CN113253705B - Fault diagnosis method and device for air blower - Google Patents
Fault diagnosis method and device for air blower Download PDFInfo
- Publication number
- CN113253705B CN113253705B CN202110538008.3A CN202110538008A CN113253705B CN 113253705 B CN113253705 B CN 113253705B CN 202110538008 A CN202110538008 A CN 202110538008A CN 113253705 B CN113253705 B CN 113253705B
- Authority
- CN
- China
- Prior art keywords
- data
- blower
- sensor
- matrix
- statistic
- 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
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0221—Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Control Of Positive-Displacement Air Blowers (AREA)
Abstract
The invention relates to a fault diagnosis method and device for an air blower, which are used for carrying out real-time fault diagnosis on the air blower. The method comprises the following steps: the method comprises the steps of collecting blower operation data under normal conditions, processing the existing data by using a median average filtering method, forming a matrix, and constructing an initial model according to a recursive dynamic principal component analysis method to obtain a statistic control limit. And collecting real-time operation data of the air blower for monitoring, judging whether the air blower breaks down according to whether the obtained data statistic exceeds a statistic control limit, and judging the cause of the fault according to the contribution rate of each statistic. The problem that the blower system cannot be accurately diagnosed due to the operation interference noise of the blower and the large data volume of the traditional method is solved.
Description
Technical Field
The invention relates to the technical field of fault detection and diagnosis, in particular to a fault diagnosis method and a fault diagnosis device for an air blower.
Background
The blower operates eccentrically by means of offset rotor inside cylinder and makes the volume change between blades in the rotor slot to suck, compress and discharge air, and is used widely in sewage treatment, smelting blast furnace, coal washing plant, mine flotation, chemical gas making, vacuum and other fields. In the operation process of equipment, the blower fault can cause gas transmission errors, the blower burns out and other serious consequences, so that the blower needs good operation state and accurate fault positioning performance when the fault occurs. The existing fault diagnosis method needs to accurately know the relevant model of the blower and has high cost, so a feasible method needs to be provided to solve the problem.
The idea of principal component analysis is to divide the collected data into principal component subspace representing the principal components of the system and residual subspace representing the noise of the system, and the projection of the original data in the two subspaces can be used to judge the change of the data. The related documents are:
soil corrosivity evaluation method based on principal component analysis method (application publication number: CN 109115675A)
Equipment energy consumption ratio early warning analysis method based on principal component analysis (application publication number CN 110751217A)
Compared with the related documents, the invention considers the influence of the noise signal inevitably generated by the blower in the operation process, correspondingly filters the noise signal, improves the accuracy of the sampled data, requires a large amount of data for corresponding processing in the related documents, correspondingly simplifies the processed data by utilizing median filtering and a recursive algorithm, and simultaneously improves the accuracy of the result. The situation that the system is dynamic is not considered in the related documents, the blower is considered as a dynamic system, and the relation between the past data and the current data is established, so that the accuracy of the result is improved. In conclusion, the invention improves the rapidity in data processing and the accuracy of results.
Disclosure of Invention
The technical problem of the invention is solved: aiming at the problems that a large amount of noise is generated in the operation process of the existing air blower system, the data volume of information acquired by the traditional fault diagnosis method is large, and the like, the fault diagnosis method and the fault diagnosis device for the air blower are provided, the influence of the measured noise in the actual operation process is eliminated by filtering the acquired sensor data of the air blower, the generation of a false alarm phenomenon caused by high-frequency noise can be effectively avoided, and the problems that the air blower system cannot be accurately diagnosed due to the operation interference noise of the air blower and the large data volume of the traditional method are solved.
The technical solution of the invention is as follows: a fault diagnosis method of a blower, comprising the steps of:
the method comprises the following steps: the method comprises the steps of collecting different sensor data of the air blower under the normal operation condition, wherein the number of the different sensor data is 7, and the different sensor data are respectively used for collecting current data of a motor of the air blower by a current sensor, collecting voltage data of the motor of the air blower by a voltage sensor, collecting vibration data of a shell of the air blower by a vibration sensor and measuring a drum by a temperature sensorThe method comprises the following steps that temperature data of a fan motor part, rotating speed data of a blower motor measured by a speed sensor, blower gas pressure data measured by a pressure sensor and blower gas flow speed data measured by a flow sensor are obtained; the method comprises the steps of forming a matrix after median average filtering is carried out on collected data of different sensors, processing the matrix according to a recursive dynamic principal component analysis method, carrying out linear projection on an original data space, dividing the original data into a principal component subspace and a residual error subspace, wherein the two fault detection methods adopted are Hotelling's T2Statistical quantity and Squared Prediction Error (T) method2The change condition of new data in the principal component subspace is measured statistically, and the projection change of a sampling vector on a residual subspace is measured statistically by a Square Prediction Error (SPE); obtaining T under normal operation condition according to the matrix obtained after processing2And SPE statistic control limits;
step two: collecting the real-time operation data of the 7 blower sensors in the step one, forming a matrix, and calculating statistics corresponding to the real-time operation data of the blower sensors at the moment and T in the step one2Comparing with the SPE statistic control limit to judge whether the blower fails;
step three: calculating T obtained by utilizing real-time operation data of the blower sensor in the step two2And comparing the statistic and the SPE statistic with the statistic control limit under the normal operation condition in the step one, and if the T calculated by utilizing the real-time operation data of the blower sensor in the step two2If the statistic and the SPE statistic do not exceed the statistic control limit under the normal operation condition, indicating that the blower operates normally, and if the statistic and the SPE statistic exceed the statistic control limit under the normal operation condition, indicating that the blower fails;
step four: after the fault of the blower occurs, the ratio of the statistic of each sensor to the statistic of all sensors is calculated, namely the cumulative contribution rate of each sensor, and the fault of a certain sensor in the blower is judged according to the obtained ratio.
In the first step, the median average filtering is performed on the collected blower sensor data in the normal operation state to form a matrix, and the following concrete implementation is realized:
selecting different sensor data of the blower in the normal condition with the data length M, and forming a matrix X. Selecting a data length k far smaller than M, calculating the median of data in each data length to serve as a data source for average filtering processing, then selecting a data length M far smaller than k, carrying out averaging processing according to a formula (2), and obtaining a data matrix X 'after median average filtering, wherein X' belongs to R7*n,
In the formula:
xithe sample data is the sample data after median average filtering;
xueach of m data lengths;
m is the data length selected for carrying out the average filtering algorithm;
performing a recursive algorithm on the filtered data matrix, specifically implementing the following steps:
in the formula:
A=[1,…,1]T;
a1the average value of each row of data in the data matrix after median average filtering;
n is the number of data in each column;
x' is a data matrix after median average filtering;
t represents the transposition of the matrix;
the matrix is data normalized as:
in the formula:
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
calculating the normalized matrix according to a recursive algorithm to obtain a recursive matrix Xk+1Comprises the following steps:
in the formula:
A=[1,…,1]T;
a1the average value of each row of data in the data matrix after median average filtering;
n is the number of data in each column;
k is the matrix order in which the recursive algorithm is performed;
t represents the transpose of the matrix;
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
carrying out dynamic principal component algorithm calculation on the matrix subjected to recursive algorithm operation to obtain a matrix Q:
in the formula:
xi,jvalues representing all rows and columns in the matrix Q;
i and j are the row number and the column number of the matrix respectively;
q is a matrix after filtering and recursive dynamic algorithm operation;
collecting data of 7 sensors under normal conditions, obtaining a matrix after filtering and recursive dynamic algorithm operation according to the operation, and further obtaining a statistic control limit of the blower under the normal conditions of the 7 sensors:
in the formula:
k represents the sequential order of the feature vectors;
a represents 7 sensors selected in sequence;
n is a normal distribution intermediate variable;
θi,i=1,2,…、h0representing an intermediate variable of the SPE statistic obeying standard normal distribution;
Cα、Fαis the confidence of the normal distribution.
In the second step, a formula (9) is adopted, and when the SPE statistic or T is used2If any one of the statistics exceeds the control limit of the statistics, the blower is in failure
In the formula:
T2represents T2Statistics;
SPE represents SPE statistic;
In the fourth step: when a blower fault occurs, each sensor pair T is calculated at each moment2And the contribution of SPE statistics, as equation (10), from the T obtained for each sensor2Value l of statistic and statistic SPEiComparing the total contribution value L to calculate each sensor pair T2And the accumulated contribution rate H of the SPE statistic, and further judging that a certain sensor of the air blower breaks down:
the total contribution L is the sum of the statistics obtained by each of the 7 sensors.
The innovation point of the method is that aiming at the running data of the blower sensor under the complex working condition, a median average filtering method is used, and a median is selected from the existing data length to carry out filtering operation, so that non-median data are reduced, and data compression is carried out. In addition, the collected noise signals can be effectively filtered, and the accuracy in the subsequent calculation process is improved. The fault diagnosis method of the recursive dynamic principal component analysis firstly carries out recursive operation according to the obtained data information, carries out prediction operation on the data, considers the influence of the past time data on the current data for a dynamic system of the air blower, and integrally improves the rapidity and the accuracy of fault diagnosis.
The invention provides a fault diagnosis device of a blower, comprising: the system comprises a power supply management part, a power driving part, a blower body part, a sensor information acquisition part and a core control part; wherein:
the power management part comprises an alternating current-to-direct current module used for converting 220V alternating current into bus voltage required by the work of the blower, and a switching power supply is used for converting 48V into 3.3V, 5V and +/-15V through the direct current voltage conversion module and respectively supplying power to the power driving part, the sensor information acquisition part and the core control part;
the power driving part comprises a buffer circuit, a braking circuit, an IGBT unit, a driving circuit and an optical coupling isolation part;
when the power-driven part normally works, firstly, the power management part needs to supply power correctly, and a PWM signal generated by the core control part reaches the driving circuit and the IGBT unit after optical coupling isolation, so that the voltage of the blower under normal work is obtained, and the blower is in the power-off and speed-down process; PWM signals generated by the core control part reach the brake circuit part after passing through the optical coupling isolation and drive circuit for carrying out speed reduction operation of the blower, and the buffer circuit is used for preventing the IGBT unit from being burnt out by suddenly increased voltage and avoiding the damage of devices;
the blower body part comprises a blower and a load; the three-phase current and the three-phase voltage generated by the power driving part can drive the air blower to normally work, and loads such as wind abrasion, friction and the like can be generated in the operation process;
the sensor information acquisition part comprises 7 sensors including a voltage sensor, a current sensor, a vibration sensor, a temperature sensor, a speed sensor, a pressure sensor and a flow sensor; the 7 sensors can respectively acquire voltage data, current data, vibration data, temperature data, speed data, pressure data and flow data under the running state of the air blower, and after the data are acquired, the data are input into a signal conditioning circuit in the control part to be processed;
the core control part comprises an upper computer, a key control part, a display module, a signal conditioning circuit and a DSP; the DSP chip is a chip capable of realizing a digital signal processing technology, and the DSP chip also comprises input and output interfaces and signals such as ECAN, PWM, GPIO, SPI, AD, ECAP and the like; the signal conditioning circuit can process the sensor signals acquired by the sensor signal acquisition part, the sensor signals are converted into signal data which can be processed by a DSP chip through an AD (analog-to-digital) and ECAP (electro-magnetic resonance) input module, the DSP is an indispensable device for processing data and achieving an expected effect, the DSP processes the data acquired by the AD and ECAP to generate corresponding PWM (pulse-width modulation) signals, and the PWM signals are input to the power driving part to generate expected voltage so as to achieve the purpose of perfectly controlling the work of the blower; the host computer is the computer that directly sends control command, and ECAN output module is connected with the host computer in the DSP, reaches simple and easy convenient complicated operation of completion, and GPIO is general output interface, is connected with key control, can use the button to accomplish the relevant operation in the DSP, and SPI is fast-speed, full-duplex, synchronous communication bus, links to each other with display module through the SPI output, whether real-time detection air-blower breaks down and certain sensor breaks down.
The realization process of the invention is as follows:
(1) performing median average filtering on the collected blower sensor data in normal operation state
(2) And (4) determining a recursive algorithm and a dynamic time lag order v by using the filtered data.
(3) Constructing PCA mathematical model by using the processed data, and adopting square prediction error SPE statistic and T2The statistics detect the data.
(4) And judging the fault position of the blower sensor after the fault according to the obtained statistic contribution rate.
Compared with the prior art, the invention has the advantages that:
(1) the data-based median filtering algorithm provided by the invention can effectively reduce noise signals existing in the blower sensor, effectively improves the phenomenon of data misinformation caused by directly using system data in the traditional principal component analysis method, and improves the precision of fault diagnosis.
(2) Compared with the traditional principal component analysis method, the recursive dynamic principal component analysis method has the advantages that the calculation time can be effectively reduced, the data can be replaced according to the existing model, and the diagnosis accuracy is effectively improved according to the actual running state. The dynamic analysis can be used for effectively detecting and diagnosing the fault of the dynamic system of the air blower, and the fault diagnosis rate is further improved.
(3) Using SPE and T2The two statistics are used for fault diagnosis together, and the fault position can be judged quickly and accurately.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a diagram of a control device of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and examples.
(1) As shown in fig. 1, in the running process of the blower, the invention uses sensors to obtain data of the running of the blower in real time, and firstly, 7 sensors of the blower under the normal running condition are collected, namely, the current sensor collects the current data of the blower motor, the voltage sensor collects the voltage data of the blower motor, the vibration sensor collects the vibration data of the blower shell, the temperature sensor measures the temperature data of the blower motor part, the speed sensor measures the rotating speed data of the blower motor, the pressure sensor measures the air pressure data of the blower, and the flow sensor measures the air flow rate data of the blower; selecting different sensor data of the blower in the normal condition with the data length M, and forming a matrix X. Selecting a data length k far smaller than M, calculating the median of data in each data length to serve as a data source for average filtering processing, then selecting a data length M far smaller than k, and carrying out averaging processing according to a formula (2) to obtain a data matrix X ', X' belongs to R after median average filtering7*n,
In the formula:
xithe sample data after median average filtering;
xueach of m data lengths;
m is the data length selected for carrying out the average filtering algorithm;
(2)X'∈R7*nthe matrix is subjected to filtering processing, and the average value of each column of data in the data matrix after median average filtering is as follows:
in the formula:
A=[1,…,1]T;
a1the average value of each column of data in the data matrix after median average filtering;
n is the number of data in each column;
x' is a data matrix after median average filtering;
t represents the transposition of the matrix;
the matrix is normalized to data:
in the formula:
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
when the blower normally operates, data can be continuously input, the kth data length is set to be acquired, if k is 1, the kth data length corresponds to the 1 st data length, iterative calculation is started when k is larger than 1, and a sample of the next acquired data lengthCalculating a by using a recursive methodk+1、Xk+1;
For the (k +1) th data length data, the mean vector ak+1And akThe following relationships exist:
calculating the resulting recursive matrix Xk+1Comprises the following steps:
in the formula:
A=[1,…,1]T;
a1the average value of each row of data in the data matrix after median average filtering;
n is the number of data in each column;
k is the matrix order in which the recursive algorithm is performed;
t represents the transposition of the matrix;
miis the data length size;
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
(3) in a dynamic system of a blower, data at the current moment partially depends on data at the past moment, and the direct relationship between the data and the past moment needs to be mastered so as to judge the system state more accurately, so that an optimal time lag order v is introduced as a condition for determining the relationship between the data and the past moment, and the method specifically comprises the following steps:
1) setting a parameter u as one of the principal elements in the principal element subspace, respectively representing parameters capable of replacing the main information of the acquired data, enabling an initial time-lag order v to be 0, carrying out principal element analysis according to the data obtained after the recursive algorithm, and solving each principal element score;
2) if the u-th principal element can be linearly analyzed, subtracting 1 from the selected u-th principal element, judging the relationship between the next principal elements, and adding 1 to the number of linearly analyzable elements with time lag until the linear analysis can not be performed;
In the formula:
rnew(v) the number of linear analysis without time lag;
r (v) is the number of linear analyses possible in the presence of time lag;
if rnew(v) If the value is less than 0, stopping calculation, otherwise, adding one to the time lag order v on the existing basis, and returning to the step 2), thereby obtaining the time lag order v;
(4) after the steps (2) and (3), obtaining a matrix Q subjected to recursive dynamic analysis processing by using xi,jData representing all rows and columns in the matrix Q; normalizing the data in the matrix Q according to a formula (10) to obtain a normalized data matrix Y;
in the formula:
μ is the mean of all sample data;
σ is the standard deviation of all sample data;
xi,jthe values of all rows and columns in the matrix Q;
(5) solving a covariance matrix S of the data matrix Y by using a formula (11), and performing singular value decomposition on the obtained covariance matrix S according to a formula (12);
S=VΛVT (12)
wherein, Λ is epsilon to RI×IIs diagonal matrix, the main diagonal element values of A are reduced in turn, and the matrix P formed from correspondent characteristic vectors is defined as load matrix Pk;
For a sample vector x ∈ RnProjecting the load vector to obtain the score of x in each characteristic vector direction, wherein the vector t formed by the scores is Pk Tx;
(6) Determining the number k of the principal elements by utilizing a method of accumulating the variance contribution rate, as shown in formula (13):
(7) calculating statistic SPE and T according to the obtained principal component information2And statistic control limit Jth,SPE、
T2=xTPkΛ-1 kPk Tx (14)
SPE=||(I-PkPk T)x||2=xT(I-PkPk T)x (15)
In the formula:
Pka matrix composed of the first k eigenvectors;
Λ∈RI×Iis a diagonal matrix;
i is an identity matrix;
T2the statistics obey an F distribution;
Wherein k represents the sequential order of the feature vectors;
a represents 7 sensors selected in sequence;
θi、h0representing an intermediate variable of the SPE statistic obeying standard normal distribution;
then the standard normal distribution is followed,
thus, the following results were obtained:
in the formula:
k represents the sequential order of the feature vectors;
a represents 7 sensors selected in sequence;
n is a normal distribution intermediate variable;
θi,i=1,2,…、h0representing SPE systemIntermediate variables that obey a standard normal distribution are metered.
Cα、FαIs the confidence of normal distribution;
calculating SPE statistic and T under the real-time running state at the moment according to the formula (18) and the formula (19)2And (3) comparing the statistic obtained by calculation at the moment with the statistic control limit, and comparing the SPE statistic or the T statistic as shown in formula (20)2If any of the statistics exceeds the control limit of the statistics, it is an indication that the blower system is malfunctioning.
T2=xTPkΛ-1 kPk Tx (18)
SPE=||(I-PkPk T)x||2=xT(I-PkPk T)x (19)
In the formula:
T2represents T2Statistics;
SPE represents SPE statistic;
Λ∈RI×Iis a diagonal matrix;
Pkrepresenting a matrix formed by the first k eigenvectors in the eigenvectors;
x represents a new sample vector;
in the formula:
T2represents T2Statistics;
SPE represents SPE statistic;
(8) When a fault occurs, each variable pair T is calculated at each moment2And the contribution of SPE statistics, as equation (21), from the T obtained for each sensor2Statistic and value l of statistic SPEiComparing with the total contribution value L, and calculating each variable pair T2And the accumulated contribution rate H of the SPE statistic, and further judging that a certain sensor of the air blower breaks down:
the total contribution value L is the sum of statistics obtained by each sensor of 7 sensors;
as shown in fig. 2, a failure diagnosis apparatus of a blower of the present invention includes a power management part, a power driving part, a blower body part, a sensor information acquiring part, and a core control part. The power management part comprises an AC-DC conversion module for converting 220V AC into bus voltage required by the work of the blower. The 48V is converted into 3.3V, 5V and +/-15V by a switching power supply through a direct-current voltage conversion module and is respectively used for supplying power to the power driving part, the sensor information acquisition part and the core control part. The power driving part comprises a buffer circuit, a braking circuit, an IGBT unit, a driving circuit and an optical coupling isolation five part. The normal work of the power driving part firstly needs the correct power supply of the power management part, the PWM signal generated by the core control part reaches the driving circuit and the IGBT unit after being isolated by the optical coupler, so that the voltage of the blower in normal work is obtained, and in the process of power-off and speed reduction of the blower, the PWM signal generated by the core control part reaches the braking circuit part after being isolated by the optical coupler and the driving circuit, so that the speed reduction operation of the blower is carried out, the buffer circuit is used for preventing the IGBT unit from being burnt out by the suddenly increased voltage, and the damage of devices is avoided. The blower body part comprises a blower and a load. The three-phase current and the three-phase voltage generated by the power driving part can drive the air blower to normally work, and loads like wind abrasion, friction and the like can be generated in the operation process. The sensor information acquisition part comprises seven sensors including a voltage sensor, a current sensor, a vibration sensor, a temperature sensor, a speed sensor, a pressure sensor and a flow sensor. The seven sensors can respectively collect voltage data, current data, vibration data, temperature data, speed data, pressure data and flow data under the running state of the air blower, and after the data are collected, the data are input into a signal conditioning circuit in the core control part to be processed. The DSP chip is a chip capable of realizing a digital signal processing technology, and the DSP chip also comprises input and output interfaces and signals such as ECAN, PWM, GPIO, SPI, AD, ECAP and the like. The signal conditioning circuit can process the sensor signals acquired by the sensor signal acquisition part, the sensor signals are converted into signal data which can be processed by a DSP chip through the AD and ECAP input modules, the DSP is an indispensable device for data processing and achieving the expected effect, the DSP can process the data acquired by the AD and ECAP to generate corresponding PWM signals, and the PWM signals are input to the power driving part to generate expected voltage so as to achieve the purpose of perfectly controlling the work of the air blower. The host computer is the computer that can directly send out the control command, ECAN output module is connected with the host computer in the DSP, thereby reach simple and easy convenient complicated operation of completion, GPIO is general output interface, be connected with key control, thereby reach and to use the button to accomplish the relevant operation in the DSP, SPI is a fast-speed, full-duplex, synchronous communication bus, it links to each other with display module through SPI output, thereby make the staff can real-time detection current operating condition and relevant parameter.
Although the invention is a blower fault diagnosis method, the invention is also suitable for other systems similar to the blower, and an operator can flexibly and conveniently realize the functions of the blower according to the special application field by modifying software, changing hardware parameters and the like.
The invention has not been described in detail and is within the skill of the art.
The above description is only a part of the embodiments of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (4)
1. A failure diagnosis method of a blower is characterized in that: the method comprises the following steps:
the method comprises the following steps: collecting different sensor data of the air blower under the normal operation condition, wherein the number of the different sensor data is 7, and the different sensor data are respectively that a current sensor collects current data of an air blower motor, a voltage sensor collects voltage data of the air blower motor, a vibration sensor collects vibration data of an air blower shell, a temperature sensor measures temperature data of the air blower motor part, a speed sensor measures rotating speed data of the air blower motor, a pressure sensor measures air pressure data of the air blower, and a flow sensor measures air flow speed data of the air blower; the method comprises the steps of forming a matrix after median average filtering is carried out on collected data of different sensors, processing the matrix according to a recursive dynamic principal component analysis method, carrying out linear projection on an original data space, dividing the original data into a principal component subspace and a residual error subspace, wherein the two fault detection methods adopted are Hotelling's T2Statistical and squared prediction error SPE method, T2The change condition of the new data in the principal component subspace is measured statistically, and the projection change of the sampling vector on the residual subspace is measured statistically by the square prediction error SPE statistic; obtaining T under normal operation condition according to the matrix obtained after processing2And SPE statistic control limits;
step two: collecting the real-time operation data of the 7 blower sensors in the step one, forming a matrix, and calculating statistics corresponding to the real-time operation data of the blower sensors at the moment and T in the step one2Comparing the SPE statistic control limit to judge whether the blower fails;
step three: calculating T obtained by utilizing real-time operation data of the blower sensor in the step two2Statistics and SPE statistics, and step one positiveComparing the statistic control limit under the normal operation condition, and if the T calculated by the real-time operation data of the blower sensor in the step two2If the statistic and the SPE statistic do not exceed the statistic control limit under the normal operation condition, indicating that the blower operates normally, and if the statistic and the SPE statistic exceed the statistic control limit under the normal operation condition, indicating that the blower fails;
step four: after the fault of the blower occurs, calculating the ratio of the statistic of each sensor to the statistic of all sensors, namely the cumulative contribution rate of each sensor, and judging that a certain sensor in the blower has a fault according to the obtained ratio;
in the first step, the median average filtering is performed on the collected blower sensor data in the normal operation state to form a matrix, and the specific implementation is as follows:
selecting different sensor data of a blower under the normal condition with the data length of M, forming a matrix X, selecting the data length k far smaller than M, calculating the median of the data in each data length as a data source for average filtering processing, then selecting the data length M far smaller than k, carrying out averaging processing according to a formula (2), and obtaining a data matrix X 'after median average filtering, wherein X' belongs to R7*n,
In the formula:
xithe sample data after median average filtering;
xueach of m data lengths;
m is the data length selected for carrying out the average filtering algorithm;
performing a recursive algorithm on the filtered data matrix, specifically implementing the following steps:
in the formula:
A=[1,…,1]T;
a1the average value of each row of data in the data matrix after median average filtering;
n is the number of data in each column;
x' is a data matrix after median average filtering;
t represents the transposition of the matrix;
the matrix is normalized to data:
in the formula:
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
calculating the normalized matrix according to a recursive algorithm to obtain a recursive matrix Xk+1Comprises the following steps:
in the formula:
A=[1,…,1]T;
a1the average value of each row of data in the data matrix after median average filtering;
n is the number of data in each column;
k is the matrix order in which the recursive algorithm is performed;
t represents the transposition of the matrix;
X1the matrix X' is a standardized matrix with the processed mean value of 0 and the variance of 1;
∑1=diag(χ1.1,…χ1.i,…χ1.7),χ1.ithe standard deviation of each row of data is 1-7;
carrying out dynamic principal component algorithm calculation on the matrix subjected to recursive algorithm operation to obtain a matrix Q:
in the formula:
xi,jvalues representing all rows and columns in the matrix Q;
i and j are the row number and the column number of the matrix respectively;
q is a matrix after filtering and recursive dynamic algorithm operation;
acquiring data of 7 sensors under a normal condition, obtaining a matrix subjected to filtering and recursive dynamic algorithm operation according to the operation, and further obtaining a statistic control limit of the blower under the normal condition of the 7 sensors:
in the formula:
k represents the sequential order of the feature vectors;
a represents 7 sensors selected in sequence;
n is a normal distribution intermediate variable;
θi,i=1,2,…、h0representing an intermediate variable of the SPE statistic obeying standard normal distribution;
Cα、Fαis the confidence of the normal distribution.
2. The failure diagnosis method of a blower according to claim 1, characterized in that: in the second step, a formula (9) is adopted, and when the SPE statistic or T is used2If any one of the statistics exceeds the control limit of the statistics, a failure of the blower occurs:
in the formula:
T2represents T2Statistics;
SPE represents SPE statistic;
3. The failure diagnosis method of a blower according to claim 1, characterized in that: in the fourth step: when a blower fault occurs, each sensor pair T is calculated at each moment2And the contribution of SPE statistics, as equation (10), from the T obtained for each sensor2Statistic and value l of statistic SPEiComparing the total contribution value L to calculate each sensor pair T2And the accumulated contribution rate H of the SPE statistic, and further judging that a certain sensor of the air blower breaks down:
the total contribution L is the sum of the statistics obtained by each of the 7 sensors.
4. An apparatus for implementing the fault diagnosis method of the blower according to any one of claims 1 to 3, characterized in that: the system comprises a power supply management part, a power driving part, a blower body part, a sensor information acquisition part and a core control part; wherein:
the power management part comprises an alternating current-to-direct current module used for converting 220V alternating current into bus voltage required by the work of the blower, and a switching power supply is used for converting 48V into 3.3V, 5V and +/-15V through the direct current voltage conversion module and respectively supplying power to the power driving part, the sensor information acquisition part and the core control part;
the power driving part comprises a buffer circuit, a braking circuit, an IGBT unit, a driving circuit and an optical coupling isolation part;
when the air blower works normally, the power driving part firstly needs to supply power correctly by the power management part, and a PWM signal generated by the core control part reaches the driving circuit and the IGBT unit after being isolated by the optical coupler, so that the voltage of the air blower under normal working is obtained, and the air blower is in the process of power-off and speed reduction; PWM signals generated by the core control part reach the brake circuit part after passing through the optical coupling isolation and drive circuit for carrying out speed reduction operation of the blower, and the buffer circuit is used for preventing the IGBT unit from being burnt out by suddenly increased voltage and avoiding the damage of devices;
the blower body part comprises a blower and a load; the three-phase current and the three-phase voltage generated by the power driving part can drive the air blower to normally work, and similar wind abrasion and friction load generation can be generated in the operation process;
the sensor information acquisition part comprises 7 sensors including a voltage sensor, a current sensor, a vibration sensor, a temperature sensor, a speed sensor, a pressure sensor and a flow sensor; the 7 sensors can respectively acquire voltage data, current data, vibration data, temperature data, speed data, pressure data and flow data under the running state of the air blower, and after the data are acquired, the data are input into a signal conditioning circuit in the control part to be processed;
the core control part comprises an upper computer, a key control part, a display module, a signal conditioning circuit and a DSP; the DSP chip is a chip capable of realizing a digital signal processing technology, and the DSP chip also comprises an ECAN, a PWM, a GPIO, an SPI, an AD and an ECAP input/output interface and signals; the signal conditioning circuit can process the sensor signals acquired by the sensor signal acquisition part, the sensor signals are converted into signal data which can be processed by a DSP chip through an AD (analog-to-digital) and ECAP (electro-magnetic resonance) input module, the DSP is an indispensable device for processing data and achieving an expected effect, the DSP processes the data acquired by the AD and ECAP to generate corresponding PWM (pulse-width modulation) signals, and the PWM signals are input to the power driving part to generate expected voltage so as to achieve the purpose of perfectly controlling the work of the blower; the upper computer is a computer which directly sends out a control command, and an ECAN output module in the DSP is connected with the upper computer; the GPIO is a general output interface and is connected with the key control, so that the key can be used for finishing related operations in the DSP; SPI is the synchronous communication bus of high-speed full duplex, links to each other with display module through SPI output, and whether real-time detection air-blower breaks down and certain sensor breaks down.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110538008.3A CN113253705B (en) | 2021-05-18 | 2021-05-18 | Fault diagnosis method and device for air blower |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110538008.3A CN113253705B (en) | 2021-05-18 | 2021-05-18 | Fault diagnosis method and device for air blower |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113253705A CN113253705A (en) | 2021-08-13 |
CN113253705B true CN113253705B (en) | 2022-05-03 |
Family
ID=77182478
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110538008.3A Active CN113253705B (en) | 2021-05-18 | 2021-05-18 | Fault diagnosis method and device for air blower |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113253705B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115095534A (en) * | 2022-04-11 | 2022-09-23 | 中核核电运行管理有限公司 | KPCA-based CANDU6 reactor main pump fault diagnosis method |
CN115307670B (en) * | 2022-10-12 | 2023-01-13 | 蘑菇物联技术(深圳)有限公司 | Method, apparatus and medium for locating central air conditioning system anomaly sensors |
CN117150446B (en) * | 2023-10-30 | 2024-02-09 | 武汉华信数据***有限公司 | Blower operation state identification method and device, electronic equipment and storage medium |
CN117970856A (en) * | 2024-03-29 | 2024-05-03 | 杭州七所科技有限公司 | Automatic control system and method for unattended washing module |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN203643834U (en) * | 2014-01-10 | 2014-06-11 | 安徽理工大学 | Fault diagnosis system of mining main ventilator |
CN103884359A (en) * | 2014-04-17 | 2014-06-25 | 哈尔滨工业大学 | Satellite gyrounit fault diagnosis method based on principal component analysis algorithm |
CN105003453A (en) * | 2014-07-18 | 2015-10-28 | 河北联合大学 | Online monitoring and fault diagnosis system of mine fan |
CN106884809A (en) * | 2017-03-20 | 2017-06-23 | 中国矿业大学 | A kind of Coal Mine Ventilator real-time fault diagnosis and prior-warning device based on virtual instrument development platform |
CN107608335A (en) * | 2017-09-14 | 2018-01-19 | 山东科技大学 | A kind of UAV Flight Control System fault detect and the data-driven method of fault reconstruction |
CN111680725A (en) * | 2020-05-28 | 2020-09-18 | 哈尔滨工业大学 | Gas sensor array multi-fault isolation algorithm based on reconstruction contribution |
-
2021
- 2021-05-18 CN CN202110538008.3A patent/CN113253705B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN203643834U (en) * | 2014-01-10 | 2014-06-11 | 安徽理工大学 | Fault diagnosis system of mining main ventilator |
CN103884359A (en) * | 2014-04-17 | 2014-06-25 | 哈尔滨工业大学 | Satellite gyrounit fault diagnosis method based on principal component analysis algorithm |
CN105003453A (en) * | 2014-07-18 | 2015-10-28 | 河北联合大学 | Online monitoring and fault diagnosis system of mine fan |
CN106884809A (en) * | 2017-03-20 | 2017-06-23 | 中国矿业大学 | A kind of Coal Mine Ventilator real-time fault diagnosis and prior-warning device based on virtual instrument development platform |
CN107608335A (en) * | 2017-09-14 | 2018-01-19 | 山东科技大学 | A kind of UAV Flight Control System fault detect and the data-driven method of fault reconstruction |
CN111680725A (en) * | 2020-05-28 | 2020-09-18 | 哈尔滨工业大学 | Gas sensor array multi-fault isolation algorithm based on reconstruction contribution |
Also Published As
Publication number | Publication date |
---|---|
CN113253705A (en) | 2021-08-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113253705B (en) | Fault diagnosis method and device for air blower | |
CN111308991B (en) | Coal mill operation fault identification method and application | |
CN104390657B (en) | A kind of Generator Unit Operating Parameters measurement sensor fault diagnosis method and system | |
WO2016131228A1 (en) | Multi-information online detection system for machine tool multi-source energy consumption system | |
CN103324943A (en) | Identification method of complex device panel image multi-sub zone state | |
KR102067344B1 (en) | Apparatus and Method for Detecting Abnormal Vibration Data | |
CN104181385A (en) | Power detection method and power detection apparatus of air conditioner | |
CN110231529A (en) | A kind of control cabinet intelligent Fault Diagnose Systems and method for diagnosing faults | |
CN111160652B (en) | Knowledge-sensing-based equipment abnormal state comprehensive judgment and operation and maintenance method | |
CN107607342B (en) | Healthy energy efficiency detection method for air conditioner room equipment group | |
CN1321328C (en) | Wavelet diagnostic system for initial failure of electromotor and method for diagnosing malfunction of electromotor | |
Bejaoui et al. | A data-driven prognostics technique and rul prediction of rotating machines using an exponential degradation model | |
CN105629959A (en) | Industrial process fault detection method | |
CN104569814B (en) | A kind of DC traction motor health status real-time analysis method based on approximate entropy | |
CN116796261B (en) | Closed switch equipment mechanical characteristic prediction method based on artificial intelligence | |
CN109406923B (en) | Method and device for predicting ground leakage fault of train power supply system | |
CN116260250B (en) | Intelligent monitoring system for high-frequency switch direct-current power supply | |
CN112034782A (en) | Air preheater online state monitoring and fault prediction method | |
CN205173022U (en) | Nuclear power station nuclear island is with intelligent energy -conserving fan system | |
CN115081491B (en) | Running state fault diagnosis method for sanitary towel production equipment | |
CN110057587A (en) | A kind of nuclear power pump bearing intelligent failure diagnosis method and system | |
CN212550002U (en) | Coal pulverizer running state monitoring system | |
CN115329942A (en) | Bolt fastening device and method based on artificial intelligence | |
CN115343650A (en) | High-precision digital power supply system based on embedded system | |
CN211852164U (en) | Screw fan vibration data acquisition system |
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 |