CN112225026B - Elevator maintenance method on demand based on acoustic signal - Google Patents
Elevator maintenance method on demand based on acoustic signal Download PDFInfo
- Publication number
- CN112225026B CN112225026B CN202011200413.6A CN202011200413A CN112225026B CN 112225026 B CN112225026 B CN 112225026B CN 202011200413 A CN202011200413 A CN 202011200413A CN 112225026 B CN112225026 B CN 112225026B
- Authority
- CN
- China
- Prior art keywords
- elevator
- maintenance
- acoustic signal
- signals
- acoustic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
- B66B5/0006—Monitoring devices or performance analysers
- B66B5/0018—Devices monitoring the operating condition of the elevator system
- B66B5/0025—Devices monitoring the operating condition of the elevator system for maintenance or repair
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
- B66B5/0006—Monitoring devices or performance analysers
- B66B5/0037—Performance analysers
Landscapes
- Indicating And Signalling Devices For Elevators (AREA)
Abstract
The invention belongs to the field of pattern recognition, and discloses an elevator maintenance-on-demand method based on acoustic signals, which comprises the following steps: building an acoustic signal acquisition hardware platform inside a traction machine room of the elevator, on the inner wall of a well and the top of the outer side of a car, and acquiring acoustic signals of each component of the elevator in the same day in the operation process; preprocessing the acquired acoustic signals, performing multiple derivation on a frequency domain to obtain frequency domain characteristic values of the acoustic signals, and constructing an acoustic signal characteristic data set including elevator running conditions under various different conditions; measuring the running condition of the elevator, and marking and scoring the sound signals according to the measuring result; and establishing a decision-level feature fusion model based on RNN neural network and CNN network cascade, training the network by using the labeled data set, and realizing the grading of the elevator running state and the prediction of maintenance on the same day according to the output of the trained model. The method can accurately and effectively sense the running state of the elevator, and reasonable and efficient operation and maintenance on demand are realized.
Description
Technical Field
The invention relates to the field of elevator maintenance and a pattern recognition algorithm, in particular to an elevator maintenance-on-demand method based on acoustic signals.
Background
The current elevator industry realizes regular maintenance of elevators according to elevator maintenance rules. In the maintenance process, maintenance personnel need to spend a long time to check the elevator traction machine room, the hoistway and the car and replace and maintain worn or fault parts. With the construction of more and more high-rise buildings and super high-rise buildings, the demand of the elevator is increased more and more, and the workload in the later maintenance process of the elevator is increased. With the increase of the installation amount of the elevator, the regular maintenance mode leads to the situations of untimely and in-place maintenance of the elevator and the like.
According to the relevant spirit indication in the opinion on the quality safety work of the elevator issued by the state: the maintenance mode conversion is required to be promoted, maintenance according to needs is promoted by law, new modes such as 'full package maintenance', 'Internet of things + maintenance' and the like are popularized, the quality supervision and spot check of the maintenance is enhanced, and the maintenance quality is improved. Therefore, the existing elevator maintenance mode needs to be improved, a reasonable and efficient method is provided for improving the elevator maintenance efficiency, and the elevator maintenance mode is refined and efficient.
Disclosure of Invention
The invention aims to provide an elevator maintenance-on-demand method based on acoustic signals, which can accurately judge the running state of an elevator and realize efficient elevator maintenance-on-demand.
In order to achieve the above purpose, the invention provides an elevator maintenance-on-demand method based on acoustic signals, which comprises the following steps:
s1, building an acoustic signal acquisition hardware platform in a traction machine room of the elevator, the inner wall of a well and the top of the outer side of a car, and acquiring acoustic signals of components of the elevator in the same day in the operation process;
s2, preprocessing the acoustic signal acquired in the step S1, performing multiple derivation on a frequency domain to obtain a frequency domain characteristic value of the acoustic signal, and constructing an acoustic signal characteristic data set including elevator running conditions under various different conditions;
and S3, measuring the elevator running condition of the current day corresponding to the characteristic data set in the step S2, and marking and scoring the sound signals according to the measurement result.
And S4, establishing a decision-level feature fusion model based on RNN neural network and CNN network cascade connection, training the network by using the labeled data set, and realizing the grading of the elevator running state and the prediction of maintenance on the same day according to the output of the trained model.
Preferably, the sound signal acquisition hardware platform in the first step comprises a target elevator, a free field microphone arranged in the target elevator, a conversion, storage and remote transmission system and a server; the free field microphones are respectively arranged in an elevator traction room of the target elevator, on the inner wall of a well and at the top of the outer side of a car and are used for acquiring sound signal analog signals in the running process of the target elevator; a free field microphone arranged in the target elevator transmits the analog signal to a conversion, storage and remote transmission system in a wired connection mode; the core processing device in the conversion, storage and remote transmission system converts the analog signal into a digital signal and adds a time stamp, the digital signal is stored in a system hard disk in a file form after being numbered and labeled, and data remote transmission is carried out in a 4G or WiFi form through a DTU or a router. The server opens up a fixed IP port number, monitors data signals transmitted in a 4G or WiFi mode and processes the data signals.
Preferably, the frequency domain characteristic values in step S2 include the acoustic signal energy at the frequency f and the rate of change of the acoustic signal energy at the time frame t.
Preferably, in step S3: the labeling of the acoustic signal is divided into two parts:
(1) the acoustic signal in the single time frame that gathers inside the elevator traction machine room, well inner wall and car outside top is markd according to the rule, and the type of marking includes: unqualified (1 point), basically qualified (2 points) and qualified (3 points);
the unqualified condition is the condition that maintenance is urgently needed, such as abnormal sound of a tractor, abnormal sound of a reduction box, abnormal vibration of a car, abnormal swinging of a steel wire rope and the like, occurs in the time frame;
basically qualified as allowing the sound generated by the aging of the components of the elevator in the normal operation process in the time frame;
qualified that the sound in the time frame has no abnormal condition;
(2) evaluating and scoring the whole operation process of the elevator on the same day according to professional manual experience, wherein the scoring of 1-5 grades comprises the following steps:
1 minute: the elevator does not need maintenance;
and 2, dividing: the elevator needs to be maintained for half a month according to elevator maintenance rules;
and 3, dividing: elevators need to be maintained quarterly according to elevator maintenance rules;
and 4, dividing: the elevator needs to be maintained for half a year according to elevator maintenance rules;
and 5, dividing: the elevator needs to be maintained annually according to elevator maintenance rules.
Preferably, the overall framework of the decision-level feature fusion model based on the RNN neural network and CNN network cascade comprises a feature input layer, an RNN hidden layer, a CNN hidden decision layer, an output layer and a network training module; the input layer is responsible for carrying out primary processing on the characteristic values of the acoustic signals of different parts of the elevator; the RNN hidden layer is responsible for evaluating the qualification condition of the part according to the acoustic signal; the CNN hidden decision layer is used as a decision level, and the overall operation condition of the elevator is evaluated according to the qualification condition of each part; the output layer is responsible for providing the result of the predictive scoring; the BPTT optimization method is used for RNN network training, and the BP optimization method is used for CNN network training; the trained model can be output to score the running state of the elevator on the same day and predict whether maintenance is needed and the degree of maintenance needed.
Compared with the existing on-time maintenance mode, the elevator on-demand maintenance method based on the acoustic signals, provided by the invention, has the advantages that the sound collection platform is built to carry out feature extraction on the acoustic signals of the key parts of the elevator, and the decision-level feature fusion model based on the RNN neural network and the CNN network cascade is used for identifying the acoustic signals, so that the evaluation and the scoring of the running state of the elevator are realized. On the basis of ensuring the maintenance effect, the method greatly improves the overall efficiency of elevator maintenance, realizes targeted fine maintenance of the elevator, and greatly reduces elevator faults caused by overdue maintenance.
Drawings
Fig. 1 is a hardware platform schematic diagram of an acoustic signal based elevator on-demand maintenance method of the present invention;
FIG. 2 is a network structure diagram of a decision-level feature fusion model based on the cascade connection of RNN neural network and CNN network;
FIG. 3 is a diagram illustrating the process of transforming an acoustic signal into Mel frequency domain based on Mel frequency coefficients to obtain energy amplitude in the present invention;
fig. 4 is a flow chart of a specific implementation method of the elevator maintenance-on-demand method based on the acoustic signal.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and fully described below with reference to the accompanying drawings. It is apparent that the embodiments described below are only a part of the embodiments of the present invention, and not all of them. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without inventive modifications, are intended to be included within the scope of the present invention.
In order that those skilled in the art will better understand the technical solution of the present invention, the following detailed description of the embodiments of the present invention will be given with reference to the accompanying drawings
The invention provides an elevator maintenance-on-demand method based on acoustic signals, which comprises the following steps as shown in figure 1:
s1, building an acoustic signal acquisition hardware platform in a traction machine room of the elevator, the inner wall of a well and the top of the outer side of a car, and acquiring acoustic signals of components of the elevator in the same day in the operation process;
the specific processing procedure is shown in fig. 2: the acoustic signal acquisition hardware platform comprises a target elevator 101, a free field microphone arranged in the target elevator, a conversion, storage and remote transmission system 102 and a server 103; the free field microphones are respectively arranged in an elevator traction room of a target elevator 101, on the inner wall of a well and at the top of the outer side of a car and are used for acquiring sound signal analog signals in the running process of the target elevator; a free field microphone arranged in the target elevator transmits the analog signal to a conversion, storage and remote transmission system 102 in a wired connection mode; the conversion, storage and remote transmission system 102 may use a single chip, a DSP, an FPGA, etc. as core processing devices, convert analog signals into digital signals and add timestamps, store the digital signals in a system hard disk in a file form after numbering and labeling, and perform data remote transmission in a 4G or WiFi form through a DTU or a router. The server 103 opens up a fixed IP port number, listens and processes data signals transmitted in 4G or WiFi form, and the subsequent steps S2, S3, S4 and S5 are all set to run in the server.
S2, preprocessing the acoustic signal acquired in the step S1, performing multiple derivation on a frequency domain to obtain a frequency domain characteristic value of the acoustic signal, and constructing an acoustic signal characteristic data set including elevator running conditions under various different conditions;
in practical situations, people or goods enter and exit the elevator in operation, and noise and interference inevitably exist in collected acoustic signals, so that the elevator needs to be subjected to filtering pretreatment. The preprocessing filter adopts an Elliptic filter (eliptic filter), which has equal ripples at a pass band and a stop band, has minimum pass band and stop band ripples under the condition of consistent orders compared with other types of filters, and is very suitable for minimizing and filtering uncertain noise and interference in elevator operation. Its magnitude squared function is:
wherein omegapIs the bandpass cut-off angular frequency; ε is the waviness coefficient of
UN(x) Is an N-order Jacobian elliptic function, and:
in the formula: rpIs the ripple factor (dB);
(Ω is stop band angular frequency);
(
Asstopband attenuation (dB)); k (x) is the first type of elliptic integral.
After the acoustic signal is filtered to remove the environmental noise, the processed acoustic signal is subjected to framing processing every 30 seconds, and 50% of overlapped framing is required to avoid the large difference between two frames. Windowing is then performed to multiply each frame by a hamming window w (n) to increase intra-frame left-right coherence.
After multiplying by the hamming window, each frame is converted into the mel frequency domain based on the mel frequency coefficient to obtain the energy amplitude of the frame, and the specific process is as shown in fig. 3:
firstly, the sound signal of the frame is subjected to fast Fourier transform to obtain a frequency spectrum signal, and then the linear frequency spectrum signal f is converted into a Mel frequency spectrum signal mel (f) ═ 2595 log10(1+ f/700), compared with a linear spectrum, the Mel spectrum takes human auditory characteristics into consideration, and abnormal sounds in the elevator running process can be reflected more prominently. Subsequently, taking logarithm, namely log (mel (f)), of the Mel frequency spectrum signal, and conveniently performing discrete cosine transform to obtain a final output energy amplitude vector e (f, t). Finally, the acoustic signal characteristic vector of the elevator operation is formed by the acoustic signal energy e (f, t) and the first derivative thereof
The rate of change of the acoustic signal energy.
And S3, measuring the elevator running condition of the current day corresponding to the characteristic data set in the step S2, and marking and scoring the sound signals according to the measurement result.
The labeling of the acoustic signal is divided into two parts:
(1) the acoustic signal in the single time frame that gathers inside the elevator traction machine room, well inner wall and car outside top is markd according to the rule, and the type of marking includes: unqualified (1 point), basically qualified (2 points) and qualified (3 points);
the unqualified condition is the condition that maintenance is urgently needed, such as abnormal sound of a tractor, abnormal sound of a reduction box, abnormal vibration of a car, abnormal swinging of a steel wire rope and the like, occurs in the time frame;
basically qualified as allowing the sound generated by the aging of the components of the elevator in the normal operation process in the time frame;
it is qualified that the sound in the time frame has no abnormal condition.
(2) Evaluating and scoring the whole operation process of the elevator on the same day according to professional manual experience, wherein the scoring of 1-5 grades comprises the following steps:
1 minute: the elevator does not need maintenance;
and 2, dividing: the elevator needs to be maintained for half a month according to elevator maintenance rules;
and 3, dividing: the elevator needs to be maintained quarterly according to elevator maintenance rules;
and 4, dividing: the elevator needs to be maintained for half a year according to elevator maintenance rules;
and 5, dividing: the elevator needs to be maintained annually according to elevator maintenance rules.
And S4, establishing a decision-level feature fusion model based on RNN neural network and CNN network cascade connection, training the network by using the labeled data set, and realizing the grading of the elevator running state and the prediction of maintenance on the same day according to the output of the trained model.
The decision-level feature fusion model based on the RNN neural network and CNN network cascade is shown in FIG. 4, and the model still maintains the comprehensive integration of information of different key parts of an elevator while considering the internal hidden meaning of each frame of signal through a multi-layer network structure.
The characteristic input layer is used for standardizing the sound signal data characteristic values of the interior of the elevator traction machine room, the inner wall of the shaft and the top of the outer side of the car obtained in the step S2 and inputting the sound signal data characteristic values into the RNN hidden layer, and the specific formula is as follows:
the RNN neural network learns the feature expression of each frame of standardized sound signal feature value by mining the high-level meaning of the feature value of each frame of standardized sound signal, and the RNN neural network finishes the judgment of whether the sound signal in the time frame is qualified or not. The RNN neural network is composed of a plurality of neural units in series connection, and each neural unit is internally mainly composed of the following two functions:
S(t)＝φ(Ux(t)+WS(t-1)+b)
O(t)＝VS(t)+c
where x (t) is input, s (t) is information transfer unit, o (t) is output unit, U, W, V are weight matrix, b and c are offset values, and phi () is Tanh activation function. The RNN neural network may adjust the internal parameters of the weight matrix U, W, V and the bias values b, c by learning features in the timing signal based on the information delivery unit of time t. The RNN neural network is very suitable for the judging process of whether the continuous time frame is qualified or not.
And the CNN hidden decision-making layer takes the qualified condition of the sound signal frame of each elevator key part judged by the RNN neural network as input, and performs overall decision-making level evaluation on the elevator overall operation process in the same day. The CNN hidden decision layer is internally provided with three convolution layers, and the three convolution layers all comprise convolution operation of 3x3 and a linear correction unit. The CNN hidden decision layer is responsible for overall planning of the whole-day running conditions of different key parts of the elevator, the actual running state of the elevator is mined by utilizing multilayer convolution, and an internal weight matrix is adjusted through back propagation, so that the state of the elevator is accurately evaluated, and the prediction of maintenance is completed.
The output layer is a full connection layer and is responsible for mapping the output of the last layer of the convolutional layer to Y1-Y5, Y1-Y5 sequentially corresponds to the scoring condition of 1-5 minutes in the second part of the elevator running process in the day in the step S3, the operation and maintenance personnel judge according to the value of Y1-Y5, and the item with the largest value represents the running condition of the elevator in the day and the type of running maintenance required.
In the network training module, the BPTT optimization method is used for RNN network training, and the BP optimization method is used for CNN network training.
After the decision-level feature fusion model based on the RNN neural network and CNN network cascade connection is built, the data set is labeled according to the standard through professional manual identification in the step S3, and then the data set is input into the model, and the network learns the basis of professional manual identification evaluation from the data set through multiple back propagation iterative optimization, so that the training process of the model is completed. The trained model can score the running state of the elevator on the same day and effectively predict whether maintenance is needed and the maintenance degree is needed.
While there have been shown and described what are at present considered the fundamental principles and essential features of the invention and its advantages, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (3)
1. An elevator maintenance-on-demand method based on acoustic signals is characterized by comprising the following steps:
s1, building an acoustic signal acquisition hardware platform in a traction machine room of the elevator, the inner wall of a well and the top of the outer side of a car, and acquiring acoustic signals of components of the elevator in the same day in the operation process;
s2, preprocessing the acoustic signal acquired in the step S1, performing multiple derivation on a frequency domain to obtain a frequency domain characteristic value of the acoustic signal, and constructing an acoustic signal characteristic data set including elevator running conditions under various different conditions;
s3, measuring the characteristic data set in the step S2 corresponding to the running condition of the elevator on the same day, and marking and scoring the sound signals according to the measurement result;
s4, establishing a decision-level feature fusion model based on RNN neural network and CNN network cascade connection, training the network by using the labeled data set, and realizing the grading of the elevator running state and the prediction of maintenance on the same day according to the output of the trained model;
the frequency domain feature values in step S2 include: acoustic signal energy at frequency f at time frame t and rate of change of acoustic signal energy;
the acoustic signal energy e (f, t) represents that the t frame acoustic signal is converted into a frequency domain signal after being subjected to fast Fourier transform, and then the frequency domain signal is converted into a Mel frequency domain based on a Mel frequency coefficient to obtain an energy amplitude; the change rate of the acoustic signal energy is a first derivative of the acoustic signal energy e (f, t)
In the step S3:
the labeling of the acoustic signal is divided into two parts:
(1) the acoustic signal in the single time frame that gathers inside the elevator traction machine room, well inner wall and car outside top is markd according to the rule, and the type of marking includes: unqualified (1 point), basically qualified (2 points) and qualified (3 points);
the disqualification is the condition that maintenance is urgently needed when abnormal sound of a traction machine, abnormal sound of a reduction gearbox, abnormal vibration of a car and abnormal swinging of a steel wire rope occur in the time frame;
basically qualified as allowing the sound generated by the aging of the components of the elevator in the normal operation process in the time frame;
qualified that the sound in the time frame has no abnormal condition;
(2) evaluating and scoring the whole operation process of the elevator on the same day according to professional manual experience, wherein the scoring of 1-5 grades comprises the following steps:
1 minute: the elevator does not need maintenance;
and 2, dividing: the elevator needs to be maintained for half a month according to elevator maintenance rules;
and 3, dividing: the elevator needs to be maintained quarterly according to elevator maintenance rules;
and 4, dividing: the elevator needs to be maintained for half a year according to elevator maintenance rules;
and 5, dividing: the elevator needs to be maintained annually according to elevator maintenance rules.
2. The on-demand maintenance method for elevators according to claim 1, wherein the hardware platform for collecting acoustic signals in step S1 includes a target elevator, a free-field microphone disposed in the target elevator, a conversion, storage and remote transmission system and a server; the free field microphones are respectively arranged in an elevator traction room of the target elevator, on the inner wall of a well and at the top of the outer side of a car and are used for acquiring sound signal analog signals in the running process of the target elevator; a free field microphone arranged in the target elevator transmits the analog signal to a conversion, storage and remote transmission system in a wired connection mode; the core processing device in the conversion, storage and remote transmission system converts analog signals into digital signals and adds a time stamp, the digital signals are stored in a system hard disk in a file form after numbering and marking, data remote transmission is carried out in a 4G or WiFi form through a DTU or a router, a server opens up a fixed IP port number, and data signals transmitted in the 4G or WiFi form are monitored and processed.
3. The on-demand maintenance method for the elevator according to claim 1, wherein the step S4 is specifically: the overall framework of the decision-level feature fusion model based on the RNN neural network and CNN network cascade comprises a feature input layer, an RNN hidden layer, a CNN hidden decision layer, an output layer and a network training module; the input layer is responsible for carrying out primary processing on the characteristic values of the acoustic signals of different parts of the elevator; the RNN hidden layer is responsible for evaluating the qualification condition of the part according to the acoustic signal; the CNN hidden decision layer is used as a decision level, and the overall operation condition of the elevator is evaluated according to the qualification condition of each part; the output layer is responsible for providing the result of the predictive scoring; the BPTT optimization method is used for RNN network training, and the BP optimization method is used for CNN network training; the trained model can be output to score the running state of the elevator on the same day and predict whether maintenance is needed and the degree of maintenance needed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011200413.6A CN112225026B (en) | 2020-10-30 | 2020-10-30 | Elevator maintenance method on demand based on acoustic signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011200413.6A CN112225026B (en) | 2020-10-30 | 2020-10-30 | Elevator maintenance method on demand based on acoustic signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112225026A CN112225026A (en) | 2021-01-15 |
| CN112225026B true CN112225026B (en) | 2022-05-24 |
Family
ID=74121871
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011200413.6A Active CN112225026B (en) | 2020-10-30 | 2020-10-30 | Elevator maintenance method on demand based on acoustic signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112225026B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112978531A (en) * | 2021-02-07 | 2021-06-18 | 猫岐智能科技（上海）有限公司 | Elevator operation evaluation system |
| CN115009944A (en) * | 2022-05-26 | 2022-09-06 | 浙江鑫梯互联科技有限公司 | Intelligent real-time monitoring system for elevator running state |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07228443A (en) * | 1994-02-15 | 1995-08-29 | Hitachi Building Syst Eng & Service Co Ltd | Inspecting device for elevator |
| CA2810457A1 (en) * | 2013-03-25 | 2014-09-25 | Gerald Bradley PENN | System and method for applying a convolutional neural network to speech recognition |
| CN108291837A (en) * | 2015-12-09 | 2018-07-17 | 三菱电机株式会社 | The diagnostic system for deteriorating position estimation device, deteriorating position method of estimation and moving body |
| CN110407063A (en) * | 2019-08-28 | 2019-11-05 | 开滦（集团）有限责任公司电信分公司 | Mine shaft hoist system rigid cage guide method for real-time monitoring and system |
| CN110407051A (en) * | 2019-08-02 | 2019-11-05 | 杭州岁丰信息技术有限公司 | A kind of identification lift running safety monitoring of audio-video and pacify system |
| CN111325095A (en) * | 2020-01-19 | 2020-06-23 | 西安科技大学 | Intelligent equipment health state detection method and system based on sound wave signals |
| CN111354371A (en) * | 2020-02-26 | 2020-06-30 | Oppo广东移动通信有限公司 | Method, device, terminal and storage medium for predicting running state of vehicle |
| CN111526469A (en) * | 2020-04-30 | 2020-08-11 | 成都千立网络科技有限公司 | Sound amplification system squeaking point detection method based on neural network |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9190053B2 (en) * | 2013-03-25 | 2015-11-17 | The Governing Council Of The Univeristy Of Toronto | System and method for applying a convolutional neural network to speech recognition |
| US9743911B2 (en) * | 2014-09-03 | 2017-08-29 | Contextvision Ab | Methods and systems for automatic control of subjective image quality in imaging of objects |
| CN109292567A (en) * | 2018-02-28 | 2019-02-01 | 武汉大学 | A kind of elevator faults prediction technique based on BP neural network |
-
2020
- 2020-10-30 CN CN202011200413.6A patent/CN112225026B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07228443A (en) * | 1994-02-15 | 1995-08-29 | Hitachi Building Syst Eng & Service Co Ltd | Inspecting device for elevator |
| CA2810457A1 (en) * | 2013-03-25 | 2014-09-25 | Gerald Bradley PENN | System and method for applying a convolutional neural network to speech recognition |
| CN108291837A (en) * | 2015-12-09 | 2018-07-17 | 三菱电机株式会社 | The diagnostic system for deteriorating position estimation device, deteriorating position method of estimation and moving body |
| CN110407051A (en) * | 2019-08-02 | 2019-11-05 | 杭州岁丰信息技术有限公司 | A kind of identification lift running safety monitoring of audio-video and pacify system |
| CN110407063A (en) * | 2019-08-28 | 2019-11-05 | 开滦（集团）有限责任公司电信分公司 | Mine shaft hoist system rigid cage guide method for real-time monitoring and system |
| CN111325095A (en) * | 2020-01-19 | 2020-06-23 | 西安科技大学 | Intelligent equipment health state detection method and system based on sound wave signals |
| CN111354371A (en) * | 2020-02-26 | 2020-06-30 | Oppo广东移动通信有限公司 | Method, device, terminal and storage medium for predicting running state of vehicle |
| CN111526469A (en) * | 2020-04-30 | 2020-08-11 | 成都千立网络科技有限公司 | Sound amplification system squeaking point detection method based on neural network |
Non-Patent Citations (1)
| Title |
|---|
| 基于小波的声发射信号去噪研究;杨慧等;《现代电子技术》;20170701;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112225026A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112225026B (en) | Elevator maintenance method on demand based on acoustic signal | |
| CN110940539B (en) | Machine equipment fault diagnosis method based on artificial experience and voice recognition | |
| CN109599126B (en) | Voice fault identification method based on mel energy spectrum and convolutional neural network | |
| CN107527617A (en) | Monitoring method, apparatus and system based on voice recognition | |
| CN106006344A (en) | Online escalator fault early warning system and fault diagnosis method | |
| CN111275255A (en) | Construction method of concrete dam deformation monitoring and forecasting model | |
| CN113611084B (en) | Visual monitoring and early warning method, device and equipment for natural disasters | |
| CN113009566B (en) | Local earthquake motion prediction model and construction method thereof | |
| CN105424366A (en) | Bearing fault diagnosis method based on EEMD adaptive denoising | |
| CN111178732A (en) | Regional dynamic fire risk assessment method based on big data enabling condition | |
| CN113192532A (en) | Mine hoist fault acoustic analysis method based on MFCC-CNN | |
| CN114004262A (en) | Gearbox bearing fault detection method and system | |
| CN113707175B (en) | Acoustic event detection system based on feature decomposition classifier and adaptive post-processing | |
| CN107656156A (en) | A kind of equipment fault diagnosis and running status appraisal procedure and system based on cloud platform | |
| CN115687969A (en) | Low-voltage transformer fault diagnosis method based on sound characteristic analysis | |
| CN114217149A (en) | Transformer acoustic fingerprint uninterrupted power detection and state early warning method | |
| CN111178731A (en) | Social unit dynamic fire risk assessment method based on big data enabling condition | |
| CN113157663A (en) | Network traffic prediction method and device based on data reconstruction and hybrid prediction | |
| CN115959549A (en) | Escalator fault diagnosis method based on digital twinning | |
| CN208516744U (en) | Escalator lubricating status monitors system and the voice data collection device for it | |
| CN113505898A (en) | Equipment predictive maintenance method based on AI technology | |
| CN113270110A (en) | ZPW-2000A track circuit transmitter and receiver fault diagnosis method | |
| CN115034422A (en) | Wind power short-term power prediction method and system based on fluctuation identification and error correction | |
| CN109886538B (en) | Railway signal equipment quality evaluation method and device based on dynamic monitoring data | |
| CN116681281A (en) | Sudden public health event acquisition system and method based on context awareness |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |