A phase modulator remote fault diagnosis and intelligent decision system

Abstract

The application discloses a phase modifier remote fault diagnosis and intelligent decision system, comprising the following modules: a phase modifier monitoring module; a fault detection module: taking the prepared feature vector as input data set, constructing a K-D tree data structure according to the feature vector; selecting a node in the K-D tree as an initial clustering center, using the selected initial clustering center as a starting point; executing a K-means clustering algorithm, iteratively calculating the distance between each data point and the clustering center, and assigning the data point to the nearest clustering center, updating the position of the clustering center and setting it as the average value of the belonging cluster, continuing iteration until the clustering center is stable and no longer changes or the predefined iteration number is reached; checking the belonging cluster of each data point, if a certain data point is relatively far from other data points, it may indicate that the data point has an anomaly or a fault; a remote diagnosis module; a decision support module. The application provides convenience and reliability for remote control and maintenance of the phase modifier device, and improves the reliability and maintenance efficiency of the device.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a remote fault diagnosis and intelligent decision-making system for synchronous condensers. Background Technology

[0002] The domestic synchronous condenser fault diagnosis system is still in its infancy. Currently, equipment fault diagnosis relies more on front-line professionals to conduct detailed analysis and processing of signals collected on-site. This mode limits the real-time nature of fault detection and processing. However, the current system cannot automatically detect faults in synchronous condenser equipment, and there is a high reliance on manual fault detection, resulting in low accuracy and efficiency in fault location. Summary of the Invention

[0003] To address the above problems, this invention provides a remote fault diagnosis and intelligent decision-making system for synchronous condensers.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A remote fault diagnosis and intelligent decision-making system for synchronous condensers includes the following modules:

[0006] The synchronous condenser monitoring module is used to collect monitoring data from the synchronous condenser equipment and convert the data into feature vector representations.

[0007] The fault detection module takes the prepared feature vectors as input data and constructs a KD-tree data structure based on the feature vectors. It selects nodes in the KD-tree as initial cluster centers, using these initial cluster centers as the starting point. It executes the K-means clustering algorithm, iteratively calculating the distance between each data point and its cluster center, assigning each data point to the nearest cluster center, updating the position of the cluster center by setting it to the average value of its cluster, and continuing iteratively until the cluster centers stabilize or reach a predefined number of iterations. Finally, it checks the cluster to which each data point belongs; if a data point is far from other data points, it may indicate an anomaly or fault.

[0008] Remote diagnostic module: By establishing a remote connection with the synchronous condenser equipment, it acquires diagnostic data to further analyze and locate faults in the system;

[0009] Decision support module: Provides intelligent decision support based on the results of fault diagnosis and the current operating status.

[0010] Furthermore: the camera monitoring module includes:

[0011] Use sensors or monitoring devices to acquire monitoring data of the synchronous condenser equipment. The data includes current, voltage, frequency, phase angle, and power factor. Ensure that appropriate sensors or devices are selected to accurately collect data related to the operating status of the synchronous condenser equipment.

[0012] The collected monitoring data is preprocessed to ensure data quality and consistency;

[0013] Based on the characteristics and fault modes of the synchronous condenser equipment, relevant features are extracted from the pre-processed monitoring data. These features should be able to capture information related to the fault state.

[0014] The features extracted from the monitoring data are combined into a feature vector representation. Each feature vector should contain a set of feature values, where each feature value corresponds to a feature. Depending on the number and type of features, the feature vector can be a one-dimensional array or a multi-dimensional matrix, ensuring that the feature vector has an appropriate dimension and can fully describe the monitoring information of the camera condenser.

[0015] Furthermore: the fault detection module includes:

[0016] Step 1: Prepare feature vectors: Use the feature vectors obtained from the monitoring data of the camera switching equipment as the input dataset.

[0017] Step 2: Select the split dimension: Select a split dimension, that is, on which feature to split. The selection is based on the range of feature values ​​and variance index in the feature vector. The goal of selecting the split dimension is to evenly divide the data at each node.

[0018] Step 3: Select a splitting value: Based on the selected splitting dimension, select a splitting value to divide the dataset into two subspaces. The splitting value is selected based on the median of the splitting dimension in the feature vector.

[0019] Step 4: Create a node: Create a node to represent the partition dimension and the partition value. Use the partition value as the node's split point to divide the dataset into two subsets.

[0020] Step 5: Recursive construction: For each subset, repeat steps 2 to 4 to recursively construct the KD tree, select new split dimensions and split values, and create new nodes;

[0021] Step 6 Termination Condition: Recursively build until the termination condition is met. The termination condition can be that the number of data in a node reaches a predetermined threshold, or the maximum depth of the tree is reached.

[0022] Step 7: Construction complete. Once all subspaces have been constructed, the KD tree construction is complete.

[0023] Furthermore: the camera monitoring module includes:

[0024] Step 1: Assign data points: For each data point, calculate its distance to all cluster centers and assign it to the cluster to which the nearest cluster center belongs;

[0025] Step 2: Update cluster centers: For each cluster, calculate the average value of its data points and use this average value as the new cluster center location;

[0026] Step 3: Repeat the iteration: Repeat steps 1 and 2 until a predefined termination condition is reached. The termination condition is that the cluster centers are stable and no longer change or the predefined number of iterations is reached.

[0027] Step 4: Clustering is completed. When the termination condition is met, the K-means clustering algorithm is complete, and each data point is assigned to a cluster center.

[0028] Furthermore: the camera monitoring module includes:

[0029] Clustering check: For each data point, check its clustering by accessing the data point's label or cluster index to determine the cluster to which the data point belongs;

[0030] Anomaly detection: For each data point in the cluster, fault detection is performed. Anomaly data points represent potential faults or abnormal situations, including:

[0031] Calculate the Z-score for each data point: For a given data point, calculate the ratio of its standard deviation (σ) to the mean (μ) of the dataset using the following formula:

[0032] Z-score = (x-μ) / σ

[0033] Where X represents the value of the data point, μ represents the mean of the dataset, and σ represents the standard deviation of the dataset;

[0034] Set a threshold: Based on actual needs and the characteristics of the dataset, set a threshold to determine which Z-score values ​​are considered outliers;

[0035] Outlier labeling: Data points with Z-score values ​​above a threshold are labeled as outliers and are considered potential fault points.

[0036] Furthermore: the camera monitoring module includes:

[0037] Fault type identification: The fault type is determined by analyzing and processing fault data points;

[0038] Generate fault alarms: Generate fault alarms based on fault type, timestamp, and other relevant information.

[0039] Furthermore: The remote diagnostic module includes:

[0040] Establishing a remote connection: The remote diagnostic module establishes a remote connection with the synchronous condenser equipment through a remote connection protocol in order to obtain more diagnostic data;

[0041] Obtaining diagnostic data: The remote diagnostic module requests and obtains diagnostic data from the camera condenser. The diagnostic data includes log files, configuration information, performance indicators, and sensor data.

[0042] Fault analysis and location: Based on the acquired diagnostic data, the remote diagnostic module performs further fault analysis and location;

[0043] Generate diagnostic reports: Based on the results of fault analysis and location, the remote diagnostic module can generate diagnostic reports.

[0044] Furthermore, the decision support module includes:

[0045] Fault diagnosis result analysis: Based on the output results and analysis of the previous fault diagnosis module, the decision support module assesses and analyzes the severity and impact of the fault;

[0046] Maintenance recommendations: Based on the fault diagnosis results and the current operating status, the decision support module recommends appropriate maintenance measures;

[0047] Decision making and execution: Based on the recommended maintenance measures, the decision support module will formulate corresponding decision-making plans;

[0048] Monitoring and Feedback: The decision support module monitors the progress of the maintenance process and provides feedback based on the actual situation.

[0049] The technological advancements achieved by this invention compared to existing technologies are as follows:

[0050] Automated fault detection: The system uses machine learning and data analysis algorithms to automatically detect faults, reducing reliance on manual fault detection and improving the accuracy and efficiency of fault detection.

[0051] Accuracy of fault location: By analyzing monitoring data and feature vectors, the system can more accurately locate the fault, shorten the troubleshooting time, and improve maintenance efficiency.

[0052] Remote diagnostic function: The system can remotely connect and acquire diagnostic data, providing remote diagnostic support, avoiding the need for manual on-site visits, and reducing maintenance costs and time.

[0053] Intelligent Decision Support: Based on fault diagnosis results and the current situation, the system provides intelligent maintenance recommendations, reducing the subjectivity and errors of manual decision-making and improving the accuracy and efficiency of maintenance decisions. This system provides convenience and reliability for the remote control and maintenance of synchronous condenser equipment, improving equipment reliability and maintenance efficiency. Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0055] In the attached diagram:

[0056] Figure 1 This is a system structure diagram of the present invention;

[0057] Figure 2 A flowchart illustrating the construction of a KD-tree data structure for the fault detection module of this invention;

[0058] Figure 3 This is a flowchart of the K-means clustering process in the fault detection module of this invention. Detailed Implementation

[0059] The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0060] like Figure 1 As shown, a remote fault diagnosis and intelligent decision-making system for synchronous condensers includes:

[0061] The synchronous condenser monitoring module is used to collect monitoring data from the synchronous condenser equipment and convert the data into feature vector representations.

[0062] The fault detection module takes the prepared feature vectors as input data and constructs a KD-tree data structure based on the feature vectors. It selects nodes in the KD-tree as initial cluster centers, using these initial cluster centers as the starting point. It executes the K-means clustering algorithm, iteratively calculating the distance between each data point and its cluster center, assigning each data point to the nearest cluster center, updating the position of the cluster center by setting it to the average value of its cluster, and continuing iteratively until the cluster centers stabilize or reach a predefined number of iterations. Finally, it checks the cluster to which each data point belongs; if a data point is far from other data points, it may indicate an anomaly or fault.

[0063] Remote diagnostic module: By establishing a remote connection with the synchronous condenser equipment, it acquires diagnostic data to further analyze and locate faults in the system;

[0064] Decision support module: Provides intelligent decision support based on fault diagnosis results and current operational status.

[0065] Specifically, the camera monitoring module includes:

[0066] Collect monitoring data: Use sensors or monitoring equipment to acquire monitoring data from the synchronous condenser. This data may include key indicators such as current, voltage, frequency, phase angle, and power factor. Ensure that appropriate sensors or equipment are selected to accurately collect data related to the operating status of the synchronous condenser.

[0067] Data preprocessing: The collected monitoring data is preprocessed to ensure data quality and consistency. Possible preprocessing operations include data cleaning (removing outliers or erroneous data), data sampling (reducing the sampling frequency to decrease the data volume), and data smoothing (removing noise or fluctuations). Preprocessing helps improve the accuracy of subsequent feature extraction and fault detection.

[0068] Feature extraction: Based on the characteristics and fault modes of the synchronous condenser equipment, relevant features are extracted from the preprocessed monitoring data. These features should capture information related to the fault state. For example, statistical features such as the mean, standard deviation, maximum, and minimum current values ​​can be calculated, or spectral features such as power spectral density and frequency components can be extracted. Ensure that an appropriate feature extraction method is selected to fully represent the state and fault modes of the synchronous condenser equipment.

[0069] Feature vector representation: Features extracted from the monitoring data are combined into feature vectors. Each feature vector should contain a set of feature values, where each feature value corresponds to a feature. Depending on the number and type of features, the feature vector can be a one-dimensional array or a multi-dimensional matrix. Ensure that the feature vectors have appropriate dimensions and can adequately describe the monitoring information from the camera condenser.

[0070] like Figure 2 As shown, the fault detection module includes the following steps to construct the KD tree:

[0071] Step 1: Prepare feature vectors: Based on the monitoring data of the camera equipment and the feature vectors obtained in the feature extraction step, use the feature vectors as the input dataset.

[0072] Step 3: Select a splitting dimension: Choose a splitting dimension, i.e., on which feature to split. This can be selected based on metrics such as the range of feature values ​​and variance in the feature vector. The goal of selecting a splitting dimension is to evenly divide the data at each node.

[0073] Step 3: Select a splitting value: Based on the chosen splitting dimension, select a splitting value to divide the dataset into two subspaces. The splitting value can be selected based on the median or average of the splitting dimension in the feature vector.

[0074] Step 4: Create a node: Create a node representing the partition dimension and partition value. Use the partition value as the split point of the node to divide the dataset into two subsets.

[0075] Step 5: Recursive Construction: For each subset, repeat steps 2 through 4 to recursively construct the KD-tree. Select new split dimensions and split values, and create new nodes.

[0076] Step 6 Termination Condition: Recursively build until a termination condition is met. The termination condition can be that the number of data in a node reaches a predetermined threshold, or that the maximum depth of the tree is reached.

[0077] Step 7: Construction complete. Once all subspaces have been constructed, the KD tree construction is complete.

[0078] It's important to note that the specific construction process of a KD-tree may vary depending on the characteristics of the dataset and the features of the camera adjustment equipment. The construction steps may need to be adjusted and optimized based on the specific camera adjustment equipment and dataset. Furthermore, the performance of a KD-tree is also affected by factors such as tree balance and the choice of partition dimension. Therefore, in practical implementations, these factors need to be comprehensively considered to optimize the KD-tree construction process.

[0079] Select initial cluster centers, which will be used to initialize the K-means algorithm.

[0080] You can select some nodes as initial cluster centers in a KD-tree to ensure that the nodes are distributed throughout the entire dataset. For example... Figure 3 As shown, the K-means clustering of the camera monitoring module includes:

[0081] Step 1: Initialize cluster centers: Use the selected initial cluster centers as the starting point, with each cluster center representing one cluster;

[0082] Step 2: Assign data points: For each data point, calculate its distance to all cluster centers and assign it to the cluster to which the nearest cluster center belongs;

[0083] Step 3: Update cluster centers: For each cluster, calculate the average value of its data points and use this average value as the new cluster center location;

[0084] Step 4: Repeat steps 2 and 3 until a predefined termination condition is met. The termination condition can be that the cluster centers stabilize and no longer change, or that a predefined number of iterations has been reached.

[0085] Step 5: Clustering is complete when the termination condition is met. The K-means clustering algorithm is complete, and each data point is assigned to a cluster center.

[0086] It's important to note that in practice, implementing the K-means clustering algorithm may require adjustments based on the characteristics of the camera equipment and the dataset. For example, Euclidean distance can be used to calculate the distance between data points and cluster centers to improve the quality of the clustering results.

[0087] Step 1: Define Euclidean distance: For two given vectors A and B, the Euclidean distance can be calculated using the following formula:

[0088] Where d represents the Euclidean distance, n represents the dimension of the vector, and Ai and Bi represent the i-th eigenvalues ​​of vectors A and B, respectively.

[0089] Step 2: For each data point, calculate its Euclidean distance to each cluster center. Associate the data point with the nearest cluster center.

[0090] Step 3: Repeat step 2 until all data points are assigned to the nearest cluster center.

[0091] Step 4: Update the cluster center positions: For each cluster, calculate the average value of its data points and use this average value as the new cluster center position.

[0092] Step 5: Repeat steps 2 and 4 until the cluster centers stabilize and no longer change or the predefined number of iterations is reached.

[0093] Camera condenser monitoring module fault detection includes:

[0094] Clustering check: For each data point, check its clustering by accessing the data point's label or cluster index to determine the cluster to which the data point belongs;

[0095] Anomaly detection: For each data point in the cluster, fault detection is performed. Anomaly data points represent potential faults or abnormal situations, including:

[0096] Calculate the Z-score for each data point: For a given data point, calculate the ratio of its standard deviation (σ) to the mean (μ) of the dataset using the following formula:

[0097] Z-score=(X-μ) / σ

[0098] Where X represents the value of the data point, μ represents the mean of the dataset, and σ represents the standard deviation of the dataset;

[0099] Set a threshold: Based on actual needs and the characteristics of the dataset, set a threshold to determine which Z-score values ​​are considered outliers;

[0100] Outlier labeling: Data points with Z-score values ​​above a threshold are labeled as outliers and are considered potential fault points.

[0101] The Z-score method assumes that the data points follow a normal distribution. Therefore, before using Z-score for outlier detection, ensure that the data distribution is close to a normal distribution or perform appropriate data transformations. Furthermore, the Z-score method can be combined with other outlier detection methods to improve detection accuracy and robustness.

[0102] The fault alarm generated by the camera monitoring module includes:

[0103] Faulty data points detected: Based on the previous fault detection steps, if a data point is determined to be faulty or an outlier, that is, deviating from the normal data distribution, then we can mark it as a faulty data point.

[0104] Fault type identification: By analyzing and processing fault data points, the fault type is determined. This can be based on predefined fault patterns and rules, or by using machine learning algorithms to automatically identify fault types. For example, a fault classification model can be trained, taking the feature vectors of data points as input and outputting the corresponding fault type.

[0105] Generate fault alarms: Generate fault alarms based on fault type, timestamp, and other relevant information. Fault alarms can be text messages or internal system events or notifications. Fault alarms should contain sufficient information so that operators or the system can understand them and take appropriate action to handle the fault.

[0106] The specific methods and tools used to implement fault alarms can be determined based on the actual situation. For example, a logging system, an alarm notification system, or a monitoring system integrated with synchronous condenser equipment can be used to generate fault alarms. These systems typically have the function of pushing fault alarms to relevant personnel and recording fault alarm events.

[0107] Fault types and alarm mechanisms need to be defined based on the specific synchronous condenser equipment and application environment to ensure that the generated fault alarms can notify relevant personnel in a timely and accurate manner, and support rapid troubleshooting and repair.

[0108] The remote diagnostic module includes:

[0109] Establishing a remote connection: The remote diagnostic module needs to establish a remote connection with the synchronous condenser equipment to obtain more diagnostic data. This can be achieved via a network connection (such as the Internet) or a dedicated remote connection protocol (such as SSH). The method of establishing the connection depends on the specific requirements of the synchronous condenser equipment and the application environment.

[0110] Obtaining Diagnostic Data: Once a remote connection is established, the remote diagnostic module can request and obtain diagnostic data from the camera condenser via this connection. Diagnostic data may include log files, configuration information, performance metrics, sensor data, etc. Depending on specific needs, you can choose to obtain specific diagnostic data or the complete diagnostic dataset.

[0111] Data Transfer and Storage: Acquired diagnostic data needs to be transferred and stored. Network transfer protocols (such as FTP, SCP) or dedicated data transfer protocols can be used to transfer the data. After transfer, the diagnostic data is stored in an appropriate location, such as a database, file system, or a specific diagnostic storage server.

[0112] Fault Analysis and Location: Based on the acquired diagnostic data, the remote diagnostic module can perform further fault analysis and location. This may involve applying data analysis algorithms, machine learning techniques, or expert rules to parse the diagnostic data and extract key information. By analyzing and comparing the diagnostic data, the root cause of the fault or the location of the fault can be determined.

[0113] Generate Diagnostic Reports: Based on the results of fault analysis and location, the remote diagnostic module can generate diagnostic reports. The reports should include detailed information about the fault, including its type, root cause, time of occurrence, and possible solutions. Reports can be presented in text, charts, or other visual formats and sent to relevant personnel or systems for further processing.

[0114] The decision support module includes:

[0115] Fault Diagnosis Result Analysis: Based on the output and analysis of the previous fault diagnosis module, the decision support module assesses and analyzes the severity and impact of the fault. This may involve determining the urgency and impact of the fault based on factors such as fault type, location of the fault, and equipment importance.

[0116] Recommended Maintenance Measures: Based on the fault diagnosis results and current operating status, the decision support module can recommend appropriate maintenance measures. These measures can be remote operations, such as remotely restarting the synchronous condenser or making configuration changes. For severe faults or situations requiring on-site intervention, the decision support module can suggest notifying maintenance personnel for on-site repairs.

[0117] Decision Making and Execution: Based on the recommended maintenance measures, the decision support module will formulate corresponding decision-making solutions. This may involve generating maintenance work orders, sending notifications to maintenance personnel, or executing remote operation commands. The decision support module should ensure the accuracy and feasibility of the decision-making solutions and integrate with other systems (such as the maintenance management system) when necessary.

[0118] Monitoring and Feedback: Once a decision is made, the decision support module monitors the progress of the maintenance process and provides feedback based on the actual situation. It tracks the execution status of maintenance operations, collects data during the maintenance process, and monitors the recovery status of the synchronous condenser equipment. Through timely monitoring and feedback, the decision support module ensures the smooth progress of the maintenance process and allows for timely adjustments to achieve rapid recovery and troubleshooting.

[0119] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or substitute some of the technical features. Any modifications, substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A remote fault diagnosis and intelligent decision system for phase modulator, characterized in that, The modules include: Phase modifier monitoring module: for collecting monitoring data of the phase modifier device and converting the data into feature vector representation; Fault detection module: Step 1: Prepare feature vector: according to the feature vector obtained from the monitoring data of the phase modifier device, the feature vector is taken as the input data set; Step 2: Select division dimension: select a division dimension, that is, which feature to divide on, select according to the feature value range and variance index in the feature vector, the goal of selecting the division dimension is to uniformly divide the data on each node; Step 3: Select division value: according to the selected division dimension, select a division value, divide the data set into two subspaces, and select the division value according to the median of the division dimension in the feature vector; Step 4: Build node: create a node to represent the division dimension and the division value, take the division value as the split point of the node, and divide the data set into two subsets; Step 5: Recursively build: for each subset, repeat steps 2 to 4, recursively build the K-D tree, select new division dimension and division value, and create new nodes; Step 6: Termination condition: recursively build until the termination condition is met, the termination condition can be that the number of data in the node reaches a predetermined threshold, or the maximum depth of the tree is reached; Step 7: Construction is completed: when all subspaces are constructed, the K-D tree construction is completed; Remote diagnosis module: by establishing a remote connection with the phase modifier device, obtain diagnostic data to further analyze and locate the fault of the system; Decision support module: according to the results of fault diagnosis and the current running situation, provide intelligent decision support.

2. A remote fault diagnosis and intelligent decision system for a phase modifier according to claim 1, characterized in that: The phase modifier monitoring module includes: Use sensors or monitoring devices to obtain monitoring data of the phase modifier device, including current, voltage, frequency, phase angle, power factor, ensure that the selected sensors or devices can accurately collect data related to the running state of the phase modifier device; Preprocess the collected monitoring data to ensure data quality and consistency; According to the characteristics and fault modes of the phase modifier device, extract relevant features from the preprocessed monitoring data, which should be able to capture information related to the fault state; Combine the features extracted from the monitoring data into a feature vector representation, each feature vector should contain a set of feature values, each of which corresponds to a feature, according to the number and type of features, the feature vector is a one-dimensional array or a multi-dimensional matrix, ensure that the dimension of the feature vector is appropriate and can fully describe the monitoring information of the phase modifier device.

3. A remote fault diagnosis and intelligent decision system for a phase modifier according to claim 2, characterized in that: The phase modifier monitoring module includes: Step 1: Assign data points: for each data point, calculate its distance from all cluster centers and assign it to the cluster to which the nearest cluster center belongs; Step 2: Update cluster center: for each cluster, calculate the average value of its data points and take the average value as the new cluster center position; Step 3: Repeat iteration: repeat steps 1 and 2 until a predefined termination condition is reached, the termination condition is that the cluster center is stable and does not change or reaches a predefined number of iterations; Step 4: Cluster completion: When the termination condition is met, the K-means clustering algorithm is completed, and each data point is assigned to a cluster center.

4. The remote fault diagnosis and intelligent decision system for a phase modifier according to claim 3, characterized in that: The phase modulation machine monitoring module includes: Cluster check: For each data point, check its belonging cluster by accessing the data point's label or cluster index to determine the cluster the data point belongs to; Anomaly detection: For each data point in the belonging cluster, perform fault detection, and the abnormal data points represent potential faults or abnormal situations, including: Calculate the Z-score of each data point: For a given data point, calculate the ratio of its standard deviation (σ) to the mean (μ) of the data set by the following formula: Where X represents the value of the data point, μ represents the mean of the data set, and σ represents the standard deviation of the data set; Set threshold: According to the actual demand and the characteristics of the data set, set a threshold to determine which Z-score values are considered outliers; Label outliers: For data points with Z-score values higher than the threshold, label them as outliers, and the labeled data points are considered potential fault points.

5. A remote fault diagnosis and intelligent decision system for a phase modifier according to claim 4, characterized in that: The phase modulation machine monitoring module includes: Fault type identification: Determine the fault type by analyzing and processing the fault data points; Generate fault alarm: Generate a fault alarm based on the fault type, timestamp, and other related information.

6. The remote fault diagnosis and intelligent decision system for a phase modifier according to claim 1, characterized in that: The remote diagnosis module includes: Establish remote connection: The remote diagnosis module establishes a remote connection with the phase modulation machine device through a remote connection protocol to obtain more diagnostic data; Obtain diagnostic data: The remote diagnosis module requests and obtains diagnostic data from the phase modulation machine device, including log files, configuration information, performance indicators, and sensor data. Fault analysis and positioning: Based on the obtained diagnostic data, the remote diagnosis module performs further fault analysis and positioning; Generate diagnosis report: Based on the results of fault analysis and positioning, the remote diagnosis module can generate a diagnosis report.

7. A remote fault diagnosis and intelligent decision system for a phase modifier according to claim 6, characterized in that: The decision support module includes: Fault diagnosis result analysis: Based on the output results and analysis of the previous fault diagnosis module, the decision support module evaluates and analyzes the severity and impact of the fault; Maintenance measure recommendation: Based on the fault diagnosis results and the current operating conditions, the decision support module recommends appropriate maintenance measures; Decision making and execution: Based on the recommendation of maintenance measures, the decision support module will formulate the corresponding decision scheme; Monitoring and feedback: The decision support module monitors the progress of the maintenance process and provides feedback according to the actual situation.

Images