Shopping mall equipment maintenance prediction method based on time series anomaly detection

Abstract

The application discloses a shopping mall equipment maintenance prediction method based on time sequence anomaly detection, relates to the technical field of equipment operation and maintenance engineering, and comprises the following steps: obtaining standard time sequence parameters of each time period by using a single time period analysis method; obtaining maintenance order rankings and abnormal time sequence characteristics corresponding to all faults; and performing maintenance prediction on equipment in a shopping mall based on a time period in which the equipment is located, abnormal time sequence characteristics of all faults and the maintenance order rankings. The application is used for solving the problem that in the existing shopping mall equipment maintenance prediction method, the time sequence characteristics of the shopping mall equipment are not analyzed in different time periods, the running state of the equipment in different time periods is not known, the peak value of the time sequence curve of the equipment parameters and the period characteristics of the peak value interval are not analyzed in different time periods, and early period drift type hidden faults in the equipment cannot be identified in advance.

Description

Technical Field

[0001] This invention relates to the field of equipment operation and maintenance engineering technology, specifically to a method for predicting the maintenance of shopping mall equipment based on time series anomaly detection. Background Technology

[0002] Predictive maintenance of shopping mall equipment is a specific application of predictive maintenance in commercial facilities. It is an intelligent strategy that predicts shopping mall equipment failures and arranges maintenance in advance based on data analysis. Time series anomaly detection refers to an intelligent maintenance solution that uses time series anomaly detection technology as the core to predict shopping mall equipment failures. It is a typical technical application of predictive maintenance in shopping mall scenarios.

[0003] Existing methods for predicting maintenance of shopping mall equipment typically extract features from parameters generated during equipment operation, construct feature vectors for these parameter samples, and then combine them with neural networks to build a model for intelligent diagnosis of equipment operation. While this improved method can detect equipment faults, it lacks a method for specifically analyzing the periodic characteristics of equipment parameters' time-series curves, peak values, and peak intervals based on the equipment's operating status at different times within the mall. This results in the inability to identify early-stage latent faults such as periodic drift in the equipment. For example, patent application CN117303148A discloses an elevator vibration analysis method based on car acceleration. The fault diagnosis method involves extracting features from the elevator car vibration signal in the time and frequency domains, respectively, and reconstructing the feature vector of the vibration sample. This feature vector is then combined with a single-gated recurrent neural network (SGU) and a convolutional neural network (CNN). Other methods for predicting the maintenance of shopping mall equipment typically only improve real-time data acquisition. They lack methods for utilizing the temporal characteristics of shopping mall equipment, specifically analyzing the peak values ​​and periodic characteristics of equipment parameters based on their operating status at different times within the mall. This results in the inability to identify early-stage, latent faults such as periodic drift in the equipment. Therefore, it is necessary to improve existing methods for predicting the maintenance of shopping mall equipment. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art by proposing a shopping mall equipment maintenance prediction method based on time series anomaly detection. This method addresses the lack of a specific analysis method for the periodic characteristics of the time series curves of equipment parameters based on the equipment operating status at different times within the shopping mall, and the periodic characteristics of the peak values ​​and peak intervals of the equipment parameters in different time periods. This results in the inability to identify early-stage latent faults such as periodic drift in the equipment in advance.

[0005] To achieve the above objectives, this application provides a method for predicting the maintenance of shopping mall equipment based on time series anomaly detection, comprising the following steps: The day is segmented based on the mall's business hours to obtain multiple time periods. After segmentation, the single-time-period analysis method is used to analyze the equipment operation data in each time period, and the standard time sequence parameters of all equipment in each time period are obtained based on the analysis results. The system analyzes the faults of all equipment in the shopping mall that require maintenance, and obtains the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault based on the analysis results. The abnormal time sequence characteristics include idle time sequence characteristics, off-peak time sequence characteristics and peak time sequence characteristics. When making maintenance predictions for equipment in a shopping mall, the predictions are based on the time period in which the prediction takes place, the abnormal timing characteristics of all faults, and the ranking of the maintenance order.

[0006] Furthermore, based on the shopping mall's operating hours, the day is segmented to obtain multiple time periods, including: Get the shopping mall's opening and non-open hours for a day, and mark the time period corresponding to the non-open hours as the idle period; mark the time period corresponding to the opening hours as the operating period, and mark the number of hours occupied by the operating period as h after rounding up. The business hours are evenly divided into h time periods, which are denoted as business sub-segment YZ1 to business sub-segment YZ. h For any business segment: Based on the historical customer flow data recorded in the mall's AI video customer flow system, obtain the customer flow of the business segment for each of the mall's operating days, and record the average of all customer flow as the average customer flow of the business segment.

[0007] Furthermore, the process of segmenting the day into multiple time periods based on the mall's operating hours also includes: The average of the average passenger flow of all business segments is recorded as the measured passenger flow; business segments with an average passenger flow greater than the measured passenger flow are recorded as peak segments, and business segments with an average passenger flow less than or equal to the measured passenger flow are recorded as off-peak segments. The time period obtained by merging all peak sub-segments is called the peak time period, and the time period obtained by merging all off-peak time periods is called the off-peak time period.

[0008] Furthermore, single-time period analysis methods include: Obtain all equipment in the shopping mall that requires maintenance prediction, and label them as parameter analysis equipment CF1 to parameter analysis equipment CF2. nFor any parameter analysis device: among all the parameters that can be collected in the parameter analysis device, the parameters that allow the construction of time series curves are recorded as the time series monitoring parameters of the parameter analysis device, where the time series curve is the relationship curve between time and parameter; For any time period α among idle time period, off-peak time period and peak time period: For any time series monitoring parameter γ of any parameter analysis device β, based on all historical monitoring records of time series monitoring parameter γ in parameter analysis device β, obtain the historical monitoring record corresponding to the time series monitoring parameter γ when the mall is in normal operation and the time is in time period α, and record it as the time period α record of time series monitoring parameter γ.

[0009] Furthermore, single-time period analysis also includes: For any day in the time period α record, the curve showing the relationship between the time series monitoring parameter γ and time within the time period α is denoted as the time series curve γ. The curve obtained by fitting the time series curve γ corresponding to all days in the time period α record, and the area enclosed by the time series curve γ corresponding to all days in the same coordinate system are respectively denoted as the standard time series curve and standard time series region of the time series monitoring parameter γ and time period α. Acquire all time-series monitoring parameters of all parameter analysis devices in time period α and their corresponding standard time-series curves and standard time-series regions for time period α; For any parameter analysis device β, all the time-series detection parameters of the parameter analysis device β are respectively compared with the standard time-series curves and standard time-series regions of the idle period, off-peak period and peak period, and are recorded as the standard time-series parameters of the parameter analysis device.

[0010] Furthermore, the faults of all equipment requiring maintenance in the shopping mall are analyzed, and based on the analysis results, the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault are obtained, including: All faults that can be generated by the parameter analysis equipment are recorded as equipment faults. For any equipment fault, based on the historical fault sample records of the parameter analysis equipment where the equipment fault is located, the shortest duration during which the parameter analysis equipment where the equipment fault is located can operate normally at the time the equipment fault occurs is obtained and recorded as the pre-fault runtime of the equipment fault. All device faults are sorted in ascending order based on their pre-run time and denoted as device fault SG1 to device fault SG2. u Among them, the lead time for equipment failure SG1 is the shortest among all equipment failures. u Its lead time is the longest among all device failure lead times.

[0011] Furthermore, the analysis of faults corresponding to all equipment requiring maintenance in the shopping mall, and the resulting ranking of repair order for all faults, as well as the abnormal time sequence characteristics for each fault, also include: For any device fault of any parameter analysis device β: Based on any fault occurrence record of the device fault in the historical fault sample record of the parameter analysis device β, obtain the parameter records of the device fault β in the idle period, off-peak period and peak period respectively within Q days before the occurrence of the device fault, and record them as the previous idle record, the previous off-peak record and the previous peak record respectively. The preceding record analysis method was used to analyze the preceding idle records, preceding off-peak records, and preceding peak records respectively, and the abnormal parameter characteristics in the analysis results were recorded as the idle time sequence characteristics, off-peak time sequence characteristics, and peak time sequence characteristics of equipment failure respectively.

[0012] Furthermore, the pre-record analysis method includes: The time period corresponding to the records analyzed in the pre-analysis record is recorded as the pre-analysis time period, where the pre-analysis time period is the idle time period, off-peak time period or peak time period; For any time-series monitoring parameter γ of parameter analysis device β, the relationship curve between the time-series monitoring parameter γ and time for each day of the recorded Q days is recorded as the anomaly analysis curve γ; the curve obtained by fitting the Q anomaly analysis curves γ corresponding to the recorded Q days, and the area enclosed by the Q anomaly analysis curves γ in the same coordinate system are respectively recorded as the time-series anomaly curve and the time-series anomaly region. The time-series anomaly curve and the time-series monitoring parameter γ are placed in the same Cartesian coordinate system M1 as the standard time-series curve for the preceding analysis period. For the Cartesian coordinate system M1: the maximum slope of the time-series anomaly curve minus the maximum slope of the standard time-series curve is recorded as the anomaly slope difference. The minimum and maximum values ​​of the ordinates of all peaks in the anomaly time-series curve are recorded as Y1 and Y2, respectively. The maximum value of the ordinates of all peaks in the standard time-series curve is recorded as Y3. The interval [Y1-Y3, Y2-Y3] is recorded as the anomaly peak interval.

[0013] Furthermore, the pre-record analysis method also includes: The time-series anomaly region and the time-series monitoring parameter γ are placed in the same Cartesian coordinate system M2 as the standard time-series region of the previous analysis period. For the Cartesian coordinate system M2: the region in the time-series anomaly region that does not coincide with the standard time-series region is recorded as the anomaly analysis region. When there is no anomaly analysis region or the number of regions corresponding to the anomaly analysis region is 1, the analysis of the Cartesian coordinate system M2 is stopped. Obtain the midpoint of all regions corresponding to the anomaly analysis region and record it as the anomaly midpoint; record the difference of the x-coordinate of adjacent anomaly midpoints as the anomaly interval; record the interval formed by the minimum and maximum values ​​of all anomaly intervals obtained from all anomaly midpoints as the anomaly period interval. Abnormal slope difference, abnormal peak interval, and abnormal period interval are denoted as abnormal difference parameters of time series monitoring parameter γ; The abnormal differences in all time-series monitoring parameters of the parameter analysis device β are recorded as abnormal parameter characteristics.

[0014] Furthermore, when making maintenance predictions for equipment in a shopping mall, based on the time period in which the prediction takes place, the abnormal timing characteristics of all faults, and the ranking of maintenance order, the maintenance predictions for equipment in the shopping mall include: When making maintenance predictions for equipment in a shopping mall, the time period in which the prediction occurs is recorded as the predicted maintenance period. For any parameter analysis device β, the parameter records corresponding to all time-series monitoring parameters of the parameter analysis device β within the complete predicted maintenance period are obtained and recorded as parameter prediction records. For any time-series monitoring parameter γ of the parameter analysis device β, the curve showing the relationship between the time-series monitoring parameter γ and time in the parameter prediction record is recorded as the time-series prediction curve; The time-series prediction curve and time-series monitoring parameters are placed in the same Cartesian coordinate system M1 as the standard time-series curve for the predicted maintenance period. Based on the prior record analysis method, the abnormal slope difference and abnormal peak interval between the time-series prediction curve and the standard time-series curve in the Cartesian coordinate system M1 are obtained and recorded as the prediction slope difference and prediction peak interval, respectively. Place the time-series prediction curve and the time-series monitoring parameter γ in the same Cartesian coordinate system M2 as the standard time-series region for the predicted maintenance period. For the Cartesian coordinate system M2: the region in the time-series prediction curve that is not in the standard time-series region is recorded as the prediction analysis region. When the time-series prediction curve is completely in the abnormal analysis region or the number of regions corresponding to the abnormal analysis region is 1, the analysis of the Cartesian coordinate system M2 is stopped. The midpoints of all regions corresponding to the prediction analysis region are obtained and recorded as the prediction midpoints; the abnormal period intervals are obtained by using the abnormal midpoints, and the prediction period intervals are obtained based on the prediction midpoints. When any timing monitoring parameter corresponding to a device fault in parameter analysis device β simultaneously satisfies conditions 1, 2, and 3, the device fault is recorded as a predicted fault. Condition 1: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the abnormal slope difference of time series monitoring parameter γ is less than or equal to the predicted slope difference; Condition 2: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal peak interval of the time-series monitoring parameter γ is less than the median of the predicted peak interval. Condition 3: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal period interval of the time series monitoring parameter γ is less than the median of the predicted period interval. When the number of predicted faults obtained from analyzing all parameters of the device is greater than 1, all predicted faults are sorted from first to last based on the order in which they are sorted among all device faults, and fault handling is performed on all predicted faults based on the sorted order of the predicted faults.

[0015] The beneficial effects of this invention are as follows: This application first segments the time of day based on the business hours of the shopping mall to obtain multiple time periods; after segmentation, the single-time-period analysis method is used to analyze the equipment operation data in each time period, and the standard time series parameters of all equipment in each time period are obtained based on the analysis results. The advantage of this is that by segmenting the time of day based on the business hours of the shopping mall, and then using the single-time-period analysis method to analyze the equipment operation data in each time period, and then summarizing with standard time series parameters, the actual operating status of the equipment in each time period can be accurately matched, and the periodic characteristic changes such as the peak value and peak interval of the time series curve can be effectively captured. Thus, in real-time monitoring, the early identification of early hidden faults of periodic drift based on standard time series parameters can be achieved. This application also analyzes the faults of all equipment requiring maintenance in the shopping mall, and obtains the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault based on the analysis results. Finally, when predicting the maintenance of equipment in the shopping mall, the maintenance prediction is made based on the predicted time period, the abnormal time sequence characteristics of all faults, and the maintenance order ranking. The advantage of this is that by obtaining the maintenance order ranking and abnormal time sequence characteristics of the faults, the priority of handling faults when multiple faults exist can be obtained, as well as the parameters for judging the fault anomalies. Thus, when predicting the maintenance of equipment in the shopping mall, the anomaly type and fault root cause can be accurately matched based on the predicted time period, and the maintenance priority can be reasonably divided based on the established maintenance order. This not only improves the accuracy of fault prediction, but also standardizes the operation and maintenance scheduling sequence and ensures that potential equipment hazards are handled in an orderly manner. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of the method of the present invention; Figure 2 This is a schematic diagram illustrating the acquisition of the standard timing curve and standard timing region of the present invention; Figure 3 This is a schematic diagram illustrating the acquisition of the processing sequence for predicted faults according to the present invention. Figure 4 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1, please refer to Figure 1 As shown, this application provides a method for predicting the maintenance of shopping mall equipment based on time series anomaly detection, including the following steps: Step S1: Divide the time of day into segments based on the mall's business hours to obtain multiple time periods; after segmentation, use single-time-period analysis to analyze the equipment operation data in each time period, and obtain the standard time sequence parameters of all equipment in each time period based on the analysis results. Step S1 includes: Step S101, obtaining the shopping mall's business hours and non-business hours within a day, and recording the time period corresponding to the non-business hours as the idle time period; recording the time period corresponding to the business hours as the business period, and recording the value corresponding to the number of hours occupied by the business period as h after rounding up; Step S102: Divide the business hours evenly into h time periods, and label them as business sub-segments YZ1 to YZ1 respectively. h For any business segment: Based on the historical customer flow data recorded in the mall's AI video customer flow system, obtain the customer flow of each business segment for all days of mall operation, and record the average of all customer flow as the average customer flow of the business segment. In the specific implementation process, if the shopping mall is not equipped with an AI video customer flow system, other methods can be used to obtain the mall's historical customer flow, so as to divide the time period based on the customer flow in the future.

[0019] Step S1 further includes: Step S103, the average of the average passenger flow of all business segments is recorded as the measured passenger flow; business segments with average passenger flow greater than the measured passenger flow are recorded as peak segments, and business segments with average passenger flow less than or equal to the measured passenger flow are recorded as off-peak segments. Step S104: The time period obtained by merging all peak sub-segments is recorded as the peak time period, and the time period obtained by merging all off-peak time periods is recorded as the off-peak time period. In the data analysis of this embodiment, for example, after analyzing the customer flow data, the peak period is [11:00-19:00], the off-peak period is [7:00, 11:00) and (19:00, 23:00], and the idle period is (23:00, 7:00). In actual data analysis, if the peak, off-peak and idle periods in the mall have been divided according to the characteristics of the mall's customer flow, then the peak, off-peak and idle periods can be directly matched, and the single-period analysis method can be performed respectively.

[0020] Step S105, the single-period analysis method includes: Step S1051, obtaining all equipment in the mall that requires maintenance prediction, and labeling them as parameter analysis equipment CF1 to parameter analysis equipment CF1 respectively. n For any parameter analysis device: among all the parameters that can be collected in the parameter analysis device, the parameters that allow the construction of time series curves are recorded as the time series monitoring parameters of the parameter analysis device, where the time series curve is the relationship curve between time and parameter; Step S1052: For any time period α among idle time period, off-peak time period and peak time period: For any time series monitoring parameter γ of any parameter analysis device β, based on all historical monitoring records of time series monitoring parameter γ in parameter analysis device β, obtain the historical monitoring record corresponding to the time series monitoring parameter γ when the mall is in normal operation and the time is in time period α, and record it as the time period α record of time series monitoring parameter γ. In the data analysis of this embodiment, for example, in a single-period analysis method, the period α of the analysis is the peak period, the parameter analysis device is the elevator, and one of the time-series monitoring parameters of the elevator is the vertical vibration acceleration of the car. In the single-period analysis process, the historical monitoring records corresponding to the vertical vibration acceleration of the car when the mall is in normal operation and the time is during the peak period should be obtained and used as the peak period records for subsequent analysis.

[0021] The single-period analysis method also includes: step S1053, for any day in the period α record, the relationship curve between the time series monitoring parameter γ and time in the period α is recorded as the time series curve γ; the curve obtained by fitting the time series curve γ corresponding to all days in the period α record, and the area enclosed by the time series curve γ corresponding to all days in the same coordinate system are respectively recorded as the standard time series curve and standard time series area of ​​the time series monitoring parameter γ and period α. Step S1054: Obtain all time-series monitoring parameters of all parameter analysis devices in time period α and the standard time-series curve and standard time-series region of time period α; In the data analysis of this embodiment, for example, in a single data analysis, all time-series curves corresponding to the vertical vibration acceleration of the car are obtained as follows: Figure 2As shown, curves SQ1 to SQ3 are time-series curves corresponding to the vertical vibration acceleration of the car. The standard time-series curves and standard time-series regions obtained through fitting and region mapping are respectively... Figure 2 As shown in the curves BS and BQ, by obtaining the standard time series curve and standard time series region of the car's vertical vibration acceleration, the characteristics of the car's vertical vibration acceleration during peak hours can be obtained. At the same time, the characteristics of the time series curve during off-peak hours and idle hours can be distinguished so that the standard time series parameters obtained later can accurately match the parameter characteristics of various parameters in the parameter analysis equipment at different times, thereby achieving more accurate fault screening during real-time monitoring. Step S1055: For any parameter analysis device β, record all the time-series detection parameters of the parameter analysis device β and the standard time-series curves and standard time-series regions of the idle period, off-peak period and peak period respectively as the standard time-series parameters of the parameter analysis device.

[0022] Step S2: Analyze the faults of all equipment in the mall that need maintenance, and obtain the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault based on the analysis results. The abnormal time sequence characteristics include idle time sequence characteristics, off-peak time sequence characteristics and peak time sequence characteristics. Step S2 includes: Step S201, obtaining all the faults that the parameter analysis equipment can generate and recording them as equipment faults; for any equipment fault, based on the historical fault sample record corresponding to the parameter analysis equipment where the equipment fault is located, obtaining the shortest duration that the parameter analysis equipment where the equipment fault is located can operate normally when the equipment fault occurs, and recording it as the pre-running duration of the equipment fault. In the specific implementation process, historical fault sample records can be collected based on all records corresponding to equipment faults generated by the parameter analysis equipment in the shopping mall usage scenario. The acquisition of historical fault sample records is not limited to the shopping mall where fault prediction is performed, but the usage scenario of the parameter analysis equipment in the historical fault sample records should be consistent with the shopping mall. Step S202: Sort all device faults in ascending order based on the pre-run time, and record them as device fault SG1 to device fault SG2. u Among them, the lead time for equipment failure SG1 is the shortest among all equipment failures. u The lead time for failure is the longest among all device failure lead times. In the data analysis of this embodiment, by obtaining the maintenance order ranking of faults and the abnormal timing characteristics, it is possible to obtain the priority of fault handling when there are multiple faults, so as to prioritize the handling of faults that require urgent handling and avoid potential operational hazards to the equipment due to unreasonable fault handling priorities.

[0023] Step S2 also includes: Step S203, for any one device fault of any parameter analysis device β: based on any fault occurrence record of the device fault in the historical fault sample record of the parameter analysis device β, obtain the parameter records of the device fault β in the idle period, off-peak period and peak period respectively within Q days before the occurrence of the device fault, and record them as the previous idle record, the previous off-peak record and the previous peak record respectively. In the data analysis of this embodiment, Q can be determined based on the number of days before abnormal parameters can be detected in all devices before a failure occurs. In the data analysis of this embodiment, the value of Q is set to 5; that is, by default, abnormalities can be detected in various parameters of the device 5 days before a failure occurs. Step S204: Use the pre-record analysis method to analyze the pre-idle record, pre-peak record, and pre-peak record respectively, and record the abnormal parameter characteristics in the analysis results as the idle time sequence characteristics, peak time sequence characteristics, and peak time sequence characteristics of equipment failure respectively.

[0024] Step S205, the pre-record analysis method includes: Step S2051, recording the time period corresponding to the record analyzed in the pre-analysis record as the pre-analysis time period, wherein the pre-analysis time period is an idle time period, off-peak time period or peak time period; Step S2052: For any time-series monitoring parameter γ of the parameter analysis device β, the relationship curve between the time-series monitoring parameter γ and time for each day of the recorded Q days is recorded as the anomaly analysis curve γ; the curve obtained by fitting the Q anomaly analysis curves γ corresponding to the recorded Q days, and the area enclosed by the Q anomaly analysis curves γ in the same coordinate system are respectively recorded as the time-series anomaly curve and the time-series anomaly region. In the data analysis of this embodiment, the method for obtaining the time-series anomaly curves and time-series anomaly regions is the same as that for obtaining the standard time-series curves and standard time-series regions, that is, obtaining the corresponding fitted curves and the enclosed regions through multiple curves; when obtaining the time-series anomaly curves and time-series anomaly regions, one can refer to... Figure 2 The method for obtaining the standard time series curve and the standard time series region, and the processing of the anomaly analysis curve; Step S2053: Place the time-series anomaly curve and the time-series monitoring parameter γ in the same Cartesian coordinate system M1 as the standard time-series curve of the preceding analysis period. For the Cartesian coordinate system M1: the value obtained by subtracting the maximum slope of the standard time-series curve from the maximum slope of the time-series anomaly curve is recorded as the anomaly slope difference. The minimum and maximum values ​​of the ordinates of all peaks in the anomaly time-series curve are recorded as Y1 and Y2, respectively. The maximum value of the ordinates of all peaks in the standard time-series curve is recorded as Y3. The interval [Y1-Y3, Y2-Y3] is recorded as the anomaly peak interval.

[0025] The pre-record analysis method also includes: step S2054, placing the time-series anomaly region and the time-series monitoring parameter γ in the same Cartesian coordinate system M2 as the standard time-series region of the pre-analysis period; for the Cartesian coordinate system M2: the region in the time-series anomaly region that does not coincide with the standard time-series region is recorded as the anomaly analysis region, wherein, when there is no anomaly analysis region or the number of regions corresponding to the anomaly analysis region is 1, the analysis of the Cartesian coordinate system M2 is stopped; In the data analysis of this embodiment, for example, during the analysis of a "traction system failure" equipment malfunction, the abnormal time-series curve and abnormal time-series region of the time-series monitoring parameter "car vertical vibration acceleration" are obtained, and combined with the standard time-series curve and standard time-series region of the car vertical vibration acceleration; it is found that in the Cartesian coordinate system M1, the corresponding abnormal slope difference is 2m / s³, and the abnormal peak interval is [0.4m / s², 0.8m / s²]. In addition, in the Cartesian coordinate system M2, the abnormal period interval is [0.65s, 0.88s]. In summary, if the equipment malfunction in the elevator is a "traction system failure", the "traction system failure" can be effectively judged based on the abnormal slope difference of "car vertical vibration acceleration": 2m / s³, the abnormal peak interval [0.4m / s², 0.8m / s²], and the abnormal period interval [0.65s, 0.88s], thereby obtaining the abnormal detection result. Step S2055: Obtain the midpoints of all regions corresponding to the anomaly analysis region and record them as anomaly midpoints; record the difference in the x-coordinates of adjacent anomaly midpoints as anomaly intervals; record the interval formed by the minimum and maximum values ​​of all anomaly intervals obtained from all anomaly midpoints as anomaly period intervals. Step S2056: Record the abnormal slope difference, abnormal peak interval and abnormal period interval as the abnormal difference parameter of the time series monitoring parameter γ. Step S2057: Record the abnormal difference parameters of all time-series monitoring parameters of parameter analysis device β as abnormal parameter features.

[0026] Step S3: When making maintenance predictions for equipment in the shopping mall, the maintenance predictions for equipment in the shopping mall are made based on the time period in which the prediction takes place, the abnormal time sequence characteristics of all faults, and the ranking of the maintenance order. Step S3 includes: Step S301, when making maintenance predictions for equipment in the shopping mall, the time period in which the prediction occurs is recorded as the predicted maintenance time period; for any parameter analysis device β, obtain the parameter records corresponding to all time-series monitoring parameters of the parameter analysis device β within the complete predicted maintenance time period, and record them as parameter prediction records; Step S302: For any time-series monitoring parameter γ of the parameter analysis device β, record the relationship curve between the time-series monitoring parameter γ and time in the parameter prediction record as the time-series prediction curve; Step S303: Place the time series prediction curve and time series monitoring parameters and the standard time series curve for the predicted maintenance period in the same Cartesian coordinate system M1; obtain the abnormal slope difference and abnormal peak interval between the time series prediction curve and the standard time series curve in the Cartesian coordinate system M1 based on the previous record analysis method, and record them as the prediction slope difference and the prediction peak interval respectively. Step S304: Place the time series prediction curve and the time series monitoring parameter γ in the same Cartesian coordinate system M2 as the standard time series region for the predicted maintenance period; For the Cartesian coordinate system M2: the region in the time series prediction curve that is not in the standard time series region is recorded as the prediction analysis region. When the time series prediction curve is completely in the abnormal analysis region or the number of regions corresponding to the abnormal analysis region is 1, stop the analysis of the Cartesian coordinate system M2. Step S305: Obtain the midpoints of all regions corresponding to the prediction analysis region and record them as prediction midpoints; obtain the prediction period interval based on the prediction midpoints by using the method of obtaining abnormal period intervals through abnormal midpoints. In the data analysis of this embodiment, for example, during a single data analysis, the predicted maintenance period is a peak period, and the time-series monitoring parameter being analyzed is "car vertical vibration acceleration"; through analysis of the time-series prediction curve, the predicted slope difference is obtained as 3 m / s³, the predicted peak interval is [0.7 m / s², 1.3 m / s²], and the predicted period interval is [0.8 s, 1.0 s]; combined with the above-obtained abnormal slope difference of "car vertical vibration acceleration" corresponding to 2 m / s³, an abnormal... Analysis of the peak interval [0.4 m / s², 0.8 m / s²] and the abnormal period interval [0.65 s, 0.88 s] reveals that the "car vertical vibration acceleration" satisfies conditions 1, 2, and 3, indicating that the lifting amplitude, peak value, and interval between peaks of the "car vertical vibration acceleration" are abnormally increased. Therefore, for the equipment fault "traction system fault" that is related to the "car vertical vibration acceleration," the "traction system fault" can be recorded as a predicted fault. Step S306: When any timing monitoring parameter corresponding to a device fault in parameter analysis device β simultaneously satisfies conditions 1, 2 and 3, the device fault is recorded as a predicted fault. Condition 1: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the abnormal slope difference of time series monitoring parameter γ is less than or equal to the predicted slope difference; Condition 2: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal peak interval of the time-series monitoring parameter γ is less than the median of the predicted peak interval. Condition 3: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal period interval of the time series monitoring parameter γ is less than the median of the predicted period interval. Step S307: When the number of predicted faults obtained from analyzing all parameters of the device is greater than 1, all predicted faults are sorted from first to last based on the order in which they are sorted among all device faults, and fault processing is performed on all predicted faults based on the sorted order of the predicted faults. In the data analysis of this embodiment, for example, after analyzing data from a peak period, all predicted faults and the processing order of all predicted faults are obtained as follows: Figure 3 As shown.

[0027] Example 2, please refer to Figure 4 As shown, Figure 4 A schematic diagram of an electronic device is provided, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call these instructions. When the processor executes a computer-readable instruction, it performs steps similar to those in a shopping mall equipment maintenance prediction method based on time-series anomaly detection, to achieve the following functions: First, the day is segmented based on the shopping mall's business hours to obtain multiple time periods; after segmentation, a single-time-period analysis method is used to analyze the equipment operation data in each time period, and the standard time-series parameters of all equipment in each time period are obtained based on the analysis results; then, the faults corresponding to all equipment requiring maintenance in the shopping mall are analyzed, and the maintenance order ranking of all faults and the abnormal time-series characteristics of each fault are obtained based on the analysis results; finally, when predicting the maintenance of equipment in the shopping mall, maintenance predictions are made based on the predicted time period, the abnormal time-series characteristics of all faults, and the maintenance order ranking.

[0028] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0029] Example 3: This application also provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the shopping mall equipment maintenance prediction method based on time series anomaly detection provided by the above methods. The method includes: firstly, dividing the time of day into multiple time periods based on the shopping mall's business hours; after segmentation, using a single-time-period analysis method to analyze the equipment operation data in each time period, and obtaining the standard time series parameters of all equipment in each time period based on the analysis results; then analyzing the faults corresponding to all equipment in the shopping mall that need maintenance, and obtaining the maintenance order ranking of all faults and the abnormal time series characteristics of each fault based on the analysis results; finally, when making maintenance predictions for the equipment in the shopping mall, making maintenance predictions for the equipment in the shopping mall based on the predicted time period, the abnormal time series characteristics of all faults, and the maintenance order ranking.

[0030] Example 4: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs the steps of the above-mentioned shopping mall equipment maintenance prediction method based on time series anomaly detection to achieve the following functions: First, the time within a day is segmented based on the shopping mall's business hours to obtain multiple time periods; after segmentation, a single-time-period analysis method is used to analyze the equipment operation data in each time period, and the standard time series parameters of all equipment in each time period are obtained based on the analysis results; then, the faults corresponding to all equipment in the shopping mall that require maintenance are analyzed, and the maintenance order ranking of all faults and the abnormal time series characteristics of each fault are obtained based on the analysis results; finally, when making maintenance predictions for the equipment in the shopping mall, maintenance predictions are made for the equipment in the shopping mall based on the predicted time period, the abnormal time series characteristics of all faults, and the maintenance order ranking.

[0031] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0032] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0033] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for predicting the maintenance of shopping mall equipment based on time series anomaly detection, characterized in that, Includes the following steps: The day is segmented based on the mall's business hours to obtain multiple time periods. After segmentation, the single-time-period analysis method is used to analyze the equipment operation data in each time period, and the standard time sequence parameters of all equipment in each time period are obtained based on the analysis results. The system analyzes the faults of all equipment in the shopping mall that require maintenance, and obtains the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault based on the analysis results. The abnormal time sequence characteristics include idle time sequence characteristics, off-peak time sequence characteristics and peak time sequence characteristics. When making maintenance predictions for equipment in a shopping mall, the predictions are based on the time period in which the prediction takes place, the abnormal timing characteristics of all faults, and the ranking of the maintenance order.

2. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 1, characterized in that, The day is divided into segments based on the mall's operating hours, resulting in multiple time periods, including: Get the shopping mall's opening and non-open hours for a day, and mark the time period corresponding to the non-open hours as the idle period; mark the time period corresponding to the opening hours as the operating period, and mark the number of hours occupied by the operating period as h after rounding up. The business hours are evenly divided into h time periods, which are denoted as business sub-segment YZ1 to business sub-segment YZ. h For any business segment: Based on the historical customer flow data recorded in the mall's AI video customer flow system, obtain the customer flow of the business segment for each of the mall's operating days, and record the average of all customer flow as the average customer flow of the business segment.

3. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 2, characterized in that, The day is divided into segments based on the shopping mall's operating hours, resulting in multiple time periods, including: The average of the average passenger flow of all business segments is recorded as the measured passenger flow; business segments with an average passenger flow greater than the measured passenger flow are recorded as peak segments, and business segments with an average passenger flow less than or equal to the measured passenger flow are recorded as off-peak segments. The time period obtained by merging all peak sub-segments is called the peak time period, and the time period obtained by merging all off-peak time periods is called the off-peak time period.

4. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 3, characterized in that, Single-period analysis methods include: Obtain all equipment in the shopping mall that requires maintenance prediction, and label them as parameter analysis equipment CF1 to parameter analysis equipment CF2. n For any parameter analysis device: among all the parameters that can be collected in the parameter analysis device, the parameters that allow the construction of time series curves are recorded as the time series monitoring parameters of the parameter analysis device, where the time series curve is the relationship curve between time and parameter; For any time period α among idle time period, off-peak time period and peak time period: For any time series monitoring parameter γ of any parameter analysis device β, based on all historical monitoring records of time series monitoring parameter γ in parameter analysis device β, obtain the historical monitoring record corresponding to the time series monitoring parameter γ when the mall is in normal operation and the time is in time period α, and record it as the time period α record of time series monitoring parameter γ.

5. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 4, characterized in that, Single-period analysis also includes: For any day in the time period α record, the curve showing the relationship between the time series monitoring parameter γ and time within the time period α is denoted as the time series curve γ. The curve obtained by fitting the time series curve γ corresponding to all days in the time period α record, and the area enclosed by the time series curve γ corresponding to all days in the same coordinate system are respectively denoted as the standard time series curve and standard time series region of the time series monitoring parameter γ and time period α. Acquire all time-series monitoring parameters of all parameter analysis devices in time period α and their corresponding standard time-series curves and standard time-series regions for time period α; For any parameter analysis device β, all the time-series detection parameters of the parameter analysis device β are respectively compared with the standard time-series curves and standard time-series regions of the idle period, off-peak period and peak period, and are recorded as the standard time-series parameters of the parameter analysis device.

6. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 5, characterized in that, The system analyzes the faults of all equipment requiring maintenance in the shopping mall, and based on the analysis results, obtains the maintenance order ranking for all faults and the abnormal time sequence characteristics for each fault, including: All faults that can be generated by the parameter analysis equipment are recorded as equipment faults. For any equipment fault, based on the historical fault sample records of the parameter analysis equipment where the equipment fault is located, the shortest duration during which the parameter analysis equipment where the equipment fault is located can operate normally at the time the equipment fault occurs is obtained and recorded as the pre-fault runtime of the equipment fault. All device faults are sorted in ascending order based on their pre-run time and denoted as device fault SG1 to device fault SG2. u Among them, the lead time for equipment failure SG1 is the shortest among all equipment failures. u Its lead time is the longest among all device failure lead times.

7. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 6, characterized in that, The analysis covers the faults of all equipment requiring maintenance in the shopping mall, and based on the analysis results, it obtains the maintenance order ranking of all faults and the abnormal time sequence characteristics of each fault, including: For any device fault of any parameter analysis device β: Based on any fault occurrence record of the device fault in the historical fault sample record of the parameter analysis device β, obtain the parameter records of the device fault β in the idle period, off-peak period and peak period respectively within Q days before the occurrence of the device fault, and record them as the previous idle record, the previous off-peak record and the previous peak record respectively. The preceding record analysis method was used to analyze the preceding idle records, preceding off-peak records, and preceding peak records respectively, and the abnormal parameter characteristics in the analysis results were recorded as the idle time sequence characteristics, off-peak time sequence characteristics, and peak time sequence characteristics of equipment failure respectively.

8. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 7, characterized in that, Pre-record analysis includes: The time period corresponding to the records analyzed in the pre-analysis record is recorded as the pre-analysis time period, where the pre-analysis time period is the idle time period, off-peak time period or peak time period; For any time-series monitoring parameter γ of parameter analysis device β, the relationship curve between the time-series monitoring parameter γ and time for each day of the recorded Q days is recorded as the anomaly analysis curve γ; the curve obtained by fitting the Q anomaly analysis curves γ corresponding to the recorded Q days, and the area enclosed by the Q anomaly analysis curves γ in the same coordinate system are respectively recorded as the time-series anomaly curve and the time-series anomaly region. The time-series anomaly curve and the time-series monitoring parameter γ are placed in the same Cartesian coordinate system M1 as the standard time-series curve for the preceding analysis period. For the Cartesian coordinate system M1: the maximum slope of the time-series anomaly curve minus the maximum slope of the standard time-series curve is recorded as the anomaly slope difference. The minimum and maximum values ​​of the ordinates of all peaks in the anomaly time-series curve are recorded as Y1 and Y2, respectively. The maximum value of the ordinates of all peaks in the standard time-series curve is recorded as Y3. The interval [Y1-Y3, Y2-Y3] is recorded as the anomaly peak interval.

9. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 8, characterized in that, Pre-record analysis also includes: The time-series anomaly region and the time-series monitoring parameter γ are placed in the same Cartesian coordinate system M2 as the standard time-series region of the previous analysis period. For the Cartesian coordinate system M2: the region in the time-series anomaly region that does not coincide with the standard time-series region is recorded as the anomaly analysis region. When there is no anomaly analysis region or the number of regions corresponding to the anomaly analysis region is 1, the analysis of the Cartesian coordinate system M2 is stopped. Obtain the midpoint of all regions corresponding to the anomaly analysis region and record it as the anomaly midpoint; record the difference of the x-coordinate of adjacent anomaly midpoints as the anomaly interval; record the interval formed by the minimum and maximum values ​​of all anomaly intervals obtained from all anomaly midpoints as the anomaly period interval. Abnormal slope difference, abnormal peak interval, and abnormal period interval are denoted as abnormal difference parameters of time series monitoring parameter γ; The abnormal differences in all time-series monitoring parameters of the parameter analysis device β are recorded as abnormal parameter characteristics.

10. The method for predicting the maintenance of shopping mall equipment based on time series anomaly detection according to claim 9, characterized in that, When making maintenance predictions for equipment in a shopping mall, the predictions are based on the time period in which the prediction takes place, the abnormal timing characteristics of all failures, and the ranking of maintenance order. The process includes: When making maintenance predictions for equipment in a shopping mall, the time period in which the prediction occurs is recorded as the predicted maintenance period. For any parameter analysis device β, the parameter records corresponding to all time-series monitoring parameters of the parameter analysis device β within the complete predicted maintenance period are obtained and recorded as parameter prediction records. For any time-series monitoring parameter γ of the parameter analysis device β, the curve showing the relationship between the time-series monitoring parameter γ and time in the parameter prediction record is recorded as the time-series prediction curve; The time-series prediction curve and time-series monitoring parameters are placed in the same Cartesian coordinate system M1 as the standard time-series curve for the predicted maintenance period. Based on the prior record analysis method, the abnormal slope difference and abnormal peak interval between the time-series prediction curve and the standard time-series curve in the Cartesian coordinate system M1 are obtained and recorded as the prediction slope difference and prediction peak interval, respectively. Place the time-series prediction curve and the time-series monitoring parameter γ in the same Cartesian coordinate system M2 as the standard time-series region for the predicted maintenance period. For the Cartesian coordinate system M2: the region in the time-series prediction curve that is not in the standard time-series region is recorded as the prediction analysis region. When the time-series prediction curve is completely in the abnormal analysis region or the number of regions corresponding to the abnormal analysis region is 1, the analysis of the Cartesian coordinate system M2 is stopped. The midpoints of all regions corresponding to the prediction analysis region are obtained and recorded as the prediction midpoints; the abnormal period intervals are obtained by using the abnormal midpoints, and the prediction period intervals are obtained based on the prediction midpoints. When any timing monitoring parameter corresponding to a device fault in parameter analysis device β simultaneously satisfies conditions 1, 2, and 3, the device fault is recorded as a predicted fault. Condition 1: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the abnormal slope difference of time series monitoring parameter γ is less than or equal to the predicted slope difference; Condition 2: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal peak interval of the time-series monitoring parameter γ is less than the median of the predicted peak interval. Condition 3: Among the abnormal parameter characteristics corresponding to equipment failure and predicted maintenance period, the median of the abnormal period interval of the time-series monitoring parameter γ is less than the median of the predicted period interval. When the number of predicted faults obtained from analyzing all parameters of the device is greater than 1, all predicted faults are sorted from first to last based on the order in which they are sorted among all device faults, and fault handling is performed on all predicted faults based on the sorted order of the predicted faults.

Images