Method and device for judging elevator stopping based on heterogeneous multi-modal data

Abstract

The invention provides an elevator stopping judgment method and device based on heterogeneous multi-modal data, and belongs to the field of elevator intelligent monitoring. The method comprises the following steps: acquiring initial operation data of all elevators in the same place; based on position data in the initial operation data, the relation between the elevator stopping position and the non-leveling position of the elevator to be judged is judged, and if the elevator is not located at the non-leveling position, the ratio of the operation frequency of the elevator to be judged to the average operation frequency of other elevators in the same place is calculated; performing first abnormality judgment based on the ratio; the basic probability of elevator stopping caused by each fault is calculated, and the change trend of a fault mode is recognized; the basic probability is adjusted according to the change trend, and the elevator stopping probability is obtained; based on the elevator operation mode and fault features in the initial operation data, a second machine learning model is used for conducting second anomaly judgment; and comprehensively judging whether the to-be-judged elevator is in an elevator stopping state or not by combining the first abnormity judgment, the elevator stopping probability and the second abnormity judgment. The safety and reliability of the elevator can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of elevator intelligent monitoring, and particularly to a method and device for judging elevator stoppage based on heterogeneous multi-modal data. Background Art

[0002] With the acceleration of the urbanization process, elevators, as key equipment for vertical transportation, have seen a continuous increase in their usage frequency and quantity. While people's dependence on elevators is growing, the safety issues of elevators have also attracted much attention. Currently, the safety monitoring and control technology for elevator operation plays a crucial role in ensuring the reliable operation of elevators, but there are still many deficiencies in the existing technologies.

[0003] Regarding the current situation of elevator operation monitoring, elevators have been widely used in various buildings, ranging from residential buildings to commercial buildings, with an extremely wide coverage. However, the elevator operation environments in different buildings are complex and diverse, and the operating conditions vary greatly, posing challenges to the unified management and safety monitoring of elevators. At the same time, with the development of the Internet of Things technology, more and more elevators are equipped with various sensors and data acquisition devices, generating a large amount of operation data, but the utilization efficiency of these data remains to be improved.

[0004] In terms of current technologies, the intelligent monitoring systems of traditional elevators mainly rely on the elevator operation state after a fault occurs to judge whether the elevator stops. The monitoring system will collect the operation parameters of the elevator in real time, such as speed, acceleration, position, etc., as well as relevant information when a fault occurs, such as fault type, fault time, etc. By analyzing these data, the operation state of the elevator is judged. In terms of hardware, the elevator is equipped with various sensors, such as car position sensors, floor door area sensors, etc., for detecting the position and operation state of the elevator; in terms of software, simple algorithms are used to process and analyze the collected data.

[0005] However, there are significant problems with current technologies. Firstly, there is the problem of missed reports. Due to the limitations of monitoring devices, some relatively minor but potential faults are difficult to be detected in a timely manner. When the elevator is unoccupied, the system may not be able to capture small fault signals, resulting in the elevator continuing to operate with potential fault hazards, increasing the safety risk. Secondly, there is the problem of false reports. Minor faults may be misjudged as serious faults in some cases, especially when the elevator is idle or has no passengers. For example, the sensors of the elevator may generate false signals due to small vibrations or power fluctuations, leading to the system wrongly judging that the elevator stops. This will not only cause unnecessary panic but also affect the trust of maintenance units and passengers in the safety of elevators. In addition, traditional technologies are insufficient in dealing with complex faults and comprehensively analyzing multi-source data, and it is difficult to achieve a comprehensive and accurate judgment of elevator faults, unable to provide effective support for the maintenance and management of elevators. Summary of the Invention

[0006] In view of this, the present application provides a method and device for judging elevator stopping based on heterogeneous multi-modal data, which can timely identify potential faults, reduce false negatives and false positives, and improve the safety and reliability of elevators.

[0007] Specifically, the present application is implemented through the following technical solutions:

[0008] The first aspect of the present application provides a method for judging elevator stopping based on heterogeneous multi-modal data, and the method includes:

[0009] Obtain the initial operation data of all elevators in the same location;

[0010] Based on the position data in the initial operation data, judge the relationship between the elevator stopping position to be judged and the non-leveling position. If it is not in the non-leveling position, calculate the ratio of the operation times of the elevator to be judged to the average operation times of other elevators in the same location;

[0011] Perform a first anomaly judgment based on the relationship between the ratio and the operation times threshold;

[0012] Extract historical fault data from the elevator control system, maintenance records, and fault logs, calculate the basic probability of each fault causing elevator stopping based on the historical fault data, and identify the change trend of the fault mode over time based on the time series analysis method;

[0013] Adjust the basic probability according to the change trend and the first machine learning model to obtain the elevator stopping probability. The first machine learning model is used to analyze the historical fault data extracted from the elevator control system, maintenance records, and fault logs, determine the factors with an influence degree on elevator stopping greater than the threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factors output by the first machine learning model and the real-time fault information in the initial operation parameters, and adjust the basic probability based on the adjustment margin and the adjustment coefficient;

[0014] Perform a second anomaly judgment using the second machine learning model based on the elevator operation mode and fault characteristics in the initial operation data;

[0015] Combine the first anomaly judgment, elevator stopping probability, and second anomaly judgment to comprehensively judge whether the elevator to be judged is in a stopped state; among them, the comprehensive anomaly judgment result is calculated based on the weighted sum of the first anomaly judgment, elevator stopping probability, and the second anomaly judgment.

[0016] The second aspect of the present application provides a device for judging elevator stopping based on heterogeneous multi-modal data, and the device includes an acquisition module, a judgment module, a calculation module, and an adjustment module;

[0017] Among them, the obtaining module is used to obtain the initial operation data of all elevators in the same location;

[0018] The judging module is used to judge the relationship between the stopping position of the elevator to be judged and the non - leveling position based on the position data in the initial operation data. If it is not in the non - leveling position, calculate the ratio of the number of operations of the elevator to be judged to the average number of operations of other elevators in the same location;

[0019] The judging module is also used to perform a first anomaly judgment based on the relationship between the ratio and the operation - number threshold;

[0020] The calculating module is used to extract historical fault data from the elevator control system, maintenance records, and fault logs, calculate the basic probability of each fault causing the elevator to stop based on the historical fault data, and identify the change trend of the fault mode over time based on the time - series analysis method;

[0021] The adjusting module is used to adjust the basic probability according to the change trend and the first machine - learning model to obtain the stopping probability. Among them, the first machine - learning model is used to analyze the historical fault data extracted from the elevator control system, maintenance records, and fault logs, determine the factors with an influence degree on elevator stopping greater than the threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factors output by the first machine - learning model and the real - time fault information in the initial operation parameters, and adjust the basic probability based on the adjustment margin and the adjustment coefficient;

[0022] The judging module is also used to perform a second anomaly judgment using the second machine - learning model based on the elevator operation mode and fault characteristics in the initial operation data;

[0023] The judging module is also used to comprehensively judge whether the elevator to be judged is in a stopped state by combining the first anomaly judgment, the stopping probability, and the second anomaly judgment; among them, the comprehensive anomaly judgment result is calculated based on the weighted sum of the first anomaly judgment, the stopping probability, and the second anomaly judgment.

[0024] For the elevator stopping discrimination method and device for heterogeneous multi - modal data provided in this application, on the one hand, by obtaining the initial operation data of all elevators in the same location, and comprehensively determining whether there is a stopping fault of the elevator by integrating three types of data with different modalities and configurations, namely position data, number - of - operations data, and status data, from the initial operation data, it can comprehensively understand the operation state of the elevator. Compared with the traditional discrimination method that only relies on a single type of data, it can provide a more solid data basis for accurately discriminating the elevator stopping state. Through the judgment of whether it is in a non - leveling state, this dangerous situation can be discovered in time and a warning can be issued, which helps relevant personnel take measures quickly, ensure the safety of passengers' lives, and reduce the occurrence of safety accidents.

[0025] In a second aspect, corresponding to the modal data, the present invention improves the accuracy of elevator stop judgment through three-level comprehensive judgment based on different modal data. Specifically, the first-level judgment directly uses position data for quick judgment. When the position judgment is not clear, it enters the second level, that is, the number of running times data is used for judgment to obtain ratio information. The third level uses state data, and respectively uses the learning of the change trend and the direct judgment of machine learning to obtain two fault judgment results. That is, the fault judgments of the first, second, and third levels respectively use different types of fault data, improving the utilization rate of different types of data and the accuracy of stop identification. At the same time, for data with complex recognition methods, two methods of directly judging by machine learning and only learning the change trend by machine learning are used for fault recognition at the same time, further improving the accuracy of fault recognition.

[0026] Finally, through multi-dimensional judgment and comprehensive analysis, the limitations of a single judgment method are avoided, and the actual operating state of the elevator can be judged more comprehensively and accurately, effectively reducing the probability of misjudgment and missed judgment. Especially in the face of complex operating environments and potential faults, this multi-dimensional judgment method can improve the reliability of the discrimination results and provide more powerful guarantees for the safe operation of the elevator. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flowchart of the first embodiment of the elevator stop discrimination method for heterogeneous multi-modal data provided by the present application;

[0028] Figure 2 It is a schematic structural diagram of the second embodiment of the elevator stop discrimination device for heterogeneous multi-modal data provided by the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application.

[0030] The terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "the", and "said" used in the present application are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term " / and" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0031] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0032] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0033] Embodiment 1:

[0034] Figure 1 This is a flowchart of Embodiment 1 of the elevator stop judgment method for heterogeneous multimodal data provided by this application. Figure 1 , the method provided in this embodiment may include:

[0035] S101. Obtain initial operation data of all elevators in the same place.

[0036] It should be noted that the same place refers to a specific area with a certain boundary range and where multiple elevators are operating. For example, in a large shopping mall, multiple elevators installed in different locations inside the mall are located in the same place; in another example, in a high-rise office building, multiple elevators used for vertical transportation are located in the same place. The definition of the same place helps to collect elevator data with similar operating environments and demand characteristics, which is convenient for subsequent comparison and comprehensive analysis.

[0037] The initial operation data includes various information reflecting the real-time operation status of the elevator. Specifically, the initial operation data may include the location data, operation data, fault log, sensor data of the elevator to be identified, and the operation times of other elevators in the same place.

[0038] Among them, the operation data includes the collected multi-dimensional operation parameters such as the up and down speed, acceleration, deceleration, running time, and dwell time of the elevator. These parameters are key indicators for measuring the efficiency of elevator operation. For example, by analyzing the up and down speed, it can be determined whether the elevator is running at a normal speed. The changes in acceleration and deceleration can reflect whether the acceleration and deceleration process of the elevator is smooth. The running time and dwell time help to understand the busyness of the elevator and the dwelling pattern on each floor. Through the comprehensive analysis of these parameters, it is possible to preliminarily determine whether the elevator has abnormal behavior, such as abnormal speed, unstable acceleration and deceleration, which may indicate problems with the mechanical system or control system.

[0039] The inter-floor running information is also an important part of the running data, recording the starting floor and the arriving floor of each run. This is crucial for determining whether the elevator has abnormal stops (such as non-level stops). Non-level stops may pose safety hazards when passengers enter and exit the elevator. By monitoring and analyzing the inter-floor running information, such problems can be detected in a timely manner.

[0040] The fault log is extracted from the elevator control system and can include the fault type (such as mechanical faults, electrical faults, etc.), the fault occurrence time, the fault duration, and the relevant error codes. This information is an important basis for understanding the elevator's historical fault patterns and their frequencies. For example, if a certain type of fault occurs frequently, then it is necessary to focus on this fault type and analyze its causes in order to take targeted measures for prevention and repair. At the same time, the records of the fault occurrence time and duration are also helpful for evaluating the impact of the fault on the normal operation of the elevator.

[0041] Sensor data is used to monitor the running state of the elevator in real time by using various sensors equipped on the elevator (such as position sensors, load sensors, door switch sensors, etc.). The position sensor can accurately feedback the current position of the elevator, the load sensor can monitor the carrying weight of the elevator, and the door switch sensor is used to monitor the opening and closing state of the elevator door. These real-time data can provide accurate feedback on the state of the elevator during operation. For example, when the load sensor detects an overload situation, or when the door switch sensor feedbacks an abnormal situation such as the door not closing properly, corresponding measures can be taken in a timely manner to ensure the safe operation of the elevator. In addition, if the elevator is also equipped with environmental sensors (such as temperature, humidity, vibration, etc.), it is also necessary to collect the relevant data. Because environmental factors may affect the operation of the elevator. For example, high temperature may affect the performance of elevator electronic components, high humidity may cause electrical equipment to get damp, and excessive vibration may imply looseness in the elevator mechanical structure. By monitoring and analyzing the environmental data, potential fault hazards can be detected in advance and corresponding protective measures can be taken.

[0042] In addition, the maintenance and repair records of the elevator can also be obtained to analyze past repair problems and the measures taken. By analyzing the previous repair records, it can be understood which components of the elevator are prone to failure and whether the repair measures taken are effective. If it is found that certain components need to be repaired frequently, then it is necessary to consider whether an upgrade or replacement is needed to improve the reliability of the elevator.

[0043] It should be noted that for the acquisition of data of other elevators in the same location, including collecting the operation times and operation parameters of other elevators in the same building, a benchmark data can be established. By comparing the operation behaviors of the elevator to be judged and other elevators, abnormal patterns can be identified. For example, if the operation times of the elevator to be judged are significantly lower than those of other elevators, or there are large differences in its operation parameters (such as speed, acceleration, etc.) compared with other elevators, it may imply potential faults or low operation efficiency of the elevator to be judged. By analyzing the operation conditions of other elevators in the same location, a more comprehensive understanding of the overall operation status of the elevators in this location can be obtained, providing a more accurate reference basis for the fault analysis and stop judgment of the elevator to be judged.

[0044] Finally, by integrating the various types of data obtained above, a comprehensive data set is established through a database management system or a big data processing platform. The scattered and different types of data can be centrally managed, facilitating subsequent unified analysis and processing, and providing an efficient and reliable data basis for elevator stop judgment and fault diagnosis based on these data.

[0045] S102. Based on the position data in the initial operation data, judge the relationship between the stop position of the elevator to be judged and the non-leveling position. If it is not in the non-leveling position, calculate the ratio of the operation times of the elevator to be judged to the average operation times of other elevators in the same location.

[0046] It should be noted that before judging the relationship between the stop position of the elevator to be judged and the non-leveling position, preprocessing of the obtained initial operation data is also included. Specifically, the preprocessing can include data cleaning, denoising, normalization, and feature selection. Among them, the core goal of data cleaning is to ensure the high quality and consistency of the data. The methods of data cleaning can include removing duplicate values, handling missing values (deletion method, filling method), formatting data, and outlier detection, etc. The main purpose of denoising is to reduce the noise in the data, thereby improving the accuracy of the data. Common denoising methods include smoothing techniques, filters, and outlier correction. Normalization aims to scale the data to a standard range to prevent adverse effects of different dimensions or magnitudes on the subsequent analysis results. Common normalization methods include min-max scaling, Z-score standardization, etc. The purpose of feature selection is to select the features that have the most influence on fault recognition from the preprocessed data, thereby reducing the data dimension and improving the model performance. Common feature selection methods include filtering method, wrapper method, embedding method, and principal component analysis, etc.

[0047] It should be noted that for the specific implementation principles and processes of the above various preprocessing methods, please refer to the descriptions of related technologies and will not be elaborated here.

[0048] Through the above comprehensive and detailed data cleaning, denoising, normalization, and feature selection steps, a deeply processed dataset will ultimately be obtained. This dataset features high quality, low noise, standardization, and prominent key features, providing a solid and reliable data foundation for subsequent analysis, model training, and fault identification, and contributing to significantly improving the accuracy and robustness of the model.

[0049] After preprocessing, based on the position data in the initial operating data after preprocessing, the relationship between the elevator stop position to be discriminated and the non-leveling position is judged. Specifically, this judgment mainly relies on the inter-floor operation information and sensor data of the elevator to be discriminated. The inter-floor operation information records the starting floor and the arrival floor of each elevator operation, and sensor data such as position sensors can provide real-time feedback on the accurate position of the elevator. Through these data, the position where the elevator stops can be accurately determined. Among them, the sensor data includes the data detected by the position sensor located in the elevator car and / or the data detected by the floor door area sensor. In specific implementation, the actual position of the elevator when it stops is compared with the leveling position. The leveling position refers to the position where the floor of the elevator car is flush with the floor of the floor when the elevator stops normally, which is the standard position for the safe operation of the elevator and the normal entry and exit of passengers. If the elevator stop position is not at the leveling position, that is, in the non-leveling position, it means that the elevator has an abnormal stop situation.

[0050] In addition, judging the relationship between the elevator stop position to be discriminated and the non-leveling position also includes judging whether the elevator to be discriminated stops at the non-leveling position through at least one of the motor operation data of the elevator to be discriminated, the self-check data of the elevator to be discriminated, the emergency device of the elevator to be discriminated, the floor display, and the safety circuit monitoring. Among them, the motor operation data of the elevator to be discriminated and the self-check data of the elevator to be discriminated belong to the operation data of the elevator to be discriminated. Specifically, for judging whether it stops at the non-leveling position, any one or a combination of the following methods can be used:

[0051] The first method: Position sensors such as encoders or limit switches are installed inside the elevator car to monitor the position of the elevator in real time. The sensors regularly send position data to the control system. The control system compares the received position data with the preset leveling position.

[0052] The position sensor can provide high-precision position feedback and timely detect whether the elevator stops at the non-leveling position. In case of a fault, the abnormal signal sent by the sensor can quickly trigger emergency measures to ensure the safety of passengers.

[0053] The second method: Door area sensors are installed at the entrance of each floor of the elevator to detect whether the elevator car enters the door area; when the elevator arrives at a floor, the sensor will identify whether the elevator is in the correct door area; if the elevator stops outside the door area and the sensor cannot detect a signal, the system will determine it as non-leveling.

[0054] This method can effectively prevent the elevator from opening doors in a non-leveling state, reducing the risk of passenger injury. The door area sensor can also provide additional data to help maintenance personnel analyze the operating status and fault conditions of the elevator.

[0055] The third method: The elevator's motor and braking system have a self-check mechanism that continuously monitors the motor operating parameters (such as rotational speed, acceleration). The system analyzes the feedback from the motor and the brake to identify whether there are any abnormalities, such as failure to decelerate or stop as expected. If an abnormality is detected, the system will record a fault log and issue an alarm.

[0056] Through real-time monitoring, problems with the motor or braking system can be detected in a timely manner to avoid serious faults. The fault log helps with subsequent repairs and maintenance, making problem-solving more efficient.

[0057] The fourth method: The control system conducts self-checks regularly, monitoring parameters such as the elevator's acceleration, deceleration, running time, and running distance. If the actual operating parameters are found to be inconsistent with the expected values, the control system will trigger a fault handling procedure. The system will stop the elevator, issue an alarm, and record the fault information.

[0058] The self-check mechanism can improve the safety of the elevator and reduce the failure rate. By monitoring the operating data, maintenance personnel can obtain the elevator's operating history to assist with fault analysis.

[0059] The fifth method: Install emergency operation devices (such as emergency buttons) and floor displays inside and outside the elevator. When the elevator stops at a non-leveling position, the floor display will show abnormal information or error codes. Passengers can use the emergency device to notify maintenance personnel for rescue.

[0060] The emergency operation device provides a safety guarantee for passengers, enabling them to seek help in a timely manner in case of elevator failure. The abnormal status of the display can help maintenance personnel quickly diagnose elevator faults.

[0061] The sixth method: The elevator is equipped with a safety circuit (such as a leveling induction switch) to ensure that the elevator is at a leveling position before opening the door. When the elevator attempts to open the door while not at a leveling position, the safety circuit will be cut off, triggering an emergency brake. The system records such events and issues a fault alarm.

[0062] The safety circuit provides redundant protection, further reducing the risk of passenger injury. The fault alarm function enhances the safety of the elevator, enabling maintenance personnel to handle problems in a timely manner.

[0063] Through the comprehensive application of the above - mentioned multiple methods, based on factors such as the operation data, fault logs, sensor data of the elevator to be judged, and the operation times of other elevators in the same location, a more in - depth analysis can be carried out. For example, by comparing the operation data and fault logs of the elevator, elevators with high failure rates can be identified and given priority maintenance. At the same time, analyzing the operation times of other elevators can also help identify whether there are common faults or design defects, thereby improving the safety and reliability of the overall elevator system.

[0064] When it is judged that the elevator stop position is not at the leveling position, a warning operation is directly generated. This is because non - leveling docking may pose great safety hazards to passengers. For example, passengers are prone to falling or dropping when entering or leaving the elevator. The warning operation can be to send alarm messages to elevator management personnel, such as text messages, emails, or prompt messages within the system, and at the same time, emit audible and visual alarms in the elevator car and on the relevant floors to remind passengers to pay attention to safety, so as to take rescue and maintenance measures in a timely manner.

[0065] If the elevator stop position is at the leveling position, it indicates that the current docking state of the elevator is normal. At this time, calculate the ratio of the operation times of the elevator to be judged to the average operation times of other elevators in the same location. Specifically, when implementing, the operation times of the elevator to be judged can be directly obtained from its operation data, and the operation times of other elevators are also part of the initial operation data collected in step S101. By counting the operation times of all other elevators in the same location except the elevator to be judged and calculating their average value, the average operation times of other elevators in the same location can be obtained.

[0066] It should be noted that the purpose of calculating this ratio is to judge whether the operation efficiency of the elevator to be judged is normal by comparing the operation frequencies of the elevator to be judged and other elevators in the same location. If the ratio deviates significantly from 1, it may imply potential problems with the elevator to be judged. For example, if the ratio is much less than 1, it may indicate that the usage demand for the elevator to be judged is low, or there are faults causing its operation to be unsmooth; if the ratio is much greater than 1, it may mean that the elevator is overloaded and has been operating at a high load for a long time, which is prone to causing faults.

[0067] S103. Perform a first anomaly judgment based on the relationship between the ratio and the operation - times threshold.

[0068] Before performing the first anomaly judgment, the following explanations are given first. In the same location, the operation times of multiple elevators are interrelated, especially when designing, managing, and controlling their operations, this relationship is particularly obvious. This relationship can be understood from the following perspectives:

[0069] Scheduling Algorithm: Associated elevators are usually scheduled by a central control system. This system selects which elevator to perform the service based on parameters such as the current state of each elevator, the requested floors of the passengers, and their respective number of runs. In this way, the system can achieve more efficient operation, avoid service delays caused by overloading of a certain elevator, and prevent other elevators from being idle.

[0070] Load Balancing: To prevent a certain elevator from being overused, the control system will try to balance the number of runs among the elevators. By balancing the load, the service life of the elevator equipment can be extended, and the maintenance requirements of a single elevator due to frequent operation can be reduced.

[0071] Service Efficiency: If the elevator system is not designed reasonably, some elevators may be called frequently while others are relatively idle. This situation not only affects the waiting time of passengers but also may reduce the overall service efficiency. Therefore, it is necessary to reasonably allocate the number of runs among the associated elevators to reduce congestion during peak periods and the waiting time of passengers.

[0072] Energy-saving Control: Some elevator systems will dynamically adjust the number of available elevators according to the floor demand. During off-peak periods, the system may reduce the number of enabled elevators to reduce energy consumption by concentrating on using a few elevators. This dynamic adjustment not only improves energy efficiency but also optimizes the use of elevators.

[0073] In summary, the number of runs among associated elevators is interrelated, and scheduling, load balancing, and efficiency optimization all rely on the coordinated operation of elevators. Through scientific management and control, the overall performance of the elevator system and the user experience can be significantly improved.

[0074] Therefore, after calculating the ratio of the number of runs of the elevator to be judged to the average number of runs of other elevators in the same place, a first abnormal judgment is further made on the operating state of the elevator based on this ratio. Specifically, the first abnormal judgment is made based on the relationship between the ratio and the number-of-runs threshold, including: comparing the ratio with a preset number-of-runs threshold range, where the preset number-of-runs threshold range includes an upper limit and a lower limit; if the ratio is less than the lower limit or greater than the upper limit, it is determined as the first abnormality; if the ratio is between the lower limit and the upper limit, it is judged that the elevator is operating normally.

[0075] As an alternative embodiment, the preset range of the number of running times is calculated by the scheduling process of the running scheduling algorithm. Calculating the preset range of the number of running times includes: calculating the average number of running times of an elevator according to the elevator running load balancing requirement of the recognition area; calculating the highest and lowest running probabilities of the elevator according to the elevator scheduling algorithm; respectively determining the upper and lower limits of the preset range of the number of running times based on the products of the highest running probability and the lowest running probability and the average number of running times.

[0076] It should be noted that the preset range of the number of running times includes two values, an upper limit and a lower limit. These two values are comprehensively determined based on factors such as statistical analysis of a large amount of elevator running data, actual operation experience, and the design performance of the elevator. For example, by long-term monitoring of multiple elevators of the same type and in the same usage scenario, analyzing the distribution of the ratio of their running times, and finding the interval where the ratio is located when most elevators operate normally, so as to determine a reasonable upper limit and a lower limit. The preset range of the number of running times provides a standard reference for judging whether the elevator operation is abnormal, and it defines the range within which the ratio of the running times of the elevator to be judged should be under normal circumstances.

[0077] Compare the calculated ratio with the preset range of the number of running times. If the ratio is less than the lower limit, this may indicate that the running times of the elevator to be judged are much lower than the average level of other elevators in the same place, and there may be problems such as un-repaired faults, incorrect control system settings, or low usage requirements. For example, the elevator may have some minor faults, resulting in occasional stops and thus reducing the running times; or some function settings of the elevator are unreasonable, affecting its normal use. If the ratio is greater than the upper limit, it means that the running times of the elevator to be judged are much higher than the average level, and it may face excessive load pressure. Running like this for a long time is likely to cause increased equipment wear and accelerated aging of components, thereby increasing the probability of faults. For example, the number of people in the area where a certain elevator is located suddenly increases, or other elevators break down, causing passengers to concentrate on this elevator, resulting in a significant increase in its running times. When the ratio is less than the lower limit or greater than the upper limit, it is determined as the first type of abnormality.

[0078] If the ratio is between the lower limit and the upper limit, it indicates that the running times of the elevator to be judged are within a reasonable relative relationship range with the average running times of other elevators in the same place, and it can be judged that the elevator is currently in a normal running state, which means that the running frequency of the elevator adapts to the overall environment and there is no obvious abnormal fluctuation.

[0079] By comparing the ratio of the number of runs with a preset range of the number of runs threshold, it is possible to quickly and effectively determine whether there is a first abnormal situation in the elevator to be judged. This judgment method is simple and intuitive, and the threshold range determined based on a large amount of data and experience has high reliability. Timely discovery of the first anomaly allows elevator managers to take measures in advance. For the case where the ratio is less than the lower limit, it is possible to check whether there is a fault in the elevator and repair it, or adjust the operating settings of the elevator; for the case where the ratio is greater than the upper limit, it is possible to consider increasing the maintenance frequency of the elevator, or adjusting the passenger diversion strategy to ensure the safe and stable operation of the elevator.

[0080] S104. Extract historical fault data from the elevator control system, maintenance records, and fault logs, calculate the basic probability of each fault causing the elevator to stop based on the historical fault data, and identify the change trend of the fault mode over time based on the time series analysis method.

[0081] It should be noted that the elevator control system, maintenance records, and fault logs are the main sources of historical fault data. The elevator control system records various data during the operation of the elevator, including the operating status and parameters at the time of the fault; the maintenance records contain information such as the problems found during each maintenance, repair measures, and repair time; the fault logs specifically record the detailed situation of the fault occurrence, such as the fault type, occurrence time, and duration. Extracting data from these sources can comprehensively obtain information related to the historical faults of the elevator.

[0082] Specifically, calculating the basic probability of each fault causing the elevator to stop based on the historical fault data, and identifying the change trend of the fault mode over time, includes:

[0083] (1) Arrange the historical data in chronological order.

[0084] It should be noted that historical data is widely collected from the elevator control system, maintenance records, and fault logs. These data cover the operating duration of the elevator, which intuitively reflects the usage intensity of the elevator at different stages; elevator stop events and the corresponding fault events, which detail each elevator stop and the fault conditions that caused the stop; the timestamps of the fault events, which accurately mark the moment of the fault occurrence and provide a key basis for subsequent time series analysis; the fault types, such as power failures, mechanical failures, control system failures, etc., different types of faults have different generation mechanisms and influence degrees; the maintenance records contain the detailed content, time, and frequency of each maintenance, reflecting the development of the elevator maintenance work; external factor data, such as the usage frequency reflects the busyness of the elevator, weather changes (such as high temperature, humidity, etc.) may affect the performance of elevator components, and the usage situation of the building (such as increased dust during renovation) may also have a potential impact on the elevator operation.

[0085] Arrange the collected historical data strictly in chronological order. For example, arranging the annual fault logs in the order of the occurrence time of the faults can intuitively show the occurrence of faults in different years and seasons, laying a foundation for subsequent analysis of the evolution law of faults over time.

[0086] In addition, it should also be noted that after obtaining the historical data, it also includes removing error records, duplicate data, and incomplete data from the historical data to ensure data quality; converting all data into a unified format for convenient subsequent analysis; identifying key features, such as the operating mode before the elevator stops and warning signals before the fault, etc.

[0087] (2) According to the historical data, for each type of fault, use the conditional probability formula to calculate the basic probability that the fault type causes the elevator to stop.

[0088] In the actual elevator operation scenario, there may be multiple types of faults in the elevator, and the possibility of each type of fault causing the elevator to stop is not the same. To accurately evaluate the operation risk of the elevator, we quantify this possibility by calculating the basic probability that each type of fault causes the elevator to stop. Among them, the basic probability here is obtained based on historical data statistics, reflecting the probability situation that when a specific type of fault appears during the past operation process, it causes the elevator to stop.

[0089] In addition, different types of faults will have different basic probabilities of causing the elevator to stop due to their own characteristics, influence ranges, and the degree of association with the key components of the elevator. In general, by calculating the basic probabilities of various types of faults, a clearer understanding of the fault risks of the elevator can be obtained, so that the focus can be placed on those types of faults with higher basic probabilities. In this way, preventive measures can be taken in advance, such as increasing the maintenance frequency of relevant components and stocking spare parts, etc., to reduce the occurrence probability of elevator stop events and ensure the safe and stable operation of the elevator. For each type of fault, use the conditional probability formula P(stop|fault)=P(stop and fault) / P(fault) to calculate its basic probability of causing the elevator to stop. In the actual calculation process, P(stop and fault) needs to count the number of times that a specific type of fault appears simultaneously and causes the elevator to stop in the historical data, and then divide by the total number of data records; P(fault) is to count the number of times that this type of fault appears and divide by the total number of data records. For example, in 1000 historical data records, mechanical faults appear 100 times, and among them, 30 times are caused by mechanical faults resulting in elevator stops. Then the basic probability P(stop|mechanical fault)=(30 / 1000) / (100 / 1000)=0.3. By performing such calculations for each type of fault one by one, the influence degree of each fault on elevator stops can be quantified, and it can be clarified which faults are high-risk factors for causing elevator stops.

[0090] (3)Fit the change process of the basic probability over time to identify the change trend of the fault mode.

[0091] It should be noted that time series analysis methods, such as ARIMA models and moving average methods, can be applied to deeply analyze the elevator operation and fault data. ARIMA models can fully consider the characteristics of data such as autocorrelation, seasonality, and trend. By fitting historical data, they can predict the future trend of fault occurrence. For example, taking the monthly number of fault occurrences as time series data and inputting it into an ARIMA model, after model training and parameter adjustment, a prediction curve of the number of fault occurrences over time can be obtained. The moving average method smooths data fluctuations and highlights the long-term trend of data by calculating the average value of data within a certain period. For instance, by using a 12-month moving average method to process the elevator operation duration of each month of the year, the annual change trend of the operation duration and its relationship with the fault occurrence frequency can be observed more clearly.

[0092] By carefully observing the relationship between the fault occurrence frequency and time and using chart analysis tools, such as line charts and seasonal decomposition charts, identify the seasonal or periodic changes of faults. For example, by plotting a line chart of the fault occurrence frequency of each month of the year, if it is found that the power supply fault occurrence frequency significantly increases in summer (June - August) every year, this indicates that the power supply fault has obvious seasonal characteristics in summer. Using a seasonal decomposition chart to decompose time series data into components such as trend, seasonality, and residuals can more accurately quantify the influence degree of seasonal factors on the fault occurrence frequency.

[0093] In this step, by combining historical data and initial operation data, accurately calculate the basic probability of each fault causing elevator stoppage, and deeply identify the change trend of the fault mode, which provides extremely important information for the safe operation and maintenance management of elevators. It can help elevator maintenance personnel anticipate potential fault risks in advance and formulate personalized maintenance plans for different fault types and their change trends. For example, for fault types with a relatively high occurrence probability and an upward trend, increase the maintenance frequency and reserve relevant spare parts in advance; for faults with seasonal or periodic change characteristics, conduct special inspections and maintenance before the corresponding season or cycle arrives. In this way, effectively reduce the elevator fault occurrence rate, ensure the stable operation of the elevator, and improve the safety and comfort of passengers during use. At the same time, from the perspective of economic benefits, a reasonable maintenance plan can reduce unnecessary maintenance costs, extend the service life of the elevator, and provide a scientific basis for the whole-life cycle management of the elevator.

[0094] S105. Adjust the basic probability according to the change trend and the first machine learning model to obtain the elevator stoppage probability.

[0095] Among them, the first machine learning model is used to analyze historical fault data extracted from the elevator control system, maintenance records, and fault logs, determine factors with a greater impact on elevator stoppage than the threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factors output by the first machine learning model and the real-time fault information in the initial operating parameters, and adjust the basic probability based on the adjustment margin and adjustment coefficient.

[0096] It should be noted that in this step, by comprehensively considering the change trend of the fault mode and the analysis results of the first machine learning model, the basic probability calculated previously is adjusted to obtain a stoppage probability that can better reflect the current actual situation. This process helps to more accurately evaluate the possibility of elevator stoppage due to various faults, providing an important basis for subsequent abnormal judgment and comprehensive evaluation of elevator status.

[0097] Specifically, adjusting the basic probability according to the change trend and the first machine learning model to obtain the stoppage probability includes:

[0098] (1) Determine the periodic change trend of fault occurrence according to the change trend of the fault mode identified during the change process of the basic probability over time.

[0099] Based on the change of the basic probability over time obtained by using time series analysis methods (such as ARIMA model, moving average method, etc.) and chart analysis (such as line chart, seasonal decomposition chart, etc.) in step S104, identify the periodic law of fault occurrence. For example, if it is found through analysis that the frequency of a certain type of fault is higher in summer every year, this indicates that the fault has a seasonal periodic change trend. (2) Determine the current running time of the elevator to be judged.

[0100] It should be noted that obtaining the current running time of the elevator to be judged, accurate to a specific time point or time period, is convenient for subsequent analysis in combination with the periodic change trend of the fault.

[0101] (3) Based on the current running time of the elevator to be judged, find the corresponding basic probability value for each time point in the determined periodic change trend of the fault.

[0102] It should be noted that according to the determined periodic change trend of the fault, using the current running time of the elevator to be judged as an index, find the corresponding basic probability value for this time point in the corresponding change trend data. This basic probability value is calculated based on historical data and reflects the likelihood of elevator stoppage caused by similar faults at this time point.

[0103] (4) Determine the fault cause and adjustment coefficient based on the first machine learning model, and determine the adjustment margin based on the periodic change trend.

[0104] Further, after determining the basic probability value, the cause of the fault and the adjustment coefficient are determined based on the first machine learning model. Specifically, it includes three major steps: selecting a machine learning algorithm, training the model and determining important factors, and calculating the adjustment coefficient. When specifically implemented, machine learning algorithms such as random forest and decision tree are used to deeply analyze the historical fault data extracted from the elevator control system, maintenance records, and fault logs. Taking the random forest algorithm as an example, the elevator operation parameters (such as speed, acceleration, operation time, etc.), fault types, and external factor data (such as usage frequency, weather changes, etc.) in the historical data are used as input features, and the elevator stop event is used as the output label to construct a random forest model. Furthermore, the constructed model is trained. After the training is completed, by analyzing the importance of the features in the model, the factors with an impact on elevator stops greater than a pre-set threshold are determined. For example, after model training, it is found that the elevator running at high load for a long time (reflected by features such as operation time and usage frequency) and unstable power supply voltage (reflected by external factor data) are the main reasons for the elevator to stop due to electrical faults. Finally, the key factors affecting elevator stops output by the first machine learning model are compared with the real-time fault information in the initial operation parameters, and the difference between the two is calculated. According to this difference, the adjustment coefficient is calculated in combination with certain rules or algorithms. For example, if the model determines that high load operation is an important factor, and the difference between the current real-time operation load of the elevator and the historical high load situation is large, then the adjustment coefficient may increase accordingly to more significantly adjust the basic probability.

[0105] In addition, according to the periodic change trend of the occurrence of faults, such as the long-term upward or downward trend of the fault occurrence frequency, seasonal fluctuations, etc., the adjustment margin is determined. If the fault occurrence frequency shows an upward trend and the current stage is a stage with a more obvious upward trend in this trend, then the adjustment margin can be appropriately increased; conversely, if the trend is stable or downward, the adjustment margin can be correspondingly decreased. For example, the occurrence frequency of a certain type of fault has increased every year in the past few years, and the current stage is an accelerating upward stage, and the adjustment margin can be set to a relatively large value.

[0106] (5) Determine the adjustment factor based on the basic probability value, the cause of the fault, the adjustment coefficient, and the adjustment margin, and calculate the elevator stop probability according to the adjustment factor.

[0107] It should be noted that the setting of the adjustment factor needs to comprehensively consider information such as the basic probability and the cause of the failure. If the basic probability of a certain failure is relatively high, and the main cause of this failure is more likely to occur under the current circumstances (for example, the main cause of electrical failures is more likely to occur due to high temperatures in summer), and this failure has obvious seasonal variations (such as a high incidence of electrical failures in summer), then the adjustment factor needs to be increased. For example, for electrical failures in summer, assuming the basic probability is 0.3, since high temperatures in summer make the main cause of electrical failures (such as overheating of electrical components) more likely to occur, and the failure has a seasonal high-incidence characteristic, the adjustment factor can be set to 1.5. On the contrary, if the basic probability of a certain failure is relatively low, and its main cause has been effectively controlled, and there are no obvious seasonal or periodic variations, the adjustment factor can be set to a value less than 1, such as 0.8.

[0108] Specifically, considering the basic probability value, the cause of the failure, the adjustment coefficient, and the adjustment margin, determine the setting rule of the adjustment factor. When the cause of the failure is highly correlated with seasonal or periodic change factors and the basic probability is relatively high, the adjustment factor = 1 + k1×basic probability + k2×correlation coefficient (k1 and k2 are coefficients determined according to the actual situation, and the correlation coefficient represents the degree of correlation between the cause of the failure and seasonal or periodic change factors, with a value range of 0 - 1); when the cause of the failure has a low correlation with seasonal or periodic change factors and the basic probability is relatively low, the adjustment factor = 1 - k3×basic probability - k4×correlation coefficient (k3 and k4 are coefficients determined according to the actual situation). In addition, the situations where the cause of the failure is highly correlated with seasonal or periodic change factors but the basic probability is relatively low and the cause of the failure has a low correlation with seasonal or periodic change factors but the basic probability is relatively high also need to be considered. For example, when the basic probability is relatively high but the correlation with seasonal or periodic change factors is relatively low, the adjustment factor can be set to a value between the above two situations, such as the adjustment factor = 1 + k5×basic probability - k6×correlation coefficient (k5 and k6 are also determined according to the actual situation).

[0109] Finally, multiply the previously calculated basic probability by the set adjustment factor, that is, the elevator stop probability = basic probability × adjustment factor. For example, if the basic probability of a certain failure is 0.3 and the adjustment factor is 1.5, then the elevator stop probability caused by this failure is 0.3×1.5 = 0.45; if the basic probability is 0.2 and the adjustment factor is 0.8, then the elevator stop probability is 0.2×0.8 = 0.16.

[0110] It should be noted that the obtained elevator stop probability can more accurately reflect the actual possibility of the elevator stopping due to a certain failure under the current circumstances. This probability will be used as one of the important bases for subsequent judgment of whether the elevator stops, helping maintenance personnel and relevant systems to more accurately assess the operation risk of the elevator and make reasonable decisions, such as whether maintenance needs to be carried out in advance and whether monitoring needs to be strengthened.

[0111] In addition, after obtaining the elevator stopping probability, it also includes verifying the adjusted elevator stopping probability using cross-validation or time series backtesting methods to ensure that the adjusted probability can accurately predict future elevator stopping events. Regularly evaluate and update the model based on the new fault event data to ensure its accuracy and effectiveness.

[0112] In this step, by comprehensively considering factors such as the change trend of faults, fault causes, and seasonal or periodic changes, the basic probability is adjusted to obtain the elevator stopping probability, making the calculation of the elevator stopping probability more scientific and accurate. It can more comprehensively reflect the impact of various factors on elevator stopping during the elevator operation process, providing a more reliable guarantee for the safe operation of the elevator. Through accurate calculation of the elevator stopping probability, potential high-risk faults can be discovered in advance, preventive measures can be taken in a timely manner, the occurrence of elevator faults can be reduced, the reliability and safety of the elevator can be improved, and it also helps to reasonably arrange maintenance resources and reduce maintenance costs.

[0113] S106. Based on the elevator operation mode and fault characteristics in the initial operation data, use the second machine learning model to perform a second anomaly determination.

[0114] It should be noted that using the second machine learning model to perform a second anomaly determination based on the elevator operation mode and fault characteristics in the initial operation data includes:

[0115] (1) Collect the operation data and fault event records of the elevator, and preprocess the collected data.

[0116] It should be noted that collect the operation data of the elevator (such as operation time, load, frequently stopped floors, etc.) and fault event records (fault type, fault time, fault duration, etc.). Among them, the data sources can be the elevator control system, sensors (such as temperature, vibration sensors), maintenance records, etc.

[0117] Among them, the preprocessing can include handling missing values, outliers, and duplicate data to ensure the integrity and consistency of the data. Specifically, the original data can be converted into a format suitable for machine learning algorithms, such as numerical classification variables (such as fault types), normalized or standardized numerical features (such as operation time, load, etc.). Extract features related to elevator operation and faults. For example, the average operation time, load, operation frequency, etc. before the occurrence of faults. In addition, time features (such as season, month, day of the week, etc.) can also be considered, which may affect the usage pattern of the elevator.

[0118] (2) Extract the elevator operation mode and fault characteristics from the preprocessed data.

[0119] The preprocessed data contains a large amount of information from which the elevator operation mode and fault characteristics are extracted. The operation mode characteristics may include the start-stop pattern, speed change pattern, floor stop sequence, etc. of the elevator. By analyzing these characteristics, it can be judged whether the elevator is operating normally. The fault characteristics focus on the data related to faults, such as the changes in operation parameters accompanied by specific faults, the frequency and interval time of fault occurrences, etc. These characteristics are the key basis for judging whether there are potential faults in the elevator.

[0120] (3) Based on the extracted elevator operation mode and fault characteristics, select a classification algorithm to train the second machine learning model.

[0121] It should be noted that common classification algorithms such as neural networks (with powerful non-linear fitting capabilities, capable of learning complex feature relationships, such as multi-layer perceptrons, convolutional neural networks, etc.), decision trees (easy to understand and interpret, making decision classifications based on features), random forests (composed of multiple decision trees, improving the model accuracy and stability through ensemble learning, capable of handling large-scale data and evaluating feature importance), and support vector machines (performing well in small sample and high-dimensional data, classifying by finding the optimal classification hyperplane). Input the data after preprocessing and feature extraction into the selected algorithm for training. During the training process, continuously adjust the model parameters (such as the weights and biases of neural networks), and adopt optimization algorithms (such as stochastic gradient descent, Adam optimizer, etc.) to minimize the error between the prediction result and the actual result. To prevent the model from overfitting, use regularization methods (such as L1, L2 regularization) to constrain the model complexity, and at the same time use cross-validation (such as K-fold cross-validation) to evaluate the model performance to ensure that the model has good generalization ability.

[0122] (4) Deploy the trained second machine learning model to the elevator monitoring system, analyze the elevator operation data in real time, output the fault prediction result, and judge whether there is a second anomaly according to the fault prediction result.

[0123] It should be noted that the trained second machine learning model is deployed in the elevator monitoring system and can be integrated into the local monitoring software or the cloud server to ensure real-time reception of elevator operation data. When the elevator is running, the monitoring system collects data in real time and inputs it into the model. The model analyzes the data according to the patterns learned during training and outputs the fault prediction result, that is, the probability of the elevator stopping due to a fault within a certain period in the future. A reasonable threshold is set. When the predicted stopping probability of the model exceeds the threshold, it is determined that there is a second anomaly; when it is lower than the threshold, it is determined that the elevator is running normally. This data-driven anomaly judgment method can timely detect potential fault risks, provide strong support for elevator maintenance and management, and improve the safety and reliability of elevator operation. In this step, by analyzing elevator operation data in real time, abnormal situations that may exist during the elevator operation process can be timely detected, especially when no problems are found in the first anomaly judgment, further improving the accuracy and reliability of elevator anomaly judgment. Timely detection of the second anomaly allows elevator maintenance personnel to quickly take measures to prevent elevator failures and ensure the safety of passengers and the normal operation of the elevator. At the same time, by continuously optimizing and improving the second machine learning model, the performance of the model can be improved, better adapting to the operation characteristics and environmental changes of different elevators, and providing strong support for the intelligent management of elevators.

[0124] S107. Combine the first anomaly judgment, the stopping probability, and the second anomaly judgment to comprehensively determine whether the elevator to be judged is in a stopped state; wherein, calculate the comprehensive anomaly judgment result based on the weighted sum of the first anomaly judgment, the stopping probability, and the second anomaly judgment.

[0125] It should be noted that combining the first anomaly judgment, the stopping probability, and the second anomaly judgment to comprehensively determine whether the elevator to be judged is in a stopped state includes:

[0126] In different operation scenarios, calculate the corresponding first weight, second weight, and third weight respectively according to the importance of the first anomaly judgment, the stopping probability, and the second anomaly judgment in the elevator stopping determination process; the sum of the first weight, the second weight, and the third weight is 1; calculate the first anomaly judgment score based on the ratio and the first weight, calculate the stopping probability score based on the product of the stopping probability and the second weight, and calculate the second anomaly judgment score based on the second anomaly occurrence probability and the third weight; calculate the total score based on the first anomaly judgment score, the stopping probability score, and the second anomaly judgment score, and determine the elevator state based on the total score.

[0127] Among them, calculating the first anomaly judgment score includes:

[0128] Determine the size relationship between the ratio and 1. If the ratio is greater than or equal to 1, the first abnormal judgment score is equal to 1; if the ratio is less than 1, the first abnormal judgment score is equal to the difference between 1 and the product of the ratio and the first weight.

[0129] Calculate the elevator stop probability score, including:

[0130] Obtain the elevator stop probability from the elevator stop probability model constructed by using a machine learning algorithm; calculate the elevator stop probability score based on the elevator stop probability and the second weight; the elevator stop probability score is equal to the difference between 1 and the product of the elevator stop probability and the second weight.

[0131] Calculate the second abnormal judgment score, including:

[0132] Based on the analysis result of the second machine learning model, predict the probability of the second abnormality; calculate the second abnormal judgment score based on the predicted probability of the second abnormality and the third weight; the second abnormal judgment score is equal to the product of the difference between 1 and the probability of the second abnormality and the third weight.

[0133] It should be noted that according to the importance of the first abnormal judgment, the elevator stop probability, and the second abnormal judgment in the elevator stop discrimination process, the corresponding first weight, second weight, and third weight are set respectively. The sum of these three weights is 1. For example, if it is considered that the first abnormal judgment plays a crucial role in the discrimination, the first weight can be set to 0.5; the importance of the elevator stop probability is second, and the second weight is set to 0.3; the relative importance of the second abnormal judgment is slightly lower, and the third weight is set to 0.2. The specific setting of the weights needs to be determined according to the actual business requirements, data characteristics, and past experience.

[0134] Referring to the above description, when calculating the first abnormal judgment score, if the ratio of the number of running times calculated in step S103 is greater than or equal to 1, this indicates that the number of running times of the elevator to be discriminated is not significantly abnormal compared with the average number of running times of other elevators in the same place. At this time, the first abnormal judgment score is equal to 1.

[0135] If the ratio of the number of running times is less than 1, it means that the number of running times of the elevator to be discriminated is significantly less than that of other elevators, and there is a certain possibility of abnormality. The first abnormal judgment score is equal to the difference between 1 and the product of the ratio and the first weight. Expressed by the formula: if the ratio < 1, the first abnormal judgment score = 1 - ratio × first weight. For example, the first weight is 0.5 and the ratio is 0.8, then the first abnormal judgment score = 1 - 0.8 × 0.4 = 0.60.

[0136] When calculating the elevator stop probability score, obtain the elevator stop probability from the elevator stop probability model constructed using a machine learning algorithm. This model is obtained through the analysis, processing, and adjustment of historical data and initial operation data in steps S104 and S105.

[0137] Calculate the elevator stop probability score based on the elevator stop probability and the second weight. The elevator stop probability score is equal to the difference between 1 and the product of the elevator stop probability and the second weight. The formula is: Elevator stop probability score = 1 - elevator stop probability × second weight. For example, if the elevator stop probability is 0.2 and the second weight is 0.3, then the elevator stop probability score = 1 - 0.2 × 0.3 = 0.94.

[0138] When calculating the second anomaly judgment score, obtain the second anomaly prediction probability from the second anomaly judgment model constructed using a machine learning algorithm. This model is constructed through learning using the initial operation data and the elevator stop probability in step S106.

[0139] Calculate the second anomaly judgment score based on the second anomaly prediction probability and the third weight. The second anomaly judgment score is equal to the product of the difference between 1 and the second anomaly prediction probability and the third weight. The formula is: Second anomaly judgment score = (1 - second anomaly prediction probability) × third weight. For example, if the second anomaly prediction probability is 0.1 and the third weight is 0.3, then the second anomaly judgment score = (1 - 0.1) × 0.2 = 0.18.

[0140] Finally, add the first anomaly judgment score, the elevator stop probability score, and the second anomaly judgment score to obtain the total score. Determine the elevator status based on this total score. A threshold can be set. If the total score is greater than the threshold, it is determined that the elevator is in normal operation; if the total score is less than or equal to the threshold, it is determined that the elevator is in a stopped state. For example, set the threshold to 0.7. If the total score is greater than 0.7, it is determined that the elevator is normal. If the total score is between 0.4 and 0.7, it is determined that the elevator is in an unstable state and needs further monitoring. If the total score is less than 0.4, it is determined that the elevator is in a stopped state.

[0141] In this step, through this comprehensive judgment method, the information obtained in the previous steps is fully utilized, reducing the limitations and errors of single judgment, improving the accuracy and reliability of elevator stop discrimination, helping to detect elevator faults in a timely manner, and ensuring the safe operation of the elevator.

[0142] The method provided in this embodiment comprehensively improves the safety and reliability of elevator operation and reduces the operation cost from multiple aspects such as data acquisition and analysis, fault warning, probability calculation, anomaly judgment, and comprehensive decision-making. Specifically, by obtaining multi-modal initial operation data covering operation data, fault logs, sensor data, and other elevator operation times, rich information is provided for accurate discrimination. Through integrated analysis, the operation state of the elevator can be understood from multiple dimensions, and it is easier to discover potential faults compared with traditional single data sources. Based on the initial operation data, the stopping position of the elevator is judged, and a warning is directly issued when it is not at a flat floor. Stopping at a non-flat floor is dangerous. This step can detect and notify personnel in time for handling, such as sending an alarm to the management personnel and giving audible and visual reminders in the car and on the floor, which can avoid accidents such as passengers falling when entering and leaving the elevator, effectively guaranteeing the safety of passengers' lives. By performing the first anomaly judgment, when the ratio is abnormal, potential problems of the elevator can be discovered in advance. For example, too few operation times may indicate that the fault has not been repaired, and too many may indicate high-load operation that is prone to cause faults. After discovering the anomaly in time, measures such as inspection and repair, adjustment of operation settings, or diversion of passengers can be taken to prevent faults from occurring.

[0143] In addition, by combining historical and initial operation data, the basic probability is calculated using the conditional probability formula, the change trend of the fault mode is analyzed, and factors such as seasonality and periodicity are considered to adjust and obtain the stopping probability. This enables maintenance personnel to reasonably arrange maintenance for high-probability faults, such as increasing the maintenance frequency and stocking spare parts, reducing unnecessary maintenance, lowering costs, extending the service life of the elevator, and realizing scientific management of the entire life cycle. By performing the second anomaly judgment, problems missed by the first anomaly judgment can be discovered in time. Finally, by comprehensively considering the results of the first anomaly judgment, the stopping probability, and the second anomaly judgment, the elevator state is determined by setting weights and calculating scores. This multi-dimensional comprehensive judgment overcomes the limitations of single judgment, reduces misjudgment and missed judgment, and sets weights according to different situations to make the results more in line with the actual situation.

[0144] Embodiment 2:

[0145] Corresponding to the foregoing embodiment of the elevator stopping discrimination method for heterogeneous multi-modal data, the present application also provides an embodiment of an elevator stopping discrimination device for heterogeneous multi-modal data.

[0146] Figure 2 It is a schematic structural diagram of Embodiment 2 of the elevator stopping discrimination device for heterogeneous multi-modal data provided by the present application. Please refer to Figure 2 This embodiment provides a device, including an acquisition module 210, a judgment module 220, a calculation module 230, and an adjustment module 240;

[0147] Among them, the acquisition module 210 is used to acquire the initial operation data of all elevators in the same place;

[0148] The determination module 220 is configured to determine the relationship between the elevator stop position to be determined and the non-leveling position based on the position data in the initial operation data. If it is not in the non-leveling position, calculate the ratio of the number of operations of the elevator to be determined to the average number of operations of other elevators in the same location.

[0149] The determination module 220 is further configured to perform a first anomaly determination based on the relationship between the ratio and the operation number threshold.

[0150] The calculation module 230 is configured to extract historical fault data from the elevator control system, maintenance records, and fault logs, calculate the basic probability of each fault causing a stop based on the historical fault data, and identify the change trend of the fault mode over time based on the time series analysis method.

[0151] The adjustment module 240 is configured to adjust the basic probability according to the change trend and the first machine learning model to obtain the stop probability. The first machine learning model is used to analyze the historical fault data extracted from the elevator control system, maintenance records, and fault logs, determine the factors with an influence degree on the stop greater than the threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factors output by the first machine learning model and the real-time fault information in the initial operation parameters, and adjust the basic probability based on the adjustment margin and the adjustment coefficient.

[0152] The determination module 220 is further configured to perform a second anomaly determination using the second machine learning model based on the elevator operation mode and fault characteristics in the initial operation data.

[0153] The determination module 220 is further configured to comprehensively determine whether the elevator to be determined is in a stopped state by combining the first anomaly determination, the stop probability, and the second anomaly determination. The comprehensive anomaly determination result is calculated based on the weighted sum of the first anomaly determination, the stop probability, and the second anomaly determination.

[0154] The device in this embodiment can be used to execute Figure 1 the steps of the method embodiment shown. The specific implementation principle and process are similar and will not be elaborated here.

[0155] The implementation processes of the functions and roles of each unit in the above device are specifically detailed in the implementation processes of the corresponding steps in the above method and will not be elaborated here.

[0156] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the descriptions of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. A person of ordinary skill in the art can understand and implement it without creative work.

[0157] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included within the scope of protection of this application.

Claims

1. A method for distinguishing elevator stops based on heterogeneous multimodal data, characterized in that: The method comprises: Obtain initial operation data of all elevators in the same place; Based on the position data in the initial operation data, the relationship between the stop position of the elevator to be determined and the non-leveling position is determined. If the elevator is not in the non-leveling position, the ratio of the number of operations of the elevator to be determined to the average number of operations of other elevators in the same place is calculated; Perform a first abnormality judgment based on the relationship between the ratio and the operation number threshold; Extract historical fault data from the elevator control system, maintenance records and fault logs, calculate the basic probability of each fault causing the elevator to stop based on the historical fault data, and identify the change trend of the fault mode over time based on the time series analysis method; The basic probability is adjusted according to the change trend and the first machine learning model to obtain the elevator stop probability, wherein the first machine learning model is used to analyze the historical fault data extracted from the elevator control system, maintenance records and fault logs, determine the factors whose influence on the elevator stop is greater than a threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factor output by the first machine learning model and the real-time fault information in the initial operating parameters, and adjust the basic probability based on the adjustment margin and the adjustment coefficient; Based on the elevator operation mode and fault characteristics in the initial operation data, using a second machine learning model to perform a second abnormality judgment; In combination with the first abnormal judgment, the elevator stop probability and the second abnormal judgment, a comprehensive judgment is made as to whether the elevator to be judged is in a stopped state; wherein the comprehensive abnormal judgment result is calculated based on the weighted sum of the first abnormal judgment, the elevator stop probability and the second abnormal judgment.

2. The method according to claim 1, characterized in that The first abnormality judgment based on the relationship between the ratio and the operation number threshold comprises: Comparing the ratio with a preset operation number threshold range, wherein the preset operation number threshold range includes an upper limit and a lower limit; If the ratio is less than the lower limit or greater than the upper limit, it is determined to be a first abnormality; If the ratio is between the lower limit and the upper limit, it is determined that the elevator is operating normally.

3. The method according to claim 1, characterized in that The method of calculating the basic probability of each fault causing the elevator to stop based on the historical fault data and identifying the changing trend of the fault mode over time based on the time series analysis method includes: Arranging the historical data in chronological order; Based on historical data, for each fault type, the basic probability of the fault type causing the elevator to stop is calculated using the conditional probability formula; The variation process of the basic probability over time is fitted to identify the variation trend of the failure mode.

4. The method according to claim 1, characterized in that: The step of adjusting the basic probability according to the change trend and the first machine learning model to obtain the elevator stopping probability includes: Determine the periodic trend of fault occurrence based on the fault mode change trend identified in the process of basic probability changing over time; Determining the current running time of the elevator to be identified; According to the current running time of the elevator to be identified, find the basic probability value corresponding to each time point in the determined periodic change trend of the fault; Determine a fault cause and an adjustment coefficient based on the first machine learning model, and determine an adjustment margin based on the periodic change trend; An adjustment factor is determined based on the basic probability value, the fault cause, the adjustment coefficient and the adjustment margin, and the elevator shutdown probability is calculated according to the adjustment factor.

5. The method according to claim 1, characterized in that The method of performing a second abnormality judgment based on the elevator operation mode and fault characteristics in the initial operation data using a second machine learning model includes: Collect elevator operation data and fault event records, and pre-process the collected data; Extract elevator operation mode and fault characteristics from preprocessed data; Based on the extracted elevator operation mode and fault characteristics, selecting a classification algorithm to train the second machine learning model; The trained second machine learning model is deployed to the elevator monitoring system, the elevator operation data is analyzed in real time, the fault prediction results are output, and whether there is a second abnormality is determined based on the fault prediction results.

6. The method according to claim 1, characterized in that The step of comprehensively judging whether the elevator to be judged is in a stopped state by combining the first abnormality judgment, the elevator stopping probability and the second abnormality judgment comprises: In different operation scenarios, according to the importance of the first abnormality judgment, the elevator stop probability and the second abnormality judgment in the elevator stop judgment process, the corresponding first weight, second weight and third weight are calculated respectively; the sum of the first weight, the second weight and the third weight is 1; Calculate a first abnormality judgment score based on the ratio and the first weight, calculate an elevator stop probability score based on the product of the elevator stop probability and the second weight, and calculate a second abnormality judgment score based on the second abnormality occurrence probability and the third weight; A total score is calculated based on the first abnormality judgment score, the elevator stop probability score, and the second abnormality judgment score, and the elevator state is determined based on the total score.

7. The method according to claim 6, characterized in that Calculating the first abnormality judgment score includes: Determine the magnitude relationship between the ratio and 1, if the ratio is greater than or equal to 1, the first abnormality judgment score is equal to 1; If the ratio is less than 1, the first abnormality judgment score is equal to the difference between 1 and the product of the ratio and the first weight.

8. The method according to claim 6, characterized in that Calculate the elevator stop probability score, including: Obtaining the elevator stop probability from an elevator stop probability model constructed using a machine learning algorithm; An elevator stop probability score is calculated based on the elevator stop probability and the second weight; the elevator stop probability score is equal to the difference between 1 and the product of the elevator stop probability and the second weight.

9. The method according to claim 6, characterized in that The calculating the second abnormality judgment score comprises: Based on the analysis results of the second machine learning model, predict the probability of occurrence of the second anomaly; A second abnormality judgment score is calculated based on the predicted second abnormality occurrence probability and the third weight; the second abnormality judgment score is equal to the product of the difference between 1 and the second abnormality occurrence probability and the third weight.

10. An elevator stop determination device for heterogeneous multimodal data, characterized in that: The device comprises an acquisition module, a judgment module, a calculation module and an adjustment module; Wherein, the acquisition module is used to acquire the initial operation data of all elevators in the same place; The judging module is used to judge the relationship between the stopping position of the elevator to be judged and the non-leveling position based on the position data in the initial operation data, and if it is not in the non-leveling position, calculate the ratio of the running number of the elevator to be judged to the average running number of other elevators in the same place; The judgment module is further used to perform a first abnormality judgment based on the relationship between the ratio and the operation number threshold; The calculation module is used to extract historical fault data from the elevator control system, maintenance records and fault logs, calculate the basic probability of each fault causing the elevator to stop based on the historical fault data, and identify the change trend of the fault mode over time based on the time series analysis method; The adjustment module is used to adjust the basic probability according to the change trend and the first machine learning model to obtain the elevator stop probability, wherein the first machine learning model is used to analyze the historical fault data extracted from the elevator control system, maintenance records and fault logs, determine the factors whose influence on the elevator stop is greater than a threshold, determine the adjustment margin based on the change trend, calculate the adjustment coefficient based on the difference between the factor output by the first machine learning model and the real-time fault information in the initial operating parameters, and adjust the basic probability based on the adjustment margin and the adjustment coefficient; The judgment module is further used to perform a second abnormality judgment using a second machine learning model based on the elevator operation mode and fault characteristics in the initial operation data; The judgment module is also used to combine the first abnormal judgment, the elevator stop probability and the second abnormal judgment to comprehensively judge whether the elevator to be judged is in a stopped state; wherein the comprehensive abnormal judgment result is calculated based on the weighted sum of the first abnormal judgment, the elevator stop probability and the second abnormal judgment.

Images