An intelligent adaptive traction elevator power supply and energy storage system

Abstract

The present invention belongs to the technical field of elevator power supply. The present invention discloses an intelligent adaptive traction elevator power supply and electric energy storage system, including: a historical database module stores elevator operation data and corresponding date and weather information; an operation data processing module performs time series processing on historical data to obtain an operation characteristic time series matrix; an operation prediction model training module uses these data to train an elevator operation prediction model; a real-time operation prediction module uses the trained model to predict future elevator operation characteristics based on the current date and weather data; an energy recovery prediction module generates an energy recovery prediction time series matrix based on these prediction results; and an energy supply management module formulates an optimized energy storage device working strategy based on the predicted energy recovery situation and energy storage device status. This system effectively reduces the frequent charging and discharging of energy storage equipment through accurate prediction and optimization strategies, prolongs its service life, and improves energy utilization efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator power supply, and more specifically, to an intelligent adaptive traction elevator power supply and electric energy storage system. Background Art

[0002] In recent years, in order to improve energy efficiency, more and more elevators have been equipped with energy recovery devices. This device can recover energy when the elevator is heavily loaded going down or lightly loaded going up, and store the recovered energy for reuse, thereby significantly reducing the energy consumption of the elevator system.

[0003] However, the current elevator energy recovery system still faces some challenges in practical applications. The main problem is the lack of accurate prediction of the amount of energy recovery, which leads to problems such as low efficiency and shortened life of energy storage equipment. Specifically, due to the lack of accurate energy recovery prediction, the charging and discharging strategy of energy storage equipment often adopts simple threshold control or fixed time period control. This extensive management method easily leads to frequent charging and discharging of energy storage equipment, which not only reduces the service life of energy storage equipment, but also fails to fully utilize the recovered energy. At the same time, existing systems often cannot optimize energy distribution based on predicted energy recovery and elevator usage needs. This results in part of the recovered energy being wasted because it cannot be used in time, reducing the overall energy utilization efficiency. Summary of the invention

[0004] In order to overcome the problems of the prior art affecting the service life of the equipment and low energy utilization efficiency, the present invention proposes an intelligent adaptive traction elevator power supply and energy storage system to solve the above problems.

[0005] The present invention provides the following technical solutions:

[0006] An intelligent adaptive traction elevator power supply and energy storage system, comprising:

[0007] A historical database module, used to store historical data, the historical data including daily elevator operation data and corresponding date information and weather data;

[0008] The operation data processing module is used to perform time series processing on the elevator operation data in the historical data to obtain the operation characteristic time series matrix;

[0009] An operation prediction model training module is used to use historical data in a historical database to train an elevator operation prediction model for predicting an operation feature time series matrix based on date information and weather data;

[0010] The real-time operation prediction module is used to obtain real-time date information and weather data, and generate a predicted elevator operation feature time series matrix through the elevator operation prediction model;

[0011] An energy recovery prediction module, used to generate an energy recovery prediction timing matrix according to the predicted elevator operation characteristic timing matrix and a pre-trained energy recovery prediction model;

[0012] The energy supply management module formulates an optimized energy storage equipment working strategy schedule based on the energy recovery prediction timing matrix and the energy storage equipment status.

[0013] Preferably, the date data includes year, month, date, weekday and whether it is a working day label, the weather data includes temperature, wind speed, ultraviolet intensity and weather condition label, the elevator operation data includes operation direction, stop floor and real-time load, and the operation direction includes up and down.

[0014] Preferably, the time series processing includes dividing the elevator operation data into several time periods of equal length according to the daily time series, performing feature extraction on the elevator operation data of each time period to obtain corresponding operation features, and arranging the operation features of all time periods in sequence to obtain an operation feature time series matrix.

[0015] Preferably, extracting features from the elevator operation data in each time period to obtain corresponding operation features includes:

[0016] According to the change of the elevator running direction, the elevator running data within the time period is divided into multiple independent running sequences, where each running sequence represents the continuous running process of the elevator in a single direction;

[0017] The eigenvalue of each running sequence is calculated using the following formula:

[0018] ,

[0019] In the formula, represents the eigenvalue, Indicates the number of elevator operation data in the operation sequence. Indicates the first The stopping floors in the elevator operation data, Indicates the maximum floor, Indicates the first The real-time load value in the elevator operation data;

[0020] The calculated characteristic values ​​of the running sequence are divided into an upward group and a downward group according to the running direction in the running sequence;

[0021] The eigenvalues ​​of the upstream group and the downstream group are summed up to obtain the total upstream eigenvalue and the total downstream eigenvalue;

[0022] The uplink total eigenvalue and the downlink total eigenvalue are labeled with corresponding uplink and downlink labels and then combined into the running feature.

[0023] Preferably, the method of using historical data in a historical database to train an elevator operation prediction model for predicting an operation feature time series matrix according to date information and weather data comprises:

[0024] Divide historical data into training set and validation set, extract date information and weather data from historical data to construct feature vector as model input, and use the corresponding operation feature time series matrix as model output;

[0025] Design the elevator operation prediction model using a deep neural network architecture, initialize the model structure parameters, and set the model hyperparameters;

[0026] Iterative training is performed using the training set data. In each iterative training, the model prediction value is calculated through forward propagation, the error between the predicted value and the actual running feature time series matrix is ​​calculated, the model loss is calculated based on the predefined loss function, and the model parameters are updated using the backpropagation algorithm and optimizer to minimize the loss function.

[0027] Regularly use the validation set to evaluate model performance. If the performance of the model on the validation set reaches the preset threshold, the training is completed. Otherwise, the hyperparameters and model structure parameters are adjusted and the training process is repeated.

[0028] Finally, the best performing model on the validation set is saved as the elevator operation prediction model.

[0029] Preferably, generating the energy recovery prediction timing matrix according to the predicted elevator operation characteristic timing matrix and the pre-trained energy recovery prediction model comprises:

[0030] Decomposing the predicted elevator operation feature time series matrix into multiple independent operation features;

[0031] For each operating feature, a pre-trained energy recovery prediction model is used to predict and obtain a corresponding energy recovery value; wherein the energy recovery prediction model receives a single operating feature as input and outputs a corresponding energy recovery value;

[0032] The energy recovery values ​​corresponding to all operating characteristics are recombined according to the order of the original operating characteristics in the timing matrix to form an energy recovery prediction timing matrix.

[0033] Preferably, the pre-trained energy recovery prediction model includes: collecting operation characteristics in historical operation data and corresponding actual energy recovery data;

[0034] The collected data set is divided into a training set and a validation set. A machine learning algorithm is used to train the energy recovery prediction model on the training set with operating characteristics as input and actual energy recovery values ​​as output. The model performance is evaluated on the validation set, and the model parameters are optimized based on the evaluation results. When the model performance reaches the preset threshold, a trained energy recovery prediction model is obtained.

[0035] Preferably, formulating an optimized energy storage device working strategy timing table according to the energy recovery prediction timing matrix and the energy storage device state includes:

[0036] Acquire current status information of the energy storage device, the status information including current power level, ideal power level, maximum discharge power, minimum power capacity and maximum power capacity;

[0037] Extracting the predicted energy recovery amount for each time period from the energy recovery prediction time series matrix;

[0038] Based on the predicted energy recovery amount and the status information of the energy storage device in each time period, an optimization problem model is constructed;

[0039] Solve the constructed optimization problem model and generate a time series table of energy storage equipment working strategies.

[0040] Preferably, constructing the optimization problem model includes setting an objective function and constraint conditions;

[0041] The objective function is to minimize the sum of the absolute values ​​of the difference between the actual power of the energy storage device and the ideal power at the end of each time period;

[0042] The constraints include: the power of the energy storage device in any time period does not exceed the maximum power capacity, the power of the energy storage device in any time period does not fall below the minimum power capacity, and the discharge amount in each time period does not exceed the product of the maximum discharge power and the time period length;

[0043] The process of solving the optimization problem model and generating the energy storage device working strategy time sequence table includes:

[0044] The optimization problem model is solved using a quadratic programming algorithm to obtain the optimal discharge capacity in each time period;

[0045] The optimal discharge amount in each time period is organized into a time sequence table of the energy storage device working strategy.

[0046] The present invention provides an intelligent adaptive traction elevator power supply and electric energy storage system, which has the following beneficial effects:

[0047] Through the collaborative work of the operation data processing module, the operation prediction model training module and the real-time operation prediction module, the system can accurately predict the operation characteristics of the elevator. This prediction is no longer limited to simple threshold control or fixed time period control, but is based on a deep neural network model, taking into account a variety of influencing factors such as date information and weather data. The energy recovery prediction module further combines the operation characteristics with the actual energy recovery data to train an accurate energy recovery prediction model. This method significantly improves the accuracy of energy recovery prediction and provides a reliable basis for optimizing the energy storage equipment management strategy. Based on these accurate prediction results, the energy supply management module can formulate an optimized energy storage equipment working strategy schedule. This prediction-based optimization strategy effectively reduces the frequent charging and discharging of energy storage equipment and significantly extends the service life of energy storage equipment. At the same time, since the strategy is based on a forecast for a period of time in the future, rather than a simple real-time response, it can better balance short-term and long-term energy needs and further improve the efficiency of energy storage equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The module diagram of an intelligent adaptive traction elevator power supply and electric energy storage system of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0050] See also Figure 1 In this embodiment, an intelligent adaptive traction elevator power supply and energy storage system includes:

[0051] A historical database module, used to store historical data, the historical data including daily elevator operation data and corresponding date information and weather data;

[0052] The date data includes year, month, date, week and whether it is a working day label, the weather data includes temperature, wind speed, ultraviolet intensity and weather condition labels, and the elevator operation data includes operation direction, stop floor and real-time load, and the operation direction includes up and down.

[0053] In this embodiment, the implementation of the historical database module can be carried out according to the following steps: First, design and create a database structure suitable for storing the required historical data. This database structure should contain multiple interrelated tables to effectively store and manage elevator operation data, date information and weather data. Next, the operation data of the elevator can be collected by installing sensors and data acquisition equipment on the elevator to ensure that information such as the running direction, the stop floor and the real-time load can be accurately recorded. The acquisition of weather data can be achieved by establishing an API interface with a meteorological data service provider to regularly acquire and store information such as temperature, wind speed, ultraviolet intensity and weather conditions. Different weather conditions and different dates may cause changes in people's activity patterns, thereby affecting the use time distribution and overall usage of the elevator. Therefore, these characteristics can be used to predict the working conditions of the elevator.

[0054] The operation data processing module is used to perform time series processing on the elevator operation data in the historical data to obtain the operation characteristic time series matrix;

[0055] The time series processing includes dividing the elevator operation data into a number of equal-length time periods according to the daily time series, extracting features from the elevator operation data in each time period to obtain corresponding operation features, and arranging the operation features of all time periods in sequence to obtain an operation feature time series matrix.

[0056] The feature extraction of the elevator operation data in each time period to obtain the corresponding operation features includes:

[0057] According to the change of the elevator running direction, the elevator running data within the time period is divided into multiple independent running sequences, where each running sequence represents the continuous running process of the elevator in a single direction;

[0058] The eigenvalue of each running sequence is calculated using the following formula:

[0059] ,

[0060] In the formula, represents the eigenvalue, Indicates the number of elevator operation data in the operation sequence. Indicates the first The stopping floors in the elevator operation data, Indicates the maximum floor, Indicates the first The real-time load value in the elevator operation data;

[0061] The calculated characteristic values ​​of the running sequence are divided into an upward group and a downward group according to the running direction in the running sequence;

[0062] The eigenvalues ​​of the upstream group and the downstream group are summed up to obtain the total upstream eigenvalue and the total downstream eigenvalue;

[0063] The uplink total eigenvalue and the downlink total eigenvalue are labeled with corresponding uplink and downlink labels and then combined into the running feature.

[0064] In this embodiment, the operation data processing module can be implemented according to the following steps: First, extract the elevator operation data from the historical database. Then, divide the daily elevator operation data into several equal time periods in chronological order, for example, a day can be divided into 24 hours. For each time period, according to the change of the elevator operation direction, the elevator operation data in the time period is divided into multiple independent operation sequences. Each operation sequence represents the continuous operation process of the elevator in a single direction (upward or downward). Next, calculate the characteristic value of each operation sequence. This characteristic value comprehensively considers the passenger boarding and disembarking situation of the elevator during an upward or downward process, including the change of the stop floor and the real-time load. These factors are closely related to the energy recovery of the elevator, because the load change and running distance of the elevator directly affect the amount of recoverable energy. The calculated characteristic values ​​are divided into an upward group and a downward group according to the corresponding running direction. Then, the characteristic values ​​of the upward group and the downward group are summed respectively to obtain the total upward characteristic value and the total downward characteristic value. Finally, the total upward characteristic value and the total downward characteristic value are marked with corresponding direction labels to combine into the operation characteristics of the time period. The operation characteristics of all time periods are arranged in chronological order to form an operation characteristic time series matrix. This processing method can effectively transform the original elevator operation data into time series characteristics that reflect the elevator usage pattern and energy recovery potential, providing valuable input data for subsequent prediction models.

[0065] An operation prediction model training module is used to use historical data in a historical database to train an elevator operation prediction model for predicting an operation feature time series matrix based on date information and weather data;

[0066] The elevator operation prediction model for predicting the operation feature time series matrix based on date information and weather data using the historical data in the historical database includes:

[0067] Divide historical data into training set and validation set, extract date information and weather data from historical data to construct feature vector as model input, and use the corresponding operation feature time series matrix as model output;

[0068] Design the elevator operation prediction model using a deep neural network architecture, initialize the model structure parameters, and set the model hyperparameters;

[0069] Iterative training is performed using the training set data. In each iterative training, the model prediction value is calculated through forward propagation, the error between the predicted value and the actual running feature time series matrix is ​​calculated, the model loss is calculated based on the predefined loss function, and the model parameters are updated using the backpropagation algorithm and optimizer to minimize the loss function.

[0070] Regularly use the validation set to evaluate model performance. If the performance of the model on the validation set reaches the preset threshold, the training is completed. Otherwise, the hyperparameters and model structure parameters are adjusted and the training process is repeated.

[0071] Finally, the best performing model on the validation set is saved as the elevator operation prediction model.

[0072] In this embodiment, the implementation of the operation prediction model training module can be carried out according to the following steps: First, extract the required historical data from the historical database, including date information, weather data and the corresponding operation feature time series matrix. These data are randomly divided into a training set and a validation set according to a certain ratio (such as 8:2). For each piece of data, the date information and weather data are combined into a feature vector as the input of the model, and the corresponding operation feature time series matrix is ​​used as the output of the model. Next, a deep neural network architecture is designed as an elevator operation prediction model. This network may contain multiple fully connected layers, convolutional layers or recurrent neural network layers, and the specific structure is determined according to the data characteristics and the requirements of the prediction task. Initialize the structural parameters such as the weight and bias of the model, and set hyperparameters such as the learning rate, batch size, and number of training rounds. Then start iterative training using the training set data. In each iteration, the model calculates the predicted value through forward propagation, compares the predicted value with the actual operation feature time series matrix, and calculates the error. The model loss is calculated based on a pre-defined loss function (such as mean square error), and then the gradient is calculated using the back propagation algorithm, and the model parameters are updated through an optimizer (such as Adam) to minimize the loss function. During the training process, the validation set is used regularly to evaluate the model performance. If the performance of the model on the validation set (such as prediction accuracy or mean square error) reaches the preset threshold, the training is considered complete. If the threshold is not reached, it may be necessary to adjust hyperparameters such as learning rate and batch size, or modify the network structure and restart the training process. Finally, the model with the best performance on the validation set is selected as the final elevator operation prediction model, and the model parameters are saved for subsequent use.

[0073] The real-time operation prediction module is used to obtain real-time date information and weather data, and generate a predicted elevator operation feature time series matrix through the elevator operation prediction model;

[0074] In this embodiment, the implementation of the real-time operation prediction module can be carried out according to the following steps: First, obtain the current date information and current weather data. These real-time acquired date information and weather data will be combined into a feature vector as the input of the prediction model. Next, call the previously trained elevator operation prediction model to generate a predicted elevator operation feature time series matrix based on the input date and weather features. This time series matrix contains the predicted values ​​of the elevator operation features for each time period in the future (such as 24 hours).

[0075] An energy recovery prediction module, used to generate an energy recovery prediction timing matrix according to the predicted elevator operation characteristic timing matrix and a pre-trained energy recovery prediction model;

[0076] Decomposing the predicted elevator operation feature time series matrix into multiple independent operation features;

[0077] For each operating feature, a pre-trained energy recovery prediction model is used to predict and obtain a corresponding energy recovery value; wherein the energy recovery prediction model receives a single operating feature as input and outputs a corresponding energy recovery value;

[0078] The energy recovery values ​​corresponding to all operating characteristics are recombined according to the order of the original operating characteristics in the timing matrix to form an energy recovery prediction timing matrix.

[0079] The pre-trained energy recovery prediction model includes: collecting operation characteristics in historical operation data and corresponding actual energy recovery data;

[0080] The collected data set is divided into a training set and a validation set. A machine learning algorithm is used to train the energy recovery prediction model on the training set with operating characteristics as input and actual energy recovery values ​​as output. The model performance is evaluated on the validation set, and the model parameters are optimized based on the evaluation results. When the model performance reaches the preset threshold, a trained energy recovery prediction model is obtained.

[0081] In this embodiment, the implementation of the energy recovery prediction module can be carried out according to the following steps: First, receive the predicted elevator operation feature time series matrix from the real-time operation prediction module. Decompose this time series matrix into multiple independent operation features, each of which represents the operation status of the elevator in a time period. Then, call the pre-trained energy recovery prediction model. This model is obtained through historical data training, and its training process includes collecting the operation features in the historical operation data and the corresponding actual energy recovery data, dividing the data set into a training set and a validation set, using a machine learning algorithm (such as a random forest or a neural network) to train on the training set, and evaluating the model performance on the validation set. During the model training process, the parameters are continuously optimized until the performance reaches a preset threshold. For each decomposed operation feature, it is input into the energy recovery prediction model, and the model will output the corresponding predicted energy recovery value. This process is repeated for all operation features. Finally, all the predicted energy recovery values ​​obtained are recombined according to the order of the original operation features in the time series matrix to form an energy recovery prediction time series matrix. This matrix contains the predicted energy recovery value for each time period in the future. In this way, the energy recovery prediction module can generate corresponding energy recovery predictions based on the predicted elevator operation conditions, providing an important basis for subsequent energy supply management.

[0082] The energy supply management module formulates an optimized energy storage equipment working strategy schedule based on the energy recovery prediction timing matrix and the energy storage equipment status.

[0083] The step of formulating an optimized energy storage device working strategy timing table according to the energy recovery prediction timing matrix and the energy storage device state includes:

[0084] Acquire current status information of the energy storage device, the status information including current power level, ideal power level, maximum discharge power, minimum power capacity and maximum power capacity;

[0085] Extracting the predicted energy recovery amount for each time period from the energy recovery prediction time series matrix;

[0086] Based on the predicted energy recovery amount and the status information of the energy storage device in each time period, an optimization problem model is constructed;

[0087] Solve the constructed optimization problem model and generate a time series table of energy storage equipment working strategies.

[0088] The construction of the optimization problem model includes setting the objective function and constraint conditions;

[0089] The objective function is to minimize the sum of the absolute values ​​of the difference between the actual power of the energy storage device and the ideal power at the end of each time period;

[0090] The constraints include: the power of the energy storage device in any time period does not exceed the maximum power capacity, the power of the energy storage device in any time period does not fall below the minimum power capacity, and the discharge amount in each time period does not exceed the product of the maximum discharge power and the time period length;

[0091] The process of solving the optimization problem model and generating the energy storage device working strategy time sequence table includes:

[0092] The optimization problem model is solved using a quadratic programming algorithm to obtain the optimal discharge capacity in each time period;

[0093] The optimal discharge amount in each time period is organized into a time sequence table of the energy storage device working strategy.

[0094] In this embodiment, the implementation of the energy supply management module can be carried out according to the following steps: First, the current state information of the energy storage device is obtained, including the current power level, the ideal power level, the maximum discharge power, the minimum capacity and the maximum capacity. This information can be obtained by communicating with the real-time monitoring system of the energy storage device. At the same time, the predicted energy recovery amount of each time period is extracted from the energy recovery prediction timing matrix. Next, an optimization problem model is constructed based on this information. The objective function of the model is set to minimize the sum of the absolute values ​​of the difference between the actual power of the energy storage device and the ideal power at the end of each time period, so that the power level of the energy storage device can be as close to the ideal state as possible. The constraints include ensuring that the power of the energy storage device in any time period does not exceed the maximum capacity, is not less than the minimum capacity, and the discharge amount in each time period does not exceed the product of the maximum discharge power and the length of the time period. These constraints ensure the safe operation of the energy storage device. Then, the quadratic programming algorithm is used to solve the optimization problem model. The solution process usually involves multiple iterative calculations until the optimal solution that meets all conditions is found. The solution result can give the optimal discharge amount for each time period. Finally, these optimal discharge amounts are organized into a time sequence table of energy storage device working strategies in chronological order. This time table specifies in detail the charging and discharging behaviors that the energy storage device should take in each time period in the future. In this way, the energy supply management module can control the operation of the energy storage battery in a planned manner, increase the service life of the energy storage unit and improve energy utilization efficiency.

[0095] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0096] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0097] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent adaptive traction elevator power supply and energy storage system, characterized in that: include: A historical database module, used to store historical data, the historical data including daily elevator operation data and corresponding date information and weather data; The operation data processing module is used to perform time series processing on the elevator operation data in the historical data to obtain the operation characteristic time series matrix; An operation prediction model training module is used to use historical data in a historical database to train an elevator operation prediction model for predicting an operation feature time series matrix based on date information and weather data; The real-time operation prediction module is used to obtain real-time date information and weather data, and generate a predicted elevator operation feature time series matrix through the elevator operation prediction model; An energy recovery prediction module, used to generate an energy recovery prediction timing matrix according to the predicted elevator operation characteristic timing matrix and a pre-trained energy recovery prediction model; The energy supply management module formulates an optimized energy storage equipment working strategy schedule based on the energy recovery prediction timing matrix and the energy storage equipment status; The time series processing includes dividing the elevator operation data into a number of equal-length time periods according to the daily time series, extracting features from the elevator operation data in each time period to obtain corresponding operation features, and arranging the operation features of all time periods in order to obtain an operation feature time series matrix; The feature extraction of the elevator operation data in each time period to obtain the corresponding operation features includes: According to the change of the elevator running direction, the elevator running data within the time period is divided into multiple independent running sequences, where each running sequence represents the continuous running process of the elevator in a single direction; The eigenvalue of each running sequence is calculated using the following formula: In the formula, En represents the characteristic value, m represents the number of elevator operation data in the operation sequence, and H i represents the stopping floor in the operation data of the i-th elevator in the operation sequence, MH represents the maximum floor, W i Indicates the value of the real-time load in the operation data of the i-th elevator in the operation sequence; The calculated characteristic values ​​of the running sequence are divided into an upward group and a downward group according to the running direction in the running sequence; The eigenvalues ​​of the upstream group and the downstream group are summed up to obtain the total upstream eigenvalue and the total downstream eigenvalue; The uplink total eigenvalue and the downlink total eigenvalue are labeled with corresponding uplink and downlink labels and then combined into the running feature.

2. The intelligent adaptive traction elevator power supply and energy storage system according to claim 1, characterized in that: The date information includes the year, month, date, week and whether it is a working day label, the weather data includes the temperature, wind speed, ultraviolet intensity and weather condition label, the elevator operation data includes the operation direction, stop floor and real-time load, and the operation direction includes up and down.

3. The intelligent adaptive traction elevator power supply and energy storage system according to claim 2, characterized in that: The elevator operation prediction model for predicting the operation feature time series matrix based on date information and weather data using the historical data in the historical database includes: Divide historical data into training set and validation set, extract date information and weather data from historical data to construct feature vector as model input, and use the corresponding operation feature time series matrix as model output; Design the elevator operation prediction model using a deep neural network architecture, initialize the model structure parameters, and set the model hyperparameters; Iterative training is performed using the training set data. In each iterative training, the model prediction value is calculated through forward propagation, the error between the predicted value and the actual running feature time series matrix is ​​calculated, the model loss is calculated based on the predefined loss function, and the model parameters are updated using the backpropagation algorithm and optimizer to minimize the loss function. Regularly use the validation set to evaluate model performance. If the performance of the model on the validation set reaches the preset threshold, the training is completed. Otherwise, the hyperparameters and model structure parameters are adjusted and the training process is repeated. Finally, the best performing model on the validation set is saved as the elevator operation prediction model.

4. The intelligent adaptive traction elevator power supply and energy storage system according to claim 3, characterized in that: The step of generating the energy recovery prediction timing matrix according to the predicted elevator operation characteristic timing matrix and the pre-trained energy recovery prediction model comprises: Decomposing the predicted elevator operation feature time series matrix into multiple independent operation features; For each operating feature, a pre-trained energy recovery prediction model is used to predict and obtain a corresponding energy recovery value; wherein the energy recovery prediction model receives a single operating feature as input and outputs a corresponding energy recovery value; The energy recovery values ​​corresponding to all operating characteristics are recombined according to the order of the original operating characteristics in the timing matrix to form an energy recovery prediction timing matrix.

5. The intelligent adaptive traction elevator power supply and energy storage system according to claim 4, characterized in that: The pre-trained energy recovery prediction model includes: collecting operation characteristics in historical operation data and corresponding actual energy recovery data; The collected data set is divided into a training set and a validation set. A machine learning algorithm is used to train the energy recovery prediction model on the training set with operating characteristics as input and actual energy recovery values ​​as output. The model performance is evaluated on the validation set, and the model parameters are optimized based on the evaluation results. When the model performance reaches the preset threshold, a trained energy recovery prediction model is obtained.

6. The intelligent adaptive traction elevator power supply and energy storage system according to claim 5, characterized in that: The step of formulating an optimized energy storage device working strategy timing table according to the energy recovery prediction timing matrix and the energy storage device state includes: Acquire current status information of the energy storage device, the status information including current power level, ideal power level, maximum discharge power, minimum power capacity and maximum power capacity; Extracting the predicted energy recovery amount for each time period from the energy recovery prediction time series matrix; Based on the predicted energy recovery amount and the status information of the energy storage device in each time period, an optimization problem model is constructed; Solve the constructed optimization problem model and generate a time series table of energy storage equipment working strategies.

7. The intelligent adaptive traction elevator power supply and energy storage system according to claim 6, characterized in that: The construction of the optimization problem model includes setting the objective function and constraint conditions; The objective function is to minimize the sum of the absolute values ​​of the difference between the actual power of the energy storage device and the ideal power at the end of each time period; The constraints include: the power of the energy storage device in any time period does not exceed the maximum power capacity, the power of the energy storage device in any time period does not fall below the minimum power capacity, and the discharge amount in each time period does not exceed the product of the maximum discharge power and the time period length; The process of solving the constructed optimization problem model and generating a time sequence table of the energy storage device working strategy includes: The optimization problem model is solved using a quadratic programming algorithm to obtain the optimal discharge capacity in each time period; The optimal discharge amount in each time period is organized into a time sequence table of the energy storage device working strategy.

Images