Elevator car operation process monitoring method based on fuzzy logic

Abstract

The invention discloses an elevator car operation process monitoring method based on fuzzy logic, which belongs to the field of engineering application, and comprises the following steps: S1, detrending data in the vertical direction of the vibration acceleration of a car through a Fourier fitting method; s2, calculating RMS, a peak-to-peak value and a peak value factor for the vibration signal subjected to trend removal, extracting characteristics reflecting the state of the lift car, and smoothing the characteristics; s3, dividing the RMS, the peak-to-peak value and the peak factor into intervals according to historical data and expert knowledge; and S4, constructing a membership function of the RMS, the peak-to-peak value and the peak factor for the interval by adopting fuzzy logic, and calculating a comprehensive index. Based on the comprehensive index change of the elevator car, preventive maintenance is realized, the maintenance cost is reduced, and the maintenance quality is improved.

Description

Technical Field

[0001] The present invention relates to a method for monitoring the operation process of an elevator car based on fuzzy logic, specifically, detrending of vibration signals based on Fourier curve fitting, extraction of car operation characteristics, and determination of equipment status based on fuzzy logic algorithms, belonging to the field of engineering applications. Background Art

[0003] At present, for the automatic status monitoring of elevator cars on the market, only partial event codes can be provided, such as: door opening / closing failures, overshooting, crashing, and uneven leveling, etc. The process status of the equipment cannot be provided, and predictive maintenance still requires on-site inspection by elevator maintenance personnel.

[0004] The present invention installs vibration sensors on the elevator car, and based on the collected car vibration signals, through detrending, operation feature extraction, and fuzzy logic algorithms, a fusion score reflecting the status of the car is obtained. Based on the score change of the elevator car, predictive maintenance is realized, the maintenance cost is reduced, and the maintenance quality is improved. Summary of the Invention

[0005] In view of the above deficiencies, the present invention provides a method for monitoring the operation process of an elevator car based on fuzzy logic.

[0006] The present invention is realized through the following technical solutions: A method for monitoring the operation process of an elevator car based on fuzzy logic is composed of detrending of vibration signals, extraction of operation characteristics, determination of fuzzy logic membership degrees, and calculation of comprehensive indicators. The detrending of the signals is to fit the axial vibration acceleration signals through Fourier curve fitting and subtract the fitted trend from the original signals. The extraction of the car operation characteristics is to calculate the RMS, peak-to-peak value, and peak factor characteristics for the three directions of the detrended signals respectively. The determination of the equipment status by the fuzzy logic algorithm is based on national standards, industry standards, and historical data of the car's healthy state, to determine the equipment status corresponding to the fluctuation range of each characteristic and obtain the membership function. The calculation of the comprehensive indicator is to calculate the mean value of the three characteristic values based on the membership function to reflect the status of the car.

[0007] It is characterized in that:

[0008] (1) Detrending the vertical direction data of the car vibration acceleration;

[0009] (2) Extracting the car operation characteristics;

[0010] (3) Dividing the intervals of RMS, peak-to-peak value, and peak factor according to historical data and expert knowledge;

[0011] (4) Determining the membership degree by fuzzy logic and calculating the comprehensive indicator.

[0012] The method 1: detrending the vertical direction data of the car vibration acceleration.

[0013] S1.1: Calculate the constant term . The input vibration signal is X, which has three channels: horizontal, vertical, and perpendicular. First, define the time axis t and calculate the signal period , and the corresponding angular frequency . Then, construct the input signal in complex form , where j represents the imaginary unit. Finally, calculate the constant term :

[0014] .

[0015] S1.2: Calculate the Fourier coefficients and . For each order k of the Fourier series, calculate the coefficient of each k successively through a loop to obtain the Fourier coefficients of the corresponding positive and negative frequencies and :

[0016]

[0017]

[0018] S1.3: Construct the trend signal. According to the Fourier series expansion result, superimpose all the calculated Fourier coefficients to construct the trend signal , and its formula is as follows:

[0019]

[0020] where n represents the number of terms of the Fourier series, which is used to control the accuracy of the trend signal.

[0021] S1.4: Obtain the detrended signal. The detrended signal in the vertical direction is as follows:

[0022]

[0023] where a0 is a constant, taking the real and imaginary parts of as the two components of . Subtract from the original signal in the vertical direction to obtain the detrended signal .

[0024] The method 2: Extract the running characteristics of the car and perform smoothing processing, and calculate the comprehensive characteristic value.

[0025] S2.1: Calculate the characteristics of each direction of . The three directions are respectively , and . For the signals in each direction, the root mean square (RMS), peak-to-peak value, and peak factor are calculated. The calculation formulas for the three features are as follows:

[0026] (1) RMS calculation: Calculate the RMS value to reflect the energy of vibration. The definition is as follows:

[0027]

[0028] where s represents , or signal, and N is the number of sampling points of the signal.

[0029] (2) Peak-to-peak value calculation: According to the national standard, the peak-to-peak value should be calculated based on the interval between every three zero-crossing points. The specific calculation process is as follows: First, detect the zero-crossing points in the signal, that is, the transition points where the signal changes from positive to negative or from negative to positive; Second, for every three consecutive zero-crossing points found, calculate the peak-to-peak value of this segment (the maximum value minus the minimum value of this segment); Finally, count the peak-to-peak values of all segments and select the maximum value in the peak-to-peak value sequence as the final peak-to-peak value.

[0030] (3) Peak factor calculation: The peak factor is the ratio of the maximum absolute value of the signal to the RMS, which is used to reflect the sharpness of the signal. The definition is as follows:

[0031]

[0032] S2.2: Calculate the comprehensive value of each feature value in the three directions to obtain the comprehensive features of the car operation. The calculation process is as follows:

[0033] (1) RMS comprehensive: Calculate the RMS value of the three RMSs, which represents the overall vibration energy feature of the car operation.

[0034]

[0035] where x, y, and z represent the RMS values in the three directions.

[0036] (2) Peak-to-peak value comprehensive: Take the maximum value of the peak-to-peak values in the x, y, and z directions, which represents the maximum amplitude change during the car operation.

[0037]

[0038] (3) Peak factor comprehensive: Take the maximum value of the peak factors in the x, y, and z directions, which represents the intensity of vibration change during the car operation.

[0039]

[0040] Method 3: Determine the fluctuation range of each feature and divide the intervals when the device changes from healthy to faulty by combining historical data and expert knowledge.

[0041] S3.1: Use RMS to reflect the state of the device. First, perform mean smoothing on the historical comprehensive RMS values when the car is healthy. Secondly, based on the historical comprehensive RMS values, divide the RMS into 5 intervals through Gaussian normal distribution, and use these intervals to divide the device state. The calculation formula for dividing the RMS values into the following 3 intervals [t1, t2, t3, t4, t5] is as follows:

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] Among them, n is the number of historical data, and x is the comprehensive RMS value. t1 is the comprehensive RMS value when the car is normal without jitter, t2 is the comprehensive RMS value when the car is normally jittering (left and right; up and down), t3 is the comprehensive RMS value when the car is slightly jittering, t4 is the comprehensive RMS value when the car is significantly jittering, and t5 is the comprehensive RMS value when the car is abnormally jittering (door opening and closing failure, guide rail and guide shoe looseness, etc.).

[0048] S3.2: Use the comprehensive peak-to-peak value to determine the state of the device. Determine the comprehensive peak-to-peak value pt1 in the normal state of the car based on the 25th percentile of the historical comprehensive peak-to-peak values when the car is healthy; the maximum peak-to-peak value of 0.3 specified by the national standard for the vibration of the car is the comprehensive peak-to-peak value pt5 when the car is abnormal (door opening and closing failure); through linear interpolation between pt1 and pt5, obtain pt2, pt3, and pt4 that reflect the state of the car based on the peak-to-peak value.

[0049] S3.3: Use the peak factor (PI) to determine the state of the device. Determine the comprehensive peak factor pit1 that reflects the normal state of the car based on the median of the historical comprehensive peak factors when the car is healthy; determine the comprehensive peak factor pit5 when the car is faulty (abnormal door opening and closing, guide rail and guide shoe looseness) based on the comprehensive peak factor; through linear interpolation between pit1 and pit5, obtain pit2, pit3, and pit4 that reflect the state of the car based on the peak factor.

[0050] Method 4: Use the fuzzy logic algorithm to infer the membership function.

[0051] S4.1: Define fuzzy sets and triangular membership functions. Five intervals can be set for the fuzzy sets according to the characteristics reflecting the car state. For example, divide RMS into three fuzzy sets: low, medium, and high, and use triangular membership functions to describe the membership degrees of each set. Definition of fuzzy sets: The low range is [t1, t2, t3], the medium range is [t2, t3, t4], and the high range is [t3, t4, t5]. We use triangular membership functions to describe the membership degrees of each interval. The definitions are as follows:

[0052] Low:

[0053]

[0054] Medium:

[0055]

[0056] High:

[0057]

[0058] S4.2: Define fuzzy rules for the device state. According to different intervals of the characteristic values, define the following fuzzy rules to infer the operating state of the device. Taking the RMS characteristic as an example: Rule 1: If the fuzzy set corresponding to the RMS value is low, then the device state is normal; Rule 2: If the fuzzy set corresponding to the RMS value is medium, then the device state is a warning; Rule 3: If the fuzzy set corresponding to the RMS value is high, then the device state is abnormal. The same applies to the peak-to-peak value and the peak factor.

[0059] S4.3: Defuzzification. Substitute RMS c , PTP max , PI max into their corresponding membership functions respectively, and obtain the membership degrees through the minimum rule; realize defuzzification based on the weighted average decision method to obtain the state evaluation index reflecting the car.

[0060] S.4.4 Set the threshold of the state evaluation index. When the state evaluation index is lower than the preset threshold, it indicates that the performance of the device may deteriorate, there are potential abnormal or failure risks, and further analysis or preventive maintenance measures need to be taken to ensure the safe operation and reliability of the device.

[0061] The beneficial effects of the present invention are as follows:

[0062] (1) Accurately monitor the running state of the car. By detrending, feature extraction, and smoothing processing of vibration data, the trend interference of vibration during the elevator operation can be eliminated, and features such as RMS, peak-to-peak value, and peak factor can accurately describe the running state and vibration characteristics of the car, so as to monitor whether there are abnormalities.

[0063] (2) Improve the safety of elevator operation. By calculating fuzzy logic and comprehensive evaluation indicators, combining historical data and expert experience, the elevator operation status is classified into "healthy", "abnormal", and "faulty". Potential faults during the car operation can be detected in a timely manner. When the operation status is "abnormal" or "faulty", maintenance and intervention can be carried out in advance to avoid the further expansion of potential safety hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is the overall method flow chart

[0065] Figure 2 is the original vibration signal of the elevator car in the axial direction

[0066] Figure 3 The car signal after Fourier fitting and detrending

[0067] Figure 4 is the RMS membership function graph

[0068] Figure 5 is the peak-to-peak membership function graph

[0069] Figure 6 is the peak factor membership function graph DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] The present invention will be described in detail below with reference to the accompanying drawings.

[0071] The method for monitoring the operation process of an elevator car based on fuzzy logic provided by the present invention comprises the following specific steps:

[0072] Step 1: Detrend the vertical direction data of the vibration acceleration of the car. When the elevator car is in the non-operating state, the amplitude of the vibration signal is small and fluctuates around the zero axis as a whole. When the elevator car is in the operating state, the vibration signal has obvious rising and falling trends, as Figure 2 shown. The detrended signal obtained through Fourier fitting is as Figure 3 shown, and it can be seen that the trend of the signal has been flattened.

[0073] Step 2: Extract comprehensive characteristic values.

[0074] S2.1: Calculate the RMS of the RMS values in the three directions of X, which represents the overall energy characteristic of the car operation, and obtain a comprehensive RMS value of 0.0432.

[0075] S2.2: The peak-to-peak values of X in the x, y, and z axis directions are 0.149, 0.129, and 0.274 respectively. Take the maximum value of 0.274 of the peak-to-peak values in the three directions of x, y, and z as the comprehensive peak-to-peak value.

[0076] S2.3: Similarly, the peak factors of the vibration signals in the three directions are 4.27, 4.38, and 3.58. The maximum peak factor among the three directions, 4.38, is taken as the comprehensive peak factor.

[0077] Step 3: Divide the state intervals based on historical data and expert knowledge.

[0078] First, according to the RMS of the historical data of the car health status, the RMS is divided into 5 intervals [t1, t2, t3, t4, t5] through normal distribution:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084] The obtained intervals of RMS are [0.0423, 0.0689, 0.0827, 0.0956, 0.1434].

[0085] Secondly, the peak-to-peak value in the car health state is 0.15. According to the national standard, the peak-to-peak value of vibration when the car is abnormal is 0.3. Therefore, the peak-to-peak value is divided into 5 intervals [0.15, 0.18, 0.23, 0.26, 0.3] between 0.15 and 0.3.

[0086] Finally, according to the historical data of the car, it is determined that the median of PI in the car health state is about 4.2, and the median of PI in the car abnormal state is about 13. Therefore, through linear interpolation, the car PI is divided into 5 intervals [4.2, 6.25, 8.5, 10.75, 13].

[0087] Step 4: Use the fuzzy logic algorithm to construct the membership function and infer the equipment state.

[0088] S4.1: Calculation of the RMS membership function. The RMS interval is divided into three fuzzy sets: low [0.0423 - 0.0827], medium [0.0689 - 0.0956], and high [0.0827 - 0.1434], as Figure 4 shown, and the corresponding membership functions are:

[0089]

[0090]

[0091]

[0092] S4.2: Divide the peak-to-peak value range into three fuzzy sets: low [0.15 - 0.23], medium [0.18 - 0.26], and high [0.23 - 0.3], as Figure 5 shown. The corresponding membership functions are:

[0093]

[0094]

[0095]

[0096] S4.3: Divide the peak factor range into three fuzzy sets: low [4.2 - 8.5], medium [6.25 - 10.75], and high [8.5 - 13], as Figure 6 shown. The corresponding membership functions are:

[0097]

[0098]

[0099]

[0100] Step 5: The RMS value of X is 0.0424, the peak-to-peak value is 0.3328, and the peak factor is 6.106. The membership degrees obtained by fuzzyfying the input variables are shown in Table 1:

[0101] Table 1 Membership Degrees of Input RMS, PTP, and PI

[0102] Fuzzy marker Status evaluation index RMS membership degree PTP membership degree PI membership degree Low 100-60 0.997 0 0.556 Medium 60-30 0 0 0 High 30-0 0 1 0

[0103] Step 6: Defuzzification. The membership degrees of RMS, PTP, and PI of X and the corresponding state evaluation indexes are used to obtain the final state evaluation index y through the weighted average decision method:

[0104]

[0105] Step 7: Determine the operating state. When the interval is in [100 - 60], the device changes from the healthy state to the slightly abnormal state. When the interval is in [60 - 30], the device changes from the slightly abnormal state to the significantly abnormal state. When the interval is in [30 - 0], the fluctuation of the device increases, and the device changes from the significantly abnormal state to the faulty state and needs to be shut down immediately for maintenance. Therefore, when the final output is 72.58, the device is in the healthy state.

Claims

1. A method for monitoring the operation process of an elevator car based on fuzzy logic. It is characterized in that, Including the following steps: S1: Detrend the vertical direction data of the car vibration acceleration. S1.1: Perform Fourier series fitting on the input signal in the vertical direction. The input vibration signal is X, which has three channels: horizontal, vertical, and perpendicular. Perform Fourier series fitting on the input signal in the vertical direction, and set the number of terms in the Fourier series expansion of the signal to n to control the detrending accuracy. The following are the specific calculation steps: First, define the time axis t and calculate the signal period , and the corresponding angular frequency . Construct the input signal in complex form , where j represents the imaginary unit. Calculate the value of the constant term : , and this integral is implemented by the numerical method simpson. S1.2: Calculate the Fourier coefficients and . For each order k of the Fourier series, calculate the Fourier coefficients for the corresponding positive and negative frequencies and : By calculating the coefficients for each k successively in a loop, the Simpson numerical integration method is used to calculate the Fourier coefficients. S1.3: Construct a trend signal. Based on the results of the Fourier series expansion, all calculated Fourier coefficients are superimposed to construct a trend signal , and its formula is as follows: Where n represents the number of terms of the Fourier series, which is used to control the accuracy of the trend signal. S1.4: Obtain the detrended signal in the vertical direction. The final trend signal is combined with the constant term to form the trend signal of Fourier fitting : Take the real and imaginary parts as the two components of the trend signal respectively, and subtract them from the original signal to obtain the detrended signal . S2: Extract the operating characteristics of the car and perform smoothing processing, and calculate the comprehensive RMS, peak-to-peak value, and peak factor. S2.1: Calculate the feature in different directions for the detrended signal to obtain the components of the signal in three directions , and , which respectively represent the vibration signals of the car on three axes. Calculate the features for the signals in each direction, including root mean square (RMS), peak-to-peak value, and peak factor. The following is the calculation method. RMS calculation: For the signals in each direction, calculate the RMS value to reflect the energy magnitude of the signal. The definition is as follows: where s represents 、 or a signal, and N is the number of sampling points of the signal. Peak-to-peak value calculation: According to the national standard, the peak-to-peak value should be calculated based on the interval between every three zero-crossing points. The specific calculation process is as follows: First, detect the zero-crossing points in the signal, that is, the transition points where the signal changes from positive to negative or from negative to positive. Second, for every three consecutive zero-crossing points found, extract the signal segment there. Third, calculate the peak-to-peak value of this segment, that is, the difference between the maximum value and the minimum value of the signal in this segment. Finally, perform statistics on the peak-to-peak values of all segments, and the maximum value of the peak-to-peak value sequence of each segment can be selected as the final peak-to-peak value. Peak factor calculation: Calculate the peak factor, which is the ratio of the maximum absolute value of the signal to the RMS, used to reflect the sharpness of the signal. The definition is as follows: Extract the RMS, peak-to-peak value, and peak factor from the historical data of the elevator car, and perform median smoothing processing, which can reduce the influence of accidental fluctuations and make the extracted features more stable and reliable. S2.2: Combine the eigenvalue in three directions to obtain the comprehensive characteristics of the car operation. The following is the calculation process of the RMS, peak-to-peak value, and peak factor. RMS synthesis: Take the RMS of the RMS values in the x, y, and z directions, that is, calculate the RMS value of the three RMS values, which represents the overall energy characteristics of the car operation. Maximum peak-to-peak value: Take the maximum value of the peak-to-peak values in the x, y, and z directions, which represents the maximum amplitude change during the car operation. Maximum peak factor: Take the maximum value of the peak factors in the x, y, and z directions, which represents the maximum sharpness during the car operation. S3: Divide the RMS, peak-to-peak value, and peak factor into intervals according to the historical data and expert knowledge. S3.1: In order to use the root mean square value (RMS) to determine the state of the equipment, according to the statistical characteristics of the historical data, the RMS can be divided into 5 intervals, and these intervals are used to divide the equipment state. The RMS value of the equipment follows a normal distribution. Based on the mean and standard deviation of the normal distribution, the RMS value can be divided into the following 5 intervals [t1, t2, t3, t4, t5]. The calculation formulas for the five thresholds are as follows: Where t1 is the mean of the RMS value, n is the number of data points, and xi is the i-th RMS value. Where t2 is the mean of the RMS plus three times the standard deviation, and u is the mean of the data. Where the values 1.2 and 1.5 are obtained from expert knowledge. S3.2: In order to use the comprehensive peak-to-peak value to determine the state of the equipment, according to the historical data, the 25th percentile is used to determine that the maximum peak-to-peak value in the three directions of the car is 0.15, and according to the national standard, the peak-to-peak value of the car is 0.

3. The peak-to-peak value is divided into 5 intervals [0.15, 0.18, 0.23, 0.26, 0.3]. S3.3: To determine the status of the device using the peak factor (PI), the normal fluctuation is determined to be around 4.2 based on the median of historical data and divided into 5 intervals: [4.2, 6.25, 8.5, 10.75, 13]. S4: Fuzzy logic determines the membership degree and calculates the operating status of the device through comprehensive features. S4.1: Define fuzzy sets and triangular membership functions. The fuzzy sets can be set according to the five intervals of the RMS value. The RMS is divided into three fuzzy sets: low, medium, and high, and the triangular membership function is used to describe the membership degree of each set. Fuzzy set definition: The low range is [t1, t2, t3], the medium range is [t2, t3, t4], and the high range is [t3, t4, t5]. We use the triangular membership function to describe the membership degree of each interval. The definition is as follows: Low: Medium: High: S4.2: Define the fuzzy rules for the device status. According to different intervals of the RMS value, the following fuzzy rules are defined to infer the operating status of the device: Rule 1: If the RMS value is low, the device status is normal. Rule 2: If the RMS value is medium, the device status is early warning. Rule 3: If the RMS value is high, the device status is abnormal. Refer to 5.1 and 5.2 to determine the operating status of the device using the fuzzy logic algorithm based on the peak-to-peak value and the peak factor. S4.3: Through the calculation and processing of the car operation data, key characteristic parameters such as RMS, peak-to-peak value, and peak factor are extracted, and a status evaluation index of the device is generated based on these characteristic parameters. On the basis of referring to national standards, industry standards, and historical data, this index system sets a reasonable threshold interval for dynamically monitoring the operating status of the device. When any characteristic index is lower than the preset threshold, it indicates that the performance of the device may deteriorate, there are potential abnormal or failure risks, and further analysis or preventive maintenance measures need to be taken to ensure the safe operation and reliability of the device.

Images