Method for detecting electrolytic cell ground insulation fault based on voltage phase

Abstract

This application relates to the technical field of insulation fault detection. Specifically, it discloses a method for detecting the insulation fault of an electrolytic cell to the ground based on voltage phase. First, an AC excitation signal is applied to each electrolytic cell in the electrolytic cell sequence, and at the same time, the state parameters of each electrolytic cell are collected. Then, a neural network model based on deep learning is used to perform embedded coding and global dynamic aggregation on the state parameters of each electrolytic cell to capture the overall operating state mode of the electrolytic cell sequence. Furthermore, taking the embedded representation of the state parameters of the electrolytic cell to be detected as a query feature, based on the semantic offset between it and the overall operating state mode of the electrolytic cell sequence, to intelligently identify whether the electrolytic cell to be detected has an insulation fault to the ground. In this way, accurate judgment of the insulation fault of the electrolytic cell to the ground can be achieved, and the adaptability to different working condition fluctuations and noise environments is improved, thereby effectively reducing the risk of misjudgment or missed judgment.

Description

Technical Field

[0001] The present application relates to the technical field of insulation fault detection, and more specifically, to a method for detecting insulation fault of an electrolytic cell to ground based on voltage phase. Background Art

[0002] The electrolytic cell is an important device used for electrochemical reactions in the chemical and metallurgical industries. Its performance is directly related to production efficiency and product quality. During operation, the insulation performance of the electrolytic cell to the ground is one of the necessary conditions to ensure safe operation and maintain efficient production. Once the electrolytic cell has an insulation failure to the ground, it may cause current leakage, affect the stability and safety of the electrolysis process, and may even cause a safety accident.

[0003] Traditional methods for detecting electrolytic cell insulation faults to the ground mainly include insulation resistance measurement method, leakage current detection method, etc. However, these methods have many limitations. The insulation resistance measurement method requires the electrolytic cell to be powered off for offline measurement, which cannot achieve real-time online monitoring, and the measurement results are easily affected by environmental factors and have low accuracy. The leakage current detection method has low detection sensitivity for tiny leakage currents, making it difficult to detect problems in time at the early stage of a fault.

[0004] In recent years, although some improved detection methods have emerged, for example, the invention patent with publication number CN116106697A discloses a voltage phase-based electrolytic cell insulation fault detection method, which mainly loads an AC excitation signal to the electrolytic cell, and synchronously measures the cell AC voltage of each electrolytic cell, the cell-to-ground AC voltage, the phase difference between the AC voltages, and the AC voltage difference, and determines whether there is an insulation fault to the ground based on the preset voltage setting value, pressure difference setting value, and phase difference setting value.

[0005] However, although this method achieves the purpose of fault detection to a certain extent, it mainly relies on fixed voltage setting values, pressure difference setting values ​​and phase difference setting values ​​for fault judgment. The analysis and processing of parameters are relatively limited, and the overall operating status mode of the electrolytic cell is not fully considered. It is easily affected by operating fluctuations and noise, resulting in misjudgment or missed judgment.

[0006] Therefore, an optimized voltage phase-based electrolytic cell insulation fault detection method is expected. Summary of the invention

[0007] To solve the above technical problems, the present application is proposed. An embodiment of the present application provides a method for detecting the ground insulation fault of an electrolytic cell based on voltage phase. First, an AC excitation signal is applied to each electrolytic cell in the electrolytic cell sequence, and the state parameters of each electrolytic cell are collected simultaneously. Then, a neural network model based on deep learning is used to perform embedding coding and global dynamic aggregation on the state parameters of each electrolytic cell to capture the overall operating state pattern of the electrolytic cell sequence. Furthermore, taking the embedding representation of the state parameters of the electrolytic cell to be detected as a query feature, based on the semantic offset between it and the overall operating state pattern of the electrolytic cell sequence, to intelligently identify whether the electrolytic cell to be detected has a ground insulation fault. In this way, an accurate judgment of the ground insulation fault of the electrolytic cell can be realized, and the adaptability to different working condition fluctuations and noise environments is improved, thereby effectively reducing the risk of misjudgment or missed judgment.

[0008] According to one aspect of the present application, there is provided a method for detecting the ground insulation fault of an electrolytic cell based on voltage phase, which includes:

[0009] Applying an AC excitation signal to each electrolytic cell in the electrolytic cell sequence and determining the state parameters of each electrolytic cell in the electrolytic cell sequence;

[0010] Performing vector quantization coding on the state parameters of each electrolytic cell to obtain a sequence of electrolytic cell state parameter embedding coding vectors;

[0011] Performing feature dynamic aggregation based on potential energy distribution on the sequence of electrolytic cell state parameter embedding coding vectors to obtain a significantly aggregated coding vector of the electrolytic cell sequence state parameters;

[0012] Extracting the electrolytic cell state parameter embedding coding vector of the electrolytic cell to be detected from the sequence of electrolytic cell state parameter embedding coding vectors as an electrolytic cell state parameter query coding vector;

[0013] Performing semantic offset measurement on the electrolytic cell state parameter query coding vector and the significantly aggregated coding vector of the electrolytic cell sequence state parameters to determine whether the electrolytic cell to be detected has a ground insulation fault.

[0014] Compared with the prior art, the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase provided by the present application first applies an AC excitation signal to each electrolytic cell in the electrolytic cell sequence, and simultaneously collects the state parameters of each electrolytic cell. Then, a neural network model based on deep learning is used to perform embedded coding and global dynamic aggregation on the state parameters of each electrolytic cell to capture the overall operating state pattern of the electrolytic cell sequence. Furthermore, taking the embedded representation of the state parameters of the electrolytic cell to be detected as a query feature, based on the semantic offset between it and the overall operating state pattern of the electrolytic cell sequence, to intelligently identify whether the electrolytic cell to be detected has a ground insulation fault. In this way, an accurate judgment of the ground insulation fault of the electrolytic cell can be achieved, and the adaptability to different working condition fluctuations and noise environments is improved, thereby effectively reducing the risk of misjudgment or missed judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of the present application will become more obvious. The accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a flowchart of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application.

[0017] Figure 2 It is a schematic diagram of data flow of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application.

[0018] Figure 3 It is a flowchart of sub-step S3 of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application.

[0019] Figure 4 It is a flowchart of sub-step S31 of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application.

[0020] Figure 5 It is a flowchart of sub-step S5 of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] As shown in this application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0022] Although this application makes various references to certain modules in the system according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative, and different aspects of the system and method can use different modules.

[0023] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0024] Next, example embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here.

[0025] It is worth noting that in this application, all actions of obtaining data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where it is located and obtaining the authorization given by the owner of the corresponding device.

[0026] As mentioned in the above background art, Patent CN116106697A proposes a method for detecting the ground insulation fault of an electrolytic cell based on voltage phase. It mainly loads an AC excitation signal to the electrolytic cell, synchronously measures the AC voltage of each electrolytic cell, the AC voltage between the cell and the ground, the phase difference between the AC voltages, and the AC voltage difference, and determines whether there is a ground insulation fault according to the preset voltage setting value, pressure difference setting value, and phase difference setting value.

[0027] Although the above-mentioned electrolytic cell ground insulation fault detection method based on voltage phase achieves the purpose of fault detection to a certain extent, it mainly relies on fixed voltage setting values, differential pressure setting values, and phase difference setting values for fault determination. The analysis and processing of parameters are relatively limited, and the overall operating state mode of the electrolytic cell is not fully considered, making it vulnerable to operating condition fluctuations and noise, resulting in misjudgment or missed judgment. To address this technical problem, this application proposes an optimized electrolytic cell ground insulation fault detection method based on voltage phase. First, an AC excitation signal is applied to each electrolytic cell in the electrolytic cell sequence, and the state parameters of each electrolytic cell are collected simultaneously. Then, a neural network model based on deep learning is used to perform embedded coding and global dynamic aggregation on the state parameters of each electrolytic cell to capture the overall operating state mode of the electrolytic cell sequence. Furthermore, using the embedded representation of the state parameters of the electrolytic cell to be detected as a query feature, based on the semantic shift between it and the overall operating state mode of the electrolytic cell sequence, to intelligently identify whether the electrolytic cell to be detected has a ground insulation fault. In this way, accurate judgment of the electrolytic cell ground insulation fault can be achieved, and the adaptability to different operating condition fluctuations and noise environments is improved, thus effectively reducing the risk of misjudgment or missed judgment.

[0028] Figure 1 FIG. is a flowchart of an electrolytic cell ground insulation fault detection method based on voltage phase according to an embodiment of the present application. Figure 2 FIG. is a schematic diagram of data flow of an electrolytic cell ground insulation fault detection method based on voltage phase according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the electrolytic cell ground insulation fault detection method based on voltage phase includes the steps of: S1, applying an AC excitation signal to each electrolytic cell in the electrolytic cell sequence and determining the state parameters of each electrolytic cell in the electrolytic cell sequence; S2, performing vector quantization coding on the state parameters of each electrolytic cell to obtain a sequence of electrolytic cell state parameter embedded coding vectors; S3, performing feature dynamic aggregation based on potential energy distribution on the sequence of electrolytic cell state parameter embedded coding vectors to obtain a significantly aggregated coding vector of the electrolytic cell sequence state parameters; S4, extracting the electrolytic cell state parameter embedded coding vector of the electrolytic cell to be detected from the sequence of electrolytic cell state parameter embedded coding vectors as an electrolytic cell state parameter query coding vector; S5, performing semantic shift measurement on the electrolytic cell state parameter query coding vector and the significantly aggregated coding vector of the electrolytic cell sequence state parameters to determine whether the electrolytic cell to be detected has a ground insulation fault.

[0029] In the above electrolytic cell ground insulation fault detection method based on voltage phase, in step S1, an AC excitation signal is applied to each electrolytic cell in the electrolytic cell sequence, and the state parameters of each electrolytic cell in the electrolytic cell sequence are determined. Among them, the state parameters include the cell AC voltage of each electrolytic cell, the cell-to-ground AC voltage of each electrolytic cell, the phase difference between the cell AC voltage and the cell-to-ground AC voltage of each electrolytic cell, and the AC voltage difference between the cell AC voltage of this electrolytic cell and the cell AC voltage of the next electrolytic cell. It should be understood that by applying an AC excitation signal, the dynamic behavior response of the electrolytic cell during actual operation can be more accurately simulated. In an ideal fault-free situation, the voltage and current inside the electrolytic cell are in phase or have a very small phase difference (depending on factors such as the specific design of the electrolytic cell, the materials used, and the operating conditions). At this time, the electrolytic cell maintains good insulation with the ground, and no abnormal current path will be generated. When a ground insulation fault occurs in the electrolytic cell, such as damage or aging of the insulation material, an additional path will be formed between the electrolytic cell and the ground. This path usually has capacitive or inductive characteristics, which will cause a new current path, thereby changing the original current distribution and phase relationship. Based on this, in this application, by collecting the cell AC voltage (i.e., the voltage across the two electrodes of the electrolytic cell), the cell-to-ground AC voltage (i.e., the voltage of the electrolytic cell relative to the ground), the phase difference between the cell AC voltage and the cell-to-ground AC voltage, and the AC voltage difference between the cell AC voltage of this electrolytic cell and the cell AC voltage of the next electrolytic cell of each electrolytic cell, it is possible to evaluate whether the electrolytic cell has a ground insulation fault. Among them, the cell AC voltage directly reflects the energy input during the electrolysis process and is one of the basic conditions to ensure the normal progress of the electrolysis reaction. The cell-to-ground AC voltage reveals the insulation status between the electrolytic cell and the ground. The phase difference between the cell AC voltage and the cell-to-ground AC voltage reflects the impedance characteristics in the circuit and helps to analyze whether there are abnormal capacitive or inductive paths in the circuit. The AC voltage difference can provide the voltage balance between adjacent electrolytic cells and helps to identify whether there are local short circuits or grounding faults. Through comprehensive analysis of the above state parameters, the insulation status of the electrolytic cell can be more comprehensively evaluated, thereby improving the accuracy of fault detection.

[0030] To ensure the accuracy and real-time performance of the electrolytic cell ground insulation fault detection, it is necessary to accurately collect the cell AC voltage, the cell-to-ground AC voltage, the phase difference between the voltages of each cell, and the AC voltage difference between adjacent cells of each electrolytic cell. The following is a detailed description of the specific collection method:

[0031] First, in terms of hardware selection and installation, sensors with high precision and suitable for industrial environments need to be selected to measure the required voltage parameters. For the measurement of the slot AC voltage and the slot-to-ground AC voltage, voltage dividers or current transformers and isolated voltage sensors are used respectively to ensure safety and reduce the impact of electromagnetic interference. Phase difference measurement relies on phase detection devices such as digital phase-locked loops (DLLs) or other dedicated equipment, while the voltage difference between adjacent slots is directly measured by a differential amplifier to ensure signal linearity and accuracy. According to the actual arrangement of the electrolytic cells, plan the sensing network to ensure that there is enough space around each electrolytic cell to arrange the sensors, and the connection lines are as short as possible to reduce noise introduction. All signal transmission lines are shielded, and communication protocols with strong anti-interference capabilities (such as RS485, CAN bus) are selected. When necessary, fiber optic transmission is considered to avoid the influence of strong electric fields. Install filters between the sensors and the data acquisition platform to filter out high-frequency noise and other irrelevant components and improve the signal quality.

[0032] In the actual installation process, the hardware selection is not limited to the several types mentioned above. Considering the different requirements of different application scenarios, customized selection may be required for specific conditions. For example, for electrolytic cells operating in high-temperature environments, sensors made of high-temperature-resistant materials should be selected; while in corrosive environments, products with anti-corrosion designs are needed. In addition, to ensure long-term stable operation, the sensors should have good reliability and long-life characteristics and be able to withstand harsh working conditions without being easily damaged. Attention should also be paid to the position layout of the sensors during installation to ensure that they can effectively cover the target measurement area without interfering with the operation and maintenance of other equipment. For electrolytic cells in special positions such as narrow spaces, miniaturized or flexible sensors may be needed to adapt to complex structures. At the same time, grounding treatment during installation is very important, and all sensors should be correctly grounded to avoid introducing unnecessary noise and improve the accuracy of the measurement results.

[0033] Next, in the design of the data acquisition system, build a centralized or distributed data acquisition platform equipped with high-performance processors and large-capacity storage media to ensure the ability to handle the needs of a large number of high-frequency samplings. This platform supports various types of input interfaces (analog, digital, etc.), facilitates the access of different types of sensors, and has good scalability to adapt to possible future increased monitoring points. Simultaneous sampling of all channels is the key. Ensure time consistency through hardware triggering or software timing, which is particularly important for calculating the phase difference. Set reasonable sampling frequencies and periods. For parameters with large transient changes such as the phase difference, use a higher sampling rate; while for relatively stable parameters, the sampling frequency can be appropriately reduced to save resources.

[0034] When building a data acquisition platform, in addition to considering performance and compatibility, attention should also be paid to the reliability and redundancy design of the system. To prevent the entire system from failing due to a single point of failure, backup units can be set up at important nodes, such as backup power supplies, redundant controllers, etc. In addition, adopting a modular architecture can make the system easier to maintain and upgrade. When a component has a problem, it can be quickly replaced without affecting the overall function. The data transmission path should also have a certain degree of fault tolerance. For example, media with good anti-interference characteristics such as twisted pairs or optical fibers can be used, and an automatic retransmission mechanism can be configured to handle occasional data loss situations. For occasions with strong remote monitoring requirements, a wireless communication module can also be deployed, but it is necessary to ensure that it meets industrial-grade standards and provides sufficient bandwidth and security guarantees.

[0035] Enter the data acquisition process stage. Before starting the formal data acquisition, perform system initialization, including calibrating all sensors, checking the integrity of the connection lines, configuring the parameters of the data acquisition platform, etc., to ensure that everything is ready. Conduct preliminary tests to verify that the sensors are working properly and that the data acquisition platform can correctly receive and record data. Adjust any problems found until the optimal state is reached. After starting the data acquisition program, read data from each sensor at a predetermined time interval and store it in the local database. At the same time, add a timestamp mark to each record for tracking the change trend during subsequent analysis. Regularly back up the collected data to prevent data loss due to unexpected situations. For important data, it can also be uploaded to the cloud for off-site storage to further ensure data security. Establish a real-time monitoring interface to display information such as the AC voltage of each electrolytic cell, the AC voltage between the cell and the ground, the phase difference, and the voltage difference between adjacent cells, enabling operators to intuitively understand the system operation status. Set alarm thresholds so that an alarm is automatically issued when a certain parameter exceeds the set range, reminding relevant personnel to handle potential problems in a timely manner to avoid the expansion of faults.

[0036] In the initialization stage, in addition to basic hardware calibration and connection checks, a comprehensive evaluation of the synchronization performance of the entire system is also required. Since multi-channel simultaneous sampling is involved, even a tiny time deviation may affect the accuracy of the final calculation results. Therefore, it is recommended to use a precision clock source to uniformly control the sampling moments of all sensors to ensure data consistency and reliability. In addition, various configuration parameters of the data acquisition platform, such as the sampling frequency, filtering algorithm, etc., need to be optimized and adjusted to find the best combination suitable for the current application scenario. This not only helps improve data quality but also effectively extends the service life of the system. During the initial operation period, closely monitor the changes in various indicators, record the possible problems and their solutions, and accumulate experience for subsequent routine maintenance.

[0037] In the process of continuous data acquisition, it is crucial to maintain an efficient and stable data stream. To this end, a complete set of data management strategies should be established, covering all aspects from raw data acquisition to final application. First, in the front-end acquisition link, ensure that each sensor can stably output high-quality data according to the preset frequency. Second, in the intermediate transmission link, optimize the network topology to reduce latency and packet loss. Finally, in the back-end storage link, adopt efficient compression algorithms and indexing technologies to save storage space and facilitate rapid retrieval. In this way, not only can the overall quality of the data be improved, but it also helps to detect potential hidden dangers in a timely manner and take preventive measures in advance. At the same time, as the amount of data grows, regularly clean up expired or useless data to ensure that the system is always in the best working condition.

[0038] In summary, through the above integrated method, it is possible to effectively achieve the accurate collection of the cell AC voltage, cell-to-ground AC voltage, phase difference, and adjacent cell voltage difference of each electrolytic cell, providing reliable data support for the safe and stable operation of the electrolytic cell.

[0039] In the above method for detecting the ground insulation fault of an electrolytic cell based on voltage phase, in step S2, the state parameters of each electrolytic cell are vectorized and encoded to obtain a sequence of electrolytic cell state parameter embedded encoding vectors. It should be understood that in this application, considering that the original state parameters (such as voltage, phase difference, etc.) are usually multi-dimensional data existing in different data scales and units, therefore, in order to achieve the comprehensive processing and analysis of multi-dimensional data, this application further uses a state parameter embedded encoding matrix to vectorize and encode the state parameters of each electrolytic cell, so as to map parameters of different scales and units to a common feature space, discover the internal connections and patterns between multi-dimensional state parameters, and convert the multi-dimensional heterogeneous data of each electrolytic cell into low-dimensional dense vector representations respectively, thereby obtaining a sequence of electrolytic cell state parameter embedded encoding vectors, so as to achieve the comprehensive characterization of the operating state of each electrolytic cell.

[0040] In the above method for detecting the ground insulation fault of an electrolytic cell based on voltage phase, in step S3, feature dynamic aggregation based on potential energy distribution is performed on the sequence of electrolytic cell state parameter embedded encoding vectors to obtain a significantly aggregated encoding vector of the electrolytic cell sequence state parameters. It should be understood that considering that an electrolytic cell is a complex industrial system, its operating state not only depends on the behavior of a single electrolytic cell, but also needs to consider the collaborative working mode of the entire electrolytic cell sequence. Therefore, this application further performs feature aggregation analysis on the sequence of electrolytic cell state parameter embedded encoding vectors to capture the interactions between each electrolytic cell in the electrolytic cell sequence and the overall operating law, so as to achieve the global modeling of the operating state of the entire electrolytic cell sequence and provide a more comprehensive perspective for the accurate identification of the ground insulation fault of the electrolytic cell. Among them, Figure 3It is a flowchart of sub-step S3 of the electrolytic cell ground insulation fault detection method based on voltage phase according to an embodiment of the present application. As Figure 3 shown, the step S3 includes steps: S31, calculating a pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters based on the characteristic static potential energy distribution of the sequence of the electrolytic cell state parameter embedded coding vectors; S32, calculating the aggregation movement direction of each electrolytic cell state parameter embedded coding vector in the sequence of the electrolytic cell state parameter embedded coding vectors relative to the pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters to obtain a sequence of the electrolytic cell state parameter aggregation movement directions; S33, based on the sequence of the electrolytic cell state parameter aggregation movement directions, dynamically aggregating the sequence of the electrolytic cell state parameter embedded coding vectors towards the pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters to obtain the significant aggregation coding vector of the electrolytic cell sequence state parameters.

[0041] Figure 4 It is a flowchart of sub-step S31 of the electrolytic cell ground insulation fault detection method based on voltage phase according to an embodiment of the present application. As Figure 4 shown, the step S31 includes steps: S311, respectively inputting each electrolytic cell state parameter embedded coding vector in the sequence of the electrolytic cell state parameter embedded coding vectors into a characteristic static potential energy measurement network to obtain a sequence of the electrolytic cell state parameter static potential energy measurement coefficients; S312, inputting the sequence of the electrolytic cell state parameter static potential energy measurement coefficients into a characteristic energy level screening gate unit to obtain a sequence of the electrolytic cell state parameter static potential energy weight factors; S313, based on the sequence of the electrolytic cell state parameter static potential energy weight factors, calculating the weighted sum of the sequence of the electrolytic cell state parameter embedded coding vectors to obtain the pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters.

[0042] More specifically, in a specific example of the present application, the step S311 includes: performing Z-score standardization processing on the electrolytic cell state parameter embedded coding vector to obtain a standardized electrolytic cell state parameter embedded coding vector; calculating the sum of the cubes of each eigenvalue in the standardized electrolytic cell state parameter embedded coding vector divided by the characteristic scale value of the standardized electrolytic cell state parameter embedded coding vector to obtain the electrolytic cell state parameter static potential energy measurement coefficient, which is expressed by the formula:

[0043]

[0044]

[0045] where represents the sequence of the electrolytic cell state parameter embedded coding vectors, , , and respectively represent the first, second, th, and th electrolytic cell state parameter embedded coding vectors in the sequence of the electrolytic cell state parameter embedded coding vectors, is the number of feature vectors in the sequence of the electrolytic cell state parameter embedded coding vectors, represents the th eigenvalue at the th position in the th electrolytic cell state parameter embedded coding vector, and respectively represent the feature mean and feature variance of the th electrolytic cell state parameter embedded coding vector, is the feature scale value of the electrolytic cell state parameter embedded coding vector, is the static potential energy metric coefficient of the

[0046] That is, by measuring the potential energy levels of the electrolytic cell state parameter embedded coding vectors under static conditions, the importance and influence of the characteristics of each electrolytic cell state parameter in the global state parameter characteristics distribution of the electrolytic cell sequence, as well as the internal correlation between each electrolytic cell, are revealed, so as to obtain the sequence of the static potential energy metric coefficients of the electrolytic cell state parameters.

[0047] More specifically, the step S312 is expressed by the formula:

[0048]

[0049]

[0050] where represents the th normalized electrolytic cell state parameter static potential energy metric coefficient, represents the gating mask function, represents the gating threshold, represents the th electrolytic cell state parameter static potential energy weight factor.

[0051] That is, based on the gating mechanism, the sequence of the obtained electrolytic cell state parameter static potential energy metric coefficients is subjected to gating mask processing to screen out the individual electrolytic cell state characteristics that contribute more to the representation of the overall operating state of the electrolytic cell, while suppressing the individual electrolytic cell state characteristics with less contribution, and generating the corresponding weight factors. In this way, the weights of the characteristics of each electrolytic cell state parameter can be dynamically adjusted, enabling the model to pay more attention to the feature representations that have a greater impact on the overall operating state.

[0052] More specifically, the step S313 can be expressed by the formula:

[0053]

[0054] where represents the pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters.

[0055] That is, based on the obtained weight factors, the sequence of the electrolyzer state parameter embedding and encoding vectors is weighted and aggregated to reveal the characteristic distribution trend center of the electrolyzer sequence operation state, and a pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters is generated, thereby providing effective guidance for the subsequent feature aggregation process.

[0056] Specifically, in the step S32, the aggregation movement direction of each electrolyzer state parameter embedding and encoding vector in the sequence of the electrolyzer state parameter embedding and encoding vectors relative to the pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters is calculated to obtain a sequence of electrolyzer state parameter aggregation movement directions. Among them, the electrolyzer state parameter aggregation movement direction is the arccosine function value between the electrolyzer state parameter embedding and encoding vector and the pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters, and is expressed by the formula:

[0057]

[0058] where represents the norm of the vector, is the arccosine function, represents the th aggregation movement direction of the electrolyzer state parameter embedding and encoding vector relative to the pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters.

[0059] That is, by calculating the aggregation movement direction of each electrolyzer state parameter embedding and encoding vector relative to the pseudo-anchored aggregation center representation vector of the electrolyzer sequence state parameters, the change trend of each electrolyzer state feature relative to the center point is quantified, and the dynamic relationship and its evolution path of each electrolyzer state parameter feature in the global feature space are revealed, so as to facilitate the identification of the relative relationship between features and the group behavior pattern.

[0060] Specifically, the step S33 can be expressed by the formula:

[0061]

[0062] where represents the sine function, and represent the weight matrix and the bias term respectively, Represents the significant aggregation coding vector of the electrolytic cell sequence state parameters.

[0063] That is, based on the aggregation movement direction and characteristic static potential energy metric coefficient of the embedded coding vectors of each electrolytic cell state parameter, the embedded coding vectors of each electrolytic cell state parameter are dynamically aggregated towards the pseudo-anchoring aggregation center representation vector of the electrolytic cell sequence state parameters to obtain the significant aggregation coding vector of the electrolytic cell sequence state parameters, thereby realizing the accurate description of the overall operation state mode of the electrolytic cell sequence.

[0064] In the above method for detecting the ground insulation fault of an electrolytic cell based on voltage phase, in step S4, the embedded coding vector of the electrolytic cell state parameter of the electrolytic cell to be detected is extracted from the sequence of the embedded coding vectors of the electrolytic cell state parameters as the query coding vector of the electrolytic cell state parameter. It should be understood that considering that in a large electrolytic cell sequence, the performance of each electrolytic cell may be different. Therefore, for a more accurate individual evaluation of the electrolytic cell, the present application further extracts the embedded coding vector of the electrolytic cell state parameter of the electrolytic cell to be detected from the sequence of the embedded coding vectors of the electrolytic cell state parameters, and uses it as the query coding vector of the electrolytic cell state parameter. By comparing and analyzing it with the overall operation state mode of the electrolytic cell sequence, the performance of the electrolytic cell to be detected relative to the entire sequence is determined. In this way, by taking the overall operation state mode of the electrolytic cell sequence as a reference, the influence of factors such as working condition fluctuations and noise interference on the operation state of a single electrolytic cell can be effectively avoided, thereby improving the accuracy of fault detection.

[0065] In the above method for detecting the ground insulation fault of an electrolytic cell based on voltage phase, in step S5, a semantic offset metric is performed on the query coding vector of the electrolytic cell state parameter and the significant aggregation coding vector of the electrolytic cell sequence state parameters to determine whether the electrolytic cell to be detected has a ground insulation fault. Among them, Figure 5 FIG. is a flowchart of sub-step S5 of the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase according to an embodiment of the present application. As Figure 5 shown, step S5 includes steps: S51, calculating the semantic offset degree between the query coding vector of the electrolytic cell state parameter and the significant aggregation coding vector of the electrolytic cell sequence state parameters; S52, determining whether the electrolytic cell to be detected has a ground insulation fault based on the comparison between the semantic offset degree and a preset threshold.

[0066] Specifically, in step S51, the semantic offset degree between the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector is calculated. In a specific example of the present application, the semantic offset degree is the Euclidean distance value between the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector. It should be understood that the electrolyzer sequence state parameter significant aggregation coding vector reveals the group behavior pattern and average level of the electrolyzer sequence, while the electrolyzer state parameter query coding vector represents the individual state of the detected electrolyzer. By calculating the semantic offset degree between the two, a direct comparison can be established between the individual level and the group level, and the difference degree between the detected electrolyzer and the overall sequence state can be quantitatively represented, thus helping to identify the electrolyzers that deviate from the normal operation mode.

[0067] Here, when the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector respectively represent the semantic significant aggregation coding features of the state parameters of all electrolyzers and the low-dimensional semantic embedding coding features of the state parameters of a single electrolyzer, when calculating the semantic offset degree between the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector, the inconsistent feature density and feature order between the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector will cause a difference in the feature distribution space structure of the electrolyzer state parameter query coding vector relative to the electrolyzer sequence state parameter significant aggregation coding vector, affecting the calculation accuracy of the semantic offset degree.

[0068] Therefore, in a preferred example of the present application, calculating the semantic offset degree between the electrolyzer state parameter query coding vector and the electrolyzer sequence state parameter significant aggregation coding vector includes:

[0069] Optimizing the electrolyzer state parameter query coding vector, and the optimization process includes:

[0070] Calculating the square root of the sum of the absolute values and the sum of the squares of all eigenvalues of the electrolyzer state parameter query coding vector to obtain the first electrolyzer state parameter query coding space structure value and the second electrolyzer state parameter query coding space structure value, that is:

[0071]

[0072]

[0073] Wherein, represents the th eigenvalue of the electrolyzer state parameter query coding vector, and respectively represent the first electrolyzer state parameter query coding space structure value and the second electrolyzer state parameter query coding space structure value;

[0074] Calculate the sum value of the first electrolyzer state parameter query coding space structure value and the second electrolyzer state parameter query coding space structure value as the explicit metric value of the electrolyzer state parameter query coding space structure;

[0075] Multiply the characteristic mean value of the electrolyzer state parameter query coding vector by the explicit metric value of the electrolyzer state parameter query coding space structure to obtain the electrolyzer state parameter query coding statistical field value , where, represents the characteristic mean value of the electrolyzer state parameter query coding vector;

[0076] Subtract one from the electrolyzer state parameter query coding statistical field value and then divide it by the electrolyzer state parameter query coding statistical field value to obtain the explicit metric value of the partial differential space structure of the electrolyzer state parameter query coding ;

[0077] Calculate the power function of the electrolyzer state parameter query coding vector with the explicit metric value of the partial differential space structure of the electrolyzer state parameter query coding as the exponent , and multiply it by the explicit metric value of the partial differential space structure of the electrolyzer state parameter query coding to obtain the microscopic representation vector of the electrolyzer state parameter query coding , where, represents dot product;

[0078] After dot multiplying the electrolyzer state parameter query coding vector with the explicit metric value of the partial differential space structure of the electrolyzer state parameter query coding, calculate the exponential function with the natural constant as the base to obtain the macroscopic mapping vector of the electrolyzer state parameter query coding ;

[0079] After calculating the base-2 logarithm value of the microscopic representation vector of the electrolyzer state parameter query coding, perform weighted summation with the macroscopic mapping vector of the electrolyzer state parameter query coding to obtain the optimized electrolyzer state parameter query coding vector , where, and represent weighted hyperparameters, represents point addition. Finally, calculate the semantic offset degree between the optimized electrolyzer state parameter query coding vector and the significant aggregation coding vector of the electrolyzer sequence state parameters.

[0080] That is, by querying the partial low-order derivative of the statistical distribution field corresponding to the encoding vector of the electrolytic cell state parameters, it is used as the non-overlapping macroscopic feature representation behavior patch of the encoding vector of the electrolytic cell state parameters. Based on the different macroscopic behavior patch organization spaces under the anisotropic backbone structure of the encoding vector of the electrolytic cell state parameters, the dynamic sensitivity of the long-sequence microscopic complex information distribution of the encoding vector of the electrolytic cell state parameters to the class-probability macroscopic representation behavior is strengthened. In this way, the consistency of the feature distribution space structure of the encoding vector of the electrolytic cell state parameters relative to the significantly aggregated encoding vector of the electrolytic cell sequence state parameters is improved, so as to improve the calculation accuracy of the semantic deviation degree.

[0081] Specifically, in step S52, based on the comparison between the semantic deviation degree and a preset threshold, it is determined whether the detected electrolytic cell has a ground insulation fault. Specifically, if the semantic deviation degree exceeds the preset threshold, it indicates that there is a significant difference between the operating state of the detected electrolytic cell and the average state of the overall sequence. Then it can be considered that the electrolytic cell has a ground insulation fault and further inspection and maintenance are required. On the contrary, if the semantic deviation degree is lower than the preset threshold, it means that the operating state of the detected electrolytic cell conforms to the average state of the overall sequence and belongs to the normal operating fluctuation range. Through this method, the insulation condition of the electrolytic cell can be effectively monitored in real time and fault early warning can be carried out, thus ensuring the safety and stability of the electrolysis production process.

[0082] After confirming that the electrolytic cell has a ground insulation fault, it is crucial to take prompt and effective measures. The entire processing flow needs to ensure safety, reduce losses and resume normal production as soon as possible. The following is a detailed description of the specific implementation methods of the measures:

[0083] First of all, once it is confirmed that the electrolytic cell has a ground insulation fault, immediately start the emergency shutdown procedure, cut off the power supply, prevent the current from leaking continuously, and avoid possible safety accidents. At the same time, promptly notify the on-site operators, maintenance team and management to ensure that all relevant parties are aware of the situation and are ready to take further actions. To ensure safety, use isolation belts or other physical barriers to isolate the area of the faulty electrolytic cell from other work areas to prevent unauthorized personnel from entering the dangerous area; and set warning signs and indicator plates at obvious positions around the faulty electrolytic cell to remind all personnel to pay attention to safety and stay away from the faulty area.

[0084] Next, while performing an emergency shutdown, professional technicians conduct a preliminary assessment to determine the severity of the fault and its impact scope on surrounding equipment. This step helps in formulating the next repair plan. During this process, record information such as the time and location of the fault occurrence, the specific number and serial numbers of the electrolytic cells involved, etc., and save all relevant monitoring data (such as cell AC voltage, cell-to-ground AC voltage, phase difference, and voltage difference between adjacent cells) as the basis for subsequent analysis. The collection of these detailed data not only provides solutions for the current problem but also serves as a basis for future preventive measures. Technicians also need to check the environmental conditions around the electrolytic cells, such as temperature, humidity, etc., to rule out the influence of external factors on fault judgment.

[0085] With the completion of the preliminary assessment, the next is the in-depth investigation phase, aiming to understand the fault cause more comprehensively. This phase includes detailed physical inspections, such as visual inspections, tightness of connection parts, etc., and insulation resistance measurements using professional instruments (such as megohmmeters) to directly verify the ground insulation performance of the electrolytic cells. In addition, historical data and trend charts should be reviewed to look for long-term patterns or abnormal fluctuations that may have caused the fault. Through multi-source data comparison, compare the data change trends at different time points and locations to identify abnormal fluctuations or deviations from the normal range. If it is found that a certain parameter has been in an unstable state for a long time or frequently shows abnormal values, it may be an early sign of a ground insulation fault.

[0086] Based on the results of the preliminary assessment and in-depth investigation, formulate a detailed repair plan, including the required tools, materials, schedule, and list of personnel involved. Ensure that the plan takes into account all potential risk points and corresponding preventive measures are formulated. When carrying out the repair work according to the established plan, focus on repairing or replacing damaged components, such as insulation materials, connectors, etc. Strictly follow safety regulations during the repair process to ensure the safety of the operating personnel. For complex or large-scale repair tasks, they can be implemented in phases to gradually restore the function of the electrolytic cells. After each stage of work is completed, conduct a comprehensive function test, including but not limited to re-measuring the insulation resistance, checking the firmness of electrical connections, etc., to ensure that the performance of the repaired electrolytic cells meets the standard requirements and there are no safety hazards. In addition, detailed repair records should be kept during the repair process for future reference.

[0087] After the repair work is completed, a root cause analysis must be carried out to deeply explore the root causes leading to this failure. Tools such as Failure Mode and Effects Analysis (FMEA) and fishbone diagrams are used to assist in the analysis to find out the problems. Based on the analysis results, specific improvement measures are proposed, such as optimizing the design, strengthening daily maintenance, and updating aging equipment, etc., to prevent similar failures from occurring again. In response to the problems exposed in this incident, organize employees for special training to improve the safety awareness and technical level of all employees, and ensure that everyone can correctly handle emergencies. In addition, establish a regular inspection and maintenance system, strengthen the monitoring of key equipment, discover and solve problems in advance, and reduce the probability of failures. Through continuous improvement, continuously optimize the production process and equipment management, and improve the overall operation efficiency and safety.

[0088] In addition to short-term repair and improvement measures, a long-term monitoring and prevention mechanism needs to be established to ensure the stable operation of the electrolytic cell system. Introduce advanced monitoring technologies, such as online monitoring systems, to track the key parameters of the electrolytic cells in real time and give early warnings of potential problems. Combine with a data analysis platform, use machine learning algorithms to predict the health status of equipment, and plan maintenance activities in advance. Conduct regular risk assessments, identify new potential risk points, and formulate corresponding emergency plans. Strengthen employee training to improve the technical capabilities and emergency handling capabilities of front-line operators. Through a series of comprehensive measures, build a safer and more reliable production environment and lay a solid foundation for the sustainable development of the enterprise.

[0089] Through the above series of steps, after confirming that there is a ground insulation fault in the electrolytic cell, actions can be taken quickly and effectively to ensure personnel safety, reduce economic losses, and provide valuable experience and guidance for future safe production.

[0090] In summary, based on the embodiments of the present application, the method for detecting the ground insulation fault of an electrolytic cell based on voltage phase is elucidated. Firstly, an AC excitation signal is loaded for each electrolytic cell in the electrolytic cell sequence, and at the same time, the state parameters of each electrolytic cell are collected. Then, a neural network model based on deep learning is used to perform embedding coding and global dynamic aggregation on the state parameters of each electrolytic cell to capture the overall operation state mode of the electrolytic cell sequence. Furthermore, taking the embedded representation of the state parameters of the electrolytic cell to be detected as a query feature, based on the semantic offset between it and the overall operation state mode of the electrolytic cell sequence, to intelligently identify whether the electrolytic cell to be detected has a ground insulation fault. In this way, an accurate judgment of the ground insulation fault of the electrolytic cell can be realized, and the adaptability to different working condition fluctuations and noise environments is improved, thus effectively reducing the risk of misjudgment or missed judgment.

[0091] The basic principles of the present invention have been described above in connection with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present invention are merely examples and not limitations, and it cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present invention. Additionally, the specific details of the above embodiments are only for the purpose of illustration and facilitating understanding, rather than limitations. The above details do not limit the present invention to necessarily adopt the above specific details for implementation.

[0092] In the above embodiments, the descriptions of each embodiment have their own emphases. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the unit division is only a logical function division, and there can be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0093] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any associated drawing reference signs in the claims should not be regarded as limiting the claimed rights.

[0094] In addition, obviously, the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0095] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting an electrolytic cell insulation fault based on voltage phase, characterized in that: include: Loading an AC excitation signal to each electrolytic cell in the electrolytic cell sequence and determining a state parameter of each electrolytic cell in the electrolytic cell sequence; Vectorizing and encoding the state parameters of each electrolytic cell to obtain a sequence of electrolytic cell state parameter embedded encoding vectors; Performing feature dynamic aggregation based on potential energy distribution on the sequence of electrolytic cell state parameter embedded coding vectors to obtain a significant aggregation coding vector of the electrolytic cell sequence state parameters; Extracting the electrolytic cell state parameter embedded coding vector of the detected electrolytic cell from the sequence of electrolytic cell state parameter embedded coding vectors as the electrolytic cell state parameter query coding vector; Performing semantic offset measurement on the electrolytic cell state parameter query coding vector and the electrolytic cell sequence state parameter significant aggregation coding vector to determine whether the detected electrolytic cell has a ground insulation fault; The step of performing a characteristic dynamic aggregation based on potential energy distribution on the sequence of the electrolytic cell state parameter embedded coding vectors to obtain a significant aggregation coding vector of the electrolytic cell sequence state parameters comprises: Calculating a pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters based on a characteristic static potential energy distribution of the sequence of the electrolytic cell state parameter embedding coding vectors; Calculating the aggregate movement direction of each electrolytic cell state parameter embedded coding vector in the sequence of electrolytic cell state parameter embedded coding vectors relative to the electrolytic cell sequence state parameter pseudo-anchored aggregate center representation vector to obtain a sequence of electrolytic cell state parameter aggregate movement directions; Based on the sequence of the electrolytic cell state parameter aggregation movement direction, the sequence of the electrolytic cell state parameter embedded coding vectors is dynamically aggregated toward the electrolytic cell sequence state parameter pseudo-anchor aggregation center representation vector to obtain the electrolytic cell sequence state parameter significant aggregation coding vector; Performing Z-score normalization on the electrolytic cell state parameter embedded coding vector to obtain a normalized electrolytic cell state parameter embedded coding vector; The sum of the cube of each eigenvalue in the standardized electrolytic cell state parameter embedded coding vector is calculated and divided by the characteristic scale value of the standardized electrolytic cell state parameter embedded coding vector to obtain the electrolytic cell state parameter static potential energy measurement coefficient, and the electrolytic cell sequence state parameter significant aggregation coding vector is determined based on the electrolytic cell state parameter static potential energy measurement coefficient.

2. The method for detecting the insulation fault of an electrolytic cell to the ground based on voltage phase according to claim 1, characterized in that: The state parameters include the cell AC voltage of each electrolytic cell, the cell-to-ground AC voltage of each electrolytic cell, the phase difference between the cell AC voltage and the cell-to-ground AC voltage of each electrolytic cell, and the AC voltage difference between the cell AC voltage of the electrolytic cell and the cell AC voltage of the next electrolytic cell.

3. The method for detecting an electrolytic cell insulation fault based on voltage phase according to claim 2, characterized in that: Based on the characteristic static potential energy distribution of the sequence of the electrolytic cell state parameter embedded coding vector, a pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameter is calculated, including: Inputting each electrolytic cell state parameter embedding coding vector in the sequence of electrolytic cell state parameter embedding coding vectors into a characteristic static potential energy measurement network to obtain a sequence of electrolytic cell state parameter static potential energy measurement coefficients; Inputting the sequence of electrolytic cell state parameter static potential energy measurement coefficients into a characteristic energy level screening gating unit to obtain a sequence of electrolytic cell state parameter static potential energy weight factors; Based on the sequence of static potential energy weight factors of the electrolytic cell state parameters, a weighted sum of the sequence of embedded coding vectors of the electrolytic cell state parameters is calculated to obtain a pseudo-anchored aggregation center representation vector of the electrolytic cell sequence state parameters.

4. The method for detecting insulation fault of an electrolytic cell to ground based on voltage phase according to claim 3, characterized in that: The aggregation movement direction of the electrolytic cell state parameter is the arc cosine function value between the electrolytic cell state parameter embedded coding vector and the electrolytic cell sequence state parameter pseudo-anchor aggregation center representation vector.

5. The method for detecting insulation fault of an electrolytic cell to ground based on voltage phase according to claim 4, characterized in that: The method of performing semantic offset measurement on the electrolytic cell state parameter query coding vector and the electrolytic cell sequence state parameter significant aggregation coding vector to determine whether the detected electrolytic cell has a ground insulation fault comprises: Calculating the semantic deviation between the electrolytic cell state parameter query coding vector and the electrolytic cell sequence state parameter significant aggregation coding vector; Based on the comparison between the semantic deviation and a preset threshold, it is determined whether the detected electrolytic cell has an insulation fault to the ground.

6. The method for detecting insulation fault of an electrolytic cell to ground based on voltage phase according to claim 5, characterized in that: The semantic deviation is the Euclidean distance value between the electrolytic cell state parameter query encoding vector and the electrolytic cell sequence state parameter significant aggregation encoding vector.

Images