CN112781903B - Fault diagnosis method for blast furnace blower and TRT unit based on digital twin system - Google Patents

Abstract

The invention discloses a blast furnace blower and TRT set fault diagnosis method based on a digital twin system, which comprises the following steps: constructing three-dimensional models of a blast furnace blower and a TRT unit system; constructing a blast furnace blower and a TRT set digital twin system by acquiring field real-time data and calculating virtual data; performing abnormal data elimination on data based on a digital twin system; extracting frequency domain features by using an improved vibration frequency spectrum ratio extraction method; constructing a neural network fault diagnosis algorithm based on an Adam algorithm according to the time-frequency domain characteristics and the operation parameters of the equipment; and finally, the output result of the neural network is stored in a digital twin system to form three-dimensional dynamic display of the fault information of the equipment fault component, generate a diagnosis report and push the diagnosis report to field management personnel. The method can effectively realize the fault diagnosis and the health analysis of the blast furnace blower and the TRT unit, and can guide the field operation by combining with the digital twin system, thereby ensuring the safe and efficient operation of field equipment.

Description

Fault diagnosis method for blast furnace blower and TRT unit based on digital twin system

technical field

The invention relates to the fields of a digital twin system, fault diagnosis and the like, in particular to a fault diagnosis method for a blast furnace blower and a TRT unit based on the digital twin system.

Background technique

Blast furnace blasting and TRT system are important technological processes of blast furnace ironmaking. They can not only use the residual pressure of blast furnace gas to generate high-efficiency power generation, but also effectively solve the noise pollution and pipeline vibration generated by the pressure reducing valve group, and also reduce the pressure on the top of the blast furnace. Stability control plays an important role. Practice has proved that the TRT power generation is about 40% of the power consumption of the blast furnace blower. Therefore, the health of the blast furnace blower and the TRT unit and its operation status directly affect the output and safety of ironmaking, and it is extremely important to troubleshoot them. . However, in the process of blast furnace ironmaking, blast furnace blowers and TRT units are often in bad working conditions, unstable, high power, heavy load and continuous operation, and vicious accidents caused by operating failures are not uncommon.

With the research and in-depth study of various intelligent algorithms, more and more algorithms are used in fault diagnosis. However, due to the complex and changeable operating conditions of blast furnace blowers and TRT units, and the many operating parameters, fault feature extraction has also become an important issue in the field of fault diagnosis. Big difficulty.

SUMMARY OF THE INVENTION

The invention overcomes the problems of complex and changeable operating conditions of the blast furnace blower and the TRT unit, numerous operating parameters and difficulty in extracting fault features, and proposes a fault diagnosis method for the blast furnace blower and the TRT unit based on a digital twin system.

In order to achieve the above object, the present invention provides the following solutions:

A fault diagnosis method for blast furnace blower and TRT unit based on digital twin system. Firstly, construct a three-dimensional model of blast furnace blower and TRT unit system. The twin system removes abnormal data from the data; uses the improved vibration spectrum ratio extraction method to extract time-frequency domain features; constructs a neural network fault diagnosis algorithm based on Adam algorithm according to the time-frequency domain features and other operating parameters; finally, the neural network The output results are stored in the digital twin system to form a three-dimensional dynamic display of the fault information of the faulty components of the equipment, generate a diagnosis report, and push it to the on-site management personnel.

In the above technical solution, preferably, in the step S1, three-dimensional modeling is performed based on Unity3D, including the three-dimensional modeling of the entire production line of the blast furnace blower and the TRT unit, and the actual scene of the equipment on site is reproduced, which is consistent with the on-site process flow.

Preferably, the digital twin system of the blast furnace blower and TRT unit can collect on-site real-time data and predict virtual data for data modeling, so as to realize a real-time system that runs synchronously with the site and a virtual twin system that can also run when the site is shut down Two scenarios, and mark the failure data of each scenario.

Preferably, the abnormal data removal is to remove abnormal data transmitted or abnormal data caused by accidental changes in the environment through a moving average algorithm, so as not to affect the accuracy of fault diagnosis. The specific method is as follows:

The operating parameter variable X at time t after removing abnormal data is denoted as X _t , θ _t is the value of operating parameter variable X at time t before removing abnormal data, β is the moving average coefficient, β∈[0,1), at β =0, do not use moving average, X _t = θ _t ; after using moving average:

X _t =β*X _t-1 +(1-β)*θ _t

Preferably, the improved vibration spectrum ratio extraction method described in step S4 includes the following steps:

(1) Obtain the vibration time-domain waveform signal x(n)=[x ₁ , x ₂ , x ₃ ... x _N ] of the equipment, where x _N is the vibration acceleration value, and N is the number of sampling points;

其中0≤k≤N-1,n为第n个数据，k为频域上第k个值，W_N＝e^-j*2*π/N是旋转因子，设N＝2^r，将x(n)分为前后各一半，得到两个长为N/2的序列，经整合计算得到：(2) Perform fast Fourier transform (FFT) on the vibration time domain waveform signal, according to the Fourier transform formula:

Where 0≤k≤N-1, n is the nth data, k is the kth value in the frequency domain, W _N =e ^-j*2*π/N is the twiddle factor, set N=2 ^r , set x (n) is divided into two halves before and after, and two sequences with a length of N/2 are obtained, which are obtained by integration and calculation:

(3) Obtain the frequency domain value of the vibration signal after FFT [f1,|X(1)|; f2,|X(2);...fN,|X(N)|], where f1, f2,...,fN is the frequency value after FFT, |X(1)|,|X(2)|,…,,|X(N)| is the frequency amplitude value at the corresponding frequency;

(4) The characteristic frequencies of the computing equipment are 0.5 times, 1 times, 1.5 times, 2 times, 2.5 times, 3 times, 3.5 times, 4 times and high times, respectively, which are recorded as: [ f _0.5X ,f _1X ,f _1.5X ,f _2X ,f _2.5X ,f _3X ,f _3.5X ,f _4X ,f _nX ], the frequency multiplier is a multiple of the fundamental frequency, and the fundamental frequency is equal to 1 frequency;

其中f_1X为1倍频的频率，Vr为设备的转速，得到设备的特征频率为：Among them, f _nX is the frequency of high frequency multiplication, that is, the frequency greater than 4 times frequency,

Where f _1X is the frequency of 1 frequency, Vr is the rotation speed of the equipment, and the characteristic frequency of the equipment is obtained as:

Here n is a high octave, indicating all frequencies greater than 4 times the frequency;

(5) Calculate the spectral proportion of the characteristic frequency:

fi为第i个频率值，则选取所有的fi对应幅值综合作为0.5 倍频的幅值，得到0.5倍频的幅值占比为：(i为第i个频率，或第i个频率对应的幅值)①If

fi is the ith frequency value, then all the corresponding amplitudes of fi are selected as the 0.5-octave amplitude, and the ratio of the 0.5-octave amplitude is: (i is the ith frequency, or the corresponding frequency of the ith frequency amplitude)

fj为第j个频率值，得到1倍频的幅值占比为：②Similarly, if

fj is the jth frequency value, and the ratio of the amplitude of 1 octave is:

fa为第a个频率值，得到1.5倍频的幅值占比为：③Similarly, if

fa is the a-th frequency value, and the ratio of the amplitude of the 1.5-octave frequency is:

fb为第b个频率值，得到2倍频的幅值占比为：④ In the same way, if

fb is the b-th frequency value, and the ratio of the amplitude of the 2-fold frequency is:

fc为第c个频率值，得到2.5倍频的幅值占比为：⑤Similarly, if

fc is the c-th frequency value, and the ratio of the amplitude of the 2.5 times frequency is:

fd为第d个频率值，得到3倍频的幅值占比为：⑥Similarly, if

fd is the d-th frequency value, and the ratio of the amplitude of the 3-fold frequency is:

fg为第g个频率值，得到3.5倍频的幅值占比为：⑦Similarly, if

fg is the g-th frequency value, and the ratio of the amplitude of the 3.5-octave frequency is:

fe为第e个频率值，得到4倍频的幅值占比为：⑧ Similarly, if

fe is the e-th frequency value, and the ratio of the amplitude of the quadruple frequency is:

fn为第n个频率值，得到高倍频的幅值占比为：⑨ Finally, if

fn is the nth frequency value, and the ratio of the amplitude of the high frequency multiplication is:

|X(N)|)中前t个最大值幅值的总和， t＝e/8。Among them, sum(|X(n)|) is (

The sum of the first t maximum magnitudes in |X(N)|), t=e/8.

Preferably, the described constructing a neural network fault diagnosis algorithm based on the Adam algorithm includes the following steps:

(1) Determine the input and output of the neural network, where the input is the operating parameter variables of the equipment and the vibration time-frequency domain data, including current, voltage, power, temperature, flow, triaxial vibration RMS, triaxial vibration time domain Index, three-axis frequency domain index; the output is the fault type of the equipment, including equipment rotor wear, unbalance, misalignment, loose base, bearing inner ring fault, bearing outer ring fault, friction fault of moving and static parts.

(2) Determine the following parameters of the neural network: the number of nodes in the input layer is 8, the number of nodes in the output layer is 1, there is 1 hidden layer, and the number of nodes in each hidden layer is 26, and the network learning rate is 0.21, the momentum coefficient is 0.01;

(3) Using the accelerated gradient algorithm (Adam algorithm) to optimize the weight of the neural network, the specific method is: on the basis of the gradient descent method with momentum, the square gradient is introduced, and the deviation of the rate is corrected;

(4) Build a complete forward and reverse computing neural network model, read the data of the fault identification in the digital twin system, and conduct training and testing;

(5) Read real-time operating data for fault diagnosis and output.

The beneficial effects of the present invention are:

The invention constructs a three-dimensional model of the blower and the TRT, and uses the improved spectrum ratio method to construct a fault diagnosis algorithm of the blower and the TRT neural network, which can realize the real-time monitoring and fault diagnosis of the blower and the TRT, and display them in a three-dimensional intuitive form, so it can effectively It solves the problem that the operation conditions of blast furnace blower and TRT unit are complex and changeable, there are many operating parameters, and it is difficult to extract fault features. The method of the invention can effectively realize the fault diagnosis and health analysis of the blast furnace blower and the TRT unit, and by combining with the digital twin system, the three-dimensional fault information of the equipment can be displayed intuitively, and the on-site operation can be guided, thereby ensuring the safe and efficient operation of the on-site equipment. , with high actual production value.

Description of drawings

FIG. 1 is a flowchart of a fault diagnosis method for a blast furnace blower and a TRT unit based on a digital twin system of the present invention.

FIG. 2 is a structural frame diagram of the blast furnace blower and the TRT digital twin system of the present invention.

FIG. 3 is a schematic diagram of the input and output structure of the neural network of the present invention.

Detailed ways

As shown in Figure 1, the present invention provides a kind of blast furnace blower and TRT unit fault diagnosis method based on digital twin system, first constructs the three-dimensional model of blast furnace blower and TRT unit system; The digital twin system of blast furnace blower and TRT unit; based on the digital twin system, the abnormal data is eliminated; the improved vibration spectrum ratio extraction method is used to extract the time-frequency domain features; according to the time-frequency domain features and other operating parameters, the Adam based The neural network fault diagnosis algorithm of the algorithm; finally, the output of the neural network is stored in the digital twin system, forming a three-dimensional dynamic display of the fault information of the faulty components of the equipment, generating a diagnosis report, and pushing it to the on-site management personnel.

Among them, the 3D model construction of blast furnace blower and TRT unit is based on Unity3D for 3D modeling, including the 3D modeling of the entire production line of blast furnace blower and TRT unit, which reproduces the actual scene of the equipment on site, which is consistent with the on-site process flow.

Among them, the digital twin system of blast furnace blower and TRT unit is shown in Figure 2, including 3D model real-time scene roaming, 3D model virtual scene roaming, real-time operation data collection and storage, virtual data real-time prediction and storage, pre-fault data labeling and classification, Functions such as fault result saving and alarm push, realize two scenarios, a real-time system that runs synchronously with the site, and a virtual twin system that can also run when the site is down.

The digital twin system of blast furnace blower and TRT unit includes two modules: operation and simulation. The operation module mainly realizes real-time data collection on site through the communication of intelligent sensors and on-site database, including blast furnace top pressure and blower control performance evaluation module, blower and blast furnace Top pressure control parameter optimization module, blower surge identification module, unit energy efficiency analysis module, unit energy flow module, turbine blade ash accumulation module, unit health diagnosis module, to realize real-time monitoring and status analysis of blast furnace blower and TRT unit; The simulation module realizes the blower control simulation module, blast furnace top pressure control simulation module, blower anti-surge control simulation module, TRT start-up process control simulation module, TRT shutdown process control simulation module, TRT start-up process control simulation module by setting relevant on-site working conditions. The establishment of emergency shutdown process control simulation module.

The invention eliminates abnormal data in transmission or abnormal data caused by accidental change factors in the environment through a moving average algorithm, so as not to affect the accuracy of fault diagnosis. The specific method is as follows:

The operating parameter variable X at time t after removing abnormal data is denoted as X _t , θ _t is the value of operating parameter variable X at time t before removing abnormal data, β is the moving average coefficient, β∈[0,1), at β =0, do not use moving average, X _t = θ _t ; after using moving average:

X _t =β*X _t-1 +(1-β)*θ _t

Among them, the improved method for extracting the proportion of vibration spectrum includes the following steps:

(1) Obtain the vibration time-domain waveform signal x(n)=[x ₁ , x ₂ , x ₃ ... x _N ] of the equipment, where x _N is the vibration acceleration value, and N is the number of sampling points;

其中0≤k≤N-1,n为第n个数据，k为频域上第k个值，W_N＝e^-j*2*π/N是旋转因子，设N＝2^r，将x(n)分为前后各一半，得到两个长为N/2的序列，经整合计算得到：(2) Perform fast Fourier transform (FFT) on the vibration time domain waveform signal, according to the Fourier transform formula:

Where 0≤k≤N-1, n is the nth data, k is the kth value in the frequency domain, W _N =e ^-j*2*π/N is the twiddle factor, set N=2 ^r , set x (n) is divided into two halves before and after, and two sequences with a length of N/2 are obtained, which are obtained by integration and calculation:

(3) Finally, the frequency domain value of the vibration signal after FFT is obtained [f1, |X(1)|; f2, |X(2);...fN,|X(N)|], where f1, f2,..., fN is the frequency value after FFT, |X(1)|,|X(2)|,…,,|X(N)| is the frequency amplitude value at the corresponding frequency;

(4) The characteristic frequencies of the computing equipment are 0.5 times, 1 times, 1.5 times, 2 times, 2.5 times, 3 times, 3.5 times, 4 times, high times (more than 4 times) The frequencies are collectively referred to as high octaves), recorded as: [f _0.5X , f _1X , f _1.5X , f _2X , f _2.5X , f _3X , f _3.5X , f _4X , f _nX ],

其中f_1X为1倍频的频率，Vr为设备的转速，得到设备的特征频率为：Among them, f _nX is the frequency of high frequency multiplication, that is, the frequency greater than 4 times frequency,

Where f _1X is the frequency of 1 frequency, Vr is the rotation speed of the equipment, and the characteristic frequency of the equipment is obtained as:

Here n is a high octave, indicating all frequencies greater than 4 times the frequency;

(5) Calculate the spectral proportion of the characteristic frequency:

fi为第i个频率值，则选取所有的fi对应幅值综合作为0.5 倍频的幅值，得到0.5倍频的幅值占比为：(i为第i个频率，或第i个频率对应的幅值)①If

fi is the ith frequency value, then all the corresponding amplitudes of fi are selected as the 0.5-octave amplitude, and the ratio of the 0.5-octave amplitude is: (i is the ith frequency, or the corresponding frequency of the ith frequency amplitude)

fj为第j个频率值，得到1倍频的幅值占比为：②Similarly, if

fj is the jth frequency value, and the ratio of the amplitude of 1 octave is:

fa为第a个频率值，得到1.5倍频的幅值占比为：③Similarly, if

fa is the a-th frequency value, and the ratio of the amplitude of the 1.5-octave frequency is:

fb为第b个频率值，得到2倍频的幅值占比为：④ In the same way, if

fb is the b-th frequency value, and the ratio of the amplitude of the 2-fold frequency is:

fc为第c个频率值，得到2.5倍频的幅值占比为：⑤Similarly, if

fc is the c-th frequency value, and the ratio of the amplitude of the 2.5 times frequency is:

fs为第d个频率值，得到3倍频的幅值占比为：⑥Similarly, if

fs is the d-th frequency value, and the ratio of the amplitude of the 3-fold frequency is:

fg为第g个频率值，得到3.5倍频的幅值占比为：⑦Similarly, if

fg is the g-th frequency value, and the ratio of the amplitude of the 3.5-octave frequency is:

fe为第e个频率值，得到4倍频的幅值占比为：⑧ Similarly, if

fe is the e-th frequency value, and the ratio of the amplitude of the quadruple frequency is:

fn为第n个频率值，得到高倍频的幅值占比为：⑨ Finally, if

fn is the nth frequency value, and the ratio of the amplitude of the high frequency multiplication is:

|X(N)|)中前t个最大值幅值的总和， t＝e/8。Among them, sum(|X(n)|) is (

The sum of the first t maximum magnitudes in |X(N)|), t=e/8.

Among them, constructing a neural network fault diagnosis algorithm based on Adam algorithm includes the following steps:

(1) Determine the input and output of the neural network, as shown in Figure 3, where the input is the equipment's operating parameter variables and vibration time-frequency domain data, including current, voltage, power, temperature, flow, three-axis vibration RMS, Three-axial vibration time domain index, three-axial frequency domain index; the output is the fault type of the equipment, including equipment rotor wear, unbalance, misalignment, loose base, bearing inner ring fault, bearing outer ring fault, friction of dynamic and static parts Fault.

Among them, time domain indicators include peak index, kurtosis index, skewness index, margin index, and pulse index; frequency domain indicators include 0.5 multipliers, 1 multipliers, 1.5 multipliers, 2 multipliers, 2.5 multipliers, and 3 multipliers frequency, 3.5 times frequency, 4 times frequency, high frequency (the frequencies greater than 4 times frequency are collectively referred to as high frequency).

(2) Determine the following parameters of the neural network: the number of nodes in the input layer is 8, the number of nodes in the output layer is 1, there is 1 hidden layer, and the number of nodes in each hidden layer is 26, and the network learning rate is 0.21, the momentum coefficient is 0.01;

(3) Using the accelerated gradient algorithm (Adam algorithm) to optimize the weight of the neural network, the specific method is to introduce the squared gradient on the basis of the gradient descent method with momentum, and to correct the deviation of the rate;

(4) Build a complete forward and reverse computing neural network model, read the data of the fault identification in the digital twin system, and conduct training and testing;

(5) Read the real-time operating data for fault diagnosis, and output and save it in the corresponding module of the digital twin system.

Claims (5)

1. a blast furnace blower and TRT unit fault diagnosis method based on digital twin system, is characterized in that, comprises the following steps: S1: Build a 3D model of the blast furnace blower and TRT unit system; S2: Build a digital twin system of blast furnace blower and TRT unit by collecting on-site real-time data and calculating virtual data; S3: Eliminate abnormal data based on the digital twin system; S4: Use the improved vibration spectrum ratio extraction method to extract time-frequency domain features; S5: Construct a neural network fault diagnosis algorithm based on the Adam algorithm according to the time-frequency domain characteristics and the operating parameters of the equipment; S6: The neural network output results are stored in the digital twin system, forming a three-dimensional dynamic display of the fault information of the equipment faulty components, generating a diagnosis report, and pushing it to the on-site management personnel; The improved vibration spectrum ratio extraction method described in step S4 specifically includes the following steps: S41: Obtain the vibration time-domain waveform signal x(n)=[x ₁ , x ₂ , x ₃ . . . x _N ] of the device, where x _N is the vibration acceleration value, and N is the number of sampling points; S42: Perform fast Fourier transform (FFT) on the vibration time-domain waveform signal, according to the Fourier transform formula:

where 0≤k≤N-1, n is the n-th sampled data, k is the k-th data in the frequency domain, W _N =e ^-j*2*π/N is the twiddle factor, set N=2 ^r , Divide x(n) into half before and after, and get two sequences of length N/2, which are calculated by integration:

S43: Obtain the frequency domain value of the vibration signal after FFT [f1, |X(1)|; f2, |X(2); ...fN, |X(N)|], where f1, f2, ..., fN are The frequency value after FFT, |X(1)|,|X(2)|,…,,|X(N)| is the frequency amplitude at the corresponding frequency; S44: The characteristic frequencies of the computing equipment, which are 0.5 times, 1 times, 1.5 times, 2 times, 2.5 times, 3 times, 3.5 times, 4 times, and high times, respectively, denoted as: [f _0.5X ,f _1x ,f _1.5X ,f _2X ,f _2.5X ,f _3X ,f _3.5X ,f _4X ,f _nX ], the frequency multiplier is a multiple of the fundamental frequency, and the fundamental frequency is equal to 1 frequency; Among them, f _nX is the frequency of high frequency multiplication, that is, the frequency greater than 4 times frequency,

Where f _1X is the frequency of 1 frequency, Vr is the rotation speed of the equipment, and the characteristic frequency of the equipment is obtained as:

Here n is a high octave, indicating all frequencies greater than 4 times the frequency; S45: Calculate the spectral proportion of the eigenfrequency: ①fi is the ith frequency value, if

Then select all the corresponding amplitudes of fi as the 0.5-octave amplitude, and the ratio of the 0.5-octave amplitude is:

②Similarly, if

fj is the jth frequency value, and the ratio of the amplitude of 1 octave is:

③Similarly, if

fa is the a-th frequency value, and the ratio of the amplitude of the 1.5-octave frequency is:

④ In the same way, if

fb is the b-th frequency value, and the ratio of the amplitude of the 2-fold frequency is:

⑤Similarly, if

fc is the c-th frequency value, and the ratio of the amplitude of the 2.5 times frequency is:

⑥Similarly, if

fd is the d-th frequency value, and the ratio of the amplitude of the 3-fold frequency is:

⑦Similarly, if

fg is the g-th frequency value, and the ratio of the amplitude of the 3.5-octave frequency is:

⑧ Similarly, if

fe is the e-th frequency value, and the ratio of the amplitude of the quadruple frequency is:

⑨ Finally, if

fn is the nth frequency value, and the ratio of the amplitude of the high frequency multiplication is:

where sum(|X(n)|) is

The sum of the amplitudes of the first t maximum values, t=e/8. 2. a kind of blast furnace blower and TRT unit fault diagnosis method based on digital twin system according to claim 1, is characterized in that, in described step S1, carries out three-dimensional modeling based on Unity3D, comprises blast furnace blower and TRT unit whole. The 3D modeling of the production line reproduces the actual scene of the equipment on site, which is consistent with the on-site process flow. 3. a kind of blast furnace blower and TRT unit fault diagnosis method based on digital twin system according to claim 1, is characterized in that, blast furnace blower described in step S2 and TRT unit digital twin system are used to carry out the collection of on-site real-time data As well as the prediction of virtual data for data modeling, two scenarios of a real-time system running synchronously with the site and a virtual twin system that continues to run when the site is shut down are realized, and the fault data of each scenario is marked. 4. a kind of blast furnace blower and TRT unit fault diagnosis method based on digital twin system according to claim 1, is characterized in that, in step S3, removes abnormal data by moving average algorithm, concrete method is as follows: The operating parameter variable X at time t after removing abnormal data is denoted as X _t , θ _t is the value of operating parameter variable X at time t before removing abnormal data, β is the moving average coefficient, β∈[0,1), at β =0, do not use moving average, X _t = θ _t ; after using moving average: X _t =β*X _t-1 +(1-β)*θ _t . 5. a kind of blast furnace blower and TRT unit fault diagnosis method based on digital twin system according to claim 1, is characterized in that, the described construction of step S5 is based on the neural network fault diagnosis algorithm of Adam algorithm comprises the following steps: S51: Determine the input and output of the neural network, where the input is the operating parameter variables of the equipment and the vibration time-frequency domain data, including current, voltage, power, temperature, flow, triaxial vibration RMS, and triaxial vibration time domain indicators , three-axis frequency domain indicators; the output is the fault type of the equipment, including equipment rotor wear, unbalance, misalignment, loose base, bearing inner ring fault, bearing outer ring fault, friction fault of dynamic and static parts; S52: Determine the parameters of the neural network: the number of nodes in the input layer is 8, the number of nodes in the output layer is 1, there is 1 hidden layer, and the number of nodes in each hidden layer is 26, the network learning rate is 0.21, the momentum The coefficient is 0.01; S53: Use the Adam algorithm to optimize the weight of the neural network. The specific method is as follows: on the basis of the gradient descent method with momentum, the square gradient is introduced, and the deviation of the rate is corrected; S54: Build a complete forward and reverse computing neural network model, read the data marked with faults in the digital twin system, and conduct training and testing; S55: Read real-time running data for fault diagnosis and output.

Images