Elevator safety risk self-adaptive analysis method and system

Abstract

The invention relates to the technical field of elevator safety risk analysis, in particular to an elevator safety risk self-adaptive analysis method and system. The method comprises the following steps: performing misoperation code execution unit mapping on an elevator operation logic code according to an elevator historical operation log to obtain a misoperation code execution unit; performing abnormal speed change code logic analysis on the misoperation code execution unit, and performing trigger condition backtracking to obtain abnormal speed change trigger condition data; performing causal relationship network construction based on the abnormal speed change triggering condition data to obtain a potential risk causal relationship network; and building an elevator safety risk self-adaptive analysis model for the potential risk causal relationship network based on a random forest algorithm to obtain the elevator safety risk self-adaptive analysis model. The elevator safety risk analysis technology is optimized, so that the elevator safety risk analysis technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator safety risk analysis, and in particular to an elevator safety risk adaptive analysis method and system. Background Art

[0002] With the rapid development of information technology and big data analysis technology, artificial intelligence and machine learning technologies have gradually been introduced into the field of elevator safety management. These technologies provide new ideas and methods for elevator operation status monitoring, fault warning and risk assessment. Through in-depth analysis of the historical operation logs of the elevator, potential safety hazards can be discovered, which helps to achieve real-time monitoring and prediction of the system operation status. However, a traditional adaptive analysis method for elevator safety risks has a large error in the cause analysis of abnormal speed control of the elevator, resulting in low accuracy of elevator safety risk analysis. Summary of the invention

[0003] Based on this, it is necessary to provide an elevator safety risk adaptive analysis method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for adaptively analyzing elevator safety risks is provided, the method comprising the following steps: Step S1: Obtain the elevator historical operation log and the elevator operation logic code; perform a malfunction code execution unit mapping on the elevator operation logic code according to the elevator historical operation log to obtain a malfunction code execution unit; Step S2: performing abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; performing trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; Step S3: performing a potential risk code segment linkage analysis based on the abnormal speed change trigger condition data to obtain potential risk code segment linkage data; constructing a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; Step S4: construct an elevator safety risk adaptive analysis model for the potential risk causal relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0005] Preferably, step S1 comprises the following steps: Step S11: Obtain the elevator historical operation log and the elevator operation logic code; Step S12: Clean the elevator historical operation log to obtain the elevator historical operation cleaning log; Step S13: extracting the malfunction behavior from the elevator historical operation cleaning log to obtain the elevator historical malfunction behavior; Step S14: mapping the elevator operation logic code to the malfunction code execution unit according to the historical malfunction behavior of the elevator to obtain the malfunction code execution unit.

[0006] Preferably, step S2 comprises the following steps: Step S21: Identify the abnormal speed change behavior of the elevator's historical malfunction behavior to obtain the abnormal speed change behavior of the elevator; Step S22: analyzing the speed change risk factor according to the elevator speed change abnormal behavior and the elevator historical operation cleaning log to obtain the elevator speed change risk factor; Step S23: performing abnormal speed change code logic analysis on the malfunction code execution unit based on the elevator speed change risk factor to obtain abnormal speed change code logic data; Step S24: backtrack the triggering condition of the abnormal speed change code logic data to obtain abnormal speed change triggering condition data.

[0007] Preferably, step S22 comprises the following steps: Step S221: extracting the load and personnel density of the elevator historical operation cleaning log to obtain the elevator load data and the elevator personnel density data; Step S222: performing speed change jerking fluctuation analysis on the abnormal speed change behavior of the elevator to obtain the speed change jerking fluctuation data of the elevator; Step S223: Estimating the dynamic change of the inertial gravity of the elevator load data and the elevator personnel density data according to the elevator speed change jerking fluctuation data, and obtaining the inertial gravity change data of the elevator; Step S224: obtaining the wire traction material data of the elevator traction sheave; Step S225: quantifying the indirect tension imbalance of the wire traction material data according to the ascending / descending inertial gravity change data to obtain the traction indirect tension imbalance data; Step S226: Analyze the speed change risk factor of the elevator speed change jerking fluctuation data according to the traction indirect tension imbalance data to obtain the elevator speed change risk factor.

[0008] Preferably, step S23 includes the following steps: Step S231: locating the speed change risk abnormal code of the malfunction code execution unit based on the elevator speed change risk factor to obtain the speed change risk abnormal code; Step S232: Perform syntax reverse engineering analysis on the speed change risk abnormality code to obtain syntax analysis data of the speed change abnormality code; Step S233: performing speed change abnormal cycle condition setting analysis on the speed change abnormal code syntax parsing data to obtain the speed change abnormal cycle setting conditions; Step S234: performing variable parameter overflow control analysis on the speed change abnormality code syntax analysis data according to the speed change abnormality cycle setting condition to obtain variable parameter overflow control data; Step S235: performing function exception throwing analysis based on the speed change abnormal cycle setting conditions and the variable parameter overflow control data to obtain function exception throwing data; Step S236: Based on the speed change abnormal cycle setting conditions, variable parameter overflow control data and function abnormal throwing data, the speed change risk abnormal code is logically analyzed to obtain abnormal speed change code logic data.

[0009] Preferably, step S3 comprises the following steps: Step S31: normalizing the abnormal speed change trigger condition data to obtain abnormal speed change trigger condition normalized data; Step S32: performing potential risk code segment linkage analysis on the abnormal speed change code logic data based on the abnormal speed change trigger condition normalization data to obtain potential risk code segment linkage data; Step S33: construct a causal relationship network for the abnormal speed change code logic data according to the potential risk code segment linkage data to obtain a potential risk causal relationship network.

[0010] Preferably, step S4 comprises the following steps: Step S41: performing risk characteristic analysis on the potential risk causal relationship network to obtain a potential risk causal characteristic relationship network; Step S42: assigning risk causal relationship weights to the potential risk causal feature relationship network to obtain a potential risk relationship weight relationship network; Step S43: construct an elevator safety risk adaptive analysis model for the potential risk relationship weight relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0011] Preferably, the present invention further provides an elevator safety risk adaptive analysis system for executing the elevator safety risk adaptive analysis method as described above, the elevator safety risk adaptive analysis system comprising: The malfunction code mapping module is used to obtain the elevator historical operation log and the elevator operation logic code; according to the elevator historical operation log, the elevator operation logic code is mapped to the malfunction code execution unit to obtain the malfunction code execution unit; The trigger condition backtracking module is used to perform abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; perform trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; A causal relationship network construction module is used to perform a potential risk code segment linkage analysis based on abnormal speed change trigger condition data to obtain potential risk code segment linkage data; and to construct a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; The adaptive analysis model construction module is used to construct an elevator safety risk adaptive analysis model for the potential risk causal network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; the elevator safety risk adaptive analysis model is sent to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0012] The beneficial effect of the present invention is that, firstly, by obtaining the historical operation log of the elevator, the operation status of the elevator in actual use can be deeply understood, including the records of normal operation and abnormal operation. At the same time, by comparing and analyzing these historical data with the operation logic code of the elevator, the malfunction code execution unit can be identified. Through this mapping relationship, we can reveal which operation logic causes the elevator to be improperly or misoperated, and provide basic data for subsequent optimization and fault analysis, and provide important evidence for the stability and safety of the elevator. After identifying the malfunction code, the next step is to perform abnormal speed change code logic analysis on it. This process helps to analyze the abnormal speed change phenomenon that occurs during operation, and further obtains the logic data of the abnormal speed change code. These data will be used to trace back the trigger conditions and understand under what circumstances the elevator has abnormal speed changes. By deeply digging into these trigger conditions, the logic behind the elevator behavior can be better understood, and data support can be provided for subsequent risk prediction and improvement measures, thereby improving the safety and reliability of the elevator. After obtaining the abnormal speed change trigger condition data, this step will perform linkage analysis of potential risk code segments around these data. By analyzing the relationship between each code segment, the potential risks within the system and their interrelationships can be identified. In addition, the construction of the causal network can provide us with a more comprehensive perspective to observe how various code segments interact with each other and the safety hazards caused under specific conditions. Such analysis not only helps to understand the current elevator safety situation, but also provides data basis for future risk prediction, which helps the relevant technical team to take measures to avoid risks in advance. Finally, based on the constructed potential risk causal network, the random forest algorithm is used to build a model to achieve adaptive analysis of elevator safety risks. The application of this machine learning technology can provide a more intelligent solution for elevator safety management. By creating an elevator safety risk adaptive analysis model, the elevator operation status can be monitored and evaluated in real time, and the analysis results can be fed back to the elevator operation control center in a timely manner, so that the operator can obtain risk warning information in time. Such implementation not only improves the efficiency of elevator safety monitoring, but also makes elevator management more systematic and intelligent, and further protects the lives of passengers. Therefore, the present invention is an optimization process made to a traditional elevator safety risk adaptive analysis method, which solves the problem that the error of the cause analysis of abnormal speed change control of the elevator in the traditional elevator safety risk adaptive analysis method is large, thereby causing the problem of low accuracy of elevator safety risk analysis, reduces the error of the cause analysis of abnormal speed change control of the elevator, and improves the accuracy of elevator safety risk analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of the steps of an elevator safety risk adaptive analysis method; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. DETAILED DESCRIPTION

[0014] See also Figures 1 to 3 , an elevator safety risk adaptive analysis method, the method comprising the following steps: Step S1: Obtain the elevator historical operation log and the elevator operation logic code; perform a malfunction code execution unit mapping on the elevator operation logic code according to the elevator historical operation log to obtain a malfunction code execution unit; Step S2: performing abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; performing trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; Step S3: performing a potential risk code segment linkage analysis based on the abnormal speed change trigger condition data to obtain potential risk code segment linkage data; constructing a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; Step S4: construct an elevator safety risk adaptive analysis model for the potential risk causal relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0015] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of an elevator safety risk adaptive analysis method of the present invention. In this example, the elevator safety risk adaptive analysis method includes the following steps: Step S1: Obtain the elevator historical operation log and the elevator operation logic code; perform a malfunction code execution unit mapping on the elevator operation logic code according to the elevator historical operation log to obtain a malfunction code execution unit; In the embodiment of the present invention, the historical operation log of the elevator and the operation logic code of the elevator are first obtained. The historical operation log of the elevator is recorded by the elevator control system or related sensors, and contains various operation data of the elevator in different time periods, including up and down floors, operation speed, load, passenger density, etc. After obtaining these historical operation logs, it is necessary to analyze the operation logic code of the elevator. The elevator operation logic code is generally an instruction code run by the elevator controller, and these instructions determine the specific operation mode of the elevator. The operation logic code of the elevator generally includes various state switching, floor control, speed control, fault handling, etc. of the elevator. Next, by analyzing the historical operation log of the elevator, the misoperation behavior recorded in the log is mapped with the elevator operation logic code, and the misoperation in the history of the elevator is associated with the corresponding code execution unit. The misoperation code execution unit generally refers to a control instruction segment or code unit triggered in a specific operation scenario. Specifically, firstly, by extracting the misoperation behavior from the historical operation log of the elevator, the misoperation or abnormal operation behavior is identified, such as the elevator not stopping at a certain floor, abnormal acceleration and deceleration, etc. Then, based on these misoperation behaviors, they are matched with the elevator operation logic code and mapped to the corresponding code segment. The core of this process is to locate the code execution unit that causes the malfunction by comparing various abnormal behaviors in the elevator's historical operation log with the operating steps in the elevator program code.

[0016] Step S2: performing abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; performing trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; In the embodiment of the present invention, the abnormal speed change code logic of the malfunction code execution unit is first analyzed. Speed ​​change abnormality refers to abnormal phenomena that occur during the acceleration and deceleration process of the elevator, such as sudden acceleration, uneven deceleration or sudden acceleration. These abnormal speed change behaviors are caused by hardware failures, software logic errors or external interference of the elevator. Therefore, at this stage, the code segments related to speed change in the malfunction code execution unit are first analyzed to find the logic code that causes abnormal speed change, such as functions and control instructions involving speed regulation, acceleration, and deceleration. After obtaining the code related to speed change abnormality, the trigger condition backtracking of the abnormal speed change code logic is further performed to find the specific conditions that trigger the speed change abnormality. For example, it will be found that some improper sensor input data, external load changes, and the current operating state of the elevator (such as high load state) and other factors cause the elevator to have abnormal speed change. The core method of trigger condition backtracking is to reversely calculate the specific time and conditions of the speed change abnormality based on the elevator's historical operation cleaning log, and match the relevant parameters in the historical log (such as the elevator's current load, speed, floor, etc.) with the relevant trigger conditions in the control code to obtain the trigger condition data of the abnormal speed change.

[0017] Step S3: performing a potential risk code segment linkage analysis based on the abnormal speed change trigger condition data to obtain potential risk code segment linkage data; constructing a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; In the embodiment of the present invention, firstly, the abnormal speed change trigger condition data is used to perform linkage analysis of potential risk code segments. The purpose of linkage analysis is to find out which code segments or modules interact with each other under the trigger condition of abnormal speed change during the operation of the elevator, thereby causing other potential risks to the elevator. In the implementation process, the trigger condition data of abnormal speed change is first extracted from the elevator control system, including a series of related parameters such as the speed, load, and running time of the elevator. In order to perform effective linkage analysis on these data, a linkage analysis model based on graph theory is constructed, in which each code segment or module is used as a node in the graph, and the mutual relationship and data transfer between code segments are used as edges in the graph. Then, based on the abnormal speed change trigger condition data, the linkage relationship between each module is checked when a specific speed change abnormality occurs, and by calculating the similarity or dependency relationship between different code segments, it is identified which code segments have linkage effects, and then the potential risk code segment linkage data is obtained. Specifically, by using data flow analysis technology, the linkage analysis model is constructed by examining the execution order of each code module during the abnormal speed change period and the data interaction between them. This process can calculate the data flow dependency and function call graph between code segments through an algorithm, thereby analyzing the existing potential risk linkage path. Finally, based on the linkage analysis results, a causal network is constructed, which can accurately describe the causal influence between the elevator modules and facilitate further evaluation of risk sources and their extended impact.

[0018] Step S4: construct an elevator safety risk adaptive analysis model for the potential risk causal relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0019] In an embodiment of the present invention, based on the constructed potential risk causal network, a random forest algorithm is used to establish an elevator safety risk adaptive analysis model. First, each node in the causal network is regarded as a feature, representing various potential risks or abnormal events that occur during the operation of the elevator. The features of the node may include the operating status of the elevator, abnormal speed change, load information, external interference, etc. Subsequently, these features are combined with the data in the historical operation log of the elevator to construct a training data set. Each training sample contains feature data at different time points and whether a safety risk event (such as a fault, accident, etc.) has occurred at that time point. In order to train the random forest model, data preprocessing is required, including data standardization and missing value processing to ensure data quality. The random forest algorithm performs classification and regression by constructing multiple decision trees, and each decision tree is trained by randomly selecting a subset of the training data and a subset of the features. The training process calculates the error rate of each tree and synthesizes the final prediction result through a voting mechanism or an average value. For the elevator safety risk analysis problem, each tree will judge whether there is a potential safety risk based on a combination of different features. After the model training is completed, the trained random forest model is applied to the elevator operation control center to monitor the operating status of the elevator in real time. Specifically, the elevator operation control system will collect the elevator operation data (such as floors, speed, load, etc.) in real time and input it into the trained random forest model for prediction. The output of the model will be the safety risk assessment result of the elevator in its current state. The control center will make corresponding adjustment measures based on this result, such as alarm notification, stopping the elevator operation, etc.

[0020] Step S1 includes the following steps: Step S11: Obtain the elevator historical operation log and the elevator operation logic code; Step S12: Clean the elevator historical operation log to obtain the elevator historical operation cleaning log; Step S13: extracting the malfunction behavior from the elevator historical operation cleaning log to obtain the elevator historical malfunction behavior; Step S14: mapping the elevator operation logic code to the malfunction code execution unit according to the historical malfunction behavior of the elevator to obtain the malfunction code execution unit.

[0021] In the embodiment of the present invention, the historical operation log and operation logic code of the elevator are first obtained from the elevator control system. The historical operation log of the elevator records all the operation data generated by the elevator during the past operation process, which usually includes each start, stop, acceleration, deceleration, floor stop, load condition, door switch status, fault alarm information, etc. of the elevator. The operation log is generally stored in the database of the elevator control system in a time series manner, and the format is usually CSV or database table format. The elevator operation logic code is the source code for writing the elevator control system program. These codes define the elevator's operating rules, scheduling algorithms, variable speed control strategies, fault detection and response strategies, etc. The elevator control system manages various behaviors of the elevator through these codes, such as determining how the elevator starts and how to stop at floors under different circumstances. At this time, the way to obtain the historical operation log of the elevator is usually to extract the required historical data from the log table through a database query statement, and the elevator operation logic code can extract the source code or compiled control instructions from the elevator programming development environment. First, the acquired historical operation log of the elevator is cleaned. The goal of data cleaning is to remove erroneous data and irrelevant data in the original log, so that the log data can be more effectively used for subsequent analysis and processing. The cleaning process includes several steps: First, check the data integrity, identify and repair missing values ​​or blank values. Missing value processing methods include filling missing values ​​or deleting records with missing values. Second, check the format and type of the data to ensure that the format of all data meets the expectations (for example, the timestamp field should be in date and time format, and the speed field should be numeric). Next, detect and process outliers, identify values ​​outside the normal range through statistical methods (such as box plot method or 3σ principle), and correct or delete them according to the specific situation. In addition, check the duplicate records in the log, especially when the elevator fails or restarts, duplicate data will be generated, which needs to be deduplicated. After the above data cleaning, the obtained elevator historical operation cleaning log can be used for subsequent analysis and modeling. Extract the erroneous behavior of the elevator historical operation cleaning log. Erroneous behavior refers to abnormal behavior caused by control logic errors, hardware failures or external interference during the operation of the elevator. For example, the elevator accelerates or decelerates abnormally without clear instructions, or stops at the wrong floor. In order to extract these erroneous behaviors, it is necessary to define the standards for abnormal behaviors first. For example, during the acceleration process, if the speed of an elevator changes beyond a predetermined range, or if the time it stops on a specific floor is abnormal, it can be considered a misoperation behavior. Misoperation behavior extraction can be done by writing a rule engine to identify these abnormal situations in the log. The specific operation is based on the speed, floor, time, acceleration and other data in the log, and is compared using the set threshold rules (for example, speed fluctuation range, floor error threshold, etc.). When the data exceeds the threshold, the behavior is recorded as a misoperation.In addition, machine learning algorithms, such as anomaly detection algorithms (such as isolation forest algorithms or clustering-based algorithms), are used to further extract potential misoperation behaviors. These misoperation behavior data can help to find potential problems in the elevator operation logic in the future. The elevator operation logic code is mapped to the misoperation code execution unit according to the historical misoperation behavior of the elevator. The purpose of misoperation code execution unit mapping is to associate the identified misoperation behavior with the specific code modules in the elevator control system and find out the specific code segments that cause these misoperations. In this step, it is first necessary to perform static analysis on the elevator operation logic code to identify the key modules that control the elevator behavior, such as the acceleration control module, the parking control module, the floor scheduling module, etc. Through static analysis, the functions and operating conditions of these modules can be extracted. Then, according to the historical misoperation behaviors of the elevator extracted in the above steps, the triggering conditions of these behaviors in the elevator control logic are found. For example, a misoperation behavior is related to the parameter setting of the acceleration control module of the elevator, or to the floor parking rules under a certain condition. In order to establish a mapping relationship, it is necessary to compare the historical malfunction behaviors of the elevator with the operating logic code one by one to determine which specific code execution units (such as function calls, conditional judgments, etc.) are directly related to the malfunction behaviors. The mapping process can be completed by writing code mapping rules, analyzing the event logs and abnormal behavior triggering conditions of the elevator control system, and combining manual analysis with automated tools (such as static code analysis tools). Ultimately, the malfunction code execution units obtained refer to specific code segments or modules. When these modules are running in the elevator control system, they are prone to cause malfunction behaviors and become the focus of subsequent analysis and optimization.

[0022] Step S2 includes the following steps: Step S21: Identify the abnormal speed change behavior of the elevator's historical malfunction behavior to obtain the abnormal speed change behavior of the elevator; Step S22: analyzing the speed change risk factor according to the elevator speed change abnormal behavior and the elevator historical operation cleaning log to obtain the elevator speed change risk factor; Step S23: performing abnormal speed change code logic analysis on the malfunction code execution unit based on the elevator speed change risk factor to obtain abnormal speed change code logic data; Step S24: backtrack the triggering condition of the abnormal speed change code logic data to obtain abnormal speed change triggering condition data.

[0023] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Identify the abnormal speed change behavior of the elevator's historical malfunction behavior to obtain the abnormal speed change behavior of the elevator; In the embodiment of the present invention, firstly, the abnormal speed change behavior is identified for the historical malfunction behavior of the elevator. The abnormal speed change behavior refers to the acceleration and deceleration phenomenon that does not conform to the predetermined operation logic during the operation of the elevator, such as the elevator accelerating too fast or decelerating too slowly. By analyzing the historical malfunction behavior log of the elevator, the acceleration, speed and position data during the operation of the elevator are first identified, and combined with the elevator operation specification, it is determined what kind of acceleration or deceleration behavior can be regarded as abnormal. For example, if the acceleration of the elevator at the time of starting exceeds the designed maximum acceleration value, or the deceleration time during the parking process is too long, it can be marked as abnormal speed change behavior. In order to achieve this identification, the technology used includes an abnormality detection algorithm based on time series, such as a clustering-based abnormality detection method (such as a K-means clustering algorithm) or a threshold-based method. By real-time monitoring of the speed change and acceleration change data in the elevator operation log, the acceleration of the elevator in each cycle is calculated and compared with the design threshold. When the acceleration exceeds the set threshold, it is marked as abnormal speed change behavior, and the detailed information such as the time, acceleration, and speed change of the event is recorded for subsequent analysis.

[0024] Step S22: analyzing the speed change risk factor according to the elevator speed change abnormal behavior and the elevator historical operation cleaning log to obtain the elevator speed change risk factor; In the embodiment of the present invention, the speed change risk factor is analyzed according to the abnormal speed change behavior of the elevator and the historical operation cleaning log of the elevator to obtain the speed change risk factor of the elevator. The speed change risk factor refers to the factors that affect the occurrence of abnormal speed change behavior of the elevator, such as load, personnel density, floor difference, etc. First, the background analysis of each abnormal speed change behavior is performed according to the acceleration, deceleration, speed and load data in the historical operation cleaning log of the elevator. By analyzing the performance of the elevator when running between different loads and different floors, the factors causing the abnormal speed change are found. For example, if the frequency of abnormal speed change of the elevator under high load is high, the load is a key risk factor. Further, a statistical analysis method (such as linear regression or correlation analysis) is used to calculate the correlation between each factor and the abnormal speed change behavior. The specific operation includes comparing each abnormal speed change behavior record in the historical operation cleaning log of the elevator with the corresponding load, operation time, floor difference and other data, and calculating the contribution of each factor to the abnormal speed change. These factors can be expressed in a weighted manner, and the result obtained is the speed change risk factor. The analysis of the elevator speed change risk factor is helpful for subsequent risk assessment and control measures.

[0025] Step S23: performing abnormal speed change code logic analysis on the malfunction code execution unit based on the elevator speed change risk factor to obtain abnormal speed change code logic data; In an embodiment of the present invention, the abnormal speed change code logic analysis is performed on the malfunction code execution unit based on the elevator speed change risk factor to obtain the abnormal speed change code logic data. The purpose of the abnormal speed change code logic analysis is to find out the code units related to the abnormal speed change behavior in the elevator control system, especially when the abnormal speed change behavior occurs, whether there are problems or defects in the elevator control logic. First, according to the speed change risk factor analysis results, select the risk factors with greater influence (such as load, speed change, etc.), and then map these factors to the corresponding control modules in the elevator operation logic. By analyzing the program logic in the elevator control system (for example, the elevator acceleration and deceleration control algorithm), find out the code segments related to the abnormal speed change behavior. The specific operation steps include: static code analysis of the elevator control code to identify functions or conditional statements related to variables such as speed, acceleration, and load. Then, according to the risk factor data, these code segments are dynamically analyzed to check their execution in the actual operation of the elevator, and identify the logical defects or improper algorithms that cause abnormal speed change. For example, if the change of load factor is not considered in the acceleration algorithm, or the parameters of the deceleration algorithm are not adjusted under high load conditions, abnormal speed change will occur. Finally, through these analyses, the abnormal speed change code logic data is obtained, including specific code modules, conditional judgments and other information.

[0026] Step S24: backtrack the triggering condition of the abnormal speed change code logic data to obtain abnormal speed change triggering condition data.

[0027] In the embodiment of the present invention, the trigger condition backtracking of the abnormal speed change code logic data is performed to obtain the abnormal speed change trigger condition data. The purpose of the trigger condition backtracking is to find out the specific control conditions and triggering mechanisms that lead to the abnormal speed change behavior. According to the abnormal speed change code logic data obtained in step S23, various control conditions in the elevator operation process are backtracked. For example, when the elevator starts, is there an erroneous acceleration instruction, or when the elevator decelerates, is there an inappropriate deceleration condition judgment. First, according to the acceleration or deceleration function defined in the abnormal speed change code logic data, the trigger timing and execution parameters of these functions under different conditions are analyzed. By backtracking the relevant data (such as load, speed, acceleration, etc.) in the elevator historical operation cleaning log, it is determined in what specific situation the abnormal speed change behavior is triggered. For example, if the elevator starts under high load and the acceleration is too fast, it can be backtracked to the judgment condition in the acceleration function, and it is found that there is a lack of dynamic adaptability to the load data. Further, the acceleration and deceleration process of the elevator can be backtracked in combination with the time series analysis method to analyze the specific reasons for triggering the abnormal speed change behavior within a specific time node. Through these retrospective analyses, we can obtain abnormal speed change trigger condition data, including specific control parameters, trigger timing and other information, providing a basis for subsequent optimization and adjustment.

[0028] Step S22 includes the following steps: Step S221: extracting the load and personnel density of the elevator historical operation cleaning log to obtain the elevator load data and the elevator personnel density data; Step S222: performing speed change jerking fluctuation analysis on the abnormal speed change behavior of the elevator to obtain the speed change jerking fluctuation data of the elevator; Step S223: Estimating the dynamic change of the inertial gravity of the elevator load data and the elevator personnel density data according to the elevator speed change jerking fluctuation data, and obtaining the inertial gravity change data of the elevator; Step S224: obtaining the wire traction material data of the elevator traction sheave; Step S225: quantifying the indirect tension imbalance of the wire traction material data according to the ascending / descending inertial gravity change data to obtain the traction indirect tension imbalance data; Step S226: Analyze the speed change risk factor of the elevator speed change jerking fluctuation data according to the traction indirect tension imbalance data to obtain the elevator speed change risk factor.

[0029] In the embodiment of the present invention, the load and population density of the elevator historical operation cleaning log are first extracted to obtain the elevator load data and the elevator population density data. The elevator historical operation cleaning log contains multiple important parameters during the elevator operation process, including information such as the operating speed, acceleration, number of floors, load, and population density in the elevator. First, by parsing the load-related fields in the elevator operation cleaning log, the load data of each elevator operation process is extracted. The load data can be obtained through the original signal provided by the elevator load sensor. Usually, the load sensor of the elevator will provide real-time data when the elevator is running, such as the load value at each start, the load value when reaching the target floor, etc. After filtering, denoising, etc., these data are processed to obtain an accurate load value. Then, the population density data is extracted from the log. The population density data is calculated based on the number of people in the elevator and the carrying capacity of the elevator. The change in population density is calculated by the ratio of the personnel counter in the elevator to the total load of the elevator. The specific operation includes extracting the data of the number of people getting on and off on each floor during each operation of the elevator, and calculating the population density in combination with the maximum load of the elevator. Finally, the load and population density data are sorted into a data set in chronological order for subsequent analysis. The speed change jerking fluctuation analysis is performed on the abnormal speed change behavior of the elevator to obtain the speed change jerking fluctuation data of the elevator. The speed change jerking fluctuation analysis aims to detect whether there is an abrupt speed change during the speed change process of the elevator, especially whether there is an unsteady or violent fluctuation during the acceleration or deceleration process. First, based on the speed and acceleration data in the historical operation cleaning log of the elevator, the original data is denoised using smoothing filtering technology (such as Savitzky-Golay filter) to remove short-term fluctuations caused by sensor noise or environmental interference. Next, the acceleration change rate of the elevator during acceleration and deceleration is calculated to identify the mutation point of speed or acceleration. The excessive change rate during acceleration or deceleration is the speed change jerking fluctuation. The stability during the acceleration process is judged by calculating the standard deviation of the elevator acceleration. When the standard deviation exceeds the set threshold, it can be determined as a speed change jerking fluctuation. Further, a fluctuation index (for example, defined as the ratio of the difference between the maximum acceleration and the minimum acceleration during the speed change process to the average acceleration) is used to quantify the speed change jerking fluctuation. The fluctuation data will be used as an indicator of speed change abnormality and input into the subsequent risk analysis. According to the elevator speed change and jerking fluctuation data, the elevator load data and the elevator population density data are used to estimate the dynamic changes of the rising / falling inertial gravity, and the rising / falling inertial gravity change data is obtained. The estimation of the dynamic changes of the rising / falling inertial gravity is based on the acceleration and speed changes of the elevator in the vertical direction, combined with the influence of load and population density, to analyze the inertial changes of the elevator during acceleration or deceleration. First, the elevator's acceleration sensor data is used to obtain the elevator's vertical acceleration, and the inertial force is calculated based on the elevator's mass (including changes in load and population density). The inertial force can be expressed by the following formula: ,in, is the inertial force, is the mass of the elevator (including data on load and personnel density), is the acceleration. On this basis, combined with the acceleration data of the elevator during the ascent and descent process, the inertia change during the ascent and descent process is calculated. For example, during the ascent of the elevator, the mass of the elevator will be affected by gravity, so the change in inertia force depends on the magnitude and direction of the acceleration. During the descent, the inertia force is affected by the load and the mass of the elevator itself. Therefore, the change in the inertia force of the elevator presents certain dynamic characteristics. By combining the change factors of load and personnel density with the acceleration and speed changes of the elevator, the change in inertia force during the ascent and descent process can be estimated, and the ascent / descent inertia gravity change data can be obtained. This data can provide a basis for subsequent traction force analysis. Obtain the wire traction material data of the elevator traction sheave. As an important part of the elevator power system, the quality and characteristics of its wire traction material are crucial to the safe operation of the elevator. First, it is necessary to obtain detailed data on the wire traction material used in the elevator, including the material of the wire (such as the strength and wear resistance of the wire), diameter, manufacturing process, traction capacity, etc. Extract relevant traction material information through the elevator maintenance record or the material specification provided by the equipment. The tension bearing capacity of the steel wire is crucial to the safe operation of the elevator. Therefore, obtaining this data helps to analyze whether the elevator has insufficient traction under different loads and speed changes. The traction material data can be obtained through the technical documents provided by the elevator manufacturer or maintenance unit, and is distinguished in detail according to the elevator model. The indirect tension imbalance of the steel wire traction material data is quantified according to the rising / falling inertial gravity change data to obtain the traction indirect tension imbalance data. The traction indirect tension imbalance refers to the uneven distribution of tension in the traction system due to the imbalance of inertial force caused by speed change setbacks or the aging of the steel wire traction material during the operation of the elevator. In order to quantify this imbalance, it is first necessary to calculate the tension distribution of the traction system during the operation of the elevator based on the rising / falling inertial gravity change data in step S223 and the characteristics of the steel wire. By analyzing the acceleration and speed data at each moment during the operation of the elevator, the traction force changes between different floors are estimated, and the tension limit of the steel wire is considered. The specific calculation method is: during the acceleration and deceleration of the elevator, the traction force demand at each moment is calculated and compared with the maximum load-bearing tension of the elevator steel wire. If the traction force exceeds the maximum load-bearing capacity of the steel wire, tension imbalance occurs. Finally, the indirect traction tension imbalance data is obtained as an important indicator of elevator operation safety. The speed change risk factor is analyzed for the elevator speed change jerking fluctuation data based on the indirect traction tension imbalance data to obtain the elevator speed change risk factor. The purpose of speed change risk factor analysis is to find out the root cause of the abnormal speed change behavior of the elevator and link it with traction imbalance. First, by correlating the traction indirect tension imbalance data with the speed change jerking fluctuation data, the influence of traction imbalance on the jerking fluctuation during the acceleration and deceleration of the elevator is identified.By tracing back the acceleration and deceleration behaviors between different floors, it is found that traction imbalance causes abnormal speed changes in elevators under certain specific conditions. After combining traction imbalance with speed fluctuation data, risk factor analysis methods (such as principal component analysis or regression analysis) are used to derive speed change risk factors. These risk factors can help identify high-risk conditions for abnormal elevator speed changes and provide guidance for subsequent elevator safety control.

[0030] Step S225 includes the following steps: Perform inertial gravity time series incremental difference processing on the ascending / descending inertial gravity change data to obtain inertial gravity time series incremental difference data; According to the inertial gravity time series incremental difference data, the instantaneous gravity tensile load-bearing evaluation is performed on the wire traction material data to obtain the instantaneous gravity tensile load-bearing data of the material; Based on the material instantaneous gravity tensile bearing data, a discontinuous inertial gravity similarity interval is restricted to obtain a gravity tensile discontinuous similarity interval; According to the Simpson integral method and the gravity tensile discontinuity similarity interval, the material instantaneous gravity tensile load data is subjected to the tensile load piecewise constraint integration to obtain the tensile load piecewise integral data; The indirect tension imbalance data of the wire traction material is quantified according to the tensile load segment integral data to obtain the indirect tension imbalance data of traction.

[0031] In the embodiment of the present invention, the ascending / descending inertial gravity change data is firstly processed by inertial gravity time series increment difference. Inertial gravity time series increment difference processing eliminates trend changes and low-frequency noise by performing differential operations on the continuous moments of gravity change data, thereby highlighting more subtle instantaneous changes. Specifically, for the inertial gravity data at each moment, , perform incremental difference processing and calculation to obtain differential data ,That For the moment Inertial gravity value. In this way, the incremental differential data obtained can be used to capture the small and rapidly changing gravity changes during the operation of the elevator, which are critical to the safety of the elevator system. Next, the instantaneous gravity tensile load of the wire traction material is evaluated based on the inertial gravity time series incremental differential data. In this step, the tensile load capacity of the wire traction material under instantaneous changes is calculated based on the gravity changes experienced by the elevator when it rises or descends during operation. The tensile load evaluation formula is as follows: ,in, It represents the instantaneous tension (bearing capacity) of the steel wire traction material at time t, that is, the tensile force that the steel wire material is subjected to at that moment. It represents the inertial gravity increment differential data at time t, reflecting the instantaneous change of gravity during the operation of the elevator. It represents the mass of the elevator, that is, the total weight of the elevator. Represents the cross-sectional area of ​​the steel wire, usually in square millimeters or square meters. It determines the stress distribution of the steel wire when it is subjected to tension. It reflects the change in gravity. Through this evaluation, the load-bearing capacity of the steel wire traction material at each instant in the operation of the elevator can be accurately predicted to avoid safety problems caused by overload stretching. Then, based on the instantaneous gravity tensile load-bearing data of the material, the discontinuous inertial gravity similarity interval is restricted to obtain the gravity stretching discontinuous similarity interval. In this step, by analyzing the instantaneous gravity tensile load-bearing data, the discontinuous inertial gravity change pattern that occurs in the elevator in certain specific time periods is determined. These discontinuous changes are usually caused by factors such as changes in elevator load, people getting on and off the elevator, or the opening and closing of elevator doors. By extracting these specific interval data, they can be regarded as potential abnormal fluctuation intervals. These intervals are usually shown as sudden changes or violent fluctuations in the gravity change diagram. Then, according to the Simpson integral method and the gravity stretching discontinuous similarity interval, the material instantaneous gravity tensile load-bearing data is subjected to the tensile load segmented constraint integration. Simpson integration method is a common numerical integration method, which is suitable for calculating the area under an irregular curve. In this step, the gravity stretching data is segmented and integrated by Simpson integration method. The specific calculation formula is as follows: ,in, Indicates that in the interval Intrinsic Functions The integral of a physical quantity (such as carrying capacity, force, etc.) is usually used to calculate the cumulative value of a physical quantity (such as carrying capacity, force, etc.) within a certain interval. Indicates the starting and ending points of the integral interval, that is, the upper and lower bounds of the interval. h represents the length of the small interval of each interval, which is calculated as follows: , where n is the number of small segments divided into the interval. Represents the function to be integrated, which is the instantaneous gravity tensile load data or other functions that reflect the behavior of the elevator. Through this method, the cumulative tensile load of the wire traction material in each discontinuous gravity change interval can be obtained. This result is crucial for analyzing the performance of the elevator under a specific load.

[0032] Finally, according to the tensile load segmented integral data, the wire traction material data is indirectly quantified for tension imbalance to obtain the indirect tension imbalance data of traction. This step quantifies the tension imbalance in the traction system by combining the tensile load segmented integral results with the physical properties of the wire traction material (such as the elastic modulus and yield strength of the material). The specific calculation method is to compare the tensile load in each integral interval with the theoretical load-bearing limit of the material to obtain the tension imbalance degree. The formula for the indirect tension imbalance data of traction is expressed as: ,in Indicates the tension imbalance in the elevator traction system, and quantifies the degree of tension imbalance of the wire traction material during elevator operation. The larger the value, the more serious the imbalance in the traction system. It represents the tensile load data in the i-th time period, reflecting the tensile force borne by the steel wire traction material in each time period during the operation of the elevator. It indicates the maximum load-bearing capacity of the wire traction material, that is, the maximum tension that the wire can withstand, which is usually the physical limit value of the material. Represents the sum of all time periods i, and calculates the total tension imbalance. Each Fi corresponds to the tensile load data in a specific time period. The existence of tension imbalance causes abnormal fluctuations during the operation of the elevator system, which in turn affects the safety of the elevator. By quantifying this data, a reliable input can be provided for subsequent risk analysis. The above steps ultimately obtain the traction indirect tension imbalance data, which is used to further analyze the speed change anomalies and safety risks that occur during the operation of the elevator system, and help realize the adaptive analysis and prediction of elevator safety risks.

[0033] Step S23 includes the following steps: Step S231: locating the speed change risk abnormal code of the malfunction code execution unit based on the elevator speed change risk factor to obtain the speed change risk abnormal code; Step S232: performing syntax reverse engineering analysis on the speed change risk abnormality code to obtain syntax analysis data of the speed change abnormality code; Step S233: performing speed change abnormal cycle condition setting analysis on the speed change abnormal code syntax parsing data to obtain the speed change abnormal cycle setting conditions; Step S234: performing variable parameter overflow control analysis on the speed change abnormality code syntax analysis data according to the speed change abnormality cycle setting condition to obtain variable parameter overflow control data; Step S235: performing function exception throwing analysis based on the speed change abnormal cycle setting conditions and the variable parameter overflow control data to obtain function exception throwing data; Step S236: Based on the speed change abnormal cycle setting conditions, variable parameter overflow control data and function abnormal throwing data, the speed change risk abnormal code is logically analyzed to obtain abnormal speed change code logic data.

[0034] In the embodiment of the present invention, the value of the elevator speed change risk factor is first used to analyze and locate the malfunction code execution unit. The elevator speed change risk factor is obtained through the real-time monitoring data of the inertia force, load state and operation mode of the elevator during the speed change process. These data include the acceleration, speed change, load condition of the wire traction system, etc. of the elevator. During the analysis process, the speed change risk factor is mapped to different parts of the elevator's operation logic code, especially the code lines related to speed change and acceleration / deceleration control logic. By matching the elevator operation state with the control logic, the logic unit that causes the speed change abnormality is identified. The specific operation is to accurately locate the speed change control module with potential risks by comparing the historical speed change abnormal behavior with the corresponding function call, judgment conditions, etc. in the elevator control logic code line by line, and finally obtain the speed change risk abnormal code. The positioning process adopts a rule matching method based on data mining, and cross-validates in combination with the time series characteristics of the abnormal data to ensure the accurate identification of the abnormal code. The speed change risk abnormal code located in step S231 is reverse engineered and parsed. Specifically, the abnormal code is parsed by decompiling technology to reversely infer its source code structure. The key to reverse engineering is to recover the corresponding high-level language code by parsing assembly code, machine code or bytecode. First, the control flow graph and data flow graph of the target code are extracted through static analysis tools to analyze the execution path, function call relationship and conditional judgment statements of the program. Then, the target code is restored to an abstract syntax tree (AST) using a syntax parsing algorithm (such as recursive descent parsing, LL(1) grammar parsing, etc.), thereby extracting the logical structure, loop body, conditional statements and function calls in the code. Finally, based on the recovered syntax information, the syntax parsing data of the speed change exception code is generated, including variables, constants, control statements and their relationships, providing a basis for subsequent logical analysis. By further analyzing the syntax parsing data of the speed change exception code, the loop structure and its conditions involving the speed change operation are identified. First, the loop structure (such as for, while, do-while, etc.) in the speed change exception code is extracted, and the starting condition, loop body and termination condition of each loop are analyzed. The focus is on analyzing the loop body part that affects the speed change process, especially the logic of controlling acceleration, deceleration, and parking operations. During the analysis, combined with the actual speed change requirements of the elevator, determine whether the key variables in the loop conditions, such as the speed, position, acceleration, etc. of the elevator are subject to unreasonable restrictions or trigger conditions. For example, if there are acceleration thresholds or deceleration rates that have not been reasonably verified in the loop conditions, these will cause the elevator to change speed abnormally under certain conditions. Static analysis tools are used to detect and simulate the performance of loop conditions under different input parameters, and finally the setting conditions of the speed change abnormal loop are extracted. The variables of the speed change abnormality code are analyzed, especially the control mechanism of its parameter overflow is checked.First, identify the key variables used to control acceleration, deceleration, position and speed in the speed change abnormal code, which are affected by the boundary values ​​of the input data. Through static data flow analysis, trace the transmission process of these variables in the code to detect whether their values ​​are affected by the limit values ​​of external input or internal calculation, especially those involving important parameters such as acceleration threshold, braking torque, and wire traction. Subsequently, analyze the overflow protection measures in the code (such as range checking, overflow detection, etc.) to determine whether there are overflow vulnerabilities. For unhandled overflow problems, use symbolic execution methods to simulate the changes in variable values ​​under different input conditions to detect whether there are abnormal behaviors caused by overflow, such as unlimited acceleration or deceleration. Finally, according to the detection results of the overflow control mechanism, the variable parameter overflow control data is obtained to identify the link that causes overflow in the code. Based on the speed change abnormal loop setting conditions and variable parameter overflow control data, further analyze the function exception throwing logic in the code. First, in the speed change abnormal code, locate all functions related to speed change operations, especially those involving acceleration, deceleration, and stopping operations. Analyze the error handling mechanisms of these functions, such as whether exceptions are thrown or error handling is performed when the input parameters are illegal or the calculation exceeds the predetermined range. Combined with the loop setting conditions and overflow control data analyzed in steps S233 and S234, check whether there is a situation where the exception thrown due to improper loop conditions or variable overflow is not captured in time. Use static code analysis tools to simulate input scenarios and detect whether the function can throw corresponding errors or take protective measures under abnormal circumstances. Finally, based on these exception throwing analyses, function exception throwing data is generated to indicate under what circumstances the abnormal code will trigger uncontrollable speed change problems. Based on the results of the above analysis, an overall logical analysis of the speed change risk abnormal code is performed. According to the speed change abnormal loop setting conditions, variable parameter overflow control data and function exception throwing data, the execution logic of the speed change risk abnormal code is systematically analyzed. Specifically, first, based on the condition judgment and overflow control analyzed above, re-examine each logical node in the code, analyze its performance under different input conditions, and identify potential risk points. Then, combined with the exception throwing analysis data, check whether there is a scenario where the speed change abnormality is caused by the abnormal situation not being captured in time. Finally, by merging all relevant analysis results, we can obtain complete logic data of speed change abnormality code, including the logical association of all speed change operations, control conditions, overflow control, error throwing and other links, forming complete analysis data of elevator speed change risks, providing a basis for subsequent risk assessment and correction.

[0035] Step S3 includes the following steps: Step S31: normalizing the abnormal speed change trigger condition data to obtain abnormal speed change trigger condition normalized data; Step S32: performing potential risk code segment linkage analysis on the abnormal speed change code logic data based on the abnormal speed change trigger condition normalization data to obtain potential risk code segment linkage data; Step S33: construct a causal relationship network for the abnormal speed change code logic data according to the potential risk code segment linkage data to obtain a potential risk causal relationship network.

[0036] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: normalizing the abnormal speed change trigger condition data to obtain abnormal speed change trigger condition normalized data; In an embodiment of the present invention, the abnormal speed change trigger condition data is first normalized to eliminate the differences between different data dimensions and unify the scale to facilitate subsequent analysis. The normalization process is performed by converting the original data into the range of [0,1]. The common method used in this process is maximum-minimum normalization or Z-score normalization. Specifically, for each abnormal speed change trigger condition data, the minimum and maximum values ​​of the data are first calculated. Then, the minimum value is subtracted from each data point and divided by the difference between the maximum and minimum values ​​to obtain the normalized data. This normalization process ensures that all trigger condition data are processed in a consistent manner and effectively avoids analysis bias caused by inconsistent data dimensions. The normalized data will be used for risk linkage analysis and causal network construction in subsequent steps.

[0037] Step S32: performing potential risk code segment linkage analysis on the abnormal speed change code logic data based on the abnormal speed change trigger condition normalization data to obtain potential risk code segment linkage data; In an embodiment of the present invention, the normalized abnormal speed change trigger condition data and the abnormal speed change code logic data are linked and analyzed to identify potential risk code segments. The specific operation is to first analyze the relationship between the abnormal speed change trigger condition data and the execution status of each code module in the control system to determine the mutual influence between the trigger condition and the code logic. The correlation analysis and multi-dimensional matching methods are adopted to compare the abnormal speed change trigger conditions with different code modules one by one by constructing a data association matrix. For each trigger condition data, the abnormal behavior caused by it in different code segments is analyzed, the potential risk code segments are recorded, and weighted evaluation is performed according to the degree of data association. For example, if a certain speed change abnormal trigger condition is highly correlated with the speed change function and acceleration regulation module in the elevator control system, the function or module is marked as a potential risk code segment. This process adopts an association rule analysis method similar to that in data mining, such as the Apriori algorithm, which further improves the recognition accuracy of the potential risk code segment by calculating the support and confidence between the data. The potential risk code segment linkage data finally obtained will provide a basis for the subsequent causal relationship network construction.

[0038] Step S33: construct a causal relationship network for the abnormal speed change code logic data according to the potential risk code segment linkage data to obtain a potential risk causal relationship network.

[0039] In the embodiment of the present invention, based on the potential risk code segment linkage data obtained in step S32, a causal relationship network is further constructed. The construction of the causal relationship network adopts a directed graph model in graph theory, and the mutual influence relationship between code segments is represented as the relationship between nodes and edges. In the figure, each code segment represents a node, and the influence relationship between the trigger condition and the risk code segment is represented by a directed edge. By calculating the causal dependency between the code segments, a causal network containing potential risks is constructed. In order to accurately identify the causal relationship between the code segments, the potential risk code segments are first analyzed in detail, and their impact on other code segments is analyzed in combination with their execution order, input and output variables, and state changes. By constructing a causal graph, the causal chain between the abnormal speed change trigger condition and its corresponding code segment can be visualized, and it is clear which code segments are the root cause of the abnormal speed change. For example, if the output state of an acceleration module directly affects the working mode of the speed change control module, there will be a directed edge connecting the two modules in the causal relationship network. In this process, a causal inference algorithm (such as Granger causality test or Bayesian network inference) is used to determine the causal relationship between the code segments. This causal relationship network not only reveals the interaction between potential risk code segments, but also helps locate the key code segments that cause abnormal speed changes. The resulting potential risk causal relationship network provides a clear basis for subsequent risk analysis and control.

[0040] Step S33 includes the following steps: Step S331: Perform linkage multi-layer nesting analysis on the potential risk code segment linkage data to obtain risk code linkage multi-layer nesting data; Step S332: construct a risk code logic diagram based on the risk code linkage multi-layer nested data; Step S333: performing causal relationship reasoning on the abnormal speed change code logic data based on the risk code logic diagram to obtain causal relationship logic reasoning data; Step S334: performing relationship clustering processing on the risk code linkage multi-layer nested data based on the causal relationship logical reasoning data to obtain risk code causal relationship clustering data; Step S335: Construct a causal relationship network based on the risk code causal relationship clustering data and the causal relationship logical reasoning data to obtain a potential risk causal relationship network.

[0041] In the embodiment of the present invention, a multi-layer nested analysis is first performed on the potential risk code segment linkage data to capture the complex linkage relationship between potential code segments. By constructing a multi-level linkage analysis framework, the code modules at different levels are analyzed hierarchically according to their degree of association. The specific operation is to use the nested structure in graph theory to divide the code segments into multiple levels according to their impact on the system. For example, for the relationship between high-level control logic and low-level hardware control, a multi-layer nested structure can be constructed, in which each layer represents the state changes of different modules in the system and their impact on other modules. This process first calculates the impact range and action chain of each potential risk code segment, and then arranges these code segments hierarchically according to the impact relationship. In order to ensure the accuracy of the linkage relationship, data flow analysis technology is also used to take the input, output, state change of each code segment and its impact on other modules as the basis of the linkage relationship. The risk code linkage multi-layer nested data finally generated can clearly show the linkage level of each code segment and the complex relationship between them, providing basic data for subsequent risk assessment. Based on the risk code linkage multi-layer nested data obtained in step S331, a complete risk code logic diagram is constructed. The construction of the logic graph adopts a directed graph model, in which each node represents a risk code segment, and each edge represents the potential linkage relationship between two code segments. First, according to the results of the multi-layer nesting analysis, the level of each code segment is determined, and the hierarchical structure of the graph is designed based on this. Then, according to the relationship between different levels, the edges in the graph are gradually filled to define the causal influence relationship between the code segments. During the construction process, it is considered that there is not only a direct causal relationship between code modules, but also an indirect influence. In order to more comprehensively reflect the logical relationship between each code segment in the system, the "hierarchical progressive" method is adopted, that is, on the basis of the low-level module, the influence of the upper-level module on it is gradually derived. The risk code logic graph generated in the end can intuitively show the logical connection between each module in the elevator system and help identify potential risk points and critical paths. Based on the constructed risk code logic graph, causal reasoning is performed on the abnormal speed change code logic data. The core of causal reasoning is to deduce the causal chain between each code segment by analyzing the directed edges in the code logic graph. The specific method is to use algorithms such as Bayesian network reasoning and DAG (directed acyclic graph) traversal to simulate the execution order and state changes of code segments and identify the potential causes of abnormal speed changes. Whenever a potential abnormal speed change trigger condition appears in a certain code segment, the system will track the relevant upstream and downstream code segments and gradually deduce the causal chain that causes the speed change abnormality. The process first analyzes each node in the logic diagram, identifies its input, output, and internal state changes, and uses conditional probability for reasoning to deduce the causal dependency between events. Ultimately, the causal logic reasoning data obtained through the reasoning process will provide an effective basis for subsequent clustering analysis and network construction.Based on the causal logic reasoning data, the risk code linkage multi-layer nested data obtained in step S331 is clustered and analyzed to obtain the risk code causal clustering data. The clustering process adopts an unsupervised learning method based on similarity, such as K-means clustering or spectral clustering. First, by analyzing the causal reasoning data of each risk code segment, the similarity between each code segment is calculated. This similarity is measured by comparing the state changes, input-output relationships and causal influence chains between the code segments. Then, the risk code segments are divided into several clusters according to the similarity. The code segments in each cluster have a strong correlation in causal relationships and can jointly trigger a certain type of abnormal speed change. In the clustering process, the influence of the multi-level structure is also considered to ensure that the code segments at the same level can be clustered into the same cluster, and the cross-level linkage relationship is included in the corresponding clustering calculation. Finally, the risk code causal clustering data obtained by cluster analysis can help identify which code segment combinations trigger similar abnormal behaviors, thereby providing a basis for further risk warning. Combined with the risk code causal clustering data obtained in step S334 and the causal logic reasoning data obtained in step S333, a potential risk causal network is finally constructed. The construction process uses a weighted directed graph to visualize the code segments in different clusters and their relationships through graph theory methods. In this network, each clustering result is taken as a node, and the edges between the nodes represent the causal relationship between the code segments. The weight of the edge is derived from the causal reasoning result in step S333, reflecting the intensity and possibility of mutual influence between the code segments. The final causal network clearly shows the potential risk paths in the elevator system, and can intuitively identify the key code segments that cause abnormal speed changes and their interrelated risk factors. Through this causal network, it can provide an accurate basis for risk warning, fault diagnosis and safety assessment during elevator operation.

[0042] Step S4 includes the following steps: Step S41: performing risk characteristic analysis on the potential risk causal relationship network to obtain a potential risk causal characteristic relationship network; Step S42: assigning risk causal relationship weights to the potential risk causal feature relationship network to obtain a potential risk relationship weight relationship network; Step S43: construct an elevator safety risk adaptive analysis model for the potential risk relationship weight relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0043] In the embodiment of the present invention, the potential risk causal network is firstly analyzed in depth to identify the key risk features in the system. Specifically, the risk features related to elevator safety are extracted by analyzing the nodes (i.e., the risk code segments) in the network and the causal relationships between them. These features include the execution frequency of the risk code segments, the degree of impact on the overall performance of the system, the propagation path of potential faults, and the interdependence of the risk nodes. Through network structure analysis, the code modules that become safety hazards and their interaction characteristics with other modules are identified. The centrality, importance, and force of each code segment in the causal network are calculated by using graph analysis technology, so as to find the key code segments that cause greater risks. At the same time, the causal network is divided using a community detection algorithm (such as the Louvain algorithm or the hierarchical clustering algorithm) to identify the collaborative risk features between different modules. The output of this process is a potential risk causal feature network, which includes the feature description of each risk node and the strength of its association with other nodes, as the basis for subsequent risk analysis and modeling. Based on the potential risk causal feature network analyzed in step S41, each causal relationship is weighted to quantify the importance and influence of each risk factor in the entire causal network. First, a weighted graph model is used to represent each edge in the causal network, and the weight of each edge represents the causal influence from one risk code segment to another risk code segment. The edge weight calculation adopts a statistically based weighting method, including but not limited to maximum likelihood estimation, correlation coefficient calculation and other technologies. The impact of each risk factor on the elevator system will be quantified based on historical data, the complexity of the code segment, the frequency of fault occurrence and other factors, so as to assign a numerical weight to each risk causal relationship. Specifically, if multiple code segments in a causal chain work together and generate a higher risk, the weight of the chain will be assigned a larger value. The weight assignment not only takes into account the directness and strength of the causal relationship, but also includes time delay and fault propagation effects, ensuring that the dynamic changes of the causal relationship can be accurately reflected in actual operation. The final output is a potential risk relationship weight relationship network containing all causal relationships and their corresponding weights, providing a quantitative assessment of each risk factor in the elevator safety system. The potential risk relationship weight relationship network obtained in step S42 is modeled using the random forest algorithm to construct an elevator safety risk adaptive analysis model. Random forest is an ensemble learning method that is widely used to deal with problems with multiple input variables and complex decision boundaries. First, the input data set of the decision tree is constructed by taking each causal node and edge in the potential risk relationship weight relationship network as input features. The training goal of each decision tree is to determine the impact of different risk features and causal relationships on elevator safety by learning from the historical data of the elevator system.Specifically, the weights of each node in the risk causal relationship network are used as feature inputs, combined with the operation data of the elevator (such as load, speed, acceleration, etc.) for training, to predict the probability of elevator safety risks under different causal relationship combinations. Each tree is trained according to a randomly selected feature set, so that it has a strong generalization ability when processing complex multidimensional data. Finally, the prediction results of each decision tree are aggregated through a voting mechanism to obtain the final evaluation model of elevator safety risk. The model can automatically evaluate the safety status of the elevator based on real-time data and make adaptive adjustments for different operating conditions. The constructed elevator safety risk adaptive analysis model will be transmitted to the elevator operation control center through the network. The control center can monitor and evaluate the safety status of the elevator in real time through the received model. The specific operation is to deploy the model to the central processing system of the control center, the system regularly receives elevator operation data, and sends this data as input to the constructed random forest model for analysis. In this way, the elevator control system can evaluate the potential risks in the elevator operation process in real time and make corresponding safety measures based on the output results of the model. For example, when the model detects that a risk factor exceeds the set safety threshold, the control center can automatically trigger a safety alarm or initiate an emergency stop operation to avoid accidents. In addition, the elevator operation control center can also continuously optimize the control strategy based on the adaptive characteristics of the model to ensure the safety of the elevator under different operating conditions.

[0044] The present invention also provides an elevator safety risk adaptive analysis system, which is used to execute the elevator safety risk adaptive analysis method as described above, and the elevator safety risk adaptive analysis system comprises: The malfunction code mapping module is used to obtain the elevator historical operation log and the elevator operation logic code; according to the elevator historical operation log, the elevator operation logic code is mapped to the malfunction code execution unit to obtain the malfunction code execution unit; The trigger condition backtracking module is used to perform abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; perform trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; A causal relationship network construction module is used to perform a potential risk code segment linkage analysis based on abnormal speed change trigger condition data to obtain potential risk code segment linkage data; and to construct a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; The adaptive analysis model construction module is used to construct an elevator safety risk adaptive analysis model for the potential risk causal network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; the elevator safety risk adaptive analysis model is sent to the elevator operation control center to perform elevator safety risk adaptive analysis.

[0045] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An elevator safety risk adaptive analysis method, characterized in that: The following steps are involved: Step S1: Obtain the elevator historical operation log and the elevator operation logic code; According to the elevator historical operation log, the elevator operation logic code is mapped to the malfunction code execution unit to obtain the malfunction code execution unit; Step S2: performing abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; performing trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; Step S3: performing a potential risk code segment linkage analysis based on the abnormal speed change trigger condition data to obtain potential risk code segment linkage data; Construct a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; Step S4: constructing an elevator safety risk adaptive analysis model for the potential risk causal relationship network based on a random forest algorithm to obtain an elevator safety risk adaptive analysis model; The elevator safety risk adaptive analysis model is sent to the elevator operation control center to perform elevator safety risk adaptive analysis.

2. The elevator safety risk adaptive analysis method according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Obtain the elevator historical operation log and the elevator operation logic code; Step S12: Clean the elevator historical operation log to obtain the elevator historical operation cleaning log; Step S13: extracting the malfunction behavior from the elevator historical operation cleaning log to obtain the elevator historical malfunction behavior; Step S14: mapping the elevator operation logic code to the malfunction code execution unit according to the historical malfunction behavior of the elevator to obtain the malfunction code execution unit.

3. The elevator safety risk adaptive analysis method according to claim 2 is characterized in that: Step S2 includes the following steps: Step S21: Identify the abnormal speed change behavior of the elevator's historical malfunction behavior to obtain the abnormal speed change behavior of the elevator; Step S22: analyzing the speed change risk factor according to the elevator speed change abnormal behavior and the elevator historical operation cleaning log to obtain the elevator speed change risk factor; Step S23: performing abnormal speed change code logic analysis on the malfunction code execution unit based on the elevator speed change risk factor to obtain abnormal speed change code logic data; Step S24: backtrack the triggering condition of the abnormal speed change code logic data to obtain abnormal speed change triggering condition data.

4. The elevator safety risk adaptive analysis method according to claim 3 is characterized in that: Step S22 includes the following steps: Step S221: extracting the load and personnel density of the elevator historical operation cleaning log to obtain the elevator load data and the elevator personnel density data; Step S222: performing speed change jerking fluctuation analysis on the abnormal speed change behavior of the elevator to obtain the speed change jerking fluctuation data of the elevator; Step S223: Estimating the dynamic change of the inertial gravity of the elevator load data and the elevator personnel density data according to the elevator speed change jerking fluctuation data, and obtaining the inertial gravity change data of the elevator; Step S224: obtaining the wire traction material data of the elevator traction sheave; Step S225: quantifying the indirect tension imbalance of the wire traction material data according to the ascending / descending inertial gravity change data to obtain the traction indirect tension imbalance data; Step S226: Analyze the speed change risk factor of the elevator speed change jerking fluctuation data according to the traction indirect tension imbalance data to obtain the elevator speed change risk factor.

5. The elevator safety risk adaptive analysis method according to claim 3 is characterized in that: Step S23 includes the following steps: Step S231: locating the speed change risk abnormal code of the malfunction code execution unit based on the elevator speed change risk factor to obtain the speed change risk abnormal code; Step S232: Perform syntax reverse engineering analysis on the speed change risk abnormality code to obtain syntax analysis data of the speed change abnormality code; Step S233: performing speed change abnormal cycle condition setting analysis on the speed change abnormal code syntax parsing data to obtain the speed change abnormal cycle setting conditions; Step S234: performing variable parameter overflow control analysis on the speed change abnormality code syntax analysis data according to the speed change abnormality cycle setting condition to obtain variable parameter overflow control data; Step S235: performing function exception throwing analysis based on the speed change abnormal cycle setting conditions and the variable parameter overflow control data to obtain function exception throwing data; Step S236: Based on the speed change abnormal cycle setting conditions, variable parameter overflow control data and function abnormal throwing data, the speed change risk abnormal code is logically analyzed to obtain abnormal speed change code logic data.

6. The elevator safety risk adaptive analysis method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: normalizing the abnormal speed change trigger condition data to obtain abnormal speed change trigger condition normalized data; Step S32: performing potential risk code segment linkage analysis on the abnormal speed change code logic data based on the abnormal speed change trigger condition normalization data to obtain potential risk code segment linkage data; Step S33: construct a causal relationship network for the abnormal speed change code logic data according to the potential risk code segment linkage data to obtain a potential risk causal relationship network.

7. The elevator safety risk adaptive analysis method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing risk characteristic analysis on the potential risk causal relationship network to obtain a potential risk causal characteristic relationship network; Step S42: assigning risk causal relationship weights to the potential risk causal feature relationship network to obtain a potential risk relationship weight relationship network; Step S43: construct an elevator safety risk adaptive analysis model for the potential risk relationship weight relationship network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; send the elevator safety risk adaptive analysis model to the elevator operation control center to perform elevator safety risk adaptive analysis.

8. An elevator safety risk adaptive analysis system, characterized in that: Used to execute the elevator safety risk adaptive analysis method according to claim 1, the elevator safety risk adaptive analysis system comprises: The malfunction code mapping module is used to obtain the elevator historical operation log and the elevator operation logic code; according to the elevator historical operation log, the elevator operation logic code is mapped to the malfunction code execution unit to obtain the malfunction code execution unit; The trigger condition backtracking module is used to perform abnormal speed change code logic analysis on the malfunction code execution unit to obtain abnormal speed change code logic data; perform trigger condition backtracking on the abnormal speed change code logic data to obtain abnormal speed change trigger condition data; A causal relationship network construction module is used to perform a potential risk code segment linkage analysis based on abnormal speed change trigger condition data to obtain potential risk code segment linkage data; and to construct a causal relationship network based on the potential risk code segment linkage data to obtain a potential risk causal relationship network; The adaptive analysis model construction module is used to construct an elevator safety risk adaptive analysis model for the potential risk causal network based on the random forest algorithm to obtain an elevator safety risk adaptive analysis model; the elevator safety risk adaptive analysis model is sent to the elevator operation control center to perform elevator safety risk adaptive analysis.

Images