Diagnosis method and system for short-circuit port insulation state of electrolytic cell based on grey correlation method

Abstract

The application discloses a kind of based on grey correlation method's electrolytic cell short-circuit mouth insulation state diagnostic method and system, the present application method includes detecting the partial discharge signal of electrolytic cell short-circuit mouth, from partial discharge signal extraction partial discharge signal feature, partial discharge signal feature standardization processing and generation to-be-diagnosed feature vector;Based on grey correlation method calculates the grey correlation coefficient and grey correlation degree of to-be-diagnosed feature vector and pre-set multiple typical state standard feature vector;Find the typical state standard feature vector with the greatest grey correlation degree of to-be-diagnosed feature vector, and the insulation state of electrolytic cell short-circuit mouth corresponding to typical state standard feature vector is as the diagnostic result output corresponding to to-be-diagnosed feature vector.The present application can extract effective partial discharge feature, realize insulation state accurate grading, strong anti-interference capability, can solve the technical problems, such as existing diagnostic method lag, low accuracy, fuzzy classification.

Description

Technical Field

[0001] This invention relates to the field of insulation monitoring technology for electrolytic equipment, and specifically to a method and system for diagnosing the insulation status of short-circuit points in electrolytic cells based on the grey relational analysis method. Background Technology

[0002] Aluminum electrolytic cells are the core equipment in aluminum electrolytic production. The short-circuit port, as a critical connection point to the electrolytic cell busbar, directly determines the safe and stable operation of the entire electrolytic cell series. The short-circuit port of an aluminum electrolytic cell is subjected to extreme conditions for extended periods, including high temperatures (80~120℃), strong magnetic fields (≥100mT), high levels of metallic dust (aluminum, carbon, and alumina dust), and corrosion from electrolytic flue gas. This causes the insulation board to age and break, and dust to accumulate, leading to gradual deterioration of insulation performance. This can then trigger partial discharge, which, if not diagnosed and warned of in time, can further develop into surface flashover, "explosion," or even a major accident, causing huge economic losses and safety hazards. Currently, insulation diagnosis of aluminum electrolytic cell short-circuit ports mainly relies on manual inspection, infrared thermography, and offline insulation resistance testing, which have significant drawbacks: manual inspection is inefficient and subjective, failing to detect early insulation deterioration; infrared thermography can only detect localized overheating caused by mid-to-late-stage insulation deterioration, resulting in a delayed diagnosis; and offline insulation resistance testing cannot achieve real-time online monitoring and can only provide an overall assessment of the insulation status, lacking precise grading. Partial discharge detection is an effective means of early diagnosis of insulation degradation. By extracting partial discharge characteristic parameters, the degree of insulation degradation can be reflected. However, the aluminum electrolysis cell site is subject to strong electromagnetic interference and high dust levels, resulting in low accuracy of diagnosis based on a single partial discharge characteristic parameter. Furthermore, existing technologies lack precise grading methods for the insulation status of short circuits, making it impossible to effectively distinguish between the four typical states of normal, slight degradation, moderate degradation, and severe degradation. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for diagnosing the insulation status of electrolytic cells at short circuit points based on the grey relational analysis method, which can extract effective partial discharge features, achieve accurate insulation status classification, and has strong anti-interference ability, thereby solving the technical problems of lagging, low accuracy, and ambiguous classification in existing diagnostic methods.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for diagnosing the insulation status of an electrolytic cell short-circuit point based on grey relational analysis includes the following steps: S1, detecting partial discharge signals at the electrolytic cell short-circuit point, extracting partial discharge signal features from the partial discharge signals, standardizing the partial discharge signal features, and generating a feature vector to be diagnosed; S2, calculating the grey relational coefficient and grey relational degree between the feature vector to be diagnosed and several preset typical state standard feature vectors based on grey relational analysis; S3, finding the typical state standard feature vector with the highest grey relational degree to the feature vector to be diagnosed, and outputting the insulation status of the electrolytic cell short-circuit point corresponding to the typical state standard feature vector as the diagnostic result corresponding to the feature vector to be diagnosed.

[0005] Optionally, when extracting partial discharge signal features from the partial discharge signal in step S1, the partial discharge signal features include the initial discharge voltage Ust, the apparent discharge quantity q, the discharge frequency n, the distribution dispersion, and the nq distribution entropy H. n The term q can be a subset of the following: the initial discharge voltage Ust is the bus voltage corresponding to the occurrence of the first partial discharge pulse after the short-circuit bus voltage signal is acquired and processed by bandpass filtering and wavelet threshold denoising; the apparent discharge quantity q is the apparent discharge quantity calculated by combining the amplitude of the partial discharge pulse acquired by the high-frequency voltage sensor with the sensor calibration coefficient; the discharge frequency n is the total number of partial discharge pulses within a set time period, converted into the number of discharges per unit time; and the dispersion of the distribution is the standard deviation of the apparent discharge quantity q. The nq distribution entropy H nq Let be the Shannon entropy of the two-dimensional joint distribution of apparent discharge quantity q and discharge frequency n under a certain fixed initial discharge voltage Ust.

[0006] Optionally, in step S1, when standardizing the partial discharge signal features and generating the feature vector to be diagnosed, the standardization function expression for the initial discharge voltage Ust is: ; For apparent discharge quantity q, discharge frequency n, distribution dispersion and nq distribution entropy H n The standardized processing function expression for q is: ; in, Let be the original characteristic value of the j-th characteristic parameter of the i-th partial discharge signal. for Standardized eigenvalues The j-th feature parameter is the maximum value among all typical state standard feature vectors. The j-th feature parameter is the minimum value among all typical state standard feature vectors.

[0007] Optionally, step S2 may include constructing a variety of preset typical state standard feature vectors, which are typical state standard feature vectors composed of all standardized partial discharge signal features under four typical states: normal state, slightly deteriorated state, moderately deteriorated state, and severely deteriorated state.

[0008] Optionally, step S2, which calculates the grey relational coefficient and grey relational degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states based on the grey relational method, includes: S2.1, for various typical state standard feature vectors, calculate the absolute difference of each partial discharge signal feature under each typical state standard feature vector according to the following formula: Δ k (j)=|X 0k (j)-X i (j)|; Where, Δ k (j) represents the absolute difference between the j-th partial discharge signal feature of the feature vector to be diagnosed and the partial discharge signal feature in the preset k-th typical state standard feature vector, X 0k (j) The standardized eigenvalue of the j-th partial discharge signal feature under the standard eigenvector of the k-th typical state, X i (j) is the standardized feature value of the j-th partial discharge signal feature of the feature vector to be diagnosed; S2.2, calculate the maximum absolute difference Δmax and minimum absolute difference Δmin of the absolute differences of each partial discharge signal feature under the standard eigenvectors of all typical states; Δmax=max(max(Δ k (j))); Δmin=min(min(Δ k (j))); Where max represents the maximum value and min represents the minimum value; max(max(Δ k (j))) and min(min(Δ k In (j), the innermost layer takes the maximum or minimum value of all partial discharge signal characteristics, and the outermost layer takes the maximum or minimum value of all typical state standard feature vectors. S2.3, Calculate the grey relational coefficient: ξ k (j)=(Δmin+λ·Δmax) / (Δ k (j)+λ·Δmax); Where, ξ k (j) is the gray correlation coefficient between the j-th partial discharge signal feature of the feature vector to be diagnosed and the preset k-th typical state standard feature vector, and λ is the resolution coefficient; S2.4, Calculate the grey relational degree: r k =(1 / J)·Σξ k (j); Where, r k J represents the grey correlation degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states, and J is the number of partial discharge signal features.

[0009] Optionally, in step S3, when finding the typical state standard feature vector with the largest gray correlation degree with the feature vector to be diagnosed, the maximum value of the gray correlation degrees r1, r2, r3, and r4 between the feature vector to be diagnosed and the standard feature vectors of four typical states (normal state, slightly deteriorated state, moderately deteriorated state, and severely deteriorated state) is taken to obtain r. max If r max =r1, then the insulation state of the short circuit point is determined to be normal; if r max If r = r2, then it is determined to be a slightly deteriorated state; if r max If r = r3, then it is judged as a moderately deteriorated state; if r max If the value is r4, it is determined to be in a severely degraded state, and a fault warning signal will be issued.

[0010] Optionally, when detecting the partial discharge signal at the short circuit point of the electrolytic cell in step S1, the partial discharge signal at the short circuit point of the electrolytic cell is collected using a high-frequency differential voltage sensor. The bandwidth of the high-frequency differential voltage sensor is 10kHz~3MHz. During the collection process, electromagnetic shielding and wavelet denoising are used to suppress interference caused by the strong magnetic field of the aluminum electrolytic cell, rectified harmonics and metal dust.

[0011] The present invention also provides a short-circuit insulation status diagnosis system for electrolytic cells based on the grey relational analysis method, comprising a microprocessor and a memory interconnected thereto, wherein the microprocessor is programmed or configured to execute the short-circuit insulation status diagnosis method for electrolytic cells based on the grey relational analysis method.

[0012] The present invention also provides a computer-readable storage medium storing a computer program or instructions that are programmed or configured to execute the electrolytic cell short-circuit insulation status diagnosis method based on grey relational analysis by a processor.

[0013] The present invention also provides a computer program product, including a computer program or instructions, which are programmed or configured to execute, via a processor, the electrolytic cell short-circuit insulation state diagnosis method based on grey relational analysis. Compared with existing technologies, the present invention mainly achieves the following beneficial effects: Addressing the technical problems in existing aluminum electrolytic cell short-circuit insulation diagnosis, such as difficulty in extracting partial discharge signals under strong interference environments, ambiguous insulation status classification, low diagnostic accuracy, and inability to achieve early warning, the present invention provides an aluminum electrolytic cell short-circuit insulation status diagnosis method based on grey relational analysis. This method extracts key partial discharge features, constructs standard feature vectors, and uses grey relational analysis to achieve accurate insulation status classification. The grey relational analysis method has advantages such as low sample size requirements, strong anti-interference ability, and no need for explicit mathematical models. It can achieve multi-feature fusion diagnosis under complex operating conditions, improve diagnostic accuracy and early warning capability, and ensure the safe operation of the electrolytic cell. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the basic process of the method in an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions of the present invention will be further described in detail below with reference to the accompanying drawings in the embodiments of the present invention.

[0016] like Figure 1 As shown, the electrolytic cell short-circuit insulation state diagnosis method based on grey relational analysis in this embodiment includes the following steps: S1, detecting the partial discharge signal at the electrolytic cell short-circuit, extracting partial discharge signal features from the partial discharge signal, standardizing the partial discharge signal features, and generating a feature vector to be diagnosed; S2, calculating the grey relational coefficient and grey relational degree between the feature vector to be diagnosed and a variety of preset typical state standard feature vectors based on grey relational analysis; S3, finding the typical state standard feature vector with the largest grey relational degree to the feature vector to be diagnosed, and outputting the electrolytic cell short-circuit insulation state corresponding to the typical state standard feature vector as the diagnosis result corresponding to the feature vector to be diagnosed.

[0017] In step S1 of this embodiment, when detecting the partial discharge signal at the short-circuit port of the electrolytic cell, a high-frequency differential voltage sensor is used to collect the partial discharge signal at the short-circuit port. The bandwidth of the high-frequency differential voltage sensor is 10kHz~3MHz. During the acquisition process, electromagnetic shielding and wavelet denoising are used to suppress interference from the strong magnetic field, rectifier harmonics, and metal dust in the aluminum electrolytic cell. Specifically, in this embodiment, a high-frequency differential voltage sensor (bandwidth 10kHz~3MHz) is installed at the bus of the short-circuit port of the aluminum electrolytic cell to collect the partial discharge signal at the short-circuit port and the bus voltage signal. During the acquisition process, the signal is subjected to 300kHz center frequency bandpass filtering and wavelet threshold denoising (using sym4 wavelet, 7-level decomposition, soft thresholding) to suppress interference from the strong magnetic field, rectifier harmonics, and metal dust, eliminate interference pulses generated by the electrolytic cell itself, and retain the effective partial discharge signal.

[0018] In step S1 of this embodiment, when extracting partial discharge signal features from the partial discharge signal, the partial discharge signal features include the initial discharge voltage Ust, the apparent discharge quantity q, the discharge frequency n, the distribution dispersion, and the nq distribution entropy H. n q. Among them: (1) The initial discharge voltage Ust is the bus voltage corresponding to the first partial discharge pulse identified after the short-circuit bus voltage signal is collected and processed by bandpass filtering and wavelet threshold noise reduction; the unit is V; the more severe the insulation degradation, the lower the initial discharge voltage, because after the insulation performance deteriorates, a lower voltage can trigger partial discharge.

[0019] (2) Apparent discharge quantity q is the apparent discharge quantity obtained by combining the amplitude of partial discharge pulse collected by high frequency voltage sensor with the sensor calibration coefficient; the unit is pC; for example, the average value of apparent discharge quantity within 1 minute can be taken as this characteristic parameter. The larger the apparent discharge quantity, the more serious the insulation degradation.

[0020] (3) The discharge frequency n is the total number of partial discharge pulses within a set time period, which is converted into the number of discharges per unit time. Specifically, in this embodiment, the total number of partial discharge pulses within 1 minute is counted and converted into the number of discharges per unit time (per second), with the unit being times / second. The higher the discharge frequency, the more active the partial discharge and the higher the degree of insulation degradation.

[0021] (4) The dispersion of the distribution is the standard deviation of the apparent discharge quantity q. The standard deviation σ of the apparent discharge quantity is used to characterize the dispersion of the distribution, with the unit being pC. σ is calculated as the square root of the sum of squares of the deviations of all apparent discharge quantities from the mean within 1 minute. The larger σ is, the more dispersed the partial discharge pulse distribution is, and the more serious and uneven the insulation defect is.

[0022] (5) nq distribution entropy H nq Let H be the Shannon entropy of the two-dimensional joint distribution of apparent discharge quantity q and discharge frequency n under a fixed initial discharge voltage Ust. Construct a two-dimensional distribution spectrum of discharge frequency n and apparent discharge quantity q, and calculate the information entropy of this spectrum. nq The value range of H is 0~1; nq The larger the value, the more disordered the distribution of discharge frequency and discharge quantity, the more unstable the insulation condition, and the higher the degree of degradation. The apparent discharge quantity is... q Divided into k The interval, the first i indivual q The discharge frequency within the interval is n i The total number of discharges is: N total =∑ni Then the first i The frequency probability of interval discharge is: ; Define the entropy of the nq distribution: ; After normalization, we get: , Its value range: .

[0023] Specifically, in this embodiment, a high-frequency differential voltage sensor is used to acquire the partial discharge signal at the short circuit point. After filtering and noise reduction, five key characteristic parameters are extracted: Ust = 80V, q = 35pC, n = 3 times / second, σ = 8pC, H nq =0.2.

[0024] Because the five characteristic parameters have different dimensions (e.g., Ust is in V, q is in pC, and n is in times / second), direct grey relational analysis will produce errors. Therefore, it is necessary to standardize the five characteristic parameters of the collected samples to be diagnosed to eliminate the dimensional differences and bring each characteristic parameter to the same order of magnitude. In this embodiment, the standardization process uses a normalization method to eliminate the dimensional differences between different characteristic parameters. In step S1 of this embodiment, when standardizing the partial discharge signal features and generating the feature vector to be diagnosed, the standardization function expression for the initial discharge voltage Ust is: ; For apparent discharge quantity q, discharge frequency n, distribution dispersion and nq distribution entropy H n The standardized processing function expression for q is: ; in, Let be the original characteristic value of the j-th characteristic parameter of the i-th partial discharge signal. for Standardized eigenvalues The j-th feature parameter is the maximum value among all typical state standard feature vectors. The j-th feature parameter is the minimum value among all typical state standard feature vectors. In this embodiment, during standardization, the feature parameters are standardized, and the max(x) of each feature parameter is calculated. j ) and min(x j ): Ust: max=75, min=20; q: max=600, min=50; n: max=60, min=5; σ: max=70, min=10; H nq : max=0.9, min=0.3.

[0025] The feature vector to be diagnosed after standardization is: X i =[0.09, 0.027, 0.036, 0.033, 0.167].

[0026] Before step S2 in this embodiment, the method further comprises constructing a plurality of preset typical state standard feature vectors, which respectively include typical state standard feature vectors formed by all standardized partial discharge signal features under four typical states of normal state, slight deterioration state, moderate deterioration state and severe deterioration state. Through statistical analysis, fitting optimization of a large amount of test data of stub connections of 300~600kA aluminum reduction cells from multiple large domestic aluminum smelters, and in combination with on-site operation and maintenance experience, standard feature vectors of four typical insulation states of normal, slight deterioration, moderate deterioration and severe deterioration are constructed. Each state corresponds to a 5-dimensional standard feature vector composed of the five characteristic parameters described in step 1, and then a 4×5 standard feature matrix is formed. Specifically, the construction method of standard feature vectors for four typical insulation states in this embodiment is as follows: (1) Normal state: the insulation performance of the stub connection is good, no obvious partial discharge, the standard feature vector is [Ust1, q1, n1, , H nq1 , where Ust1≥75V, q1≤50pC, n1≤5 times / second, ≤10pC, H nq1 ≤0.3; (2) Slight deterioration state: a small amount of dust accumulates at the stub connection, slight partial discharge occurs, the standard feature vector is [Ust2, q2, n2, , H nq2 , where 50V≤Ust2<75V, 50pC<q2≤150pC, 5 times / second<n2≤20 times / second, 10pC< ≤30pC, 0.3<H nq2 ≤0.6; (3) Moderate deterioration state: a large amount of dust accumulates at the stub connection, the insulation performance decreases significantly, and partial discharge is active. The standard feature vector is [Ust3, q3, n3, , H nq3 , where 30V≤Ust3<50V, 150pC<q3≤500pC, 20 times / second<n3≤50 times / second, 30pC< ≤60pC, 0.6<H nq3≤0.8; (4) Severely deteriorated state: the insulation at the short circuit point is severely damaged, a large amount of dust accumulates, partial discharge is severe, and surface flashover is likely to occur. The standard eigenvector is [Ust4,q4,n4, H nq4 [Where Ust4 < 30V, q4 > 500pC, n4 > 50 cycles / second] >60pC, H nq4 >0.8; The parameter values ​​in the standard feature vector were obtained through statistical analysis and fitting optimization of partial discharge data under different insulation states at the short-circuit points of a large number of aluminum electrolytic cells. The standard feature vectors (before standardization) for four typical states are as follows: Normal state X 01 =[75,50,5,10,0.3]; Slight degradation X 02 =[60,100,15,20,0.45]; Moderate degradation X 03 =[40,250,35,45,0.7]; Severely degraded X 04 =[20,600,60,70,0.9].

[0027] Standardized reference sequence: X 01 =[0,0,0,0,0]; X 02 =[0.27,0.091,0.181,0.167,0.25]; X 03 =[0.636,0.363,0.545,0.583,0.667]; X 03 =[1,1,1,1,1].

[0028] In step S2 of this embodiment, the calculation of the grey relational coefficient and grey relational degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states based on the grey relational method includes: S2.1, for various typical state standard feature vectors, calculate the absolute difference of each partial discharge signal feature under each typical state standard feature vector according to the following formula: Δ k (j)=|X 0k (j)-X i (j)|; Where, Δ k (j) represents the absolute difference between the j-th partial discharge signal feature of the feature vector to be diagnosed and the partial discharge signal feature in the preset k-th typical state standard feature vector, X 0k(j) The standardized eigenvalue of the j-th partial discharge signal feature under the standard eigenvector of the k-th typical state, X i (j) represents the standardized eigenvalue of the j-th partial discharge signal feature in the feature vector to be diagnosed; the standard feature vectors of four typical states are used as the reference sequence. X 0k ( k =1, 2, 3, 4, corresponding to normal, slightly deteriorated, moderately deteriorated, and severely deteriorated, respectively. The reference sequence underwent the same standardization process as the sample to be diagnosed. The standardized feature vector of the sample to be diagnosed was used as the comparison sequence. X i , X i It is a 5-dimensional vector; S2.2, calculate the maximum absolute difference Δmax and minimum absolute difference Δmin of the absolute differences of each partial discharge signal feature under the standard eigenvectors of all typical states; Δmax=max(max(Δ k (j))); Δmin=min(min(Δ k (j))); Where max represents the maximum value and min represents the minimum value; max(max(Δ k (j))) and min(min(Δ k In (j), the innermost layer takes the maximum or minimum value of all partial discharge signal characteristics, and the outermost layer takes the maximum or minimum value of all typical state standard feature vectors. S2.3, Calculate the grey relational coefficient: ξ k (j)=(Δmin+λ·Δmax) / (Δ k (j)+λ·Δmax); Where, ξ k (j) is the gray correlation coefficient between the j-th partial discharge signal feature of the feature vector to be diagnosed and the preset k-th typical state standard feature vector. λ is the resolution coefficient, which ranges from 0.1 to 0.5. Considering the characteristics of the partial discharge feature data of the short circuit of the aluminum electrolytic cell, λ=0.3 is preferred, which can ensure the resolution accuracy and avoid interference. S2.4, Calculate the grey relational degree: r k =(1 / J)·Σξ k (j); Where, r kThe gray correlation degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states is given by J, where J is the number of partial discharge signal features. Here, j = 1 to 5. The gray correlation degree between the correlation coefficients of the five feature parameters and the k-th typical state is obtained by averaging these coefficients. Specifically, in this embodiment, the absolute difference Δ is calculated. k (j), take λ=0.3, calculate the correlation coefficient and correlation degree: Δ1(j)=[0.09,0.027,0.036,0.033,0.167], Δmax=1, Δmin=0; ξ1(j)=[0.769,0.917,0.892,0.9,0.642], r1=(0.769+0.917+0.892+0.9+0.642) / 5=0.824; similarly calculate r2=0.719, r3=0.483, r4=0.385.

[0029] In step S3 of this embodiment, when finding the typical state standard feature vector with the highest gray correlation degree with the feature vector to be diagnosed, the maximum value of the gray correlation degrees r1, r2, r3, and r4 between the feature vector to be diagnosed and the standard feature vectors of four typical states—normal state, slightly deteriorated state, moderately deteriorated state, and severely deteriorated state—is taken to obtain r. max If r max If r = r1, then the insulation status of the short circuit point is determined to be normal, and no maintenance measures are required; continuous monitoring is sufficient. If r max If r = r2, it is determined to be a slightly deteriorated state, and a warning is issued. Maintenance personnel need to strengthen the inspection of this short circuit point, clean up dust in a timely manner, and prevent further insulation deterioration; if r max If r = r3, it is determined to be a moderately deteriorated state, an abnormality warning is issued, a planned shutdown is arranged, the short-circuit insulation board is cleaned and inspected, and insulation components are replaced if necessary; if r max If r = r4, it is determined to be in a severely deteriorated state, and a fault warning signal is issued. The power supply to the electrolytic cell is immediately cut off, and emergency repairs are carried out to prevent serious accidents such as "blasting" or explosions. In one embodiment, since r max =r1=0.824, the insulation condition of the short circuit is determined to be normal. In another embodiment, the signal is acquired and the characteristic parameters are extracted: Ust=20V, q=350pC, n=55 times / second, σ=65pC, H n q=0.85. The same standard feature vector is used. The standardized feature vector to be diagnosed is X. i =[1,0.545,0.909,0.917,0.917]. Calculate the correlation coefficient and correlation degree: r1=0.265, r2=0.313, r3=0.423, r4=0.746. Final r max=r4=0.746, indicating that the insulation condition of the short circuit is severely deteriorated, an emergency warning is issued, and the machine is immediately shut down for emergency repairs.

[0030] In summary, the method of this embodiment has the following technical effects: 1. Highly targeted feature extraction and outstanding anti-interference capability: This embodiment targets the extreme working conditions of strong electromagnetic fields, high dust, and high temperature at the short-circuit point of the aluminum electrolysis cell. It uses a high-frequency differential voltage sensor for data acquisition, combined with bandpass filtering and wavelet denoising to effectively eliminate interference signals. The extracted five partial discharge characteristic parameters comprehensively reflect different dimensions of insulation degradation. From the five aspects of initial discharge voltage, discharge intensity, discharge activity, distribution characteristics, and disorder, it accurately depicts the changes in insulation state, solving the problem of inaccurate diagnosis by a single characteristic parameter. 2. Precise insulation state classification and early warning: Standard feature vectors for four typical states—normal, slightly deteriorated, moderately deteriorated, and severely deteriorated—are constructed, covering the entire cycle of insulation degradation, solving the problem of ambiguous insulation state classification in existing technologies. It can effectively identify slightly deteriorated states and provide early warning of insulation degradation 7 to 30 days in advance, providing maintenance personnel with sufficient time to handle the situation and avoiding serious accidents such as "explosions". 3. The diagnostic method is scientific and reliable with strong anti-interference capabilities: Employing grey relational analysis, it eliminates the need for complex mathematical models, requires a small sample size, and can adapt to fluctuations in characteristic parameters under complex operating conditions in aluminum electrolysis cells. It effectively suppresses the influence of interference signals, achieving a diagnostic accuracy of ≥95% and a false alarm rate of <5%, significantly improving accuracy compared to traditional diagnostic methods. 4. Simple engineering implementation and strong practicality: The feature extraction, standardization, and grey relational analysis methods in this embodiment are easily implemented using industrial-grade data acquisition hosts and software. Sensor installation is convenient, requiring no modification to the original structure of the aluminum electrolysis cell. It is compatible with various aluminum electrolysis cells ranging from 300 to 600 kA and can be directly applied to the upgrading and transformation of existing aluminum plants. The low cost facilitates large-scale promotion. 5. This embodiment can form a complete diagnostic closed loop: By outputting diagnostic results through the status judgment module, combined with the display and early warning module and data storage module, a complete closed loop of "acquisition → extraction → analysis → judgment → early warning → operation and maintenance" is achieved. This improves the intelligent level of short-circuit insulation operation and maintenance in aluminum electrolysis cells, extends the service life of insulation components by 30-50%, and reduces unplanned downtime.

[0031] Those skilled in the art will understand that the technical solutions provided by this invention can take the form of a method, system, or computer program product. Therefore, this invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this invention can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention can provide a system for diagnosing the insulation status of an electrolytic cell short-circuit based on the grey relational analysis method, including a microprocessor and a memory interconnected, the microprocessor being programmed or configured to execute the method for diagnosing the insulation status of an electrolytic cell short-circuit based on the grey relational analysis method. This invention can provide a computer-readable storage medium storing a computer program or instructions programmed or configured to execute the method for diagnosing the insulation status of an electrolytic cell short-circuit based on the grey relational analysis method via a processor. This invention can provide a computer program product including a computer program or instructions programmed or configured to execute the method for diagnosing the insulation status of an electrolytic cell short-circuit based on the grey relational analysis method via a processor. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0032] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for diagnosing the insulation status of a short-circuit point in an electrolytic cell based on grey relational analysis, characterized in that, The process includes the following steps: S1, detecting the partial discharge signal at the short-circuit point of the electrolytic cell, extracting the partial discharge signal features from the partial discharge signal, standardizing the partial discharge signal features, and generating a feature vector to be diagnosed; S2, calculating the gray correlation coefficient and gray correlation degree between the feature vector to be diagnosed and a variety of preset typical state standard feature vectors based on the gray relational method; S3, finding the typical state standard feature vector with the largest gray correlation degree with the feature vector to be diagnosed, and outputting the insulation state of the short-circuit point of the electrolytic cell corresponding to the typical state standard feature vector as the diagnostic result corresponding to the feature vector to be diagnosed.

2. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis as described in claim 1, characterized in that, In step S1, when extracting partial discharge signal features from the partial discharge signal, the partial discharge signal features include the initial discharge voltage Ust, the apparent discharge quantity q, the discharge frequency n, the distribution dispersion, and the nq distribution entropy H. n The term q can be a subset of the following: the initial discharge voltage Ust is the bus voltage corresponding to the occurrence of the first partial discharge pulse after the short-circuit bus voltage signal is acquired and processed by bandpass filtering and wavelet threshold denoising; the apparent discharge quantity q is the apparent discharge quantity calculated by combining the amplitude of the partial discharge pulse acquired by the high-frequency voltage sensor with the sensor calibration coefficient; the discharge frequency n is the total number of partial discharge pulses within a set time period, converted into the number of discharges per unit time; and the dispersion of the distribution is the standard deviation of the apparent discharge quantity q. The nq distribution entropy H nq Let be the Shannon entropy of the two-dimensional joint distribution of apparent discharge quantity q and discharge frequency n under a certain fixed initial discharge voltage Ust.

3. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis according to claim 2, characterized in that, In step S1, when standardizing the partial discharge signal features and generating the feature vector to be diagnosed, the expression for the standardization function for the initial discharge voltage Ust is: ； For apparent discharge quantity q, discharge frequency n, distribution dispersion and nq distribution entropy H n The standardized processing function expression for q is: ； in, Let be the original characteristic value of the j-th characteristic parameter of the i-th partial discharge signal. for Standardized eigenvalues The j-th feature parameter is the maximum value among all typical state standard feature vectors. The j-th feature parameter is the minimum value among all typical state standard feature vectors.

4. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis according to claim 1, characterized in that, Step S2 also includes constructing a variety of preset typical state standard feature vectors, which are typical state standard feature vectors composed of all standardized partial discharge signal features under four typical states: normal state, slightly deteriorated state, moderately deteriorated state, and severely deteriorated state.

5. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis according to claim 1, characterized in that, Step S2, which calculates the grey relational coefficient and grey relational degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states based on the grey relational method, includes: S2.1, for various typical state standard feature vectors, calculate the absolute difference of each partial discharge signal feature under each typical state standard feature vector according to the following formula: Δ k (j)=|X 0k (j)-X i (j)|; Where, Δ k (j) represents the absolute difference between the j-th partial discharge signal feature of the feature vector to be diagnosed and the partial discharge signal feature in the preset k-th typical state standard feature vector, X 0k (j) The standardized eigenvalue of the j-th partial discharge signal feature under the standard eigenvector of the k-th typical state, X i (j) is the standardized feature value of the j-th partial discharge signal feature of the feature vector to be diagnosed; S2.2, calculate the maximum absolute difference Δmax and minimum absolute difference Δmin of the absolute differences of each partial discharge signal feature under the standard eigenvectors of all typical states; Δmax=max(max(Δ k (j))); Δmin=min(min(Δ k (j))); Where max represents the maximum value and min represents the minimum value; max(max(Δ k (j))) and min(min(Δ k In (j), the innermost layer takes the maximum or minimum value of all partial discharge signal characteristics, and the outermost layer takes the maximum or minimum value of all typical state standard feature vectors. S2.3, Calculate the grey relational coefficient: x k (j)=(Δmin+λ·Δmax) / (Δ k (j)+λ·Δmax); Where, ξ k (j) is the gray correlation coefficient between the j-th partial discharge signal feature of the feature vector to be diagnosed and the preset k-th typical state standard feature vector, and λ is the resolution coefficient; S2.4, Calculate the grey relational degree: r k =(1 / J)·Sq k (j); Where, r k J represents the grey correlation degree between the feature vector to be diagnosed and the preset standard feature vectors of various typical states, and J is the number of partial discharge signal features.

6. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis according to claim 1, characterized in that, In step S3, when finding the typical state standard feature vector with the highest gray correlation degree with the feature vector to be diagnosed, the maximum value of the gray correlation degrees r1, r2, r3, and r4 between the feature vector to be diagnosed and the standard feature vectors of four typical states (normal state, slightly deteriorated state, moderately deteriorated state, and severely deteriorated state) is taken to obtain r. max If r max =r1, then the insulation state of the short circuit point is determined to be normal; if r max If r = r2, then it is determined to be a slightly deteriorated state; if r max If r = r3, then it is judged as a moderately deteriorated state; if r max If the value is r4, it is determined to be in a severely degraded state, and a fault warning signal will be issued.

7. The method for diagnosing the insulation status of an electrolytic cell short circuit based on grey relational analysis according to claim 1, characterized in that, In step S1, when detecting the partial discharge signal at the short circuit point of the electrolytic cell, a high-frequency differential voltage sensor is used to collect the partial discharge signal at the short circuit point of the electrolytic cell. The bandwidth of the high-frequency differential voltage sensor is 10kHz~3MHz. During the acquisition process, electromagnetic shielding and wavelet denoising are used to suppress interference caused by the strong magnetic field of the aluminum electrolytic cell, rectified harmonics and metal dust.

8. A short-circuit insulation status diagnostic system for electrolytic cells based on grey relational analysis, comprising a microprocessor and a memory interconnected, characterized in that, The microprocessor is programmed or configured to execute the electrolytic cell short-circuit insulation status diagnosis method based on grey relational analysis as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program or instructions, characterized in that, The computer program or instructions are programmed or configured to execute, via a processor, the electrolytic cell short-circuit insulation state diagnosis method based on grey relational analysis as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, The computer program or instructions are programmed or configured to execute, via a processor, the electrolytic cell short-circuit insulation state diagnosis method based on grey relational analysis as described in any one of claims 1 to 7.

Images