Multi-door elevator car control system
By constructing the sample set of elevators and using neural network algorithms to generate control models, the problem that the multi-door elevator system cannot effectively deal with service needs on different floors is solved, and accurate prediction of elevator usage patterns and improvement of elevator operation efficiency is achieved.
Patent Information
- Application Number
- CN202510496680.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-21
- Publication Date
- 2025-06-17
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing multi-door elevator system cannot effectively handle the dynamic service needs of different floors, lacks in-depth learning and prediction of passenger behavior patterns, and cannot flexibly adjust the elevator scheduling strategy to adapt to changing usage needs.
The elevator is constructed using sample sets and iteratively trained using neural network algorithms to generate multiple weak prediction models. By weighting and combining these models, the control model is obtained, and the elevator operation strategy is dynamically adjusted to meet passenger needs.
Accurate prediction of elevator usage patterns is achieved, passenger waiting time is reduced, elevator usage efficiency is improved, elevator parking strategy is optimized, and overall operation efficiency is improved.
Smart Images
Figure CN120156972A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent elevator control systems, and more particularly, to a multi-door elevator car control system. Background Art
[0002] Existing elevator systems, especially multi-door elevator systems, are very common in high-rise buildings and provide a convenient vertical transportation method for passengers. However, with the increase in building height and passenger flow, traditional elevator systems face problems of low efficiency and long waiting times. In terms of technical principles, these systems usually rely on fixed scheduling algorithms, such as First-Come-First-Served (FCFS) or Shortest Waiting Time First (SWT). Although these algorithms are simple, they often cannot adapt to the dynamically changing passenger demands and building structures.
[0003] In a multi-door elevator system, the elevator car needs to stop at multiple floors, which increases the complexity of scheduling. Existing systems often do not take into account the specific needs of passengers and the diversity of elevator usage patterns, resulting in uneven resource allocation and low efficiency. In addition, traditional elevator control systems lack the ability to learn and adapt to passenger behavior in real time, and cannot effectively predict passenger flow during peak hours and on specific floors, thus unable to make optimized scheduling in advance.
[0004] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: existing elevator control systems cannot effectively handle the dynamic changes in service demands of multi-door elevators on different floors, lack in-depth learning and prediction of passenger behavior patterns, and cannot flexibly adjust elevator scheduling strategies to adapt to the changing usage demands. Summary of the Invention
[0005] The present invention provides a multi-door elevator car control system, including:
[0006] Based on the same building, an elevator usage sample set is constructed, and the elevator usage sample set includes a car usage target sample set of a multi-door elevator and a car usage auxiliary sample set of a single-door elevator; a floor service area set is constructed according to the positions of the floors where the multi-door elevator stops in the single-door elevator;
[0007] Based on the neural network algorithm, the elevator usage sample set is iteratively trained to obtain multiple weak prediction models; wherein each time during training, according to the prediction error rate of the current weak prediction model for the target sample set, the weight of the current weak prediction model and the sample weights of the target sample set are set, and in combination with the usage conditions and confidence probabilities of the samples in the auxiliary sample set, the sample weights and original labels of the auxiliary sample set are updated, and the next weak prediction model is trained according to the updated elevator usage sample set;
[0008] The control model is obtained by weighted combination of each weak prediction model;
[0009] Obtain the car usage data of the single - opening elevator again as the dataset to be controlled, input it into the control model, map the data to be controlled predicted as a multi - opening elevator into the floor service area set, and predict the multi - opening elevator usage demand of the single - opening elevator.
[0010] Further, each sample in the elevator usage sample set includes a usage feature vector and an original label, where the usage feature vector includes: waiting time, number of passengers, number of floors traveled, passenger age, riding period, passenger gender, whether the passenger carries large items, and passenger floor arrival rate; the original label of the target sample set includes: multi - opening elevator, single - opening elevator; the original label of the auxiliary sample set includes: single - opening elevator.
[0011] Further, the prediction error rate of the current weak prediction model for the target sample set is the sum of the sample weights of the samples in the target sample set where the predicted label and the original label are inconsistent.
[0012] Further, the weight of the current weak prediction model is set by the following formula:
[0013]
[0014] The sample weight of the target sample set is updated by the following formula:
[0015]
[0016] where, α k is the weight of the k - th weak prediction model, e k is the prediction error rate of the k - th weak prediction model for the target sample set, is the sample weight of the i - th sample in the target sample set before the training of the k - th weak prediction model, y i is the original label of the i - th sample in the target sample set, is the predicted label of the i - th sample in the target sample set in the k - th weak prediction model.
[0017] Further, combining the usage situation and confidence probability of the samples in the auxiliary sample set, updating the sample weight and original label of the auxiliary sample set includes:
[0018] Update the sample weights of the samples with the usage label of single - opening elevator in the auxiliary sample set;
[0019] Use the samples with the label of multi - door elevator in the auxiliary sample set as the samples to be processed. Obtain the sample set in the floor service area predicted by the current weak prediction model for each sample to be processed, and calculate the multi - door elevator confidence probability of each sample to be processed. After sorting the samples to be processed in descending order of the multi - door elevator confidence probability, update the original label of the samples to be processed with the top Z multi - door elevator confidence probabilities to multi - door elevator, and update the sample weights of all samples to be processed, where Z ∈ {1, 2,..., Z max - 1}, and Z max is the number of samples to be processed.
[0020] Furthermore, update the sample weights of the samples with the label of single - door elevator in the auxiliary sample set through the following formula:
[0021]
[0022] where, is the sample weight of the jn - th sample predicted as a single - door elevator in the k - th weak prediction model in the auxiliary sample set, γ is the learning rate parameter, is the indicator function. When the predicted label is not equal to the true label y jn it takes the value of 1, otherwise 0.
[0023] Furthermore, the multi - door elevator confidence probability of each sample to be processed is the proportion of the number of samples with the original label of multi - door elevator in the sample set where each sample to be processed is located.
[0024] Furthermore, update the sample weights of all samples to be processed through the following formula:
[0025]
[0026] where, is the sample weight of the ju - th sample to be processed in the k - th weak prediction model in the auxiliary sample set, δ is the confidence probability adjustment factor, and Z ju is the ranking of the multi - door elevator confidence probability of the ju - th sample to be processed.
[0027] These formulas provide more complex mathematical descriptions to adapt to the characteristics of the multi - door elevator car control system.
[0028] Furthermore, each piece of data in the dataset to be controlled includes the starting and ending floors requested by the passenger, the usage feature vector, and the original label; mapping the data to be controlled predicted as a multi - door elevator into the set of floor service areas, and predicting the multi - door elevator usage demand includes:
[0029] For the data to be controlled predicted as a multi - door elevator, successively obtain the floor service areas where each starting point and ending point are located according to the starting and ending floors of each piece of data to be controlled; increment the number of stops of the multi - door elevator corresponding to the floor service area by 1;
[0030] Summarize the number of stops of the multi - door elevator to predict the usage demand of the multi - door elevator.
[0031] Further, the system includes:
[0032] A data processing module, configured to construct an elevator usage sample set based on the same building, where the elevator usage sample set includes a car usage target sample set of multi - door elevators and a car usage auxiliary sample set of single - door elevators; construct a floor service area set according to the floor positions where multi - door elevators stop in single - door elevators; and, obtain the car usage data of single - door elevators again as the data set to be controlled;
[0033] A model training module, configured to iteratively train the elevator usage sample set based on a neural network algorithm to obtain multiple weak prediction models; where each time of training, set the weight of the current weak prediction model and update the sample weights of the target sample set according to the prediction error rate of the current weak prediction model for the target sample set, and combine the usage situation and confidence probability of the samples in the auxiliary sample set to update the sample weights and original labels of the auxiliary sample set, and train the next weak prediction model according to the updated elevator usage sample set; weighted - combine each weak prediction model to obtain a control model;
[0034] A usage demand prediction module, configured to input the data set to be controlled into the control model, map the data to be controlled predicted as a multi - door elevator into the floor service area set, and predict the usage demand of the multi - door elevator of the single - door elevator.
[0035] According to the above - mentioned embodiments of the present invention, it has at least the following beneficial effects: By constructing an elevator usage sample set and using a neural network algorithm to iteratively train the sample set, the multi - door elevator car control system can accurately predict the elevator usage pattern. The system can dynamically adjust the operation strategy of the elevator according to the prediction result, thereby reducing the waiting time of passengers and improving the usage efficiency of the elevator. In addition, by weighted - combining multiple weak prediction models, the system can improve the accuracy and robustness of the prediction, ensuring optimal elevator scheduling in different usage scenarios.
[0036] The system can further improve the performance of the prediction model by updating the sample weights and original labels of the auxiliary sample set. In this way, the system can more accurately identify the usage requirements of multi-opening elevators for single-opening elevators, thereby optimizing the elevator stop strategy, reducing unnecessary stop times, and improving the overall operation efficiency. At the same time, by updating the dataset to be controlled in real time, the system can quickly respond to changes in passenger demands and further enhance the passenger riding experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of example and not limitation, wherein:
[0038] Figure 1 is a schematic flowchart of a multi-opening elevator car control system provided by an embodiment of the present invention;
[0039] Figure 2 is a schematic structural diagram of a multi-opening elevator car control system provided by an embodiment of the present invention;
[0040] Figure 3 Schematically shows a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0041] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are given only to enable those skilled in the art to better understand and then implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to be able to convey the scope of the present invention fully to those skilled in the art.
[0042] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, equipment, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.
[0043] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.
[0044] The following refers to Figure 1 , Figure 1 is a schematic flowchart of a multi-opening elevator car control system provided by an embodiment of the present invention. As Figure 1 shown, a multi-opening elevator car control system 100 includes:
[0045] Step 101, based on the same building, construct an elevator usage sample set, where the elevator usage sample set includes a car usage target sample set for multi - door elevators and a car usage auxiliary sample set for single - door elevators; construct a floor service area set according to the floor positions where multi - door elevators stop in single - door elevators;
[0046] Step 102, perform iterative training on the elevator usage sample set based on the neural network algorithm to obtain multiple weak prediction models; where each time during training, according to the prediction error rate of the current weak prediction model for the target sample set, set the weight of the current weak prediction model and update the sample weights of the target sample set, and combine the usage situation and confidence probability of the samples in the auxiliary sample set to update the sample weights and original labels of the auxiliary sample set, and train the next weak prediction model according to the updated elevator usage sample set;
[0047] Step 103, obtain a control model by weighted combination of each weak prediction model;
[0048] Step 104, obtain the car usage data of the single - door elevator again as the data set to be controlled, input it into the control model, map the data to be controlled predicted as a multi - door elevator into the floor service area set, and predict the multi - door elevator usage demand of the single - door elevator.
[0049] It should be noted that the core of the multi - door elevator car control system lies in predicting and optimizing the elevator usage efficiency by constructing and training a sample set. This involves collecting elevator usage data, constructing a model, and improving the prediction accuracy through algorithm iteration.
[0050] Specifically, the elevator usage sample set includes a car usage target sample set for multi - door elevators and a car usage auxiliary sample set for single - door elevators. The target sample set mainly contains the usage data of multi - door elevators, while the auxiliary sample set includes the usage data of single - door elevators and is used to assist in training the model. Each sample in the sample set contains a usage feature vector and an original label. The usage feature vector may include information such as waiting time and number of passengers, and the original label indicates whether the sample belongs to a multi - door elevator or a single - door elevator.
[0051] Preferably, in the selection of the usage feature vector, the floor arrival rate of passengers can be further included, which helps the model more accurately predict the destinations of passengers. At the same time, when constructing the floor service area set, a dynamic division method based on passenger flow and elevator usage frequency can be adopted to adapt to the changes in usage demands in different time periods and on different floors. In addition, the neural network algorithm can select an ensemble learning method, such as random forest or gradient boosting machine, to improve the generalization ability and prediction accuracy of the model.
[0052] In some embodiments, each sample in the elevator usage sample set includes a usage feature vector and an original label, where the usage feature vector includes: waiting time, number of passengers, number of floors traveled, passenger age, riding period, passenger gender, whether the passenger carries large items, and passenger floor arrival rate, and the original labels of the target sample set include: multi - door elevator, single - door elevator; the original labels of the auxiliary sample set include: single - door elevator.
[0053] Specifically, each sample in the elevator usage sample set contains a usage feature vector and an original label. The usage feature vector may include information such as waiting time, number of passengers, number of floors traveled, passenger age, riding period, passenger gender, whether the passenger carries large items, and passenger floor arrival rate. These feature vectors reflect the behavior patterns of passengers using the elevator and the operating status of the elevator. The original label specifies whether the sample belongs to a multi - door elevator or a single - door elevator, which is crucial information for model training and prediction.
[0054] Preferably, when constructing the usage feature vector, more refined feature engineering methods can be adopted. For example, the riding period can be subdivided into morning rush hour, evening rush hour, and other periods to more accurately capture the elevator usage patterns at different times. At the same time, for the feature of whether the passenger carries large items, it can further include the type and quantity of large items carried, so that the model can better understand and predict the elevator usage requirements in different situations. In addition, for the processing of the original label, a multi - label learning mechanism can be introduced to allow a sample to have multiple labels simultaneously to adapt to the complex and changeable elevator usage scenarios.
[0055] In some embodiments, the prediction error rate of the current weak prediction model for the target sample set is the sum of the weights of the samples in the target sample set where the predicted label is inconsistent with the original label.
[0056] Specifically, the prediction error rate refers to the sum of the weights of the samples in the target sample set where the predicted label is inconsistent with the original label. This means that if the prediction result of a sample does not match the actual result, then the weight of this sample will be included in the error rate. The weight is usually proportional to the importance of the sample. Therefore, samples with larger weights contribute more to the error rate. This calculation method ensures that the model can focus on those samples that have a greater impact on the prediction result during the training process.
[0057] Preferably, when calculating the prediction error rate, the concept of weighted average can be introduced, where the weight of each sample depends not only on the accuracy of its prediction but also can be adjusted according to other attributes of the sample, such as the number of passengers or the waiting time. In addition, a strategy of dynamically adjusting the weights can be adopted, that is, after each iterative training, the sample weights are updated according to the performance of the model to further improve the prediction accuracy of the model. Regularization terms can also be considered to prevent the model from overfitting and improve its generalization ability by restricting the model complexity. These methods can all be used as alternative or supplementary solutions for calculating the prediction error rate.
[0058] In some embodiments, the weight of the current weak prediction model is set by the following formula:
[0059]
[0060] The sample weights of the target sample set are updated by the following formula:
[0061]
[0062] where α k is the weight of the k-th weak prediction model, e k is the prediction error rate of the k-th weak prediction model for the target sample set, is the sample weight of the i-th sample in the target sample set before the training of the k-th weak prediction model, y i is the original label of the i-th sample in the target sample set, is the predicted label of the i-th sample in the target sample set in the k-th weak prediction model.
[0063] Specifically, the formula for setting the weight of the current weak prediction model is where ∈ is the prediction error rate of the i-th weak prediction model for the target sample set. This formula indicates that if the prediction error rate is lower, the weight of the model is higher, so that this model can be given greater influence in the subsequent model combination. The formula for updating the sample weights of the target sample set is where is the sample weight of the j-th sample before the training of the i-th weak prediction model, y j is the original label of the sample, is the predicted label, and α is the learning rate parameter. This update formula takes into account the prediction correctness of the sample. If the prediction is correct, the sample weight is increased; otherwise, it is decreased.
[0064] Preferably, when setting the weights of the weak prediction model, more factors can be considered, such as the complexity of the model, training time, etc., to achieve a more comprehensive model evaluation. In addition, for the update of sample weights, a more complex adjustment mechanism can be introduced, such as considering the historical prediction records of samples or dynamically adjusting according to the difficulty of samples. These methods can further improve the robustness and prediction accuracy of the model. Nonlinear weight adjustment strategies can also be considered to better reflect the importance and prediction difficulty of samples.
[0065] In some embodiments, updating the sample weights and original labels of the auxiliary sample set by combining the usage situation and confidence probability of the samples in the auxiliary sample set includes:
[0066] Updating the sample weights of the samples with the label of single-opening elevator in the auxiliary sample set;
[0067] Taking the samples with the label of multi-opening elevator in the auxiliary sample set as samples to be processed, obtaining the sample set in the floor service area predicted by the current weak prediction model for each sample to be processed, and calculating the multi-opening elevator confidence probability of each sample to be processed; after sorting the samples to be processed in descending order of the multi-opening elevator confidence probability, updating the original labels of the top Z samples to be processed with the multi-opening elevator confidence probability to multi-opening elevator, and updating the sample weights of all samples to be processed, where Z ∈ {1, 2,..., Z max -1}, and Z max is the number of samples to be processed.
[0068] Specifically, this process includes two main parts: First, updating the sample weights of the samples with the label of single-opening elevator in the auxiliary sample set; Second, taking the samples with the label of multi-opening elevator in the auxiliary sample set as samples to be processed, obtaining the sample set in the floor service area predicted by the current weak prediction model for each sample to be processed, and calculating the multi-opening elevator confidence probability of each sample to be processed. Here, the confidence probability refers to the credibility of the model predicting the sample as a multi-opening elevator, which can be calculated based on the prediction results of the model and the actual usage situation of the sample.
[0069] Preferably, for the update of sample weights in the auxiliary sample set, a more refined algorithm can be used to adjust the weights. For example, the weights can be dynamically adjusted based on the historical prediction accuracy and usage frequency of the samples. In addition, for the calculation of the multi-opening elevator confidence probability, more features can be introduced, such as the floor arrival pattern of passengers, the operation efficiency of the elevator, etc., to improve the accuracy of the confidence probability. The integration methods in machine learning, such as random forest or gradient boosting decision tree, can also be considered to integrate the prediction results of multiple models to improve the overall prediction accuracy.
[0070] In some embodiments, the sample weights of the samples with the label of single - door elevator in the auxiliary sample set are updated by the following formula:
[0071]
[0072] where is the sample weight of the \(j_n\) - th sample predicted as a single - door elevator in the \(k\) - th weak prediction model in the auxiliary sample set, \(\gamma\) is the learning rate parameter, is the indicator function, when the predicted label is not equal to the true label \(y\) jn , its value is 1, otherwise it is 0.
[0073] Specifically, the sample weight update formula for the samples with the label of single - door elevator in the auxiliary sample set is where is the sample weight of the \(k\) - th sample in the \(i\) - th weak prediction model in the auxiliary sample set, \(\lambda\) is the learning rate parameter, is the indicator function, when the predicted label is not equal to the true label \(y\) k , its value is 1, otherwise it is 0. This update formula means that if the predicted label of the sample is inconsistent with the true label, the sample weight will decrease, otherwise it remains unchanged.
[0074] Preferably, when setting the learning rate parameter \(\lambda\), a relatively small value can be selected to ensure the stability of the model and avoid over - adjustment. In addition, a mechanism of adaptive learning rate can be introduced, that is, the learning rate is dynamically adjusted according to the historical prediction accuracy of the samples to improve the adaptability and prediction accuracy of the model. Regularization terms such as L1 or L2 regularization can also be considered to prevent the model from overfitting and improve the generalization ability of the model.
[0075] 7In some embodiments, the multi - door elevator confidence probability of each sample to be processed is the proportion of the number of samples with the original label of multi - door elevator in the sample set where each sample to be processed is located.
[0076] It should be noted that the calculation of the multi - door elevator confidence probability of the samples to be processed in the auxiliary sample set is determined based on the proportion of the number of samples with the original label of multi - door elevator in the sample set, which reflects the confidence level of the model for the samples belonging to the multi - door elevator category.
[0077] Specifically, the confidence probability of the multi-opening elevator refers to the ratio of the number of samples with the original label of multi-opening elevator in the sample set where each sample to be processed is located to the total number of samples in the auxiliary sample set. The higher this ratio, the more confident the model is in predicting that the sample belongs to a multi-opening elevator. For example, if there are 10 samples in a sample set and the original labels of 8 samples are multi-opening elevators, then the confidence probability of the multi-opening elevator in this sample set is 80%.
[0078] Preferably, in order to improve the calculation accuracy of the confidence probability, more complex statistical methods can be introduced, such as Bayesian estimation or confidence interval estimation in machine learning, to consider the influence of other relevant features in the sample set on the confidence probability. In addition, a weighted confidence probability calculation method can be adopted, where samples with different features can have different weights, and the weights can be determined according to the historical prediction accuracy of the samples or other relevant indicators.
[0079] In some embodiments, the sample weights of all samples to be processed are updated by the following formula:
[0080]
[0081] where, is the sample weight of the \(j_u\)-th sample to be processed in the \(k\)-th weak prediction model in the auxiliary sample set, \(\delta\) is the confidence probability adjustment factor, and \(Z\) ju is the ranking of the confidence probability of the \(j_u\)-th sample to be processed for the multi-opening elevator.
[0082] These formulas provide a more complex mathematical description to adapt to the characteristics of the multi-opening elevator car control system.
[0083] Specifically, the formula for updating the sample weight of the sample to be processed is where is the sample weight of the \(k\)-th sample to be processed in the \(i\)-th weak prediction model in the auxiliary sample set, \(p\) k is the ranking of the confidence probability of the \(k\)-th sample to be processed for the multi-opening elevator, and \(\beta\) is the confidence probability adjustment factor. This formula indicates that the sample weight will be adjusted according to the confidence probability ranking of the sample. The higher the ranking of the sample, that is, the higher the confidence probability of the sample, the weight will increase, and vice versa, it will decrease or remain unchanged.
[0084] Preferably, the setting of the confidence probability adjustment factor \(\beta\) can be optimized according to the performance of the model on the validation set. A \(\beta\) value that makes the prediction performance of the model best on the validation set can be selected. In addition, a non-linear weight adjustment strategy can be introduced. For example, when the confidence probability ranking \(p\) kWhen reaching above a certain threshold, the increased weight can grow in a non-linear manner to more significantly enhance the influence of high-confidence samples. It is also possible to consider periodically re-evaluating and adjusting the β value to adapt to the requirements of the model at different training stages.
[0085] In some embodiments, each piece of data in the to-be-controlled dataset includes the starting and ending floors requested by the passenger, a usage feature vector, and an original label; mapping the to-be-controlled data predicted to be a multi-opening elevator to the set of floor service areas, and predicting the usage demand of the multi-opening elevator includes:
[0086] For the to-be-controlled data predicted to be a multi-opening elevator, successively according to the starting and ending floors of each piece of to-be-controlled data, obtain the floor service area where each starting point and ending point are located; increment by 1 the number of stops of the multi-opening elevator corresponding to this floor service area;
[0087] Summarize the number of stops of the multi-opening elevator to predict the usage demand of the multi-opening elevator.
[0088] Specifically, each piece of data in the to-be-controlled dataset includes the starting and ending floors requested by the passenger, a usage feature vector, and an original label. The system will determine the floor service area where each starting point and ending point are located according to the starting and ending floors of each piece of to-be-controlled data. Then, the system will increment by 1 the number of stops of the multi-opening elevator corresponding to this floor service area. In this way, the system can count the demand for multi-opening elevators in each floor service area. Finally, the system summarizes the number of stops of the multi-opening elevator in all floor service areas, thereby predicting the usage demand of the multi-opening elevator in the entire building.
[0089] Preferably, in order to improve the prediction accuracy and response speed, the system can adopt real-time data analysis technologies, such as stream data processing, to quickly process and analyze the to-be-controlled dataset. In addition, the system can introduce intelligent optimization algorithms, such as genetic algorithms or particle swarm optimization, to optimize the division of floor service areas and the scheduling strategy of multi-opening elevators. It is also possible to consider using machine learning models, such as deep learning networks, to further analyze and predict the elevator usage patterns and demands of passengers.
[0090] The above various embodiments of the present invention have the following beneficial effects: The multi-opening elevator car control system of the present invention can improve the intelligent level of elevator scheduling by constructing an elevator usage sample set and using a neural network algorithm for iterative training. The system can more accurately predict the elevator usage demand by continuously updating and optimizing the weights of the weak prediction model and adjusting the sample weights of the sample set, thereby reducing the waiting time of passengers and improving the elevator operation efficiency. In this way, the elevator car control system can not only improve the usage efficiency of the elevator, but also optimize the riding experience of passengers, realizing a more user-friendly service.
[0091] Such asFigure 2 As shown in Figure 2 , a multi-door elevator car control system 200 of some embodiments includes:
[0092] A data processing module 201, configured to construct an elevator usage sample set based on the same building, where the elevator usage sample set includes a car usage target sample set of a multi-door elevator and a car usage auxiliary sample set of a single-door elevator; construct a floor service area set according to the floor positions where the multi-door elevator stops in the single-door elevator; and, obtain the car usage data of the single-door elevator again as a data set to be controlled;
[0093] A model training module 202, configured to perform iterative training on the elevator usage sample set based on a neural network algorithm to obtain multiple weak prediction models; where each time during training, according to the prediction error rate of the current weak prediction model for the target sample set, set the weight of the current weak prediction model and update the sample weights of the target sample set, and combine the usage conditions and confidence probabilities of the samples in the auxiliary sample set to update the sample weights and original labels of the auxiliary sample set, and train the next weak prediction model according to the updated elevator usage sample set; weighted-combine each weak prediction model to obtain a control model;
[0094] A usage demand prediction module 203, configured to input the data set to be controlled into the control model, map the data to be controlled predicted as a multi-door elevator into the floor service area set, and predict the multi-door elevator usage demand of the single-door elevator.
[0095] It can be understood that the modules described in the multi-door elevator car control system 200 correspond to the respective steps in the multi-door elevator car control system described in the reference Figure 1 description. Thus, the operations, features, and beneficial effects described above for the multi-door elevator car control system also apply to the multi-door elevator car control system 200 and the modules included therein, and will not be elaborated here.
[0096] Next, referring to Figure 3 , which shows a schematic structural diagram of a structure 300 of an electronic device suitable for implementing some embodiments of the present invention. The electronic device in some embodiments of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The terminal device shown is only an example and should not bring any limitation to the functions and usage ranges of the embodiments of the present invention.
[0097] As Figure 3As shown, the electronic device 300 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 301, which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.
[0098] Generally, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 the electronic device 300 with various devices is shown, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices may be implemented or had alternatively. Figure 3 Each block shown in the figure may represent a device or, as needed, multiple devices.
[0099] Furthermore, the storage medium of the embodiments of the present application stores program instructions capable of implementing all the above methods. Among them, the program instructions may be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. And the foregoing storage medium includes: various media capable of storing program codes such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc, or a terminal device such as a computer, a server, a mobile phone, a tablet, etc.
[0100] The above description is only some preferred embodiments of the present invention and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) disclosed in the embodiments of the present invention that have similar functions.
Claims
1. A multi-door elevator car control system, characterized in that: Follow these steps: Based on the same building, an elevator usage sample set is constructed, wherein the elevator usage sample set includes a target sample set of car usage of a multi-door elevator and an auxiliary sample set of car usage of a single-door elevator; a floor service area set is constructed according to the stop floor positions of multi-door elevators in the single-door elevator; Based on the neural network algorithm, the elevator usage sample set is iteratively trained to obtain multiple weak prediction models; During each training, the weight of the current weak prediction model is set and the sample weight of the target sample set is updated according to the prediction error rate of the current weak prediction model for the target sample set. In addition, the sample weight and original label of the auxiliary sample set are updated based on the usage and confidence probability of the samples in the auxiliary sample set, and the next weak prediction model is trained according to the updated elevator usage sample set. The control model is obtained by weighted combination of each weak prediction model; The car usage data of the single-door elevator is obtained again as the data set to be controlled and passed into the control model. The data to be controlled that is predicted to be a multi-door elevator is mapped to the floor service area set to predict the usage demand of the single-door elevator and the multi-door elevator.
2. The multi-door elevator car control system according to claim 1, characterized in that: Each sample in the elevator usage sample set includes a usage feature vector and an original label, wherein the usage feature vector includes: waiting time, number of passengers, number of floors occupied, passenger age, riding time, passenger gender, whether the passenger carries large items and passenger floor arrival rate, the original labels of the target sample set include: multi-door elevator, single-door elevator; the original labels of the auxiliary sample set include: single-door elevator.
3. The multi-door elevator car control system according to claim 2, characterized in that: The prediction error rate of the current weak prediction model for the target sample set is the sum of the weights of samples in the target sample set whose predicted labels are inconsistent with the original labels.
4. The multi-door elevator car control system according to claim 3, characterized in that: The weight of the current weak prediction model is set by the following formula: The sample weights of the target sample set are updated by the following formula: Among them, α k is the weight of the kth weak prediction model, e k is the prediction error rate of the kth weak prediction model for the target sample set, is the sample weight of the i-th sample in the target sample set before the k-th weak prediction model is trained, y i is the original label of the i-th sample in the target sample set, is the predicted label of the i-th sample in the target sample set in the k-th weak prediction model.
5. The multi-door elevator car control system according to claim 4, characterized in that: The updating of the sample weights and original labels of the auxiliary sample set in combination with the usage and confidence probability of the samples in the auxiliary sample set includes: Update the sample weights for samples labeled as single-door elevators in the auxiliary sample set; The samples labeled as multi-door elevators in the auxiliary sample set are used as samples to be processed, and the sample set in the floor service area predicted by the current weak prediction model for each sample to be processed is obtained, and the multi-door elevator confidence probability of each sample to be processed is calculated; after sorting the samples to be processed from large to small according to the multi-door elevator confidence probability, the original labels of the samples to be processed before the multi-door elevator confidence probability are updated to multi-door elevators, and the sample weights of all samples to be processed are updated, where Z∈{1,2,...,Z max -1}, Z max is the number of samples to be processed.
6. The multi-door elevator car control system according to claim 5, characterized in that: The sample weights are updated for samples labeled as single-door elevators in the auxiliary sample set using the following formula: in, is the sample weight of the jnth sample in the auxiliary sample set predicted as a single-door elevator in the kth weak prediction model, γ is the learning rate parameter, is the indicator function, when the predicted label and the true label y jn The value is 1 if they are not equal, otherwise it is 0.
7. The multi-door elevator car control system according to claim 5, characterized in that: The confidence probability of each sample to be processed being a multi-door elevator is the ratio of the number of samples whose original label is a multi-door elevator in the sample set where each sample to be processed is located.
8. The multi-door elevator car control system according to claim 5, characterized in that: Update the sample weights of all samples to be processed by the following formula: in, is the sample weight of the juth sample to be processed in the kth weak prediction model in the auxiliary sample set, δ is the confidence probability adjustment factor, Z ju Confidence probability ranking of the multi-door elevator of the juth sample to be processed. These formulas provide a more complex mathematical description to accommodate the characteristics of multi-door elevator car control systems.
9. The multi-door elevator car control system according to claim 2, characterized in that: Each piece of data in the to-be-controlled data set includes the starting and ending floors requested by the passenger, the usage feature vector and the original label; the to-be-controlled data predicted to be a multi-door elevator is mapped to the floor service area set, and the usage demand of the multi-door elevator is predicted, including: For the data to be controlled that is predicted to be a multi-door elevator, the floor service area where each starting point and end point are located is obtained according to the starting and ending floors of each data to be controlled; the number of stops of the multi-door elevator corresponding to the floor service area is increased by 1; The number of stops of the multi-door elevator is summarized to predict the use demand of the multi-door elevator.
10. The multi-door elevator car control system according to claim 1, characterized in that: The system comprises: A data processing module is used to construct an elevator usage sample set based on the same building, wherein the elevator usage sample set includes a car usage target sample set of a multi-door elevator and a car usage auxiliary sample set of a single-door elevator; construct a floor service area set according to the multi-door elevator stop floor positions in the single-door elevator; and obtain the car usage data of the single-door elevator again as a data set to be controlled; The model training module is used to iteratively train the elevator usage sample set based on the neural network algorithm to obtain multiple weak prediction models; during each training, the weight of the current weak prediction model is set and the sample weight of the target sample set is updated according to the prediction error rate of the current weak prediction model for the target sample set, and the sample weight and original label of the auxiliary sample set are updated in combination with the usage and confidence probability of the samples in the auxiliary sample set, and the next weak prediction model is trained according to the updated elevator usage sample set; the control model is obtained by weighted combination of the weak prediction models; The demand prediction module is used to pass the data set to be controlled into the control model, map the data to be controlled predicted to be multi-door elevators to the floor service area set, and predict the demand for multi-door elevators for single-door elevators.