A fault grading intelligent fault diagnosis strategy method

Abstract

The application provides a fault grading intelligent fault diagnosis strategy method. Based on the normal operation state of equipment, a plurality of sets of historical detection data of each item under the normal operation state of the equipment are obtained; the mean and standard deviation of the historical detection data are calculated according to each set of historical detection data, and a multi-level quantitative fault threshold of each item is established in combination with a normal distribution statistical rule; real-time monitoring data under the working state of a first item are obtained, and prediction data are obtained according to a prediction model; whether the real-time monitoring data and the prediction data of the first item fall within the range of the multi-level quantitative fault threshold is judged; if yes, the higher fault grade or data of the two of the real-time monitoring data and the prediction data is selected as the preliminary judgment result of the first item. Through the "real-time + prediction" dual-source high strategy, the fault identification is advanced by 2-3 time sequence periods, the false alarm rate is greatly reduced compared with the traditional fixed threshold method, and the diagnosis accuracy is improved.

Description

Technical Field

[0001] This invention relates to the technical field of tunnel boring machine fault diagnosis methods, and in particular to a fault classification-based intelligent fault diagnosis strategy method. Background Technology

[0002] Full-face tunnel boring machines (TBMs) are core and critical equipment in tunnel construction. They operate continuously under extreme conditions of high load, strong vibration, complex surrounding rock, dusty and humid environments, and severe electromagnetic interference. Core components are highly susceptible to various malfunctions, including wear, structural loosening, seal failure, transmission abnormalities, and fluctuations in hydraulic parameters. TBM malfunctions are generally characterized by high concealment, high multi-field coupling, rapid evolution, and significant cascading safety risks. Different malfunction levels require differentiated maintenance plans and construction response measures. Therefore, accurate, quantitative, and standardized fault classification and diagnosis technology is the core foundation for ensuring tunnel construction safety, controlling equipment downtime losses, and reducing maintenance costs.

[0003] Current fault diagnosis and classification technologies for tunnel boring machines still have many shortcomings: traditional fault classification methods mostly rely on the experience of staff to set fixed thresholds, resulting in highly subjective and inconsistent classification standards that cannot adapt to dynamic parameter shifts caused by changes in geological conditions and equipment load fluctuations; most diagnostic technologies rely solely on a single real-time parameter to complete threshold determination, without considering the temporal evolution of parameters, making it difficult to distinguish between instantaneous sensor noise, operating condition fluctuations, and actual equipment faults, leading to frequent misjudgments and omissions; existing technologies generally lack multi-physics parameter coupling verification mechanisms, ignoring the fault correlation characteristics between parameters such as vibration, temperature, current, and torque, and failing to identify early hidden faults formed by the co-evolution of multiple parameters; at the same time, traditional classification mechanisms are rigid and inflexible, unable to dynamically adjust fault levels based on the degree of anomalies in multiple parameters, resulting in the inability to provide early warnings for minor hidden dangers and delays in handling serious faults, which can easily induce major safety accidents such as cutterhead jamming, main bearing damage, and tunnel face instability.

[0004] In summary, existing fault classification and diagnosis technologies for tunneling machines suffer from shortcomings such as low threshold standardization, single judgment criteria, weak anti-interference ability, non-dynamic classification mechanism, and lack of multi-parameter coupling analysis, making it difficult to meet the high-precision fault classification and diagnosis requirements under complex geological conditions and long-distance tunneling. Based on these shortcomings, this invention proposes a fault classification intelligent fault diagnosis strategy and method. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a fault classification intelligent fault diagnosis strategy and method that overcomes or at least partially solves the above problems. It can solve the problems of poor anti-interference ability and low accuracy of existing fault diagnosis methods, thereby improving the accuracy of fault diagnosis.

[0006] Specifically, this invention provides a fault-level intelligent fault diagnosis strategy method, which includes: Based on the normal operating state of the equipment, acquire historical detection data for various aspects of the equipment under normal operating conditions in multiple sets; The mean and standard deviation of the historical detection data are calculated based on each group of historical detection data, and a multi-level quantitative fault threshold for each item is established in combination with the normal distribution statistical rules. Obtain real-time monitoring data under the working status of the first item, and obtain prediction data based on the prediction model; Determine whether the real-time monitoring data and the predicted data of the first item fall within the range of the multi-level quantization fault threshold; If so, the higher fault level or data between the real-time monitoring data and the predicted data is selected as the preliminary judgment result of the first item.

[0007] Optionally, the steps following the establishment of multi-level quantized fault thresholds for each institution in conjunction with normal distribution statistical rules include: Obtain real-time monitoring data under the working status of the second item, and obtain prediction data based on the prediction model; Determine whether the real-time monitoring data and the predicted data in the second item fall within the range of the multi-level quantization fault threshold; If so, the higher fault level or data between the real-time monitoring data and the predicted data shall be selected as the preliminary judgment result of the second item; If not, determine whether the faults of the first and second items are strongly correlated parameters; If so, determine whether the preliminary judgment result of the second item and the preliminary judgment result of the first item are at the same level; If so, escalate the fault level and output the corresponding maintenance strategy.

[0008] Optionally, the step following the acquisition of multiple sets of historical detection data for each item under normal operating conditions of the device, based on the device's normal operating state, includes: For each group of historical detection data, data cleaning, abnormal data removal, and time-series alignment are performed.

[0009] Optionally, the mean μ and standard deviation σ of the historical detection data; the multi-level quantified fault thresholds include the upper limit of normal operation μ+2σ, the upper limit of first-level fault μ+3σ, the upper limit of second-level fault μ+4σ, and the upper limit of third-level fault 1.2×(μ+4σ).

[0010] Optionally, the step of selecting the higher fault level or data between the real-time monitoring data and the predicted data as the preliminary determination result of the first item includes: If the real-time monitoring data and the predicted data exceed the normal working limit, the first-level fault limit, or the second-level fault limit, they are determined to be first-level fluctuation abnormalities, second-level fluctuation abnormalities, or third-level fluctuation abnormalities. If the real-time monitoring data and the predicted data continuously exceed the first-level fault limit or the second fault limit and maintain a preset time step, it is determined to be a second-level fluctuation anomaly or a third-level fluctuation anomaly. If the real-time monitoring data and the predicted data continuously exceed the upper limit of the third-level fault and maintain a preset time step, it is determined to be a fatal fault.

[0011] Optionally, the step following the determination of whether the faults of the first and second items are strongly correlated parameters further includes: If not, the preliminary judgment result of the second item shall be confirmed as the final judgment result.

[0012] In the fault classification and intelligent fault diagnosis strategy method of this invention, historical detection data of equipment operating in a steady-state, fault-free manner for 30 consecutive days are selected. The focus is on screening the vibration acceleration parameters of the main bearing, thereby providing a clean, steady-state basic data sample for constructing standardized fault thresholds and identifying interference from non-faulty operating conditions. This ensures that the subsequently established thresholds can accurately reflect the parameter fluctuation range under the equipment's healthy state, laying the foundation for adaptive quantitative classification.

[0013] A standardized, multi-level fault threshold system adapted to different operating conditions is constructed. After preprocessing the collected historical detection data, the overall mean and standard deviation are calculated, and a four-level fault threshold system is established. This system eliminates manual experience-based setting and utilizes the equipment's own big data and statistical patterns to achieve standardized thresholds that adapt to different operating conditions. This solves the problems of traditional methods' subjective grading standards and inability to adapt to varying load conditions, realizing the digitalization and standardization of fault grading.

[0014] By inputting a pre-trained deep learning prediction model, the system can predict data for the next five cycles based on historical detection data. This allows for the simultaneous acquisition of dual-source information data, including current state and future preset data, providing a richer and more forward-looking information foundation for comprehensive judgment than a single instantaneous value, effectively supporting the identification of early-stage, slowly changing faults.

[0015] A preliminary assessment of the fault level for a single parameter in a dual-source fault scenario is conducted. Real-time detection data and predicted data are independently compared against multi-level quantified fault threshold ranges for level matching. This enables independent verification of fault data from both sources, providing a reliable basis for subsequently selecting the higher fault level. Furthermore, the higher fault level determined by the real-time detection data and predicted data is selected as the preliminary diagnostic result for that single parameter. This ensures the safety of fault assessment, preventing any higher risk indicated by either side from being overlooked, and improving the ability to identify early-stage, latent faults.

[0016] In summary, when this embodiment is applied to the vibration monitoring of the main bearing of a tunnel boring machine, it successfully constructs a statistically-based adaptive threshold system. By adopting a dual-source height acquisition strategy of "real-time + prediction", the fault identification is advanced by 2-3 time cycles, the false alarm rate is greatly reduced compared with the traditional fixed threshold method, and the diagnostic accuracy is improved.

[0017] The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0018] The following sections will describe some specific embodiments of the invention in detail by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a schematic flowchart of a fault classification intelligent fault diagnosis strategy method according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of a fault classification intelligent fault diagnosis strategy method according to an embodiment of the present invention. Detailed Implementation

[0019] The following reference Figures 1 to 2 This invention describes a fault-grading intelligent fault diagnosis strategy method according to embodiments of the present invention. In this description, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" and "second" may explicitly or implicitly include at least one of that feature, that is, include one or more of that feature. In the description of the present invention, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. When a feature "includes or contains" one or more of the features it encompasses, unless otherwise specifically described, this indicates that other features are not excluded and may be further included.

[0020] Unless otherwise expressly specified and limited, the terms "set up," "install," "connect," "link," "fix," and "couple" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art should be able to understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0021] Furthermore, in the description of this embodiment, "above" or "below" the second feature can include direct contact between the first and second features, or it can include contact between the first and second features through another feature between them. That is, in the description of this embodiment, "above," "over," and "on top" of the second feature includes the first feature being directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," or "below" of the second feature can mean the first feature is directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0022] In the description of this embodiment, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0023] Figure 1 This is a schematic structural diagram of a fault classification-based intelligent fault diagnosis strategy method, such as... Figure 1 As shown, and refer to Figure 2 A fault classification-based intelligent fault diagnosis strategy method includes: S100. Based on the normal operating status of the equipment, acquire historical detection data of various items under the normal operating status of the equipment in multiple sets; S200. The mean and standard deviation of the historical detection data are calculated based on each group of historical detection data, and a multi-level quantitative fault threshold for each item is established in combination with the normal distribution statistical rules. S300: Obtain real-time monitoring data under the working state of the first item, and obtain prediction data based on the prediction model; S400. Determine whether the real-time monitoring data and prediction data of the first item fall within the range of the multi-level quantization fault threshold. S500 If so, the higher fault level or data between the real-time monitoring data and the predicted data shall be selected as the preliminary judgment result of the first item.

[0024] S510. If not, the device is determined to be not in normal operating condition.

[0025] Specifically, in step S100, historical monitoring data of the equipment under 30 consecutive days of steady-state tunneling and fault-free operation is selected, with a focus on screening the vibration acceleration parameters of the main bearing. This provides a clean, steady-state basic data sample for constructing standardized fault thresholds, and identifies interference from non-faulty operating conditions. This ensures that the subsequently established thresholds can accurately reflect the parameter fluctuation range under the equipment's healthy state, laying the foundation for adaptive quantitative grading.

[0026] Furthermore, in step S200, a standardized multi-level fault threshold system adapted to operating conditions is constructed. After preprocessing the collected historical detection data, the overall mean and standard deviation are calculated, and a four-level fault threshold is established. This eliminates manual experience-based setting and utilizes the equipment's own big data and statistical laws to achieve standardized thresholds that adapt to operating conditions. This solves the problems of traditional methods having strong subjectivity in their grading standards and being unable to adapt to varying load conditions, thus realizing the digitalization and standardization of fault grading.

[0027] Furthermore, in step S300, a pre-trained deep learning prediction model is input to predict the data for the next five cycles based on historical detection data. This allows for the simultaneous acquisition of dual-source information data from the current state and the future presets, providing a richer and more forward-looking information foundation for comprehensive judgment than a single instantaneous value, effectively supporting the identification of early-stage, slowly changing faults.

[0028] Furthermore, in step S400, a preliminary determination of the single-parameter dual-source fault level is performed. The real-time detection data and predicted data are independently compared to the multi-level quantified fault threshold range for level matching, thereby enabling independent verification of dual-source fault data and providing a reliable basis for subsequently selecting the higher level. Further, the higher fault level determined by the real-time detection data and predicted data is selected as the preliminary diagnostic result for this single parameter. This ensures the safety of fault determination, avoids overlooking higher risks indicated by either side, and improves the ability to identify early-stage latent faults.

[0029] In summary, when this embodiment is applied to the vibration monitoring of the main bearing of a tunnel boring machine, it successfully constructs a statistically-based adaptive threshold system. By adopting a dual-source height acquisition strategy of "real-time + prediction", the fault identification is advanced by 2-3 time cycles, the false alarm rate is greatly reduced compared with the traditional fixed threshold method, and the diagnostic accuracy is improved.

[0030] In some embodiments of the present invention, such as Figure 1 and Figure 2As shown, the steps following the establishment of a multi-level quantized fault threshold for each item using the normal distribution statistical rules include: S210. Obtain real-time monitoring data under the working state of the second item, and obtain prediction data based on the prediction model; S220. Determine whether the real-time monitoring data and prediction data of the second item fall within the range of the multi-level quantization fault threshold. S230. If so, the higher fault level or data between the real-time monitoring data and the predicted data shall be selected as the preliminary judgment result of the second item. S240. If not, determine whether the faults of the first and second items are strongly correlated parameters. S250. If so, determine whether the preliminary judgment result of the second item and the preliminary judgment result of the first item are at the same level. S260. If so, escalate the fault level and output the corresponding fault level maintenance strategy.

[0031] The process after step S240 also includes: S241, if not, select the highest fault level from the first and second items as the final fault output.

[0032] The process after step S250 also includes: S251, if not, directly output the fault level for processing and output the corresponding fault level maintenance strategy.

[0033] Specifically, this embodiment performs multi-parameter coupling verification and dynamic upgrade of fault levels. Multiple historical detection data points include, for example, vibration acceleration data, operating temperature data, and operating current data. These three sets of fault parameters are strongly correlated fault parameters. According to step S240 above, if two or more data points are strongly correlated, the strong correlation is combined vertically with the multi-parameter fault coupling correlation characteristics to determine that the current anomaly is not caused by fluctuations in a single parameter, but rather by an early coupled fault, such as wear of internal components of the main bearing. The initial fault level is automatically upgraded to provide a key reminder to personnel, thereby achieving dynamic adaptive upgrade of the fault level, accurately characterizing the actual severity of the fault, and forming a differentiated hierarchical operation and maintenance system of "minor anomaly monitoring, general fault maintenance, and serious fault repair."

[0034] In some embodiments of the present invention, such as Figure 1 and Figure 2 As shown, the steps following the acquisition of historical detection data for each item under multiple sets of normal operating conditions of the equipment, based on the normal operating state of the equipment, include: S110. Perform data cleaning, abnormal data removal, and time-series alignment on each set of historical detection data.

[0035] Specifically, by cleaning historical detection data, removing abnormal data, and aligning time sequences, invalid data generated by sensors due to electromagnetic interference and communication packet loss can be removed, and non-steady-state anomalies caused by accidental impacts and human error can be filtered out, ensuring that the retained data truly reflects the "steady-state normal operation state". In addition, parameters of different sampling frequencies can be unified to the same time base to solve the data misalignment problem caused by acquisition delay or clock drift.

[0036] Therefore, it significantly improves the statistical accuracy of subsequent calculations of mean and standard deviation, avoiding contamination of μ and σ by outliers. This enhances the overall robustness and reliability of the fault thresholding system.

[0037] In some embodiments of the present invention, such as Figure 1 and Figure 2 As shown, the mean μ and standard deviation σ of the historical detection data; the multi-level quantification fault thresholds include the upper limit of normal operation μ+2σ, the upper limit of first-level fault μ+3σ, the upper limit of second-level fault μ+4σ, and the upper limit of third-level fault 1.2×(μ+4σ).

[0038] Specifically, mathematical expressions for the four threshold levels are defined, combining the statistical rules of normal distribution with the requirements of engineering fault classification. For example, the mean μ and standard deviation σ of the vibration acceleration of the main bearing of a tunnel boring machine are calculated as follows: Upper limit for normal operation: μ + 2σ = 5.0 m / s²; Upper limit for Level 1 fault: μ + 3σ = 6.5 m / s²; Upper limit for Level 2 fault: μ + 4σ = 8.0 m / s²; Upper limit for Level 3 fault: 1.2 × (μ + 4σ) = 9.6 m / s².

[0039] It abandons the subjective mode of setting thresholds based on traditional human experience, and realizes the digitalization, standardization and adaptive adaptation of fault classification standards, effectively improving the uniformity and accuracy of fault classification under varying loads and geological conditions.

[0040] In some embodiments of the present invention, such as Figure 1 and Figure 2 As shown, the step of selecting the higher fault level or data between real-time monitoring data and predicted data as the preliminary judgment result of the first item includes: S600. If the real-time monitoring data and the predicted data exceed the normal working limit, the first-level fault limit, or the second-level fault limit, it is determined to be a first-level fluctuation abnormality, a second-level fluctuation abnormality, or a third-level fluctuation abnormality. S700: If real-time monitoring data and prediction data continuously exceed the first-level fault limit or the second-level fault limit and maintain the preset timing step, it is determined to be a second-level or third-level fluctuation anomaly. If the S800, real-time monitoring data, and predictive data continuously exceed the level 3 fault limit and maintain the preset timing step, it is determined to be a fatal fault.

[0041] Specifically, based on the implementation of the above steps, for example: the vibration acceleration of 7.2 m / s² exceeds the upper limit of 6.5 m / s² for a first-level fault, but does not reach the upper limit of 8.0 m / s² for a second-level fault. Combined with the continuous step size constraint, it is determined to be a second-level fault.

[0042] The step of determining a higher-level fault by continuously exceeding the fault limit and maintaining a preset time step size serves to eliminate false anomalies caused by instantaneous noise or operating condition fluctuations, thereby improving the reliability of the determination. The continuous anomaly determination step size is set to 5 time cycles. Real-time vibration data continuously exceeding the first-level fault limit but not reaching the second-level limit satisfies the conditions for a second-level continuous anomaly.

[0043] A fatal failure is a Level 4 fluctuation anomaly, which is the upper limit of Level 4 failure. It identifies the highest risk level of failure, triggers emergency measures, and is used to prevent fatal failures of equipment.

[0044] This step, combined with continuous timing step constraints, effectively distinguishes between instantaneous sensor noise, operating condition fluctuations, and actual equipment failures, significantly reducing the false positive and false negative rates and greatly improving the ability to identify early latent and fatal faults.

[0045] In some embodiments of the present invention, such as Figure 1 and Figure 2 As shown, the step following the determination of whether the faults of the first and second items are strongly correlated parameters further includes: If not, the preliminary judgment result of the second item shall be confirmed as the final judgment result.

[0046] Specifically, if the first and second data items are not strongly correlated parameters, the preliminary judgment result of the second item will be directly used as the final judgment result to avoid incorrectly upgrading the level of non-correlated parameters and ensure the rationality of the fault classification result.

[0047] For example, if the vibration parameters of the tunnel boring machine cutterhead drive have been determined to be a Level 4 fatal fault, which exceeds the upgrade range of Level 1 to 3, there is no need to perform multi-parameter level upgrade. Instead, the highest level (Level 4) of the single-parameter dual-source determination can be used as the final diagnostic result.

[0048] This step ensures the rigor and applicability of the fault level escalation mechanism, preventing excessive maintenance or false alarms due to unreasonable escalation. At the same time, it ensures that high-risk faults can be quickly and directly output to the final level, triggering emergency response mechanisms such as emergency shutdown, emergency support, and power outage protection, thereby avoiding major construction risks.

[0049] This embodiment is applied to open-face TBM tunneling in hard rock mountain tunnels, and conducts rapid graded diagnosis of fatal faults in the cutterhead drive system under high-hardness surrounding rock and high-load tunneling conditions. First, steady-state historical tunneling data of the TBM cutterhead drive system is collected, and data preprocessing and mean and standard deviation calculations are performed to construct a four-level fault threshold system adapted to hard rock tunneling conditions. Real-time acquisition of cutterhead drive vibration, load torque, and operating current parameters is combined with time-series model prediction data to complete dual-source joint judgment. The detection results show that both the real-time and predicted data of the cutterhead drive vibration parameters consistently exceed the upper limit of a level three fault, directly classifying it as a level four fatal fault. This fault level has reached the level four fatal level, exceeding the dynamic upgrade range of levels 1-3 of this invention. Therefore, multi-parameter level upgrade operations are not required, and the final diagnosis result of a level four fatal fault is directly output, simultaneously triggering emergency response mechanisms such as emergency equipment shutdown, emergency face support, and equipment power-off protection, effectively avoiding major construction risks such as cutterhead jamming, drive system damage, and surrounding rock collapse during hard rock tunneling.

[0050] Therefore, those skilled in the art should recognize that although numerous exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Thus, the scope of the present invention should be understood and construed as covering all such other variations or modifications.

Claims

1. A fault classification-based intelligent fault diagnosis strategy and method, characterized in that, include: Based on the normal operating state of the equipment, acquire historical detection data for various aspects of the equipment under normal operating conditions in multiple sets; The mean and standard deviation of the historical detection data are calculated based on each group of historical detection data, and a multi-level quantitative fault threshold for each item is established in combination with the normal distribution statistical rules. Obtain real-time monitoring data under the working status of the first item, and obtain prediction data based on the prediction model; Determine whether the real-time monitoring data and the predicted data of the first item fall within the range of the multi-level quantization fault threshold; If so, the higher fault level or data between the real-time monitoring data and the predicted data is selected as the preliminary judgment result of the first item.

2. The fault classification intelligent fault diagnosis strategy and method according to claim 1, characterized in that, The steps following the establishment of multi-level quantized fault thresholds for each institution, in conjunction with normal distribution statistical rules, include: Obtain real-time monitoring data under the working status of the second item, and obtain prediction data based on the prediction model; Determine whether the real-time monitoring data and the predicted data in the second item fall within the range of the multi-level quantization fault threshold; If so, the higher fault level or data between the real-time monitoring data and the predicted data shall be selected as the preliminary judgment result of the second item; If not, determine whether the faults of the first and second items are strongly correlated parameters; If so, determine whether the preliminary judgment result of the second item and the preliminary judgment result of the first item are at the same level; If so, escalate the fault level and output the corresponding maintenance strategy.

3. The fault classification intelligent fault diagnosis strategy and method according to claim 1, characterized in that, The steps following the acquisition of historical detection data for multiple sets of items under normal operating conditions of the equipment include: For each group of historical detection data, data cleaning, abnormal data removal, and time-series alignment are performed.

4. The fault classification intelligent fault diagnosis strategy and method according to claim 1, characterized in that, The mean μ and standard deviation σ of the historical detection data; the multi-level quantified fault thresholds include the upper limit of normal operation μ+2σ, the upper limit of first-level fault μ+3σ, the upper limit of second-level fault μ+4σ, and the upper limit of third-level fault 1.2×(μ+4σ).

5. The fault classification intelligent fault diagnosis strategy and method according to claim 4, characterized in that, The step of selecting the higher fault level or data between the real-time monitoring data and the predicted data as the preliminary determination result of the first item includes: If the real-time monitoring data and the predicted data exceed the normal working limit, the first-level fault limit, or the second-level fault limit, they are determined to be first-level fluctuation abnormalities, second-level fluctuation abnormalities, or third-level fluctuation abnormalities. If the real-time monitoring data and the predicted data continuously exceed the first-level fault limit or the second fault limit and maintain a preset time step, it is determined to be a second-level fluctuation anomaly or a third-level fluctuation anomaly. If the real-time monitoring data and the predicted data continuously exceed the upper limit of the third-level fault and maintain a preset time step, it is determined to be a fatal fault.

6. The fault classification intelligent fault diagnosis strategy and method according to claim 2, characterized in that, The step following the determination of whether the faults of the first and second items are strongly correlated parameters further includes: If not, the preliminary judgment result of the second item shall be confirmed as the final judgment result.

Images