Early fault detection method, system and equipment for induced draft fan and medium

Abstract

The invention discloses an induced draft fan early-stage fault detection method, system and device and a medium, and the method comprises the following steps: obtaining historical health time sequence data of each detection point, carrying out the information extraction of the historical health time sequence data of each detection point, and constructing an effective information memory matrix through the extracted information; according to the effective information memory matrix, estimating the operation state of the system by using a multivariate state estimation method; the operation state of the system is compared with a preset early warning criterion, early failure detection and early warning are conducted on the induced draft fan according to a comparison result, and the method, the system, the equipment and the medium can accurately predict the early failure of the induced draft fan.

Description

Technical Field

[0001] The invention belongs to the field of fault detection and relates to an induced draft fan early fault detection method, system, equipment and medium. Background Art

[0002] As a typical technology-intensive, continuous production engineering system, thermal power plants have high requirements for stability and are characterized by strong coupling, multiple variables, and nonlinearity. Among them, induced draft fans, as a large rotating machinery and important auxiliary equipment, have harsh operating conditions, changeable operating conditions, and a wide variety of faults. For power plant production with high requirements for reliability and stability, this has brought about a dual test of safety and economy to a large extent.

[0003] At present, most of the traditional methods and research results of fan research tend to be fault diagnosis. Commonly used methods include: fault diagnosis based on neural network, fault diagnosis based on gray theory, reliability-centered maintenance analysis, and fault diagnosis based on expert system. Without exception, these methods require a known and complete fault knowledge base, and can only identify faults in the fault knowledge base. The induced draft fan of a power plant is a kind of strongly coupled, multivariable, and nonlinear mechanical equipment. It has a complex mechanical structure and a wide variety of faults. There are problems such as fuzzy fault development mechanism, that is, one fault may have multiple signs, and one sign may correspond to multiple faults. There is no clear correspondence between signs and faults, so it is difficult to build a complete fault knowledge base. In addition, the monitored data is easily interfered by external random factors and can easily drown out the real signal. Therefore, the fault diagnosis of induced draft fans at home and abroad is mostly limited to the judgment of vibration signals, and other signals only play an auxiliary role in judgment, and there are problems such as low fault diagnosis accuracy. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method, system, device and medium for detecting early failure of an induced draft fan, which can accurately predict early failure of the induced draft fan.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme:

[0006] In one aspect, the present invention provides a method for detecting early faults of an induced draft fan, comprising the following steps:

[0007] Acquire historical health time series data of each measuring point, extract information from the historical health time series data of each measuring point, and construct an effective information memory matrix using the extracted information;

[0008] estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix;

[0009] The operating status of the system is compared with a preset warning criterion, and an early fault detection warning of the induced draft fan is performed according to the comparison result.

[0010] Furthermore, the process of extracting information from the historical health time series data of each measuring point is as follows:

[0011] Information is extracted from the historical health time series data of each measuring point based on a discrete particle swarm algorithm.

[0012] Furthermore, the measuring points include X-axis vibration, Y-axis vibration, induced draft fan front bearing temperature, induced draft fan middle bearing temperature, induced draft fan rear bearing temperature, motor front bearing temperature, motor rear bearing temperature, motor stator coil temperature, induced draft fan oil filter pressure difference, induced draft fan oil tank temperature, motor lubricating oil pressure and motor current.

[0013] Furthermore, the process of extracting information from the historical health time series data of each measuring point based on the discrete particle swarm algorithm is as follows:

[0014] The sample span η(x), distribution uniformity λ(x) and redundancy ρ(x) are combined and calculated respectively, and the sample evaluation index is introduced to represent the state information of the running data in the matrix. The function represents the sample evaluation index, namely:

[0015]

[0016] Among them, the power exponent γ i (i=1,2,3)∈(0,1);

[0017] A discrete particle swarm algorithm is used to extract information from the historical health time series data of each measuring point according to the sample evaluation index.

[0018] Furthermore, before extracting information from the historical health time series data of each measuring point, the process further includes:

[0019] The historical health time series data of each measuring point is denoised.

[0020] Furthermore, the process of estimating the operating state of the system using the multivariate state estimation method according to the effective information memory matrix is ​​as follows:

[0021] According to the effective information memory matrix, a weight algorithm is introduced to estimate the future operating parameters of the induced draft fan system, wherein the estimated vector for:

[0022]

[0023] Where w=[w1,w2,…,w n] T Represents the weight of each sample in the prediction matrix D, and x is a new observation vector.

[0024] Furthermore, early faults are detected based on the total measurement deviation, which is:

[0025]

[0026] Among them, k = 1, 2,…, p.

[0027] In a second aspect of the present invention, the present invention provides an early fault detection system for an induced draft fan, comprising:

[0028] An acquisition module is used to acquire the historical health time series data of each measuring point, extract information from the historical health time series data of each measuring point, and construct an effective information memory matrix using the extracted information;

[0029] An estimation module, used for estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix;

[0030] The early warning module is used to compare the operating status of the system with a preset early warning criterion and to perform early fault detection and early warning of the induced draft fan according to the comparison result.

[0031] In a third aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the induced draft fan early fault detection method when executing the computer program.

[0032] In a fourth aspect of the present invention, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the induced draft fan early fault detection method are implemented.

[0033] The present invention has the following beneficial effects:

[0034] The induced draft fan early fault detection method, system, equipment and medium described in the present invention, during specific operation, utilizes a discrete particle swarm algorithm to search for a large number of information samples to obtain an effective information memory matrix, and then utilizes a multivariate state estimation method to estimate the future operating state of the system, and performs early fault detection and warning of the induced draft fan according to preset warning criteria, which has important engineering application significance and theoretical research value. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0036] Figure 1 This is a simplified structural diagram of the induced draft fan and its sensor measurement points;

[0037] Figure 2 It is a framework diagram of the early fault detection method of the induced draft fan;

[0038] Figure 3 Convex hull space filled with physical experimental data;

[0039] Figure 4 Flowchart for wavelet decomposition;

[0040] Figure 5 Flowchart of DPSO algorithm.

[0041] Among them, 1 is the induced draft fan body, 2 is the induced draft fan inlet, 3 is the induced draft fan front bearing, 4 is the induced draft fan middle bearing, 5 is the induced draft fan rear bearing, 6 is the motor, 7 is the motor rear bearing, 8 is the motor front bearing, 9 is the induced draft fan oil tank oil control valve, 10 is the hydraulic oil A oil pump, 11 is the hydraulic oil B oil pump, 12 is the cooling fan A, 13 is the cooling fan B, 14 is the motor front bearing lubricating oil valve, 15 is the motor rear bearing lubricating oil valve, 16 is the oil cooler, 17 is the three-way valve, 18 is the lubricating oil A oil pump, 19 is the lubricating oil B oil pump, 20 is the induced draft fan hydraulic oil station tank, and 21 is the induced draft fan lubricating oil tank. DETAILED DESCRIPTION

[0042] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0043] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present application belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention.

[0044] Embodiment 1

[0045] refer to Figure 2 The method for detecting early faults of the induced draft fan of the present invention comprises the following steps:

[0046] Obtaining historical health time series data of each measuring point, extracting information from the historical health time series data of each measuring point based on a discrete particle swarm algorithm, and constructing an effective information memory matrix using the extracted information, wherein the effective information memory matrix includes complete health information when the system is running;

[0047] estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix;

[0048] The operating status of the system is compared with a preset warning criterion, and an early fault detection warning of the induced draft fan is performed according to the comparison result.

[0049] In this embodiment, the measuring points include X-axis vibration, Y-axis vibration, induced draft fan front bearing temperature, induced draft fan middle bearing temperature, induced draft fan rear bearing temperature, motor front bearing temperature, motor rear bearing temperature, motor stator coil temperature, induced draft fan oil filter pressure difference, induced draft fan oil tank temperature, motor lubricating oil pressure and motor current, a total of 25 measuring points, the specific induced draft fan structure is as follows Figure 1 shown.

[0050] In this embodiment, when extracting information, the sample span η(x), distribution uniformity λ(x) and redundancy ρ(x) are combined and calculated respectively, and the sample evaluation index is introduced to represent the state information of the running data in the matrix, wherein The function represents the sample evaluation index, namely:

[0051]

[0052] Among them, the power exponent γ i (i=1,2,3)∈(0,1), which can be adjusted according to the different contribution rates of each factor.

[0053] In this embodiment, the calculation of η(x) is a simple method to evaluate the degree to which the physical experimental data fills the space to quantitatively measure the coverage of the applicability field, which can be achieved in two steps, specifically: 1) Calculate the convex hull h of the domain defined by the physical experiment N (S), where the convex hull is the smallest convex domain containing all N-dimensional physical experimental data; 2) Calculate the volume of the convex hull and the volume of the total applicable range of the experiment h N (0), such as Figure 3 As shown, to generate the sample span metric η(x), for example, using the MATLAB built-in function convhulln.m to calculate the convex hull volume, the sample span index is:

[0054]

[0055] The distribution uniformity of the sample is evaluated using the second-order difference moment λ(x), namely:

[0056]

[0057] The smaller λ(x) is, the more uniform the data distribution is. i =[x i1 ,xi2 ,…,x ip ]∈[0,1] is the sample data after normalization, n is the sample data size, and p is the data dimension.

[0058] The redundancy of the data samples is evaluated by the root mean square value of the correlation coefficient ρ(x). Generally speaking, a smaller ρ(x) indicates low redundancy and high information quality. Specifically, ρ = 1 indicates that the data is completely correlated, while ρ = 0 indicates that the data is orthogonal, that is:

[0059]

[0060]

[0061]

[0062] In this embodiment, before optimizing and extracting the overall data, it is necessary to perform denoising on the data signal. Specifically, high-frequency and bad pixel data are filtered to improve the data analysis efficiency. The denoising process uses wavelet analysis, which is essentially to perform time-frequency domain analysis on the signal. Since most of the noise is high frequency, the original signal F is decomposed and reconstructed step by step. In this process, the high-frequency signal is removed, and finally the multi-level decomposed signal is reconstructed to obtain the denoised signal to achieve the denoising purpose. Figure 4 As shown, that is:

[0063]

[0064] Among them, L n is the low-frequency signal after the nth decomposition, H i (i=1,2,3…) is the high-frequency signal stripped step by step, and the signal F after denoising new After reconstruction, we get:

[0065]

[0066] In this embodiment, a discrete particle swarm algorithm is used to optimize and screen the time series parameters to obtain the elements of the best memory matrix. The optimization and screening of the time series sample parameters is performed on the sampling points on the discrete time series, which is essentially a discrete problem. The discrete particle swarm algorithm uses the code "1" and "0" to represent whether each sampling point of each parameter on the time series is selected, that is, the sample data is only a set of binary codes containing "0" and "1"; the sample evaluation index As the fitness function, its optimal value is used as the convergence condition to obtain the sample matrix elements containing valid information.

[0067] like Figure 5 As shown, consider the memory matrix, where n sSample data, the specific acquisition process is:

[0068] 11) Particle initialization

[0069] Initialize the population position p i , speed v i , the population number is m, among which, p i =[p i1 ,p i2 ,…,p in ] T ,i=1,2,…,m,p ij ∈{0,1}, the number of 1s encoded in each individual is limited to n s That is, the number of sample data selected.

[0070] 12) Particle Update

[0071] The position p of each individual i and speed v i will be updated as follows:

[0072]

[0073]

[0074] Among them, p bi is the optimal particle individual, p g To achieve the global optimum for the entire population, rand is used to generate random numbers between (0,1). is an S-type function, specifically:

[0075]

[0076] 13) Particle processing

[0077] When the total number of coded 1s in an individual is greater or less than n s , then adjust all particles so that the number of coded 1s in each iteration is equal to n s .

[0078] 14) Individual optimal position update

[0079] p i The individual fitness value of is:

[0080]

[0081] When f(p i ) <f(p bi ), then for p bi Use p i to update.

[0082] 15) Global optimal position update

[0083] When f(p bi ) <f(p g ), then for the global optimal position p g Use the current population's optimal particle p bi to update.

[0084] 16) When the maximum number of iterations or the minimum error is not reached, go to step 12); otherwise, go to step 17);

[0085] 17) Convert the optimal particles back to the selected samples for constructing the memory matrix.

[0086] In this embodiment, the multivariate state estimation (MSET) adopts a large information sample selected by a discrete particle swarm method to construct a large information sample matrix of each measuring point in the existing healthy state of the induced draft fan system, introduces a weight algorithm, estimates the future operating parameters of the induced draft fan, and monitors and warns of possible failures.

[0087] Considering that the characteristics of the induced draft fan are related to the vector composed of p characteristic variables, the operating state at time t is expressed as an observation vector:

[0088] x(t i )=x i =[x 1i ,x 2i ,…,x pi ] T

[0089] Then, n sets of observations are collected from the available normal operation data and the memory matrix is ​​constructed, denoted as D∈R p×n :

[0090]

[0091] Each column in D represents a specific time (from t1 to t n Each row represents the value of a specific variable (from 1 to p, with p = 25 measurement points). For a new observation vector x, the prediction vector can be calculated using the weighted average of the samples in the matrix.

[0092]

[0093] Where w=[w1,w2,…,w n ] T Represents the weight of each sample in the prediction matrix D. The weight vector can be determined by the square error between the actual observation value and its predicted value, that is:

[0094]

[0095] The minimum value of the least square error is obtained by the following formula:

[0096]

[0097] Therefore, the weight vector w is:

[0098] w=(D T D) -1 ·(D T x)

[0099] Dot product (D T D) is a linear operator, which may cause linear correlation of samples, which will reduce the calculation of the inverse matrix (D T D) -1 Therefore, a nonlinear Gaussian kernel operator can be used to replace the above linear operator and the weight vector can be modified as follows:

[0100] w=K -1 g(x)

[0101] Then we have:

[0102] K={K ij |i,j=1,2,…,n}, where K ij =K h (x i ,x j ),

[0103] g(x)=[g1(x),g2(x),…,g n (x)] T , where g i (x) = K h (x i ,x)

[0104] And, K h (x i ,x j ) is a Gaussian kernel operator, namely:

[0105]

[0106] in, For two vectors x i and x j The Euclidean distance of , h is the kernel bandwidth, which can generally be taken as 1.

[0107] The final MSET model is:

[0108]

[0109] When a new sample is obtained, a prediction can be made according to the above formula. If the induced draft fan is operating normally, the sample is very likely to be in the feature space described by the memory matrix, so the model will produce a prediction consistent with the accuracy of the memory matrix. On the contrary, if the induced draft fan fails, abnormal operating information is observed from the data sample, and the memory matrix will no longer be able to correctly describe its normal operating characteristics, resulting in a large deviation between the prediction result and the new sample.

[0110] In this embodiment, in order to ensure the completeness of the information required for early fault detection, the prediction deviation of all characteristic variables of the induced draft fan is integrated. Early faults are detected by measuring the total deviation, which is defined as the root mean square of the weighted sum of squares of the prediction errors, that is:

[0111]

[0112] Among them, the weights μ1,…μ p Used to eliminate the impact of error changes in different variables, it can be determined by the prediction residual of the sample:

[0113]

[0114] e k It is the average error of the validation set results of the kth variable in the prediction validation. That is, when a variable has a larger error level, it will occupy a smaller weight to balance the contribution of different variables.

[0115]

[0116]

[0117] κ∈[0,1] represents the similarity index of the health status of the induced draft fan. A larger κ means that the fan is operating normally, and a smaller κ means that the fan is in an abnormal state, which means that a fault may occur. Specifically:

[0118]

[0119] Where σ is the threshold value and the similarity index κ is a monotonically decreasing function of the deviation degree d. When the deviation degree d increases to the threshold value σ, κ decreases to the warning limit 1-δ, and the minimum value of κ is set to 0. The coefficient δ is related to the warning limit. For example, when δ = 0.3, the warning limit will be set to 0.7. If κ is less than 0.7, a failure is expected to occur.

[0120] The threshold σ can be determined based on the prediction residuals of the validation sample calculated by the established MSET model. To simplify, the prediction residuals are considered to be normally distributed with unknown mean and variance. The threshold σ is the upper limit of the mean value in the confidence interval with a confidence level of 1-α, that is:

[0121]

[0122] in, and s are the mean and variance of the deviation value of the validation sample, is a T-distribution with a confidence level of 1-α, and N is the capacity of the validation sample, where

[0123]

[0124]

[0125] Embodiment 2

[0126] The induced draft fan early fault detection system of the present invention comprises:

[0127] An acquisition module is used to acquire the historical health time series data of each measuring point, extract information from the historical health time series data of each measuring point, and construct an effective information memory matrix using the extracted information;

[0128] An estimation module, used for estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix;

[0129] The early warning module is used to compare the operating status of the system with a preset early warning criterion and to perform early fault detection and early warning of the induced draft fan according to the comparison result.

[0130] The division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present application may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0131] Embodiment 3

[0132] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the induced draft fan early fault detection method are implemented. The memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc. The processor, the network interface, and the memory are interconnected through an internal bus. The internal bus may be an industrial standard architecture bus, a peripheral component interconnection standard bus, an extended industrial standard architecture bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include a program code, and the program code includes computer operation instructions. The memory may include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0133] Embodiment 4

[0134] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the induced draft fan early fault detection method are implemented. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0135] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0136] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0137] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0138] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting early faults of an induced draft fan, characterized in that: The following steps are involved: Acquire historical health time series data of each measuring point, extract information from the historical health time series data of each measuring point, and construct an effective information memory matrix using the extracted information; estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix; The operating status of the system is compared with a preset warning criterion, and an early fault detection warning of the induced draft fan is performed according to the comparison result.

2. The method for detecting early failure of an induced draft fan according to claim 1, characterized in that: The process of extracting information from the historical health time series data of each measuring point is as follows: Information is extracted from the historical health time series data of each measuring point based on a discrete particle swarm algorithm.

3. The method for detecting early faults of induced draft fans according to claim 1, characterized in that: The measuring points include X-axis vibration, Y-axis vibration, induced draft fan front bearing temperature, induced draft fan middle bearing temperature, induced draft fan rear bearing temperature, motor front bearing temperature, motor rear bearing temperature, motor stator coil temperature, induced draft fan oil filter pressure difference, induced draft fan oil tank temperature, motor lubricating oil pressure and motor current.

4. The method for detecting early faults of induced draft fans according to claim 1, characterized in that: The process of extracting information from the historical health time series data of each measuring point based on the discrete particle swarm algorithm is as follows: The sample span η(x), distribution uniformity λ(x) and redundancy ρ(x) are combined and calculated respectively, and the sample evaluation index is introduced to represent the state information of the running data in the matrix. The function represents the sample evaluation index, namely: Among them, the power exponent γ i (i=1,2,3)∈(0,1); A discrete particle swarm algorithm is used to extract information from the historical health time series data of each measuring point according to the sample evaluation index.

5. The method for detecting early faults of induced draft fans according to claim 1, characterized in that: Before extracting information from the historical health time series data of each measuring point, the following step is also included: The historical health time series data of each measuring point is denoised.

6. The method for detecting early faults of induced draft fans according to claim 1, characterized in that: The process of estimating the operating state of the system using the multivariate state estimation method according to the effective information memory matrix is ​​as follows: According to the effective information memory matrix, a weight algorithm is introduced to estimate the future operating parameters of the induced draft fan system, wherein the estimated vector for: Where w=[w1,w2,…,w n ] T Represents the weight of each sample in the prediction matrix D, and x is a new observation vector.

7. The method for detecting early faults of induced draft fans according to claim 6, characterized in that: Early faults are detected based on the total deviation of the measurement, which is: Among them, k = 1, 2,…, p.

8. An induced draft fan early fault detection system, characterized in that: include: An acquisition module is used to acquire the historical health time series data of each measuring point, extract information from the historical health time series data of each measuring point, and construct an effective information memory matrix using the extracted information; An estimation module, used for estimating the operating state of the system using a multivariate state estimation method according to the effective information memory matrix; The early warning module is used to compare the operating status of the system with a preset early warning criterion and to perform early fault detection and early warning of the induced draft fan according to the comparison result.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the induced draft fan early fault detection method are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: The computer program, when executed by a processor, implements the steps of the induced draft fan early fault detection method.

Images