achieving efficient energy conversion through a PWM complementary mode.

[0109] During the operation of the elevator control system, the closed-loop control module monitors the system's operating status in real time, including passenger flow prediction errors, energy storage system aging rate, and grid fluctuations. When an anomaly is detected, the elevator control system automatically adjusts its optimization strategy. For example, when the prediction error is large, the elevator control system increases the EMD analysis window length; when the grid fluctuation is too high, the elevator control system increases the weight of β to optimize energy recovery efficiency.

[0110] Example 3

[0111] This embodiment focuses on application scenarios that improve energy recovery efficiency, such as shopping malls and subway stations. It provides an elevator control system based on dynamic passenger flow prediction and lifespan collaborative control, including a passenger flow prediction module, an energy decomposition module, a lifespan optimization module, and a closed-loop control module.

[0112] The passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM to construct a spatiotemporal dataset by collecting data from inside the elevator car, call buttons on each floor, and turnstiles, thereby enabling refined prediction of short-term passenger flow density.

[0113] Specifically, the passenger flow prediction system comprises an input layer, a hidden layer, and an output layer. The hidden layer employs a structure that fuses one-dimensional convolutional and long short-term memory networks. Input data includes passenger flow density, building type encoding, and time period information. Data augmentation is performed using historical data and adding Gaussian noise to train and optimize the network (Conv-LSTM model). After training, the system outputs a predicted passenger flow density sequence for the future, based on the input feature data for the current time period.

[0114] The energy decomposition module dynamically adjusts the window length W and wavelet entropy threshold Th of the Empirical Mode Decomposition (EMD) based on the predicted passenger flow density D. As a specific implementation, the window length W of the EMD algorithm can be dynamically adjusted according to the following formula:

[0115] W = 320 - 1.4 × D (unit: milliseconds)

[0116] In this embodiment, the base window length W0 is set to 320ms, and the adjustment coefficient k is set to 1.4. When the passenger flow density D is high, the window W is shortened to capture high-frequency components; when the passenger flow density D is low, the window W is lengthened to capture low-frequency components.

[0117] The wavelet entropy threshold Th is obtained by fitting historical data. One specific calculation formula is: Th = 0.88 - 0.006 × D, that is, fitting coefficients a = 0.88, b = 0.006.

[0118] The lifespan optimization module establishes a coupled model of the battery cycle life degradation factor λ_bat and the aging increment Δ_ESR of the supercapacitor's equivalent series resistance (ESR). An improved particle swarm optimization algorithm is used to optimize the weights of the three objective functions in real time. As a specific implementation method, the weight calculation formula is as follows:

[0119] α=0.45-0.015×λ_bat, that is, α0=0.45, c1=0.015.

[0120] β = 0.4, that is, β0 = 0.4.

[0121] γ=0.15+0.003×Δ_ESR, that is, γ0=0.15, c2=0.003.

[0122] The supercapacitor allocation ratio Ratio_SC can be calculated using one specific formula as follows:

[0123] Ratio_SC=0.7-0.05×Δ_ESR+0.08×λ_bat, that is, R_base=0.7, k1=0.05, k2=0.08.

[0124] The closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the work of each module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller via a CAN bus to achieve real-time data acquisition and control command issuance. The closed-loop control module uses a bidirectional DC / DC converter to connect the hybrid energy storage system and the elevator frequency converter, achieving efficient energy conversion through a PWM complementary mode.

[0125] During the operation of the elevator control system, the closed-loop control module monitors the system's operating status in real time, including passenger flow prediction errors, energy storage system aging rate, and grid fluctuations. When an anomaly is detected, the elevator control system automatically adjusts its optimization strategy. For example, when the prediction error is large, the system increases the EMD analysis window length; when the grid fluctuation is too high, the elevator control system increases the weight of β to optimize energy recovery efficiency.

[0126] Please see Figure 4 The present invention also discloses an elevator control method based on dynamic passenger flow prediction and lifespan coordinated control, applied to the aforementioned elevator control system based on dynamic passenger flow prediction and lifespan coordinated control, comprising the following steps:

[0127] Passenger flow forecasting: Based on the dynamic characteristics of elevator load, predict elevator passenger flow density in real time and output the elevator passenger flow forecast results;

[0128] Energy decomposition: Based on elevator passenger flow forecast results, the empirical mode decomposition parameters are dynamically adjusted, and the frequency domain boundary point is adjusted to achieve energy storage distribution;

[0129] Lifetime optimization: Establish a coupled aging model for energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives;

[0130] Closed-loop control: Based on passenger flow forecasts, energy storage allocation, and lifespan optimization information, the system performs coordinated control of each module and the hybrid energy storage system.

[0131] This invention also discloses an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described in any of the preceding claims. The electronic device of this invention can execute the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control of this invention, and can execute any combination of the steps of the method embodiments, possessing the corresponding functions and beneficial effects of the method.

[0132] This invention also discloses a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the steps of the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described in any of the preceding claims. The computer-readable storage medium of this invention can execute the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control of this invention, and can execute any combination of the steps of the method embodiments, possessing the corresponding functions and beneficial effects of the method.

[0133] In summary, the present invention has the following advantages and beneficial effects:

[0134] 1. This invention employs a spatiotemporal convolutional LSTM dynamic passenger flow prediction module, which integrates one-dimensional convolution and long short-term memory networks to achieve refined prediction of future short-term passenger flow density. It adapts to the dynamic characteristics of elevator load, solves the problem that traditional fixed filtering frequencies are difficult to cope with sudden passenger flow, and improves the response speed and energy efficiency of elevator control systems.

[0135] 2. This invention uses an adaptive frequency domain energy decomposition module to dynamically adjust the Empirical Mode Decomposition (EMD) window length and wavelet entropy threshold based on the predicted passenger flow density, thereby achieving adaptive calculation of the frequency domain boundary point. This overcomes the limitations of fixed filtering frequencies, solves the problem of high-frequency component response lag, fully leverages the advantages of supercapacitor's fast response, and improves the accuracy and efficiency of energy allocation.

[0136] 3. This invention establishes a coupling model between the battery cycle life decay factor and the aging increment of the equivalent series resistance of the supercapacitor. By improving the particle swarm optimization algorithm, the weights of multiple objectives are adjusted in real time to achieve life synergistic optimization. It comprehensively considers the coupled aging effect of energy storage components, solves the problem of large differences in life decay, and extends the overall life of the energy storage system.

[0137] 4. This invention integrates passenger flow forecasting with energy distribution control to form a complete closed-loop control. It uses the forecast results to guide the adjustment of power distribution strategies in real time, solving the energy efficiency loss caused by the disconnect between traditional "forecasting" and "control", and improving overall operating efficiency.

[0138] 5. This invention adaptively adjusts the boundary frequency and window length to achieve refined management of energy in different frequency bands, enabling rapid response to sudden passenger flow, smoothing load fluctuations, improving the dynamic response capability of the elevator control system, and meeting actual operational needs.

[0139] The above description is a detailed description of the preferred embodiments of the present invention. However, the embodiments are not intended to limit the scope of the patent application of the present invention. All equivalent changes or modifications made under the technical spirit disclosed in the present invention should fall within the patent scope covered by the present invention.

Claims

1. An elevator control system based on dynamic passenger flow prediction and lifespan coordinated control, characterized in that, It includes a passenger flow prediction module, an energy decomposition module, a lifespan optimization module, and a closed-loop control module; The passenger flow prediction module is used to predict elevator passenger flow density in real time based on the dynamic characteristics of elevator load, and output the elevator passenger flow prediction results. The energy decomposition module is used to dynamically adjust the empirical mode decomposition parameters and adjust the frequency domain boundary point based on the elevator passenger flow prediction results to realize the energy storage distribution. The lifetime optimization module is used to establish a coupled aging model of energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives. The closed-loop control module is used to coordinate the control of each module and the hybrid energy storage system based on passenger flow forecast results, energy storage allocation and lifespan optimization information.

2. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, The passenger flow prediction module is used to predict elevator passenger density in real time based on the dynamic characteristics of elevator load, and output the elevator passenger flow prediction results, specifically: The passenger flow prediction module uses a deep learning network based on spatiotemporal convolutional LSTM. By collecting data from inside the elevator car, the call buttons on each floor, and the turnstiles, it constructs a spatiotemporal dataset to achieve a refined prediction of the short-term passenger flow density. The passenger flow prediction module consists of an input layer, a hidden layer, and an output layer. The hidden layer adopts a structure that combines one-dimensional convolution and long short-term memory networks. The input data includes passenger flow density, building type coding, and time period information. The network is trained and optimized by using historical data and adding Gaussian noise for data augmentation. After the network is trained, it outputs the predicted passenger flow density sequence by inputting the feature data of the current moment.

3. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, The energy decomposition module is used to dynamically adjust the empirical mode decomposition parameters and frequency domain boundary points based on elevator passenger flow prediction results, thereby achieving energy storage allocation. Specifically: The energy decomposition module dynamically adjusts the window length and wavelet entropy threshold of the empirical mode decomposition based on the predicted passenger flow density, realizes the adaptive calculation of the frequency domain boundary point, and divides the frequency band into high-frequency and low-frequency parts through the boundary frequency mapping function. In the energy decomposition module, the window length W of the empirical mode decomposition algorithm is dynamically adjusted according to the predicted passenger flow density D, and the calculation formula is as follows: W = W0 - k·D; In the formula, W0 is the base window length; k is the adjustment coefficient; D is the predicted passenger flow density; when the passenger flow density is high, the window becomes shorter to capture high-frequency components; when the passenger flow density is low, the window becomes longer to capture low-frequency components. The wavelet entropy threshold Th is obtained by fitting historical data, and the calculation formula is as follows: Th = ab·D; In the formula, a and b are fitting coefficients.

4. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, The lifetime optimization module is used to establish a coupled aging model of energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of multi-objective lifetimes. Specifically: The lifespan optimization module establishes a coupled model of the battery cycle life decay factor λ_bat and the aging increment ΔESR of the supercapacitor's equivalent series resistance (ESR). The weights of the three optimization objectives are dynamically adjusted in real time through an improved particle swarm optimization algorithm to achieve collaborative optimization of multi-objective lifespan. The weights of the three optimization objectives are calculated using the following formulas: α = α0 - c1·λ_bat; β＝β0； γ = γ0 + c2·Δ_ESR; In the formula, α is used to minimize grid impact, β is used to maximize energy recovery efficiency, γ is used to minimize the aging rate of energy storage system, α0, β0, and γ0 are the basic weight values ​​of each optimization objective, and c1 and c2 are adjustment coefficients. The formula for calculating the supercapacitor power allocation ratio Ratio_SC is: Ratio_SC=R_base-k1·Δ_ESR+k2·λ_bat; In the formula, R_base is the basic allocation ratio, and k1 and k2 are fitting coefficients.

5. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, The closed-loop control module adopts a master-slave control structure, including a master controller and slave controllers. The master controller is responsible for coordinating the work of the passenger flow prediction module, energy decomposition module and lifespan optimization module, while the slave controllers are responsible for executing specific control tasks. The master controller communicates with each slave controller through the CAN bus to realize real-time data acquisition and control command issuance.

6. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, The closed-loop control module uses a bidirectional DC / DC converter to connect the hybrid energy storage system and the elevator frequency converter, and achieves efficient energy conversion through PWM complementary mode.

7. The elevator control system based on dynamic passenger flow prediction and lifespan coordinated control according to claim 1, characterized in that, During the operation of the elevator control system, the closed-loop control module monitors passenger flow prediction errors, energy storage system aging rate, and power grid fluctuation rate in real time. When an anomaly is detected, the elevator control system automatically adjusts and optimizes its strategy. When the prediction error is large, the elevator control system increases the length of the empirical mode decomposition analysis window. When the power grid fluctuation rate is high, the elevator control system increases the weight of maximizing energy recovery efficiency to optimize energy recovery efficiency.

8. An elevator control method based on dynamic passenger flow prediction and lifespan coordinated control, applied to the elevator control system based on dynamic passenger flow prediction and lifespan coordinated control as described in any one of claims 1 to 7, characterized in that, Includes the following steps: Passenger flow forecasting: Based on the dynamic characteristics of elevator load, predict elevator passenger flow density in real time and output the elevator passenger flow forecast results; Energy decomposition: Based on elevator passenger flow forecast results, the empirical mode decomposition parameters are dynamically adjusted, and the frequency domain boundary point is adjusted to achieve energy storage distribution; Lifetime optimization: Establish a coupled aging model for energy storage components, dynamically adjust the weights of optimization objectives in real time, and achieve collaborative optimization of the lifetime of multiple objectives; Closed-loop control: Based on passenger flow forecasts, energy storage allocation, and lifespan optimization information, the system performs coordinated control of each module and the hybrid energy storage system.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described in claim 8.

10. A storage medium, a computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the elevator control method based on dynamic passenger flow prediction and lifespan coordinated control as described in claim 8.

Images