A method and system for elevator fault diagnosis based on machine learning
By collecting and analyzing the various characteristics of elevator operation data, generating abnormal coefficients, and building an elevator fault diagnosis model, it solves the problem of insufficient accuracy and reliability of elevator fault diagnosis in the existing technology, and achieves more accurate elevator fault diagnosis .
Patent Information
- Application Number
- CN202411129827.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-16
AI Technical Summary
The existing elevator fault diagnosis methods cannot detect potential faults in a timely and accurate manner, especially the insufficient comprehensive analysis and processing capabilities of a variety of operating data, resulting in insufficient accuracy and reliability of diagnostic results.
By collecting elevator operation data, including motor temperature, door status and sound amplitude timing data, Fourier transform and feature extraction are performed, parts, door status and motor abnormal coefficients are generated, elevator fault diagnosis model is constructed, and comprehensive analysis is carried out to improve diagnostic accuracy. sex.
It realizes a more accurate and reliable diagnosis of elevator failures, takes into account the operating status and intuitive operating status of elevator components, and improves the accuracy and reliability of diagnostic results.
Smart Images

Figure CN118637442B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of elevator fault diagnosis, and in particular to an elevator fault diagnosis method and system based on machine learning. Background Art
[0002] With the acceleration of modern urbanization, elevators, as an indispensable vertical transportation tool in high-rise buildings, are particularly important for their stability and safety. However, due to the complexity of the elevator system and the aging and wear of components during long-term use, elevator failures are inevitable. Traditional elevator maintenance and fault diagnosis mainly rely on regular manual inspections and basic sensor alarms. This method is not only time-consuming and labor-intensive, but also often fails to detect potential faults in a timely and accurate manner, which can easily lead to instability and safety hazards in elevator operation. In addition, some existing machine learning-based fault diagnosis studies usually only target a single type of sensor data, such as motor current and voltage, and lack comprehensive analysis and processing of multiple operating data, resulting in insufficient accuracy and reliability of the diagnostic results. Therefore, how to use modern information technology, especially machine learning technology, to conduct multi-dimensional and comprehensive analysis and processing of elevator operation data has become an urgent problem to be solved in the current field of elevator fault diagnosis.
[0003] In the prior art, application number CN202110787189.3 discloses a method for elevator fault diagnosis based on machine learning. The elevator operation data is obtained through the elevator Internet of Things data acquisition system. After data preprocessing operations such as data cleaning and feature engineering, an experimental data set is obtained; the samples in the obtained experimental data set are divided into a training set and a test set; the samples in the training set are learned using a machine learning algorithm to construct an elevator fault diagnosis model; the constructed elevator fault diagnosis model is tested through the samples in the test set to determine whether the elevator has a fault and what kind of fault has occurred. This prior art reduces the difficulty of modeling the elevator fault diagnosis model and improves the elevator fault diagnosis, but the prior art still has defects. First, the data cleaning and feature engineering processing steps used in the prior art may not be able to fully extract the key features in the data. The features extracted by feature engineering in the prior art only involve the operating characteristics of the elevator and elevator door appearances, and lack analysis of the elevator operation components, so the results obtained are not accurate enough.
[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention
[0005] The purpose of the present invention is to provide an elevator fault diagnosis method and system based on machine learning to solve the problems raised in the above background technology.
[0006] To achieve the above object, the present invention provides the following technical solutions:
[0007] A method for diagnosing elevator faults based on machine learning, the specific steps comprising:
[0008] Step 1: Collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working for a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator;
[0009] Step 2: Divide the sound amplitude time series data into segment, perform Fourier transform on the amplitude time series data of each segment of sound, convert the amplitude time series data of each segment of sound from time domain to frequency domain, extract frequency domain features of the amplitude time series data of sound to obtain sound feature data; generate sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of sound, and obtain the average sound abnormality coefficient and sound stability coefficient; generate the part abnormality coefficient through the average sound abnormality coefficient and sound stability coefficient;
[0010] Step 3: Extract the features of the door state data to obtain the door state feature data, generate the door state abnormality coefficient according to the door state feature data, extract the features of the motor temperature timing data to obtain the motor temperature feature data; perform mathematical analysis on the motor temperature feature data to generate the motor abnormality coefficient;
[0011] Step 4: The parts abnormality coefficient, door status abnormality coefficient and motor abnormality coefficient of each set of elevator operation data are used to form an elevator feature set, and the elevator fault condition is used as a label to build an elevator diagnosis model. The elevator feature set is used to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault;
[0012] Step 5: Complete the diagnosis of elevator faults through the optimized elevator diagnosis model.
[0013] Furthermore, the specific logic for obtaining the sound feature data is as follows: the amplitude time series data of each sound segment is transformed into the form of the sum of a series of complex numbers through Fourier transform to obtain the spectrum , calculate the power spectrum density of the sound amplitude time series data, and the specific formula for calculating the power spectrum density is:
[0014] ;
[0015] in, is the power spectral density, is the number of frequency components, is the sound spectrum, is the index of the frequency component;
[0016] The amplitude of each frequency is compared with a preset threshold, and the number of frequencies with amplitudes greater than the threshold is recorded as the effective frequency number.
[0017] Furthermore, the specific logic for generating the part abnormality coefficient is as follows: generating the sound abnormality coefficient according to the sound feature data, statistically analyzing the sound abnormality coefficient of the amplitude time series data of each sound segment to obtain the average sound abnormality coefficient and the sound stability coefficient, and generating the part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient;
[0018] The specific formula for generating the sound anomaly coefficient is:
[0019] ;
[0020] in, is the sound abnormality coefficient, is the effective frequency number, is the power spectral density;
[0021] The specific formula used to generate the average sound anomaly coefficient is:
[0022] ;
[0023] in, is the average sound anomaly coefficient, For the The sound anomaly coefficient corresponding to the sound segment; is the number of segments of the sound amplitude time series data;
[0024] The specific formula used to generate the sound stability factor is:
[0025] ;
[0026] in, is the sound stability coefficient,
[0027] The specific formula for generating the part anomaly coefficient is:
[0028] ;
[0029] in, is the part abnormality coefficient.
[0030] Furthermore, the specific logic for generating the door state abnormality coefficient is: extracting features from the door state data to obtain door state feature data, and generating the door state abnormality coefficient based on the door state feature data. The specific logic for extracting features from the door state data is: extracting the maximum and minimum opening and closing time of the elevator door and the maximum internal and external air pressure difference, and generating the door state coefficient based on the specific formula is:
[0031] ;
[0032] ;
[0033] ;
[0034] in, is the gate state coefficient, is the maximum time deviation, is the minimum time deviation, To collect the maximum opening and closing time, To collect the minimum opening and closing time, is the rated maximum opening and closing time, is the rated minimum opening and closing time, To collect the maximum internal and external pressure difference, It is the standard maximum internal and external air pressure difference.
[0035] Furthermore, the specific logic for obtaining the motor temperature characteristic data is as follows: calculating the average value of the temperature time series data to obtain the average temperature, smoothing the temperature time series data, performing trend line fitting on the smoothed temperature data to obtain the trend line slope, and the specific formula for smoothness processing is as follows:
[0036] ;
[0037] in, is the smoothed temperature data, is the number of windows, For time forward The temperature data at the moment; the smoothed temperature data is transformed into The slope of the trend line is obtained by the least squares method. The specific formula for obtaining the slope of the trend line is:
[0038] ;
[0039] in, is the slope of the trend line, For the The time of the data point, For the The smoothed temperature of each data point.
[0040] Furthermore, the specific logic for generating the motor abnormality coefficient is: performing mathematical analysis on the motor temperature characteristic data to generate the motor abnormality coefficient; the specific formula is:
[0041] ;
[0042] in, is the motor abnormality coefficient, is the slope of the trend line. is the average temperature.
[0043] The present invention further provides an elevator fault diagnosis system based on machine learning, the system is used to execute the above-mentioned elevator fault diagnosis method based on machine learning, specifically comprising:
[0044] A data acquisition module is used to collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working within a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator;
[0045] The sound analysis module is used to divide the sound amplitude time series data into segment, perform Fourier transform on the amplitude time series data of each segment of sound, convert the amplitude time series data of each segment of sound from time domain to frequency domain, extract frequency domain features of the amplitude time series data of sound to obtain sound feature data; generate sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of sound, and obtain the average sound abnormality coefficient and sound stability coefficient; generate the part abnormality coefficient through the average sound abnormality coefficient and sound stability coefficient;
[0046] The comprehensive analysis module is used to extract the characteristics of the door state data to obtain the door state characteristic data, generate the door state abnormality coefficient according to the door state characteristic data, extract the characteristics of the motor temperature time series data to obtain the motor temperature characteristic data; perform mathematical analysis on the motor temperature characteristic data to generate the motor abnormality coefficient;
[0047] A model optimization module is used to form an elevator feature set from the parts abnormality coefficient, door state abnormality coefficient and motor abnormality coefficient of each set of elevator operation data, build an elevator diagnosis model with the elevator fault condition as a label, and use the elevator feature set to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault;
[0048] The diagnostic module is used to diagnose elevator faults through an optimized elevator diagnostic model.
[0049] Compared with the prior art, the beneficial effect of the present invention is that the present invention not only takes into account the intuitive operating status of the elevator, but also analyzes the elevator operating components and performs feature extraction to generate a part abnormality coefficient reflecting the operating status of the parts and a motor abnormality coefficient reflecting the abnormal condition of the motor, and then comprehensively analyzes the part abnormality coefficient and the motor abnormality coefficient reflecting the component abnormality with the door state abnormality coefficient that intuitively reflects the abnormal operating condition to diagnose the elevator fault condition, so that the diagnostic result obtained is more accurate and reliable. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 It is a schematic diagram of the overall method flow of the present invention;
[0051] Figure 2 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION
[0052] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.
[0053] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0054] Example:
[0055] See also Figure 1 , the present invention provides a technical solution:
[0056] A method for diagnosing elevator faults based on machine learning, the specific steps comprising:
[0057] Step 1: Collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working for a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator;
[0058] The motor temperature time series data is collected by a temperature sensor placed on the motor, the door status data includes the door opening and closing time and the air pressure data inside and outside the elevator, the door opening and closing time is obtained by a position sensor set on the door, and the air pressure data inside and outside the elevator is collected by a gas pressure sensor set inside and outside the elevator. The sound amplitude time series data is collected by a sound capturer inside the elevator.
[0059] Step 2: Divide the sound amplitude time series data into segment, perform Fourier transform on the amplitude time series data of each segment of sound, convert the amplitude time series data of each segment of sound from time domain to frequency domain, extract frequency domain features of the amplitude time series data of sound to obtain sound feature data; generate sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of sound, and obtain the average sound abnormality coefficient and sound stability coefficient; generate the part abnormality coefficient through the average sound abnormality coefficient and sound stability coefficient;
[0060] The specific logic for generating the part abnormality coefficient is as follows: generate the sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient of the amplitude time series data of each sound segment to obtain the average sound abnormality coefficient and the sound stability coefficient, and generate the part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient; the specific formula for generating the sound abnormality coefficient is as follows: transform the amplitude time series data of each sound segment into the form of the sum of a series of complex numbers through Fourier transform to obtain the spectrum , calculate the power spectral density of the sound amplitude time series data,
[0061] The specific formula for calculating the power spectral density is:
[0062] ;
[0063] in, is the power spectral density, is the number of frequency components, is the sound spectrum, is the index of the frequency component;
[0064] The amplitude of each frequency is compared with the preset threshold, and the frequency number with amplitude greater than the threshold is recorded as the effective frequency number. Power spectral density Reflects the energy density of sound signals at different frequencies. When an elevator fails, it usually produces abnormal noise. The power spectrum density Will increase.
[0065] The amplitude of each frequency is compared with the preset threshold, and the frequency with an amplitude greater than the threshold is recorded as the effective frequency. When an elevator fails, abnormal noise is usually generated. The power spectrum density will increase, and the number of effective frequencies will increase.
[0066] The specific logic for generating the part abnormality coefficient is as follows: generating the sound abnormality coefficient according to the sound feature data, performing statistical analysis on the sound abnormality coefficient at each sampling time point to obtain the average sound abnormality coefficient and the sound stability coefficient, and generating the part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient;
[0067] The specific formula for generating the sound anomaly coefficient is:
[0068] ;
[0069] in, is the sound abnormality coefficient, is the effective frequency number, is the power spectrum density; sound abnormality coefficient Reflects abnormal sound conditions. The larger the value, the more abnormal sounds there are inside the elevator.
[0070] The specific formula used to generate the average sound anomaly coefficient is:
[0071] ;
[0072] in, is the average sound anomaly coefficient, For the The sound anomaly coefficient corresponding to the sound segment; The number of time series data segments of the sound amplitude; the average sound abnormality coefficient It reflects the comprehensive abnormal sound situation over a period of time. The larger the value, the more abnormal sounds there are inside the elevator during this period.
[0073] The specific formula used to generate the sound stability factor is:
[0074] ;
[0075] in, is the sound stability coefficient, the sound stability coefficient It reflects the fluctuation of the sound of the elevator parts when the elevator is working. The larger the value, the greater the fluctuation. When the elevator is working normally, the sound of the parts is stable. Smaller.
[0076] The specific formula for generating the part anomaly coefficient is:
[0077] ;
[0078] in, is the part abnormality coefficient. Reflects the operating condition of the parts. The larger the value, the worse the operating condition of the parts. Part abnormality coefficient The generation of can provide an important basis for the diagnosis of elevator faults.
[0079] Step 3: Extract the features of the door state data to obtain the door state feature data, generate the door state abnormality coefficient according to the door state feature data, extract the features of the motor temperature timing data to obtain the motor temperature feature data; perform mathematical analysis on the motor temperature feature data to generate the motor abnormality coefficient;
[0080] The door state data is feature extracted to obtain the door state feature data, and the door state abnormality coefficient is generated according to the door state feature data. The specific logic for feature extraction of the door state data is as follows: the maximum and minimum opening and closing time of the elevator door and the maximum internal and external pressure difference are extracted, and the specific formula for generating the door state coefficient is as follows:
[0081] ;
[0082] ;
[0083] ;
[0084] in, is the gate state coefficient, is the maximum time deviation, is the minimum time deviation, To collect the maximum opening and closing time, To collect the minimum opening and closing time, is the rated maximum opening and closing time, is the rated minimum opening and closing time, To collect the maximum internal and external pressure difference, It is the standard maximum internal and external pressure difference. Maximum time deviation and minimum time deviation It reflects the degree of deviation of the elevator door opening and closing time from the specified range. The larger the value, the greater the deviation of the elevator door opening and closing time from the standard range, and the greater the probability of natural failure. Reflects the changes in internal and external air pressure during the door opening and closing process. For a normal elevator, the door opening and closing process is relatively smooth and the speed is uniform. The difference between the internal and external air pressure caused by each opening and closing is not too large. In the case of a fault, whether it is too fast or too slow, the difference between the internal and external pressure and the standard will be large. Door state coefficient Reflects the operating status of the door. The larger the value, the worse the door operating status. The generation of can provide an important basis for the diagnosis of elevator faults.
[0085] The specific logic for obtaining the motor temperature characteristic data is as follows: calculate the average value of the temperature time series data to obtain the average temperature, smooth the temperature time series data, perform trend line fitting on the smoothed temperature data to obtain the trend line slope, and the specific formula for smoothness processing is as follows:
[0086] ;
[0087] in, is the smoothed temperature data, is the number of windows, For time t forward The temperature data at the moment; the smoothed temperature data is transformed into The slope of the trend line is obtained by the least squares method. The specific formula for obtaining the slope of the trend line is:
[0088] ;
[0089] in, is the slope of the trend line, For the The time of the data point, For the The smoothed temperature of the data points. The slope of the trend line Reflects the rise in elevator temperature over a long period of time. The larger the value, the greater the overall trend of the elevator temperature rise, the higher the motor temperature will be, and the more likely the motor will fail. The slope of the trend line It can provide important basis for the diagnosis of elevator failures.
[0090] The specific logic for generating the motor abnormality coefficient is: mathematically analyzing the motor temperature characteristic data to generate the motor abnormality coefficient; the specific formula is:
[0091] ;
[0092] in, is the motor abnormality coefficient, is the slope of the trend line. is the average temperature. Motor abnormality coefficient Reflects the abnormal condition of the motor. The larger the value, the higher the degree of abnormality of the motor.
[0093] Step 4: The parts abnormality coefficient, door status abnormality coefficient and motor abnormality coefficient of each set of elevator operation data are used to form an elevator feature set, and the elevator fault condition is used as a label to build an elevator diagnosis model. The elevator feature set is used to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault;
[0094] The specific logic for constructing the elevator diagnosis model in this embodiment is: the parts abnormality coefficient, door state abnormality coefficient and motor abnormality coefficient of each set of elevator operation data constitute an elevator feature set, and the elevator feature set is expressed as:
[0095] ;
[0096] in, The number of historical data sets collected.
[0097] A support vector machine model is adopted, and the elevator fault condition is used as a label. The elevator feature set is used as a support vector to train and optimize the elevator diagnosis model. The elevator fault condition includes 0 for a normal elevator and 1 for an elevator fault. It should be noted that the elevator fault condition is used as a label and the elevator feature set is used to train and optimize the elevator diagnosis model as an existing technology, which specifically includes selecting a radial basis kernel function, finding a hyperplane, calculating the distance between each sample point in the training set and the hyperplane, and maximizing the distance from the hyperplane to the support vector.
[0098] Step 5: Complete the diagnosis of elevator faults through the optimized elevator diagnosis model.
[0099] See also Figure 2 The present invention further provides an elevator fault diagnosis system based on machine learning, the system is used to execute the above-mentioned elevator fault diagnosis method based on machine learning, specifically comprising:
[0100] A data acquisition module is used to collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working within a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator;
[0101] The sound analysis module is used to divide the sound amplitude time series data into segment, perform Fourier transform on the amplitude time series data of each segment of sound, convert the amplitude time series data of each segment of sound from time domain to frequency domain, extract frequency domain features of the amplitude time series data of sound to obtain sound feature data; generate sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of sound, and obtain the average sound abnormality coefficient and sound stability coefficient; generate the part abnormality coefficient through the average sound abnormality coefficient and sound stability coefficient;
[0102] The comprehensive analysis module is used to extract the characteristics of the door state data to obtain the door state characteristic data, generate the door state abnormality coefficient according to the door state characteristic data, extract the characteristics of the motor temperature time series data to obtain the motor temperature characteristic data; perform mathematical analysis on the motor temperature characteristic data to generate the motor abnormality coefficient;
[0103] A model optimization module is used to form an elevator feature set from the parts abnormality coefficient, door state abnormality coefficient and motor abnormality coefficient of each set of elevator operation data, build an elevator diagnosis model with the elevator fault condition as a label, and use the elevator feature set to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault;
[0104] The diagnostic module is used to diagnose elevator faults through an optimized elevator diagnostic model.
[0105] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.
[0106] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0107] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.
Claims
1. A method for elevator fault diagnosis based on machine learning, characterized in that: The specific steps include: Step 1: Collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working for a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator; Step 2: Divide the amplitude time series data of the sound into n segments, perform Fourier transform on the amplitude time series data of each segment of the sound, convert the amplitude time series data of each segment of the sound from the time domain to the frequency domain, extract the frequency domain features of the amplitude time series data of the sound to obtain the sound feature data; generate the sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of the sound, and obtain the average sound abnormality coefficient and the sound stability coefficient; generate the part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient; Step 3: Extract the features of the door state data to obtain the door state feature data, generate the door state abnormality coefficient according to the door state feature data, extract the features of the motor temperature timing data to obtain the motor temperature feature data; perform mathematical analysis on the motor temperature feature data to generate the motor abnormality coefficient; Step 4: The parts abnormality coefficient, door status abnormality coefficient and motor abnormality coefficient of each set of elevator operation data are used to form an elevator feature set, and the elevator fault condition is used as a label to build an elevator diagnosis model. The elevator feature set is used to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault; Step 5: Complete the diagnosis of elevator faults through the optimized elevator diagnosis model; The specific logic for obtaining the sound feature data is as follows: transform the amplitude time series data of each sound segment into the form of the sum of a series of complex numbers through Fourier transform, obtain the spectrum X(k), calculate the power spectrum density of the amplitude time series data of the sound, and the specific formula for calculating the power spectrum density is: Where P(k) is the power spectral density, z is the number of frequency components, X(k) is the sound spectrum, and k is the index of the frequency component; The amplitude of each frequency is compared with a preset threshold, and the number of frequencies with amplitudes greater than the threshold is recorded as the effective frequency number; The specific logic for generating the part abnormality coefficient is as follows: generate the sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient of the amplitude time series data of each sound segment to obtain the average sound abnormality coefficient and the sound stability coefficient, and generate the part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient; the specific formula for generating the sound abnormality coefficient is as follows: E=M*P(k) Among them, E is the sound abnormality coefficient, M is the effective frequency number, and P(k) is the power spectrum density; The specific formula used to generate the average sound anomaly coefficient is: in, is the average sound abnormality coefficient, E j is the sound anomaly coefficient corresponding to the jth sound segment; n is the number of sound amplitude time series data segments; The specific formula used to generate the sound stability factor is: Among them, Ee is the sound stability coefficient, The specific formula for generating the part anomaly coefficient is: Among them, Zj is the part abnormality coefficient.
2. The elevator fault diagnosis method based on machine learning according to claim 1, characterized in that: The specific logic for generating the door status abnormality coefficient is as follows: The door state data is feature extracted to obtain the door state feature data, and the door state abnormality coefficient is generated according to the door state feature data. The specific logic for feature extraction of the door state data is as follows: the maximum and minimum opening and closing time of the elevator door and the maximum internal and external pressure difference are extracted, and the specific formula for generating the door state coefficient is as follows: Among them, ME is the gate state coefficient, f(I max ) is the maximum time deviation, f(I min ) is the minimum time deviation, I max To collect the maximum opening and closing time, I min To collect the minimum opening and closing time, I 0max is the rated maximum opening and closing time, I 0min is the rated minimum opening and closing time, To collect the maximum internal and external pressure difference, It is the standard maximum internal and external air pressure difference.
3. The elevator fault diagnosis method based on machine learning according to claim 1, characterized in that: The specific logic for obtaining the motor temperature characteristic data is as follows: The average value of the temperature time series data is calculated to obtain the average temperature, the temperature time series data is smoothed, and the trend line is fitted on the smoothed temperature data to obtain the trend line slope. The specific formula for smoothness processing is: in, is the smoothed temperature data, N is the number of windows, and T(ti*Δt) is the temperature data at time i*Δt before time t; The smoothed temperature data is transformed into The slope of the trend line is obtained by the least squares method. The specific formula for obtaining the slope of the trend line is: Where b is the slope of the trend line, t i is the time of the ith data point, T(t i ) is the smoothed temperature of the ith data point.
4. The elevator fault diagnosis method based on machine learning according to claim 1, characterized in that: The specific logic for generating the motor abnormality coefficient is as follows: The motor temperature characteristic data is mathematically analyzed to generate the motor abnormality coefficient; the specific formula is as follows: Among them, ET is the motor abnormality coefficient, b is the slope of the trend line, is the average temperature.
5. An elevator fault diagnosis system based on machine learning, characterized in that: The system is used to implement the elevator fault diagnosis method based on machine learning according to any one of claims 1 to 4, specifically comprising: A data acquisition module is used to collect the operation data of the elevator when it is working normally and when it fails. The operation data includes the time series data of the motor temperature, the door status data and the amplitude time series data of the sound when the elevator is working within a period of time. The door status data includes the opening and closing time of the elevator door and the pressure difference between the inside and outside of the elevator; The sound analysis module is used to divide the amplitude time series data of the sound into n segments, perform Fourier transform on the amplitude time series data of each segment of the sound, convert the amplitude time series data of each segment of the sound from the time domain to the frequency domain, and perform frequency domain feature extraction on the amplitude time series data of the sound to obtain sound feature data; generate a sound abnormality coefficient according to the sound feature data, perform statistical analysis on the sound abnormality coefficient corresponding to each segment of the sound, and obtain an average sound abnormality coefficient and a sound stability coefficient; generate a part abnormality coefficient through the average sound abnormality coefficient and the sound stability coefficient; The comprehensive analysis module is used to extract the characteristics of the door state data to obtain the door state characteristic data, generate the door state abnormality coefficient according to the door state characteristic data, extract the characteristics of the motor temperature time series data to obtain the motor temperature characteristic data; perform mathematical analysis on the motor temperature characteristic data to generate the motor abnormality coefficient; A model optimization module is used to form an elevator feature set from the parts abnormality coefficient, door state abnormality coefficient and motor abnormality coefficient of each set of elevator operation data, build an elevator diagnosis model with the elevator fault condition as a label, and use the elevator feature set to train and optimize the elevator diagnosis model; the elevator fault condition includes 0 for normal elevator and 1 for elevator fault; The diagnostic module is used to diagnose elevator faults through an optimized elevator diagnostic model.
