Elevator operation risk scoring method based on time sequence anomaly detection and correlation analysis

Abstract

The elevator operation risk scoring method based on time sequence anomaly detection and correlation analysis comprises the steps that multi-source data of the Internet of Things, the Internet and the like of an elevator are obtained, a plurality of initial risk indexes are constructed, and good and bad samples are divided by calculating risk initial values; calculating the correlation between each index and the good and bad attributes of the sample, and adopting chi-square test and V value quantification and screening out effective risk indexes; determining weights and scores of the indexes in multiple dimensions such as identity features and operation capability according to the correlation, forming a scoring rule, and calculating the score of each dimension; and carrying out weighted summation on the dimension scores to obtain a total risk score, and dividing the elevators into different risk grades according to the total elevator total score distribution. According to the elevator risk assessment method, objective, dynamic and efficient assessment of elevator risks is achieved through a data-driven full-process automatic model, and a scientific basis is provided for precise supervision and preventive maintenance.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of elevator safety technology, and in particular to an elevator operation risk scoring method based on time series anomaly detection and association analysis. BACKGROUND

[0002] With the acceleration of urbanization and the popularity of high-rise buildings, elevators have become an indispensable vertical transportation tool in modern society, and their safe operation is directly related to the safety of people's lives and property. Therefore, it is crucial to scientifically and accurately assess the risks of elevators. Currently, the mainstream elevator risk assessment method in the industry mainly relies on periodic on-site inspection and expert experience. This method usually establishes a risk matrix model, combines failure mode analysis of each subsystem of the elevator and safety management condition assessment, and qualitatively or semi-quantitatively classifies the risk, such as classifying the risk level into first, second, third, and fourth levels. This assessment mode has played an important role in ensuring the safe operation of elevators within a certain period of time and has become the basis of the traditional elevator safety supervision system.

[0003] However, the above-mentioned traditional risk assessment method has its inherent limitations. First, the assessment process heavily relies on the professional knowledge and practical experience of the assessors, making the assessment results susceptible to subjective factors and leading to differences in the assessment conclusions of the same elevator by different assessors, which makes it difficult to ensure the objectivity and consistency of the assessment. Second, this method mainly relies on static historical archive data and periodic on-site inspection data, which has the problems of single data dimension and time lag, making it difficult to capture the risk evolution process caused by changes in operating conditions, environment, and gradual degradation of components in real time and dynamically. In addition, the assessment process requires a large amount of manual participation, which is not only inefficient and costly, but also difficult to meet the needs of fine and grid management of urban-level mass elevators, and cannot realize continuous monitoring and early warning of risks. SUMMARY

[0004] This section aims to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the above-mentioned existing problems, the present application is proposed. Therefore, the present application provides an elevator operation risk scoring method based on time series anomaly detection and association analysis to solve the problems proposed in the background art.

[0006] To solve the above technical problems, the present application provides the following technical solutions: an elevator operation risk scoring method based on time series anomaly detection and association analysis, comprising: Acquire multi-source data on elevators, construct multiple initial risk indicators based on the multi-source data, select key indicators from the initial risk indicators, calculate the initial risk value for each elevator, and divide the elevator samples into good samples and bad samples according to the initial risk value. Calculate the correlation between the initial risk indicators and the good / bad attributes of the samples to obtain the quantitative value of the correlation of each initial risk indicator, and screen out the effective risk indicators whose quantitative value of correlation is greater than a preset threshold. Based on the quantified relevance values ​​of the effective risk indicators, the weight, maximum score, and score range of each effective risk indicator within its respective risk dimension are determined to form a scoring system rule. Based on the scoring system rule, the dimension scores of the elevator to be evaluated on multiple risk dimensions are calculated. The scores of each risk dimension are weighted and summed to obtain the total risk score of the elevator to be evaluated. The elevator to be evaluated is then rated according to the distribution of the total risk scores of all elevators.

[0007] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the method for acquiring multi-source elevator data includes acquiring real-time operation data from the Internet of Things, Internet data, big data platform data, smart camera detection event data, and operation and maintenance data; the method for constructing multiple initial risk indicators includes processing the multi-source data using time-series data anomaly detection and correlation mining techniques.

[0008] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the method involves: calculating the correlation between the initial risk indicators and the good / bad attributes of the samples to obtain a quantitative value of the correlation degree of each initial risk indicator, including: A cross-tabulation is constructed between the values ​​of the initial risk indicators and the good / bad attributes of the samples; Calculate the chi-square value based on the actual observed frequency and theoretical frequency of the cross-tabulation; Based on the chi-square value, sample size, and number of rows and columns in the crosstab, a statistic V value is calculated to characterize the correlation strength, and the V value is used as the quantification of the correlation degree.

[0009] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, key indicators are selected from the initial risk indicators, including: The weights of the three indicators—last month's operating frequency, last month's average passenger volume, and maximum operating interval—were determined using the analytic hierarchy process (AHP), and these three indicators were then used as the key indicators.

[0010] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the determination of the maximum score and score range of each effective risk indicator includes: The preset total score and total range of each dimension are multiplied by the weight of the effective risk indicator within its respective risk dimension to obtain the maximum bonus value and score range of the effective risk indicator.

[0011] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the step of forming the scoring system rules includes: Each effective risk indicator is divided into multiple sub-items; Calculate the mean initial risk value for each elevator sample corresponding to each sub-item; Based on the magnitude relationship of the average initial risk values, the bonus value corresponding to each sub-item is determined.

[0012] As a preferred embodiment of the elevator operation risk scoring method based on temporal anomaly detection and correlation analysis described in this invention, the step of calculating the dimensional scores of the elevator to be evaluated across multiple risk dimensions includes: Assign values ​​to each effective risk indicator of the elevator to be evaluated according to the scoring system rules; The scores of all effective risk indicators under the same risk dimension are summed to obtain the dimension score of that risk dimension.

[0013] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the multiple risk dimensions include identity feature dimension, operation capability dimension, risk history dimension, operation feature dimension, and correlation relationship dimension.

[0014] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, wherein: the weighted summation of the scores of each risk dimension based on the weights of each risk dimension includes: Calculate the average value of the correlation of all valid risk indicators within each risk dimension; The weight of each risk dimension is determined based on its proportion in the sum of the average values ​​of all dimensions. The total risk score is obtained by multiplying the dimensional scores of each risk dimension by their corresponding weights and then summing the results.

[0015] As a preferred embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis described in this invention, the step of risk rating the elevator to be evaluated based on the total risk score distribution of all elevators includes: The total risk score of all elevators is divided into multiple score intervals, each score interval corresponds to a risk level, and the score intervals corresponding to each risk level are adjusted until the number distribution of all elevators in each risk level approximates a normal distribution.

[0016] Compared with existing technologies, the beneficial effects of this solution are: 1. This invention constructs good and bad samples and uses statistical methods such as chi-square test and V-value to quantify the correlation between indicators and risks. Based on this correlation, the weight of indicators and scoring rules are objectively determined, overcoming the shortcomings of traditional assessment methods that rely on expert experience, are highly subjective, and have poor consistency of results. This invention achieves objectivity and standardization in elevator risk assessment.

[0017] 2. This invention integrates multi-source heterogeneous data, which can dynamically capture the elevator's operating status, usage mode and potential risks from multiple dimensions such as identity characteristics and operating capabilities. It solves the problems of single data dimensions and delayed timeliness of traditional methods, and improves the comprehensiveness and real-time performance of risk assessment.

[0018] 3. In addition, the scoring method proposed in this invention is automatically completed by data-driven and computational models throughout the entire process. From data acquisition and indicator construction to scoring and rating, it can be executed automatically, which greatly improves the evaluation efficiency, reduces labor costs, can meet the needs of refined and routine risk monitoring of a large number of elevators at the city level, and can ensure the stability and universality of the rating system by dynamically adjusting the distribution of total scores. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating the overall process of an elevator operation risk scoring method based on time-series anomaly detection and correlation analysis according to an embodiment of the present invention. Figure 2 This is the score of a certain elevator (sample) in each dimension of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis according to an embodiment of the present invention; Figure 3 This is a risk score distribution diagram of all elevators in an embodiment of the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis according to one embodiment of the present invention. Figure 4 This is a distribution diagram of the number of elevators with different risk ratings in the elevator operation risk scoring method based on time-series anomaly detection and correlation analysis according to an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0024] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0025] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Example

[0026] Reference Figures 1 to 4This is the first embodiment of the present invention, which provides a method for elevator operation risk scoring based on time-series anomaly detection and correlation analysis, including: S1. Obtain multi-source data on elevators, and construct multiple initial risk indicators based on the multi-source data. Select key indicators from the initial risk indicators, calculate the initial risk value for each elevator, and divide the elevator samples into good samples and bad samples according to the initial risk value.

[0027] It should be noted that this step aims to construct a comprehensive and multi-dimensional elevator risk assessment index system through the fusion and processing of multi-source heterogeneous data, and to establish a preliminary risk measurement standard through a data-driven approach, providing labeled training samples for the subsequent construction of the scoring model.

[0028] Furthermore, we will acquire multi-source data and construct initial risk indicators.

[0029] Specifically, the process begins by accessing and integrating data from multiple sources, including but not limited to: real-time elevator operation data collected by Internet of Things (IoT) devices, publicly available internet data, enterprise big data platform data, event data detected by smart cameras deployed inside the elevator, and daily elevator maintenance data. Subsequently, data mining techniques are used to process this multi-source data to construct initial risk indicators. For example, time-series data anomaly detection technology is applied to analyze real-time IoT data to identify indicators such as abnormal operating speed and abnormal door opening / closing times. Relationship mining technology is applied to analyze maintenance data and enterprise credit data to construct indicators such as the "correlation degree between maintenance units and manufacturing enterprises." In a specific embodiment of this invention, the above methods initially construct 52 initial risk indicators from five dimensions: identity characteristics, operational capabilities, risk history, operational characteristics, and relationships.

[0030] Specifically, the identity feature dimension represents the elevator's basic attributes, such as whether it has a registration number, management registration period, platform registration period, and whether the manufacturing company has specific qualifications.

[0031] Specifically, the operational capability dimension measures the elevator's operational performance and activity level. Examples include the proportion of the elevator's service life to its design life, the month-on-month growth rate of the number of operations last month, and the average number of daily operations last month.

[0032] Specifically, the risk history dimension is used to quantify the elevator's historical risk record, such as whether there are records of malfunctions, the credit status of related companies, the historical average number of runs, and the maximum running interval.

[0033] Specifically, the operational characteristics dimension reflects the elevator's usage patterns and maintenance status, such as the operating frequency of the previous month, the distribution of main operating types, the frequency of elevator maintenance, and the proportion of risky usage behaviors (such as overloading and blocking doors).

[0034] Specifically, the relationship dimension is used to assess the risk of elevators being associated with other entities, such as the average risk score of elevators of the same brand and model, the number of associated types, and the monthly increase in relationships.

[0035] Furthermore, key indicators are selected from the multiple initial risk indicators constructed above, and the initial risk value for each elevator is calculated.

[0036] Specifically, to ensure the objectivity and scientific rigor of the initial risk value calculation, the present invention preferably employs the Analytic Hierarchy Process (AHP) to determine key indicators and their weights. In a specific embodiment of the present invention, three indicators are selected as key indicators: "last month's operating frequency," "last month's average passenger volume," and "maximum operating interval time." A pairwise comparison matrix is ​​constructed between these three indicators through expert scoring or historical data analysis. This matrix is ​​then normalized column-wise, and the normalized matrix is ​​summed row-wise. Finally, the summed vector is normalized again to obtain the weight of each key indicator. For example, the calculated weights for "last month's operating frequency," "last month's average passenger volume," and "maximum operating interval time" are 0.28, 0.64, and 0.07, respectively. Based on this, corresponding scoring rules are set for different value ranges of each key indicator, and the initial risk value for each elevator is calculated in conjunction with its weight. Refer to Tables 1 to 4.

[0037] Table 1 Key Indicators of the Comparison and Judgment Matrix

[0038] Table 2 Key Indicators of the Normalized Judgment Matrix

[0039] Table 3 Key Indicators of the Normalized Judgment Matrix After Summation

[0040] Table 4. Weights of Key Indicators

[0041] Furthermore, based on the calculated initial risk value, the entire elevator sample is divided into good samples and bad samples. This provides training data with clear "good" or "bad" labels for the subsequent supervised learning model (i.e., the scorecard model).

[0042] Specifically, the classification is based on the distribution of initial risk values. In one specific embodiment of the invention, elevators with initial risk values ​​in the top quantile (e.g., a score of 1000) are defined as good samples, while elevators with initial risk values ​​in the bottom quantile (e.g., a score less than 350) are defined as bad samples. In this way, a data-driven, proprietary evaluation ecosystem that reflects the actual risk status of elevators is constructed. See Tables 5 and 6.

[0043] Table 5. Weights and Scoring Rules for Key Indicators

[0044] Table 6 Extraction of "Good" and "Bad" Modeling Samples

[0045] S2. Calculate the correlation between the initial risk indicators and the quality of the samples, obtain the quantitative value of the correlation of each initial risk indicator, and screen out the effective risk indicators whose quantitative value of correlation is greater than the preset threshold. It should be noted that the core objective of this step is to perform feature screening on the 52 initial risk indicators constructed in step S1. This involves identifying indicators that are significantly correlated with the actual risk of the elevator (represented by good and bad sample labels) and eliminating indicators with weak or no correlation. This reduces the complexity of the subsequent model and improves its stability and predictive accuracy. Furthermore, this process is entirely data-driven, ensuring the objectivity of the indicator screening.

[0046] Furthermore, the chi-squared test is used to analyze the association strength between each initial risk indicator and the good / bad attribute of the sample. Since the good / bad attribute of the sample is a categorical variable ("good" or "bad"), while the initial risk indicators include both categorical variables (such as "whether it has a registration number") and continuous variables (such as "last month's operating frequency"), for uniform processing, the continuous variables are first binned to transform them into categorical variables. Then, for each initial risk indicator, the following operations are performed: First, a cross-tabulation is constructed. For each initial risk indicator, its different values ​​or binning intervals are used as rows in the cross-tabulation, and the two attributes defined in step S1, "good sample" and "bad sample," are used as columns. Each cell in the cross-tabulation records the number of elevator samples that simultaneously meet the conditions of the corresponding row and column, i.e., the actual observation frequency.

[0047] Secondly, the chi-squared value is calculated based on the cross-tabulation, which is... It should be explained that the chi-square value measures the degree of difference between the actual observed frequency and the theoretical frequency. The greater the difference, the stronger the correlation between the row and column variables. The chi-square value is calculated as follows: (1) in, This is the chi-square statistic. For the first in the cross table line, number The actual observation frequency of the column cells. This represents the theoretical frequency for the corresponding cell, indicating the expected frequency under the null hypothesis that row and column variables are independent. Its calculation method is expressed as follows: (2) in, For the first Total frequency of rows. For the first The total frequency of the column. This represents the total sample size.

[0048] Furthermore, to eliminate the influence of sample size on the chi-square value and obtain a standardized relevance measure, we also need to calculate the Cramér's V value and use it as a quantification of the relevance of each initial risk indicator. The Cramér's V value ranges from [0, 1], with a value closer to 1 indicating a stronger correlation between the indicator and the good / bad attribute of the sample. Its calculation method is as follows: (3) in, This is the Cramer V value, which is the quantification value of the degree of correlation in the scheme of this invention. The chi-square value is calculated according to formula (1). This represents the number of rows in the crosstab. This represents the number of columns in the crosstab.

[0049] Finally, all initial risk indicators are screened. The calculated V values ​​of all initial risk indicators are sorted in descending order to form an indicator importance ranking table. A preset relevance threshold, such as 0.06, is set. Indicators with V values ​​greater than this preset threshold are selected as effective risk indicators for subsequent scoring model construction. Indicators with V values ​​less than or equal to this threshold are considered insufficiently relevant to risk and are eliminated. In a specific embodiment of the present invention, after the above calculation and screening of the initially screened 52 indicators, 11 indicators had a relevance quantification value lower than 0.06 and were therefore eliminated, ultimately retaining 41 effective risk indicators. See Table 7.

[0050] Table 7. Ranking of Indicator Importance

[0051] S3. Based on the quantification of the relevance of effective risk indicators, determine the weight, maximum score and score range of each effective risk indicator in its respective risk dimension, form a scoring system rule, and calculate the dimension score of the elevator to be evaluated in multiple risk dimensions according to the scoring system rule. It should be noted that this step aims to build a quantitative and operational scorecard model based on the effective risk indicators selected in step S2.

[0052] Furthermore, the weights of each effective risk indicator within its respective risk dimension are determined. It's important to explain that the purpose of these weights is to quantify the relative contribution of different effective risk indicators to the risk of that dimension. The weights are determined based on the correlation quantification value (V value) calculated in step S2. Specifically, the V value of a given effective risk indicator is divided by the sum of the V values ​​of all effective risk indicators within its respective risk dimension.

[0053] Specifically, in one specific embodiment of the present invention, five risk dimensions are taken as examples: identity characteristics, operational capabilities, risk history, operational characteristics, and related relationships. Assume that the "risk history" dimension includes multiple valid risk indicators, such as "last month's operating frequency" and "last month's average passenger volume." The V values ​​calculated for these indicators in step S2 are summed to obtain the total V value for this dimension. See Table 8.

[0054] Table 8. Weights of the Five Risk Dimensions

[0055] The weight of this indicator within the "Risk History" dimension is calculated as follows: (4) in, For the first In the dimension of the first The weight of each effective risk indicator. For the first The correlation quantification value (V value) of each effective risk indicator. For the first The total number of effective risk indicators included in each dimension. For the first The sum of the V values ​​of all effective risk indicators within each dimension.

[0056] It should be noted that, according to formula (4), the weight of the "last month's operating frequency" indicator in the "risk history" dimension can be calculated as follows: This process will be performed on all 41 valid risk indicators to determine the weight of each indicator within its respective dimension.

[0057] Furthermore, the maximum bonus value and score range of each effective risk indicator are determined to transform the abstract weights into specific scoring ranges.

[0058] Specifically, the pre-set total score and total range of each dimension are multiplied by the weight of the effective risk indicator within its respective risk dimension. In a specific embodiment of the present invention, the total score for each dimension can be set to 950 points, and the total range to 600 points. See Table 9. It should be explained that "range" refers to the difference between the highest and lowest scores obtainable by the indicator, reflecting the magnitude of the indicator's influence. Generally, it is believed that the greater the influence, the larger the range value will be.

[0059] Table 9. Weights, maximum bonus points, and ranges of the five risk dimensions.

[0060] It should be noted that the current maximum elevator points bonus is 0.136 × 950 = 129; the current elevator points range is 0.136 × 600 = 82.

[0061] Furthermore, a specific, queryable score comparison table is established for each effective risk indicator.

[0062] Specifically, firstly, each effective risk indicator is divided into multiple sub-items (i.e., different value ranges or categories). Then, for each sub-item, all elevator samples belonging to that sub-item are selected from the samples divided in step S1, and the mean of the initial risk values ​​for these samples is calculated. Finally, the specific bonus value corresponding to each sub-item is determined based on the relationship between these mean initial risk values. Generally, a sub-item with a higher mean initial risk value represents a lower risk, and therefore should be assigned a higher bonus value, not exceeding the maximum bonus value for that indicator. For example, for the indicator "average passenger volume last month," its sub-items, mean initial risk values, and corresponding bonus values ​​are shown in Table 10 below.

[0063] Table 10. Detailed breakdown of indicators for average passenger volume last month, mean initial risk value, and corresponding bonus points.

[0064] By repeating this process for all 41 valid risk indicators, a complete scoring system rule that includes the scores of all indicator sub-items can be formed.

[0065] Finally, according to the scoring system rules, the dimensional scores of the elevator to be evaluated across multiple risk dimensions are calculated. For any elevator to be evaluated, its specific performance values ​​on 41 valid risk indicators are first obtained. Then, according to the scoring system rules formed above, each of these 41 indicators is searched and assigned a corresponding score. Finally, the scores of all valid risk indicators belonging to the same risk dimension are summed to obtain the elevator's dimensional score on that risk dimension. For example, if the sum of all indicator scores for an elevator under the "identity characteristics" dimension is 735, then 735 is its "identity characteristics dimension score." This calculation is repeated for all five dimensions to obtain the elevator's score on all risk dimensions. Refer to Tables 11 and 12.

[0066] Table 11 Scoring of Various Elevator Indicators

[0067] Table 12 Total Scores of Elevators in Each of the Five Dimensions

[0068] S4. The scores of each risk dimension are weighted and summed to obtain the total risk score of the elevator to be evaluated. The elevator to be evaluated is then rated according to the total risk score distribution of all elevators. It should be noted that this step aims to integrate the multi-dimensional scores calculated in step S3 to form a single total score that can comprehensively reflect the overall risk level of the elevator. Based on this total score, the elevator is classified into different levels, thereby providing an intuitive and clear basis for supervision and maintenance decisions.

[0069] Furthermore, the scores of each risk dimension are weighted and summed based on their respective weights to obtain the total risk score of the elevator to be evaluated.

[0070] It should be emphasized that, since the weight of each risk dimension represents the degree of contribution of different dimensions to the overall risk of the elevator, the determination of this weight is based on the average correlation strength of the effective risk indicators contained in each dimension.

[0071] Specifically, the arithmetic mean of the correlation values ​​of all effective risk indicators within each risk dimension (i.e., the V value calculated in S2) is calculated to obtain the average correlation coefficient for each risk dimension. Then, the weight of each risk dimension is determined based on its proportion in the sum of the average correlation coefficients of all dimensions. The calculation formula is as follows: (5) in, For the first The weights of each risk dimension. For the first The average correlation coefficient of each risk dimension, that is, the average value of all effective risk indicators V under that dimension. The total number of risk dimensions is 5 in this embodiment. It is the sum of the average correlation coefficients of all risk dimensions.

[0072] Specifically, in one specific embodiment of the present invention, the average correlation coefficients and calculated weights of the five dimensions are shown in Table 13 below. Taking the "identity feature" dimension as an example, its weight calculation process is as follows: .

[0073] Table 13 Average correlation coefficients and weights of the five risk dimensions

[0074] Furthermore, after determining the weights of each risk dimension, the scores for each dimension obtained by the elevator to be evaluated in step S3 are multiplied by their corresponding weights. Then, all products are summed to obtain the elevator's total risk score. The calculation formula is as follows: (6) in, This represents the total risk score for the elevator to be evaluated. For the elevator to be evaluated in the Dimensional scores on each risk dimension. For the first The weights of each risk dimension.

[0075] For example, the scores for each dimension of an elevator (number 200056505) are as follows: Identity Characteristics 735 points, Operational Capability 682 points, Risk History 543 points, Operational Characteristics 589 points, and Association Relationships 564 points. Therefore, its total risk score is calculated as follows: . refer to Figure 2 .

[0076] Furthermore, a risk rating is assigned to the elevators to be evaluated based on the total risk score distribution of all elevators. This process aims to map continuous numerical scores to discrete risk levels with clear business implications.

[0077] Specifically, steps S1 to S4 are performed on all elevators covered by the system to obtain the total risk score for each elevator. Subsequently, statistical analysis is performed on the total risk scores of all elevators, and a score distribution chart (e.g., a histogram or kernel density estimation chart) is plotted. Finally, based on the distribution of the total risk scores, the total score range is divided into multiple consecutive score intervals, each interval corresponding to a risk level. In a specific embodiment of the present invention, the risk levels are divided into five levels: poor, moderate, good, excellent, and very good, with reference to...Figure 3 To ensure the rating results have good discriminative power and statistical significance, the boundary of the score intervals for each level can be adjusted until the distribution of the total number of elevators across all risk levels approximates a normal distribution. (Refer to...) Figure 4 This classification method avoids imbalances in rating standards caused by a few extremely high or low scores, ensuring the stability and universality of the rating system. Ultimately, the final risk rating is determined based on the score range to which the elevator being evaluated falls within its total risk score.

[0078] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process.Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0083] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for scoring elevator operation risk based on time-series anomaly detection and correlation analysis, characterized in that, include: Acquire multi-source data on elevators, construct multiple initial risk indicators based on the multi-source data, select key indicators from the initial risk indicators, calculate the initial risk value for each elevator, and divide the elevator samples into good samples and bad samples according to the initial risk value. Calculate the correlation between the initial risk indicators and the good / bad attributes of the samples to obtain the quantitative value of the correlation of each initial risk indicator, and screen out the effective risk indicators whose quantitative value of correlation is greater than a preset threshold. Based on the quantified relevance values ​​of the effective risk indicators, the weight, maximum score, and score range of each effective risk indicator within its respective risk dimension are determined to form a scoring system rule. Based on the scoring system rule, the dimension scores of the elevator to be evaluated on multiple risk dimensions are calculated. The scores of each risk dimension are weighted and summed to obtain the total risk score of the elevator to be evaluated. The elevator to be evaluated is then rated according to the distribution of the total risk scores of all elevators.

2. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, The acquisition of multi-source data for the elevator includes acquiring real-time operational data from the Internet of Things, Internet data, big data platform data, smart camera detection event data, and operation and maintenance data; the construction of multiple initial risk indicators includes processing the multi-source data using time-series data anomaly detection and correlation mining techniques.

3. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, Calculate the correlation between the initial risk indicators and the good / bad attributes of the samples to obtain a quantitative value of the correlation degree of each initial risk indicator, including: A cross-tabulation is constructed between the values ​​of the initial risk indicators and the good / bad attributes of the samples; Calculate the chi-square value based on the actual observed frequency and theoretical frequency of the cross-tabulation; Based on the chi-square value, sample size, and number of rows and columns in the crosstab, a statistic V value is calculated to characterize the correlation strength, and the V value is used as the quantification of the correlation degree.

4. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, Key indicators are selected from the initial risk indicators, including: The weights of the three indicators—last month's operating frequency, last month's average passenger volume, and maximum operating interval—were determined using the analytic hierarchy process (AHP), and these three indicators were then used as the key indicators.

5. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, Determine the maximum bonus value and score range for each effective risk indicator, including: The preset total score and total range of each dimension are multiplied by the weight of the effective risk indicator within its respective risk dimension to obtain the maximum bonus value and score range of the effective risk indicator.

6. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, The steps for forming the rules of the scoring system include: Each effective risk indicator is divided into multiple sub-items; Calculate the mean initial risk value for each elevator sample corresponding to each sub-item; Based on the magnitude relationship of the average initial risk values, the bonus value corresponding to each sub-item is determined.

7. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, The calculation of the elevator to be evaluated's dimensional scores across multiple risk dimensions includes: Assign values ​​to each effective risk indicator of the elevator to be evaluated according to the scoring system rules; The scores of all effective risk indicators under the same risk dimension are summed to obtain the dimension score of that risk dimension.

8. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1 or 7, characterized in that, The multiple risk dimensions include identity characteristics, operational capabilities, risk history, operational features, and correlation.

9. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, The weighted summation of the scores for each risk dimension based on their respective weights includes: Calculate the average value of the correlation of all valid risk indicators within each risk dimension; The weight of each risk dimension is determined based on its proportion in the sum of the average values ​​of all dimensions. The total risk score is obtained by multiplying the dimensional scores of each risk dimension by their corresponding weights and then summing the results.

10. The elevator operation risk scoring method based on time-series anomaly detection and correlation analysis as described in claim 1, characterized in that, The risk rating of the elevator to be evaluated based on the total risk score distribution of all elevators includes: The total risk score of all elevators is divided into multiple score intervals, each score interval corresponds to a risk level, and the score intervals corresponding to each risk level are adjusted until the number distribution of all elevators in each risk level approximates a normal distribution.

Images