CN114893428B - Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator - Google Patents
Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator Download PDFInfo
- Publication number
- CN114893428B CN114893428B CN202210351665.1A CN202210351665A CN114893428B CN 114893428 B CN114893428 B CN 114893428B CN 202210351665 A CN202210351665 A CN 202210351665A CN 114893428 B CN114893428 B CN 114893428B
- Authority
- CN
- China
- Prior art keywords
- diagnosis
- ventilator
- module
- signal
- indexes
- 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
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04D—NON-POSITIVE-DISPLACEMENT PUMPS
- F04D27/00—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids
- F04D27/001—Testing thereof; Determination or simulation of flow characteristics; Stall or surge detection, e.g. condition monitoring
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04D—NON-POSITIVE-DISPLACEMENT PUMPS
- F04D27/00—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04D—NON-POSITIVE-DISPLACEMENT PUMPS
- F04D27/00—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids
- F04D27/008—Stop safety or alarm devices, e.g. stop-and-go control; Disposition of check-valves
Abstract
The invention discloses a device and a system for fine pre-diagnosis and precise operation and maintenance of a ventilator, wherein the device comprises a state monitoring unit, a state monitoring unit and a state monitoring unit, wherein the state monitoring unit is used for acquiring economic indexes, performance indexes and safety indexes of the ventilator, displaying the change trend related to signals in real time and calculating and displaying health degree indexes and total efficiency index values; the signal analysis unit is used for dynamically calculating and displaying the characteristic values and the change curves of all measuring points contained in each physical signal, predicting the change trend of each physical signal and dynamically calculating the alarm threshold value of each signal; the early warning diagnosis unit is used for giving out a fault diagnosis conclusion of the ventilator in parallel by utilizing a refined early warning diagnosis method and a big data artificial intelligence early warning diagnosis method, and confirming the fault of the ventilator or artificial diagnosis intervention according to the diagnosis conclusion; and the accurate operation and maintenance unit is used for predicting the service life of key parts in the ventilator and developing a fine pre-diagnosis report of the ventilator according to the prediction result and the statistical result of the point inspection/maintenance record.
Description
Technical Field
The invention relates to the technical field of detection of ventilators, in particular to a fine pre-diagnosis and accurate operation and maintenance device and system for a ventilator.
Background
The ventilator is a key device for gas circulation and discharge in industrial production, and is widely applied to important fields of ferrous metallurgy, power petrochemical industry, traffic transportation and the like. The running environment of the ventilator is severe and complex, abrasion and corrosion are caused by long-time running, accelerated aging and failure of parts are caused, and the failure of the ventilator is caused. Once a ventilator breaks down, normal work of equipment and a production line where the ventilator is located is influenced and economic loss is caused, and a safety production accident is caused. Therefore, in order to ensure the running safety of the ventilator, reduce the fault shutdown loss and avoid safety accidents, effective methods and technical means are adopted to carry out state monitoring and health pre-diagnosis on the ventilator and key parts thereof.
At present, in the field operation and maintenance work of the ventilator, the major bottleneck problems of complex failure mode, difficult prediction of early failure, weak guidance of evaluation and prediction on spare part management and the like exist. Although a monitoring and protecting system for a ventilator can be seen in the prior art, a technology for providing a system solution for the above-mentioned serious bottleneck problem is not seen yet, and mainly includes the following points:
1. the existing ventilator monitoring and protecting system only achieves the functions of field data acquisition and communication, maximum early warning value setting depending on experience and simple diagnosis, lacks of refined and customized ventilator signal analysis and feature extraction technology and is supported based on massive multi-physical signal data and artificial intelligence technology, and cannot achieve ventilator part fault positioning and ventilator complex failure mode discrimination.
2. In the field operation and maintenance work of the ventilator, equipment failure often results from the failure of each part due to abrasion and corrosion. Therefore, monitoring the working state of each key part of the ventilator is crucial to early finding and early diagnosing the early failure of the ventilator, and the prior art cannot realize the point, namely the health pre-diagnosis, trend prediction and service life prediction of the ventilator based on the early failure parts cannot be realized.
3. Two technologies of trend prediction and service life prediction are not available in the existing ventilator monitoring and protecting system, so that the existing technology cannot guide a user to carry out spare part management on key parts of the ventilator. Specifically, users can not know the prediction result of the running trend and the service life of the ventilator, the arrangement of maintenance work according to situations can not be realized, and the maintenance plan of periodic disassembly inspection and part replacement is still made on site by depending on experience.
4. At present, a fine blower pre-diagnosis and precise operation and maintenance system relating to three technologies of fine blower complex failure mode pre-diagnosis, running trend and service life prediction of early failure driving of parts and key part spare part management based on prediction results is not available.
Disclosure of Invention
In order to solve the technical problems, the invention provides a fine pre-diagnosis and precise operation and maintenance device and system for a ventilator, which can be used for calculating a plurality of characteristic parameters of a signal time domain and a signal frequency domain through continuous acquisition of multiple physical field signals such as vibration, temperature, pressure, flow and the like in the normal operation process of the ventilator, combining a big data analysis mining technology and an artificial intelligence technology, obtaining the real-time working state of key parts on line and carrying out fine pre-diagnosis on early failure of the key parts. The method is characterized in that big data analysis is carried out on massive multi-physical field signals, the health degree of the ventilator is predicted and scored by means of an artificial intelligence technology, and the service life prediction of parts is carried out in a self-adaptive mode on the basis of a refined prediction result of early failure of key parts. And (4) utilizing the health degree prediction result of the ventilator and the service life prediction result of the key parts to make a maintenance plan and manage spare parts.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a ventilator fine pre-diagnosis and accurate operation and maintenance device comprises a state monitoring unit, a signal analysis unit, a pre-warning diagnosis unit and an accurate operation and maintenance unit, wherein,
the state monitoring unit is used for acquiring the economic index, the performance index and the safety index of the ventilator, displaying the change trend of the economic index, the performance index and the safety index related to signals in real time, calculating and displaying the health degree index according to the performance index and the safety index, and calculating and displaying the total efficiency index value according to the economic index and the performance index;
the signal analysis unit is used for dynamically calculating and displaying characteristic values and change curves of all measuring points contained in each physical signal in the economic index, the performance index and the safety index, predicting the change trend of each physical signal and dynamically calculating each signal alarm threshold;
the early warning diagnosis unit is used for giving out a fault diagnosis conclusion of the ventilator in parallel by utilizing a fine early warning diagnosis method and a big data artificial intelligence early warning diagnosis method according to the calculation result of the signal analysis unit, confirming the fault of the ventilator if the diagnosis conclusion is the same, and otherwise sending out prompt information needing artificial diagnosis;
and the accurate operation and maintenance unit is used for predicting the service life of key parts in the ventilator and making a ventilator fine pre-diagnosis report according to the prediction result and the statistical result of the point inspection/maintenance record.
Preferably, the points of economic indicators include current, voltage and electrical power, the points of performance indicators include speed, inlet pressure, outlet pressure and differential pressure, and the points of safety indicators include vibration, temperature, oil level and flow, wherein the vibration includes driveshaft motor end bearing vibration, driveshaft impeller end bearing vibration, motor load end bearing vibration and motor fan end bearing vibration, and the temperature includes driveshaft motor end bearing temperature and driveshaft impeller end bearing temperature.
Preferably, the signal analysis unit comprises a signal preprocessing module, a time domain feature extraction module and a frequency domain feature extraction module, wherein,
the signal preprocessing module is used for performing communication interruption data point linkage and singular value removal of economic indexes, performance indexes and safety indexes related to signal data, and sending the processed signal data to the signal time domain analysis module and the signal frequency domain analysis module;
the time domain feature extraction module is used for extracting time domain features, wherein the time domain features comprise root mean square values, skewness factors, kurtosis factors and peak value factors;
the frequency domain characteristic extraction module is used for extracting frequency domain characteristics, the frequency domain characteristics comprise characteristic parameters of a power spectrum and an envelope spectrum, and the characteristic parameters of the power spectrum comprise center of gravity frequency, mean square frequency, root mean square frequency, frequency variance, frequency standard deviation and 1 st-5 th frequency band relative energy; the characteristic parameters of the envelope spectrum comprise the passing frequency of a bearing inner ring, the passing frequency of a bearing outer ring, the passing frequency of a bearing rolling body and the characteristic frequency of a bearing retainer.
Preferably, the early warning diagnosis unit comprises a fine early warning diagnosis module, a big data artificial intelligence early warning diagnosis module and a judgment module, wherein,
the fine early warning diagnosis module comprises a first sub-module, a second sub-module and a conclusion module, wherein,
the first submodule is used for carrying out time domain detection based on the dynamic time domain alarm threshold value and sending a detection result to the conclusion module;
the second submodule is used for carrying out real-time frequency domain feature detection according to the Lauda rule and sending a detection result to the conclusion module;
the conclusion module is used for respectively receiving the detection results of the first sub-module or the second sub-module, judging whether fault characteristics exist in the measuring point positions of the corresponding signals or not according to the detection results of the first sub-module or the second sub-module, and giving out a ventilator fault diagnosis conclusion;
the big data artificial intelligence early warning diagnosis module is used for performing data characteristic fusion on historical data of various physical signals by using a convolutional neural network to establish a state classification model, directly inputting all on-line signal data into the state classification model in an original signal form in real time to obtain a classification label to which the signal belongs, and further giving the working state of the ventilator, if the state is judged to be normal, returning to the state monitoring unit, and otherwise, giving a corresponding ventilator fault diagnosis conclusion;
and the judgment module is used for acquiring the diagnosis conclusion of the refined early warning diagnosis module and the big data artificial intelligence early warning diagnosis module, determining the fault of the ventilator if the diagnosis conclusion is the same, and prompting the intervention of artificial diagnosis if the diagnosis conclusion is not the same.
Preferably, the dynamic time domain alarm threshold is obtained by inputting historical data of various physical signals related to economic indexes, performance indexes and safety indexes into a convolutional neural network and a hidden markov model respectively for data fusion and state transition model establishment, establishing a trend prediction model according to fusion characteristic information and the state transition model, calculating the signal trend in real time through the trend prediction model, and calculating according to the Lauda rule.
Preferably, in the state transition model, a Softmax function is used to calculate the membership degree, and the expression is as follows:
in the formula, x (i) Is a firsti signal data packets, U s For the weight vectors connected to the s-th neuron in the convolutional neural network classification layer, b s Is the bias of the neuron or neurons,
is from x (i) The feature vector h (x) extracted in (i) ) Is x (i) Membership to tag y (i) The probability of (a) of (b) being,
the state transition matrix A in the state transition model has the following calculation formula:
A＝{a ij }＝P[HS(t+1)＝HS j |HS(t)＝HS i ]，
where i and j are the ith and jth hidden states in the hidden Markov model, respectively, a ij HS (t) represents the hidden state at time t, which is the probability of the transition from the ith hidden state to the jth hidden state.
Preferably, the precision operation and maintenance unit is used for inputting historical data of various physical signals related to economic indexes, performance indexes and safety indexes into a convolutional neural network and a hidden markov model respectively for data fusion and state transition model establishment, then establishing a part life prediction model by combining a part damage failure mechanism, inputting time domain characteristics and frequency domain characteristics of the current moment into the part life prediction model, predicting the residual service life of the part, and forming a fine ventilator pre-diagnosis report according to the predicted residual service life of the part and point inspection/maintenance records.
Preferably, the spot check/maintenance record includes maintenance schedule time, completion time, maintenance duration, equipment code, fault symptom and treatment mode.
Preferably, the precision operation and maintenance unit is further configured to formulate a spare part rule of the corresponding part according to the predicted remaining service life of the part, so as to form a spare part management scheme of the spare part.
A fine pre-diagnosis and accurate operation and maintenance system for a ventilator comprises: the intelligent ventilator monitoring system comprises a data acquisition unit, a cloud database, a client computer and a monitoring center platform, wherein the data acquisition unit, the client computer and the monitoring center platform are all connected with the cloud database, and the client computer is provided with the ventilator fine pre-diagnosis and accurate operation and maintenance device.
Based on the technical scheme, the invention has the beneficial effects that: the method comprises the steps of utilizing a convolutional neural network and a hidden Markov model to mine and reflect characteristic information of each physical field signal of the ventilator from massive historical data, further establishing a machine learning model, predicting the time domain variation trend of each physical field signal, dynamically setting an overrun alarm threshold according to the trend prediction result, and realizing the overrun alarm function based on self-adaptive threshold setting of each index of the ventilator; once the ventilator safety index is out of limit for alarming, early warning diagnosis is synchronously carried out by using two modes of fine early warning diagnosis and big data artificial intelligence diagnosis. The refined early warning diagnosis comprises two diagnosis methods of time domain overrun alarm triggering fault spectrum detection and fault spectrum direct detection. And the big data artificial intelligence early warning diagnosis directly judges the ventilator fault category reflected by the current original signal according to the established ventilator fault intelligent diagnosis model. The three diagnostic methods are synchronously judged in the background. And the time domain overrun alarm triggers two methods of fault frequency spectrum detection and fault frequency spectrum direct detection, and which method firstly determines the fault, the result gives a diagnosis conclusion of the ventilator. And then, comparing the data with a conclusion given by a big data artificial intelligence early warning diagnosis method for synchronously judging the fault types in the background. If the two conclusions are the same, confirming to diagnose the fault of the ventilator, otherwise, recommending manual diagnosis intervention; and storing the maintenance time and the equipment information, together with the intelligent diagnosis conclusion and the manual diagnosis record, into a background database. And (4) combining the prediction result of the residual service life of the key parts of the ventilator, developing a fine diagnosis report of the ventilator, and making a regular spare part management scheme of the key parts so as to guide the accurate operation and maintenance of ventilator equipment.
Drawings
FIG. 1 is a block diagram of a ventilator fine pre-diagnosis and precision operation and maintenance system in one embodiment;
FIG. 2 is an algorithmic tree of a signal analysis unit in one embodiment;
FIG. 3 is a block diagram of an early warning diagnosis unit in one embodiment;
FIG. 4 is a block diagram of a refined early warning module in the early warning diagnosis unit in one embodiment;
FIG. 5 is a flow diagram of a time domain over-limit alarm method in a first sub-module in one embodiment;
FIG. 6 is a flow diagram of a method for early warning diagnosis in an early warning diagnosis unit in one embodiment;
fig. 7 is a flowchart of a method for precision operation and maintenance in a precision operation and maintenance unit according to an embodiment.
In the figures, the various reference numbers are:
1. a data acquisition unit; 2. a cloud database; 3. a client computer; 31. a state monitoring unit; 32. a signal analyzing unit; 321. a signal preprocessing module; 322. a time domain feature extraction module; 323. a frequency domain feature extraction module; 33. an early warning diagnosis unit; 331. a refined early warning diagnosis module; 3311. a first sub-module, 3312, a second sub-module; 3313. a conclusion module; 332. a big data artificial intelligence early warning diagnosis module; 333. a judgment module; 34. an accurate operation and maintenance unit; 4. and monitoring the central platform.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention.
Referring to fig. 1, fig. 1 is a block diagram of a ventilator fine pre-diagnosis and precise operation and maintenance system. The system collects signals related to three types of indexes of the ventilator through a data collection module (a sensor and an instrument), wherein the three types of indexes are economic indexes, performance indexes and safety indexes, the economic indexes comprise current, voltage and electric power, the performance indexes comprise rotating speed, inlet pressure, outlet pressure and pressure difference, the safety indexes comprise vibration, temperature, oil level and flow, the vibration comprises transmission shaft motor end bearing vibration, transmission shaft impeller end bearing vibration, motor load end bearing vibration and motor fan end bearing vibration, and the temperature comprises transmission shaft motor end bearing temperature and transmission shaft impeller end bearing temperature. And after the signals are collected, intermittently sending data packets to the cloud database 2 through the gateway and storing the data packets. The monitoring center platform 4 and the client computer 3 can both read data from the cloud database 2.
The present embodiment provides a fine pre-diagnosis and precise operation and maintenance device for a ventilator, which comprises four modules, namely a state monitoring unit 31, a signal analysis unit 32, an early warning diagnosis unit 33 and a precise operation and maintenance unit 34, all of which are installed on a client computer 3, so as to realize fine diagnosis and precise operation and maintenance of multi-mode early faults of the ventilator and its parts. The system can reduce the fault occurrence rate of the ventilator and the production line outage rate caused by ventilator faults, improve the overall operation and maintenance level of ventilator users, save the operation cost for enterprises, play an important demonstration role in improving the intelligence level, and have important social and economic values. A fine pre-diagnosis and accurate operation and maintenance device of a ventilator is specifically described as follows:
the state monitoring unit 31 is used for acquiring the economic index, the performance index and the safety index of the ventilator, displaying the change trend of the economic index, the performance index and the safety index related to signals in real time, calculating and displaying the health degree index according to the performance index and the safety index, and calculating and displaying the total efficiency index value according to the economic index and the performance index;
the signal analysis unit 32 is used for dynamically calculating and displaying characteristic values and change curves of all measuring points contained in each physical signal in the economic index, the performance index and the safety index, predicting the change trend of each physical signal, and dynamically calculating each signal alarm threshold;
the early warning diagnosis unit 33 is configured to provide a ventilator fault diagnosis conclusion in parallel by using a fine early warning diagnosis method and a big data artificial intelligence early warning diagnosis method according to the calculation result of the signal analysis unit 32, determine a ventilator fault if the diagnosis conclusion is the same, and prompt an artificial diagnosis intervention if the diagnosis conclusion is not the same;
and the accurate operation and maintenance unit 34 is used for predicting the service life of key parts in the ventilator and making a ventilator fine pre-diagnosis report according to the prediction result and the statistical result of the point inspection/maintenance record.
Referring to fig. 2, fig. 2 is a tree diagram of the algorithm of the signal analysis unit 32 of the ventilator fine pre-diagnosis and precision operation and maintenance device implemented in the present invention. The signal analysis unit 32 includes a signal preprocessing module 321, a time domain feature extraction module 322, and a frequency domain feature extraction module 323, wherein,
the signal preprocessing module 321 is configured to perform communication interruption data point linkage and singular value removal on the economic indicator, the performance indicator and the safety indicator related to the signal data, and send the processed signal data to the signal time domain analysis module and the signal frequency domain analysis module;
the time domain feature extraction module 322 is configured to extract time domain features, where the time domain features include a root mean square value, a skewness factor, a kurtosis factor, and a peak factor, and the calculation formulas are respectively:
wherein x (N) is the nth point value of the N point signal,
the root mean square value of the N point signal is obtained;
in the formula (I), the compound is shown in the specification,
the skewness of the N point signal is obtained, and S is a skewness factor;
in the formula (I), the compound is shown in the specification,
is the kurtosis of the N-point signal, K V Is a kurtosis factor;
in the formula (I), the compound is shown in the specification,
the maximum value of the N-point signal and C the crest factor.
The frequency domain feature extraction module 323 is configured to extract frequency domain features, where the frequency domain features include feature parameters of a power spectrum and an envelope spectrum, where the power spectrum S (f) 27 includes six feature extraction algorithms of a barycentric frequency, a mean square frequency, a root-mean-square frequency, a frequency variance, a frequency standard deviation, and relative energies of 1 st to 5 th frequency bands, and the calculation formulas are respectively:
center of gravity frequency:
mean square frequency:
root mean square frequency:
frequency variance:
standard deviation of frequency:
relative energy of 1 st to 5 th bands:
wherein B is the frequency conversion order, B =5,E b Is the energy of the b-th order band,
is the b th orderBand relative energy.
The envelope spectrum comprises four characteristic extraction algorithms of bearing inner ring passing frequency, bearing outer ring passing frequency, bearing rolling body passing frequency and bearing retainer characteristic frequency, and the calculation formulas are respectively as follows:
inner ring passing frequency:
outer ring passing frequency:
rolling element passing frequency:
cage characteristic frequency:
wherein Z is the number of rolling elements, f i And f o The rotational frequency of the inner ring and the outer ring, D and D are divided into the diameter of the rolling body and the diameter of the pitch circle,
is the contact angle.
As shown in fig. 3 to 6, in an embodiment of the ventilator fine pre-diagnosis and fine operation and maintenance apparatus, the early warning diagnosis unit 33 includes a fine early warning diagnosis module 331, a big data artificial intelligence early warning diagnosis module 332, and a judgment module 333, wherein,
the fine early warning diagnostic module 331 includes a first sub-module 3311, a second sub-module 3312, and a conclusion module 3313, wherein,
referring to fig. 5, fig. 5 is a flowchart of a time domain over-limit alarm method in the first sub-module 3311. The first sub-module 3311 is configured to input historical data of various physical signals related to economic indicators, performance indicators, and safety indicators into a convolutional neural network and a hidden markov model respectively for data fusion and state transition model establishment, establish a trend prediction model according to fusion characteristic information and the state transition model, calculate a signal trend in real time through the trend prediction model, and then calculate and obtain a dynamic time-domain alarm threshold according to the rale, which is configured to perform numerical comparison between the time-domain alarm threshold and a data packet corresponding to a current signal, if the current data value is greater than the alarm threshold, enter a time-domain overrun alarm, otherwise enter a judgment of a next signal data packet and send a detection result to the conclusion module 3313, where the time-domain overrun alarm threshold is set according to the calculation formula of the rale:
σ th ＝3σ n ，
in the formula, σ n Is the mean, σ, of ten packets of the time-domain waveform of the signal th Is a time domain alarm threshold;
the second sub-module 3312 is configured to perform real-time frequency domain feature detection according to the lai-ta rule and send a detection result to the conclusion module 3313, specifically, whether there is a frequency component of a vibration feature frequency of a component with an outstanding energy is found, and the condition that the fault frequency spectrum detection of the second sub-module 3312 is triggered is that, according to the lai-ta rule, if a fault feature frequency with an outstanding energy is detected first, a ventilator fault diagnosis conclusion corresponding to the fault feature frequency is given. If no fault signature frequency with a prominent energy is detected, the return to the condition monitoring unit 31 continues the ventilator condition monitoring.
The conclusion module 3313 receives the detection results of the first sub-module 3311 or the second sub-module 3312, respectively, obeys the logic of early warning who starts diagnosis, and determines whether a fault characteristic exists at the position of a measurement point corresponding to a signal according to the detection result given first in the first sub-module 3311 or the second sub-module 3312, and gives a ventilator fault diagnosis conclusion;
the big data artificial intelligence early warning diagnosis module 332 is used for performing data characteristic fusion on historical data of various physical signals by using a convolutional neural network to establish a state classification model, directly inputting all online signal data into the state classification model in an original signal form in real time to obtain a classification label to which the signal belongs, further giving the working state of the ventilator, returning to the state monitoring unit 31 if the state is judged to be normal, otherwise giving a corresponding ventilator fault diagnosis conclusion, wherein a Softmax function is adopted in the state transfer model to calculate membership, and an expression is as follows:
in the formula, x (i) For the ith signal data packet, U s For the weight vectors connected to the s-th neuron in the convolutional neural network classification layer, b s Is the bias of the neuron or neurons,
is from x (i) The feature vector h (x) extracted in (1) (i) ) Is x (i) Membership to label y (i) The probability of (a) of (b) being,
the judgment module 333 is configured to obtain diagnosis results of the fine early warning diagnosis module 331 and the big data artificial intelligence early warning diagnosis module 332, confirm diagnosis of the ventilator fault if the diagnosis results are the same, and prompt artificial diagnosis intervention if the diagnosis results are not the same.
Referring to fig. 7, fig. 7 is a flowchart illustrating an accurate operation and maintenance method in the accurate operation and maintenance unit 34 according to the present invention. The accurate operation and maintenance unit 34 is configured to input historical data of various physical signals related to economic indicators, performance indicators, and safety indicators into the convolutional neural network respectively to perform feature fusion on the data. The algorithm and the expression of the classification label to which the signal belongs is determined by the convolutional neural network, which are the same as the algorithm and the expression of the classification label to which the signal belongs determined by the convolutional neural network in the big data artificial intelligence early warning diagnosis method, and are not described again.
And (5) establishing a state transition model by using a hidden Markov model while fusing the data characteristics. The core of the state transition model is to calculate a state transition matrix A, and the calculation formula is as follows:
A＝{a ij }＝P[HS(t+1)＝HS j |HS(t)＝HS i ]，
where i and j are the ith and jth hidden states, respectively, in the hidden Markov model 40, a ij Is as followsThe probability of the transition from the i hidden state to the j hidden state, and HS (t) represents the hidden state at time t.
And (4) establishing a service life prediction model of the part by combining a damage failure mechanism of the part. Reading the current data packet of the physical signal corresponding to each measuring point on the part, performing time domain analysis and frequency domain analysis on the signal, and calculating to obtain a related characteristic value. And inputting a life prediction model to predict the residual service life of the parts. And (4) reading the point inspection/maintenance records (maintenance planning time, completion time, maintenance duration, equipment codes, fault symptoms and processing modes) from the cloud database 2 to form a fine pre-diagnosis report of the ventilator. Meanwhile, in combination with the remaining service life, a spare part rule of the corresponding component is formulated in the accurate operation and maintenance unit 34, and finally a spare part management scheme of the spare part is formed.
The above description is only a preferred embodiment of the fine pre-diagnosis and precise operation and maintenance device and system for the ventilator disclosed in the present invention, and is not intended to limit the scope of the embodiments of the present disclosure. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the embodiments of the present disclosure should be included in the protection scope of the embodiments of the present disclosure.
Claims (10)
1. A ventilator fine pre-diagnosis and accurate operation and maintenance device is characterized by comprising a state monitoring unit, a signal analysis unit, an early warning diagnosis unit and an accurate operation and maintenance unit, wherein,
the state monitoring unit is used for acquiring signals related to economic indexes, performance indexes and safety indexes of the ventilator and displaying the change trends of the economic indexes, the performance indexes and the safety indexes related to the signals in real time, wherein the economic indexes comprise current, voltage and electric power, the performance indexes comprise rotating speed, inlet pressure, outlet pressure and pressure difference, the safety indexes comprise vibration, temperature, oil level and flow, the vibration comprises transmission shaft motor end bearing vibration, transmission shaft impeller end bearing vibration, motor load end bearing vibration and motor fan end bearing vibration, the temperature comprises transmission shaft motor end bearing temperature and transmission shaft impeller end bearing temperature, the health degree indexes are calculated and displayed according to the performance indexes and the safety indexes, and the total efficiency index value is calculated and displayed according to the economic indexes and the performance indexes;
the signal analysis unit is used for dynamically calculating and displaying characteristic values and change curves of all measuring points contained in each physical signal in the economic index, the performance index and the safety index, predicting the change trend of each physical signal and dynamically calculating each signal alarm threshold;
the early warning diagnosis unit is used for giving out a fault diagnosis conclusion of the ventilator in parallel by utilizing a fine early warning diagnosis method and a big data artificial intelligence early warning diagnosis method according to the calculation result of the signal analysis unit, confirming the fault of the ventilator if the diagnosis conclusion is the same, and otherwise sending out prompt information needing artificial diagnosis;
and the accurate operation and maintenance unit is used for predicting the service life of key parts in the ventilator and making a ventilator fine pre-diagnosis report according to the prediction result and the statistical result of the point inspection/maintenance record.
2. A ventilator fine pre-diagnosis and precision operation and maintenance device according to claim 1, wherein said economic indicator points comprise current, voltage and electrical power, said performance indicator points comprise rotational speed, inlet pressure, outlet pressure and differential pressure, and said safety indicator points comprise vibration, temperature, oil level and flow, wherein said vibration comprises propeller shaft motor end bearing vibration, propeller shaft end bearing vibration, motor load end bearing vibration and motor fan end bearing vibration, and said temperature comprises propeller shaft motor end bearing temperature and propeller shaft end bearing temperature.
3. A ventilator fine pre-diagnosis and precise operation and maintenance device according to claim 1, wherein the signal analysis unit comprises a signal pre-processing module, a time domain feature extraction module and a frequency domain feature extraction module, wherein,
the signal preprocessing module is used for carrying out communication interruption data point connection and singular value removal of economic indexes, performance indexes and safety indexes related to signal data and sending the processed signal data to the signal time domain analysis module and the signal frequency domain analysis module;
the time domain feature extraction module is used for extracting time domain features, wherein the time domain features comprise root mean square values, skewness factors, kurtosis factors and peak value factors;
the frequency domain characteristic extraction module is used for extracting frequency domain characteristics, the frequency domain characteristics comprise characteristic parameters of a power spectrum and an envelope spectrum, and the characteristic parameters of the power spectrum comprise a center of gravity frequency, a mean square frequency, a root mean square frequency, a frequency variance, a frequency standard deviation and 1 st-5 th frequency band relative energy; the characteristic parameters of the envelope spectrum comprise the passing frequency of a bearing inner ring, the passing frequency of a bearing outer ring, the passing frequency of a bearing rolling body and the characteristic frequency of a bearing retainer.
4. The fine pre-diagnosis and accurate operation and maintenance device of the ventilator according to claim 3, wherein the early warning diagnosis unit comprises a fine early warning diagnosis module, a big data artificial intelligence early warning diagnosis module and a judgment module,
the fine early warning diagnosis module comprises a first sub-module, a second sub-module and a conclusion module, wherein,
the first submodule is used for carrying out time domain detection based on the dynamic time domain alarm threshold value and sending a detection result to the conclusion module;
the second submodule is used for carrying out real-time frequency domain feature detection according to the Lauda rule and sending a detection result to the conclusion module;
the conclusion module is used for respectively receiving the detection results of the first sub-module or the second sub-module, judging whether fault characteristics exist in the measuring point positions of the corresponding signals or not according to the detection results of the first sub-module or the second sub-module, and giving out a ventilator fault diagnosis conclusion;
the big data artificial intelligence early warning diagnosis module is used for performing data characteristic fusion on historical data of various physical signals by using a convolutional neural network to establish a state classification model, directly inputting all on-line signal data into the state classification model in an original signal form in real time to obtain a classification label to which the signal belongs, and further giving the working state of the ventilator, if the state is judged to be normal, returning to the state monitoring unit, and otherwise, giving a corresponding ventilator fault diagnosis conclusion;
and the judgment module is used for acquiring the diagnosis conclusion of the refined early warning diagnosis module and the big data artificial intelligence early warning diagnosis module, determining the fault of the ventilator if the diagnosis conclusion is the same, and prompting the intervention of artificial diagnosis if the diagnosis conclusion is not the same.
5. The fine pre-diagnosis and precise operation and maintenance device of the ventilator according to claim 4, wherein the dynamic time domain alarm threshold is obtained by inputting historical data of various physical signals related to economic indexes, performance indexes and safety indexes into a convolutional neural network and a hidden Markov model respectively for data fusion and state transition model establishment, establishing a trend prediction model according to fusion characteristic information and the state transition model, calculating signal trends in real time through the trend prediction model, and calculating according to the Lauda rule.
6. The fine pre-diagnosis and accurate operation and maintenance device for the ventilator according to claim 5, wherein a Softmax function is adopted in the state transition model to calculate the membership degree, and the expression is as follows:
in the formula, x (i) For the ith signal packet, U s For the weight vectors connected to the s-th neuron in the convolutional neural network classification layer, b s Is the bias of the neuron or neurons,
is from x (i) The feature vector h (x) extracted in (1) (i) ) Is x (i) Membership to label y (i) The probability of (a) of (b) being,
the state transition matrix A in the state transition model has the following calculation formula:
A＝{a ij }＝P[HS(t+1)＝HS j |HS(t)＝HS i ]，
wherein i and j are respectively the ith and jth hidden states in the hidden Markov model, a ij HS (t) represents the hidden state at time t, which is the probability of the transition from the ith hidden state to the jth hidden state.
7. The fine pre-diagnosis and precise operation and maintenance device of the ventilator according to claim 1, wherein the precise operation and maintenance unit is configured to input historical data of various physical signals related to economic indicators, performance indicators and safety indicators into a convolutional neural network and a hidden markov model respectively for data fusion and state transition model establishment, and then, in combination with a component damage failure mechanism, establish a component life prediction model, and configured to input time domain characteristics and frequency domain characteristics of a current time into the component life prediction model, predict remaining service life of the component, and form a fine pre-diagnosis report of the ventilator according to the predicted remaining service life of the component and a spot inspection/maintenance record.
8. The fine pre-diagnosis and accurate operation and maintenance device of the ventilator according to claim 7, wherein the point inspection/maintenance records comprise maintenance planning time, completion time, maintenance duration, equipment codes, fault symptoms and treatment modes.
9. The fine pre-diagnosis and precise operation and maintenance device for the ventilator according to claim 7, wherein the precise operation and maintenance unit is further configured to make spare part rules of corresponding parts according to the predicted remaining service lives of the parts, so as to form a spare part management scheme of the spare parts.
10. The utility model provides a fine prediagnosis of ventilation blower and accurate fortune dimension system which characterized in that includes: the ventilator precise pre-diagnosis and precise operation and maintenance device comprises a data acquisition unit, a cloud database, a client computer and a monitoring center platform, wherein the data acquisition unit, the client computer and the monitoring center platform are all connected with the cloud database, and the client computer is provided with the ventilator precise pre-diagnosis and precise operation and maintenance device according to claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210351665.1A CN114893428B (en) | 2022-04-02 | 2022-04-02 | Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210351665.1A CN114893428B (en) | 2022-04-02 | 2022-04-02 | Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114893428A CN114893428A (en) | 2022-08-12 |
| CN114893428B true CN114893428B (en) | 2023-02-21 |
Family
ID=82716062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210351665.1A Active CN114893428B (en) | 2022-04-02 | 2022-04-02 | Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114893428B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105003453A (en) * | 2014-07-18 | 2015-10-28 | 河北联合大学 | Online monitoring and fault diagnosis system of mine fan |
| CN109506921A (en) * | 2018-12-24 | 2019-03-22 | 西安科技大学 | A kind of rotary machinery fault diagnosis and method for early warning |
| CN112343810A (en) * | 2020-11-04 | 2021-02-09 | 广州高澜节能技术股份有限公司 | Water pump health monitoring and diagnosing method for circulating water cooling system |
| CN113155469A (en) * | 2021-03-19 | 2021-07-23 | 黑龙江机智通智能科技有限公司 | Engine fault diagnosis alarm system and device |
| CN114201989A (en) * | 2021-11-29 | 2022-03-18 | 江苏大学 | Alternating current motor bearing fault diagnosis method adopting convolutional neural network and bidirectional long-time and short-time memory network |
-
2022
- 2022-04-02 CN CN202210351665.1A patent/CN114893428B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105003453A (en) * | 2014-07-18 | 2015-10-28 | 河北联合大学 | Online monitoring and fault diagnosis system of mine fan |
| CN109506921A (en) * | 2018-12-24 | 2019-03-22 | 西安科技大学 | A kind of rotary machinery fault diagnosis and method for early warning |
| CN112343810A (en) * | 2020-11-04 | 2021-02-09 | 广州高澜节能技术股份有限公司 | Water pump health monitoring and diagnosing method for circulating water cooling system |
| CN113155469A (en) * | 2021-03-19 | 2021-07-23 | 黑龙江机智通智能科技有限公司 | Engine fault diagnosis alarm system and device |
| CN114201989A (en) * | 2021-11-29 | 2022-03-18 | 江苏大学 | Alternating current motor bearing fault diagnosis method adopting convolutional neural network and bidirectional long-time and short-time memory network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114893428A (en) | 2022-08-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108460144B (en) | Coal equipment fault early warning system and method based on machine learning | |
| CN108681747B (en) | Rotary machine fault diagnosis and state monitoring system and method based on deep learning | |
| CN109543210A (en) | A kind of Wind turbines failure prediction system based on machine learning algorithm platform | |
| CN113469060A (en) | Multi-sensor fusion convolution neural network aeroengine bearing fault diagnosis method | |
| CN110529202B (en) | Steam turbine fault diagnosis detection and early warning method and system based on power plant data | |
| CN111651505B (en) | Equipment operation situation analysis and early warning method and system based on data driving | |
| EP4063875A1 (en) | Multi-information fusion-based fault early warning method and device for converter | |
| JPH08202444A (en) | Method and device for diagnosing abnormality of machine facility | |
| KR102427205B1 (en) | Apparatus and method for generating training data of artificial intelligence model | |
| CN113962299A (en) | Intelligent operation monitoring and fault diagnosis general model for nuclear power equipment | |
| CN113177646B (en) | Power distribution equipment online monitoring method and system based on self-adaptive edge proxy | |
| CN111273125A (en) | RST-CNN-based power cable channel fault diagnosis method | |
| CN115099260A (en) | Online monitoring mechanical fault real-time diagnosis method for double-screw oil transfer pump | |
| CN113007040B (en) | Online monitoring and fault early warning system and method for main shaft assembly movement of wind turbine generator | |
| CN114893428B (en) | Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator | |
| CN115524002B (en) | Operation state early warning method, system and storage medium of power plant rotating equipment | |
| CN112580858A (en) | Equipment parameter prediction analysis method and system | |
| Zhang et al. | Fault Diagnosis with Bidirectional Guided Convolutional Neural Networks Under Noisy Labels | |
| CN111563685A (en) | Power generation equipment state early warning method based on auto-associative kernel regression algorithm | |
| CN114004059B (en) | Health portrait method for hydroelectric generating set | |
| CN115774955A (en) | Intelligent health degree assessment method for wind power gear box | |
| CN110108457A (en) | A kind of First air machine shaft temperature diagnostic method based on gravitation neural network | |
| CN114091811A (en) | Maintenance decision system for circulating water pump of nuclear power plant and design method | |
| CN115456041A (en) | Equipment fault early warning method and device, computing equipment and storage medium | |
| CN111061191B (en) | Distributed oil-gas storage tank remote operation and maintenance method |
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 |