Intelligent fault alarm and analysis method and device, electronic equipment and medium

Abstract

The application provides a kind of intelligent fault alarm and analysis method, device, electronic equipment and medium, method includes real-time acquisition current processing equipment state data, the identity of current processing equipment, material batch is bound;If state data is abnormal signal, the duration proportion or frequency of abnormal signal is counted in the preset sliding time window;When duration proportion or frequency is greater than or equal to preset threshold, generate effective alarm event;Determine the upstream processing equipment of current processing equipment based on the preset process knowledge model, and according to the identity of material batch, trace back and query the historical alarm data of material batch in upstream processing equipment;Based on historical alarm data, determine the root cause equipment that triggers effective alarm event;Generate and output alarm information.The method solves the problem of invalid alarm and difficult root cause positioning in single crystal silicon processing by the cooperative use of sliding window de-bouncing alarm cleaning and upstream correlation analysis based on batch process model.

Description

Technical Field

[0001] This application relates to the field of industrial equipment monitoring technology, specifically to an intelligent fault alarm and analysis method, device, electronic equipment, and medium. Background Technology

[0002] Monocrystalline silicon processing lines (including processes such as squaring and grinding) are undergoing a transformation towards digitalization and automation, and commonly use general-purpose IoT platforms (such as ThingsBoard) for equipment monitoring. However, existing monitoring solutions have significant technical shortcomings in practical applications.

[0003] First, the strong electromagnetic environment in the workshop easily causes high-frequency jitter in sensor signals. The simple alarm mechanism of the IoT platform based on single-point thresholds will generate a massive number of instantaneous false alarms, forming an "alarm storm." This not only seriously interferes with operation and maintenance but also masks real equipment anomalies. Second, the existing system only displays single-point equipment alarms and lacks the ability to analyze the upstream and downstream correlations based on the specific process flow of monocrystalline silicon, making it difficult to locate the root cause of the fault and resulting in a lag in early warning capabilities.

[0004] In summary, there is an urgent need for a monitoring method and equipment that can filter out alarm storms and promptly locate the root cause of faults. Summary of the Invention

[0005] To address the technical problems existing in the background art, the first aspect of this application provides an intelligent fault alarm and analysis method, including:

[0006] Real-time collection of status data of the current processing equipment, and binding the identity of the current processing equipment with the identity of the material batch;

[0007] If the current status data of the processing equipment is an abnormal signal, the duration percentage or frequency of the abnormal signal is counted within the preset sliding time window.

[0008] When the duration or frequency of abnormal signals is greater than or equal to a preset threshold, a valid alarm event is generated.

[0009] In response to valid alarm events, the upstream processing equipment of the current processing equipment is determined based on a preset process knowledge model, and the historical alarm data of the material batch in the upstream processing equipment is traced and queried according to the identification of the material batch.

[0010] Based on historical alarm data, identify the root cause device that triggered the valid alarm event;

[0011] Generate and output alarm information, which includes the alarm content of the current processing equipment and the equipment information of the root cause equipment.

[0012] Optionally, the duration percentage of abnormal signals can be counted within a preset sliding time window, specifically including: adaptively configuring the length of the sliding time window and a preset threshold according to the type and / or status data of the abnormal signals;

[0013] Within the sliding time window, the duration of the abnormal signal and the ratio of the duration to the length of the sliding time window are calculated in real time to obtain the duration percentage.

[0014] A valid alarm event is generated when the duration percentage is greater than or equal to a preset threshold.

[0015] Optionally, the preset process knowledge model includes process nodes and equipment nodes in the material processing process, as well as the material flow relationship and process dependency relationship between process nodes and equipment nodes.

[0016] Optionally, based on historical alarm data, the root cause device that triggered the valid alarm event can be identified, specifically including:

[0017] Obtain the historical processing time period of the material batch on the upstream processing equipment;

[0018] Extract alarm data located within the historical processing time period;

[0019] Based on the process knowledge model, causal analysis is performed on the alarm data, and the upstream processing equipment corresponding to the alarm data that has a causal relationship with the valid alarm events of the current processing equipment is identified as the root cause equipment.

[0020] Optionally, after generating and outputting alarm information, the method further includes:

[0021] Based on the process knowledge model, determine the downstream processing equipment of the current processing equipment;

[0022] Based on valid alarm events and material batch flow information, predict the impact range of valid alarm events on downstream processing equipment and add the impact range to the alarm information.

[0023] Optionally, generate and output alarm information, specifically including:

[0024] Obtain the base of the material processing technology flow chart;

[0025] Map the alarm content of the current processing equipment and the information of the root cause equipment to the corresponding process node on the process flow chart base.

[0026] Generate and output a panoramic alarm view of the process, which can visualize the propagation path and impact range of the fault.

[0027] Optionally, after collecting the current status data of the processing equipment in real time, the method also includes:

[0028] When the duration or percentage of the duration of an abnormal signal is less than a preset threshold, the abnormal signal is determined to be an interference signal and the interference signal is filtered out.

[0029] A second aspect of this application provides an intelligent fault alarm and analysis device, comprising:

[0030] The data acquisition module is used to collect the status data of the current processing equipment in real time and bind the identity of the current processing equipment with the identity of the material batch;

[0031] The judgment module is used to determine whether the status data of the current processing equipment is an abnormal signal. If the status data of the current processing equipment is an abnormal signal, it counts the duration percentage or frequency of the abnormal signal within a preset sliding time window; and it compares the duration percentage or frequency of the abnormal signal with a preset threshold. When the duration percentage or frequency of the abnormal signal is greater than or equal to the preset threshold, a valid alarm event is generated.

[0032] The filtering module is used to respond to valid alarm events, determine the upstream processing equipment of the current processing equipment based on a preset process knowledge model, and trace and query the historical alarm data of the material batch in the upstream processing equipment according to the identification of the material batch.

[0033] The output module is used to determine the root cause device that triggered the valid alarm event based on historical alarm data; and to generate and output alarm information, which includes the alarm content of the current processing equipment and the device information of the root cause device.

[0034] Optionally, the judgment module is specifically used for:

[0035] The length of the sliding time window and the preset threshold are adaptively configured according to the type of abnormal signal;

[0036] Within the sliding time window, the duration of the abnormal signal and the ratio of the duration to the length of the sliding time window are calculated in real time to obtain the duration percentage.

[0037] A valid alarm event is generated when the duration percentage is greater than or equal to a preset threshold.

[0038] Optionally, the preset process knowledge model includes process nodes and equipment nodes in the material processing process, as well as the material flow relationship and process dependency relationship between process nodes and equipment nodes.

[0039] Optionally, the output module is specifically used for:

[0040] Obtain the historical processing time period of the material batch on the upstream processing equipment;

[0041] Extract alarm data located within the historical processing time period;

[0042] Based on the process knowledge model, causal analysis is performed on the alarm data, and the upstream processing equipment corresponding to the alarm data that has a causal relationship with the valid alarm events of the current processing equipment is identified as the root cause equipment.

[0043] Optionally, the device also includes a prediction module for determining the downstream processing equipment of the current processing equipment based on a process knowledge model; and for predicting the impact range of valid alarm events on downstream processing equipment based on valid alarm events and material batch flow information, and adding the impact range to the alarm information.

[0044] Optionally, the output module is specifically used for:

[0045] Obtain the base of the material processing technology flow chart;

[0046] Map the alarm content of the current processing equipment and the information of the root cause equipment to the corresponding process node on the process flow chart base.

[0047] Generate and output a panoramic alarm view of the process, which can visualize the propagation path and impact range of the fault.

[0048] Optionally, the judgment module is also used for:

[0049] When the duration or percentage of the duration of an abnormal signal is less than a preset threshold, the abnormal signal is determined to be an interference signal and the interference signal is filtered out.

[0050] A third aspect of this application provides an electronic device, comprising:

[0051] At least one processor; and,

[0052] A memory that is communicatively connected to at least one processor; wherein,

[0053] The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the intelligent fault alarm and analysis method provided in any of the first aspects of this application.

[0054] The fourth aspect of this application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the intelligent fault alarm and analysis method provided in any of the first aspects of this application.

[0055] The beneficial effects that this application can achieve are:

[0056] This application provides an intelligent fault alarm and analysis method that effectively solves two major technical challenges in monocrystalline silicon processing: alarm storms and difficulties in locating root causes across processes. Specifically, it utilizes a configurable sliding time window to statistically analyze the continuous proportion of abnormal signals, thoroughly filtering out instantaneous false alarms caused by electromagnetic interference, enabling alarms to accurately focus on real faults from a massive amount of invalid information. Simultaneously, based on a preset process knowledge model and material batch identification, it automatically traces historical alarm data from upstream processing equipment and performs causal analysis when a valid alarm is triggered, accurately locating the root cause equipment. Ultimately, this achieves a qualitative leap in the alarm system, transforming it from a system with numerous and complex alarms to one with fewer and more accurate alarms, significantly improving maintenance response speed and production line early warning capabilities. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 A flowchart illustrating an optional intelligent fault alarm and analysis method provided in an embodiment of this application is shown.

[0059] Figure 2 The diagram illustrates a flowchart of an intelligent fault alarm and analysis method according to a specific embodiment of this application.

[0060] Figure 3 A schematic diagram of the control logic of an optional intelligent fault alarm and analysis device provided in an embodiment of this application is shown;

[0061] Figure 4 This application provides a structural diagram of a device for intelligent fault alarm and analysis.

[0062] Figure 5 This is a schematic diagram of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0063] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0064] To address the issues of alarm storms and difficulties in locating the root cause of faults in the background technology, current technologies typically employ a sliding window-based IoT device early warning method. This method involves acquiring real-time streaming sensor data from sensors on production line equipment via edge devices, caching the sensor data using a sliding window of a streaming computing engine, and determining whether the cached sensor data within the sliding window meets a preset repetition condition. Specifically, it involves counting the number of repetitive events that reach a preset alarm threshold within the window. When the number of repetitive events reaches a preset number, an anomaly is identified in the device, triggering an alarm. This approach filters out transient fluctuation signals through the sliding window.

[0065] However, the aforementioned early warning methods only perform simple statistical noise reduction on the sensor data of single-point devices from a signal perspective, ignoring the material flow relationships and process dependencies between processes such as cutting, squaring, and grinding in the monocrystalline silicon processing line. Lacking a knowledge model supporting the process dimension, this method can only filter out short-term, instantaneous noise such as electromagnetic interference, and cannot identify cross-process, slowly changing quality risks caused by factors such as process parameter drift and equipment performance degradation. Furthermore, the alarm model of this method is static and isolated, lacking causal analysis capabilities. It can only display the alarm status of single-point devices and cannot correlate and trace the abnormal signals of the current device with historical alarm data of the same material batch in upstream processing equipment. This makes it difficult to locate the root cause of the fault, forcing maintenance personnel to rely on extensive manual investigation based on experience, which is not only inefficient but also severely delays early warning capabilities.

[0066] Based on this, the first aspect of this application provides an intelligent fault alarm and analysis method. This method aims to determine whether an alarm event is valid by introducing a debouncing algorithm based on a configurable sliding time window, which statistically analyzes the duration ratio or frequency of abnormal signals within a preset window. This effectively filters out transient noise such as electromagnetic interference at the data source, thus eliminating alarm storms. Furthermore, this application constructs a process knowledge model for the monocrystalline silicon processing line, defining the material flow relationships and process dependencies between process nodes such as cutting, squaring, and grinding, as well as equipment nodes. When a valid alarm event is triggered, it automatically traces the historical alarm data of the material batch processed by the current processing equipment back along the process route to the upstream processing equipment based on the batch identifier. Then, based on the process dependencies, it performs causal analysis to accurately locate the root cause equipment causing the current fault, thereby improving the operation and maintenance response speed and fault early warning capability of the monocrystalline silicon processing line.

[0067] This application provides an intelligent fault alarm and analysis method, such as... Figure 1 As shown, the method includes:

[0068] S1. Collect the status data of the current processing equipment in real time and bind the identity of the current processing equipment with the identity of the material batch.

[0069] Specifically, step S1 uses sensors and industrial data acquisition components deployed on equipment at each stage of the monocrystalline silicon processing production line to acquire real-time status data of the processing equipment during operation. This status data includes, but is not limited to, multi-dimensional parameters reflecting the equipment's operating condition, such as equipment vibration amplitude, spindle speed, coolant flow rate, processing temperature, and motor current. Simultaneously, it acquires the batch information of the material being processed on the equipment, extracts the batch identifier, and establishes a binding relationship between the equipment identifier and the batch identifier. This ensures that each subsequent set of status data carries clear equipment origin and batch information, enabling precise batch tracing and causal relationships along the process route. This fundamentally solves the problem of fragmented equipment and material data in traditional monitoring solutions, hindering refined traceability.

[0070] S2. If the current status data of the processing equipment is an abnormal signal, the duration percentage or frequency of the abnormal signal is counted within the preset sliding time window.

[0071] Specifically, in step S2, during the real-time acquisition of status data, it is necessary to continuously monitor whether the status data exceeds the preset normal operating threshold range. Once the status data deviates from the normal operating threshold range, it is determined that the status data at that moment is an abnormal signal. Due to strong electromagnetic interference in the monocrystalline silicon processing workshop, sensors are prone to generating instantaneous pulse-like false abnormal signals after being interfered with. If an alarm is directly triggered based on this, interference signals generated by electromagnetic fields or noise will also be classified as abnormal signals, leading to an alarm storm. Therefore, step S2 introduces a configurable sliding time window mechanism to continuously monitor and statistically analyze abnormal signals within this sliding time window. In one specific embodiment, the statistical method can be to count the cumulative duration of the abnormal signal within the sliding time window and calculate the proportion of this cumulative duration to the total duration of the sliding time window to obtain the duration percentage. In another specific embodiment, the statistical method can also be to count the number of times the abnormal signal appears within the sliding time window to obtain the abnormal frequency.

[0072] In some embodiments, the length of the sliding time window can be pre-configured based on the electromagnetic environment characteristics of the processing equipment, the process type, and the sampling frequency of the sensor. This ensures that the window size effectively covers the duration of instantaneous interference pulses, preventing interference signals from being misjudged as real faults due to an excessively small window, while also responding sensitively to real anomalies, avoiding alarm delays due to an excessively large window. This application, through sliding window statistics, incorporates instantaneous abnormal signals that would otherwise directly trigger alarms into the time dimension for re-evaluation, thereby effectively filtering out short-term random noise such as electromagnetic interference and fundamentally solving the alarm storm problem caused by instantaneous signal jitter.

[0073] S3. When the duration or frequency of abnormal signals is greater than or equal to a preset threshold, a valid alarm event is generated.

[0074] Specifically, step S3 compares the duration percentage or frequency of the abnormal signals obtained in step S2 with a pre-configured effective alarm threshold. The pre-configured threshold is a judgment standard determined comprehensively based on the process tolerance of the monocrystalline silicon processing steps, historical fault characteristics, and on-site operation and maintenance experience. In one specific embodiment, the threshold can be configured such that the duration percentage of the abnormal signal within the sliding time window reaches 80%, or the frequency of the abnormal signal reaches 80% of the total number of sampling points within the window. When the duration percentage or frequency of the abnormal signal is greater than or equal to the pre-configured threshold, it indicates that the abnormal signal is not a momentary jitter caused by electromagnetic interference, but rather a continuous abnormality reflecting a deviation of the equipment's operating state from normal conditions, and the system generates an effective alarm event accordingly.

[0075] Based on the above embodiments, this application establishes a reliable anomaly judgment checkpoint by setting a preset threshold. The quantitative indicators obtained in step S2 are compared deterministically with the preset threshold. Only anomalies that are sustained or repeated frequently enough to cross the threshold are recognized as valid alarms, fundamentally solving the problem of alarm storms interfering with operation and maintenance personnel.

[0076] S4. In response to a valid alarm event, determine the upstream processing equipment of the current processing equipment based on the preset process knowledge model, and trace and query the historical alarm data of the material batch in the upstream processing equipment according to the identification of the material batch.

[0077] Specifically, after a valid alarm event is generated in step S3, step S4 automatically triggers the cross-process alarm association and traceability process. First, the system calls a pre-built process knowledge model. This model is a knowledge base describing the material flow relationships and process dependencies between process nodes and equipment nodes in a monocrystalline silicon processing line. It internally defines complete process flow sequences such as truncation → squaring → grinding, as well as the processing equipment corresponding to each process and their upstream and downstream adjacency relationships. By querying this process knowledge model, starting from the processing equipment that currently generates the valid alarm event, the system automatically determines one or more upstream processing equipment that processed this batch of material before the current process, following the reverse path of material flow.

[0078] Furthermore, after identifying the upstream processing equipment, the system utilizes the binding relationship between the current processing equipment's identifier and the material batch's identifier established in step S1 to accurately extract alarm data related to that material batch from the upstream processing equipment's historical alarm records.

[0079] In this embodiment, the method combines a process knowledge model with batch identification to connect alarm information that was originally isolated and scattered across individual devices, according to the physical path of material flow. This breaks the fragmented nature of alarm data from different devices in traditional monitoring solutions. When any device triggers a valid alarm, the system can automatically trace upstream processes without manual intervention, accurately extracting historical alarm data from massive amounts of irrelevant data that may contain clues to faults.

[0080] S5. Based on historical alarm data, identify the root cause device that triggered the valid alarm event.

[0081] Specifically, in step S5, based on the historical alarm data of the material batch in the upstream processing equipment, root cause analysis needs to be further performed. First, the system obtains the historical processing time period of the material batch in the upstream processing equipment according to the binding relationship established in step S1. Candidate alarm data that occurred within the historical processing time period are filtered from the historical alarm data, thereby further narrowing the analysis scope to alarm events directly related to the material batch and excluding irrelevant alarm interference generated by the upstream equipment in other time periods.

[0082] Subsequently, the system invokes the process dependencies in the preset process knowledge model to perform causal analysis on the candidate alarm data. This causal analysis, based on the inter-process parameter transmission rules and fault propagation mechanisms defined in the process knowledge model, determines whether the equipment anomalies reflected in historical alarm events of upstream equipment can trigger valid alarm events in downstream equipment through material flow and process transmission chains. For example, abnormal cutting line tension in the slitting process may lead to defects on the silicon rod end face. If these defects are not detected or corrected during the squaring process, they can be transmitted to the grinding process, triggering edge chipping or dimensional deviation alarms. Based on this, the system identifies alarm data causally related to the current valid alarm event from the candidate alarm data and determines the upstream processing equipment corresponding to that alarm data as the root cause equipment.

[0083] In this embodiment, the method combines batch tracing, time constraints, and process causal reasoning to construct a complete cross-process fault causal reasoning chain. Based on a process knowledge model, the system proactively performs correlation analysis and causal inference, accurately filtering out the true source event and its corresponding equipment that caused the current fault from numerous historical alarms. This achieves precise fault root cause localization, improving the fault response efficiency and early warning capability of the monocrystalline silicon processing production line.

[0084] S6. Generate and output alarm information, which includes the alarm content of the current processing equipment and the equipment information of the root cause equipment.

[0085] Specifically, after identifying the root cause device, the processing results of the aforementioned steps are integrated into alarm information for visual output. This alarm information includes at least the alarm content of the current processing equipment and the device information of the root cause device. In this embodiment, the alarm content includes the time of alarm occurrence, alarm type, specific numerical value of the abnormal state data, and the process node where the current processing equipment is located. The device information of the root cause device includes the device number of the root cause device, its corresponding process node, historical alarm content, and the time of alarm occurrence.

[0086] Furthermore, this application uses a process flow diagram of a monocrystalline silicon processing line as a visualization base. This process flow diagram arranges the equipment nodes according to the actual process sequence, such as truncation, squaring, and grinding. The system maps the alarm content of the current processing equipment to the corresponding process node on the flow diagram for highlighting and warning. At the same time, it marks the root cause equipment differently on the flow diagram and uses graphical elements such as lines or arrows to intuitively show the path and scope of the fault propagation from the root cause equipment along the process route to the current processing equipment, ultimately generating a panoramic alarm view of the process.

[0087] In this embodiment of the application, the method improves alarm handling efficiency and production line operation and maintenance level by visualizing the output, eliminating the need for operation and maintenance personnel to check one by one in a large number of alarm lists. The location of the current fault, the root cause, and the complete fault propagation chain can be seen on the process flow diagram, thereby improving the efficiency of alarm handling and the level of production line operation and maintenance.

[0088] In some embodiments, such as Figure 2 As shown, the duration percentage of abnormal signals is statistically analyzed within a preset sliding time window, specifically including:

[0089] S21. Adaptively configure the length of the sliding time window and the preset threshold according to the type of abnormal signal and / or status data.

[0090] Specifically, in the actual operating environment of a monocrystalline silicon processing production line, abnormal signals generated by equipment in different processes exhibit differentiated temporal characteristics and fault evolution patterns. For example, abnormal signals caused by strong electromagnetic interference typically manifest as instantaneous pulse-like spikes, lasting from milliseconds to seconds, characterized by strong suddenness and rapid decay; while abnormal signals caused by equipment performance degradation or process parameter drift often present as slowly changing, gradual deviations that may last for tens of seconds or even minutes.

[0091] To address the differences in characteristics among the various anomaly types, this step first identifies the type of the currently occurring anomaly signal before performing sliding window statistics. The system automatically determines whether the anomaly signal is a transient interference type or a progressive fault type by analyzing characteristic parameters such as the amplitude change rate, fluctuation frequency, and duration of deviation from the normal operating threshold range. When identified as a transient interference type anomaly signal, the system automatically configures a shorter sliding time window length to fully cover the duration of the interference pulse and sets a higher effective alarm threshold to ensure that only repeatedly occurring strong interference triggers an alarm, thereby maximizing the filtering of random noise. When identified as a progressive fault type anomaly signal, the system automatically configures a longer sliding time window length to accommodate the slow evolution of fault parameters and appropriately lowers the effective alarm threshold to ensure sufficient detection sensitivity for gradually worsening real faults and avoid missed detections due to excessively high thresholds.

[0092] S22. Within the sliding time window, calculate the duration of the abnormal signal in real time, and the ratio of the duration to the length of the sliding time window to obtain the duration percentage.

[0093] Specifically, after detecting an abnormal signal and initiating a sliding time window as described above, real-time time-dimensional statistics of the abnormal signal are required. Whenever state data is determined to be an abnormal signal, the system records the start and end times of the abnormal signal and accumulates the duration of each abnormal signal within the sliding time window to obtain the cumulative duration. As the sliding time window advances along the time axis, the time interval covered within the window is continuously updated, and the cumulative duration changes dynamically accordingly.

[0094] At any given statistical moment, the system divides the cumulative duration within the current window by the preset length of the sliding time window to obtain the percentage of the abnormal signal's duration. This percentage is expressed as a percentage, reflecting the proportion of time the device is in an abnormal state within the current observation window. In a specific embodiment, if the sliding time window length is 30 seconds and the cumulative abnormal duration within the window is 24 seconds, then the duration percentage is 80%.

[0095] In this embodiment, the method uses duration percentage rather than simply the number of anomalies as a statistical indicator. This allows the de-jitter determination to focus not only on the frequency of anomalies but also on the persistence and stability of the abnormal state. Compared to simply counting frequency, duration percentage can more effectively distinguish between continuous steady-state anomalies and intermittent fluctuations, avoiding false alarms caused by inflated anomaly counts due to frequent sensor jumps. This improves the accuracy and robustness of the alarm de-jitter algorithm.

[0096] In some embodiments, the preset process knowledge model includes process nodes and equipment nodes in the material processing process, as well as the material flow relationships and process dependencies between process nodes and equipment nodes. Specifically, a process node refers to each processing step involved in the monocrystalline silicon processing. In one specific embodiment, a process node includes at least a truncation step, a squaring step, and a grinding step. The process nodes are arranged in the order of the actual production process, forming a complete process chain. An equipment node is one or more specific processing equipment corresponding to each process node. Each equipment node has a unique equipment identification and is bound to its respective process node. The material flow relationship between process nodes and equipment nodes indicates the order in which materials flow from a certain equipment in the previous process to a certain equipment in the next process, forming a clear upstream and downstream equipment correspondence. The process dependency relationship between process nodes and equipment nodes indicates the type of quality impact that fluctuations in processing parameters or equipment abnormalities in the upstream process may have on the downstream process.

[0097] Based on the above process knowledge model, the system can accurately grasp the upstream and downstream positions of each processing equipment in the overall production line and its process correlation characteristics. When any equipment triggers a valid alarm, it can automatically determine its upstream equipment without manually consulting process documents and perform reverse tracing according to the material flow path.

[0098] In some embodiments, such as Figure 2 As shown, based on historical alarm data, the root cause devices that triggered valid alarm events were identified, specifically including:

[0099] S41. Obtain the historical processing time period of the material batch in the upstream processing equipment.

[0100] Specifically, after identifying the upstream processing equipment of the current processing equipment, in order to further narrow down the search scope of alarm data and improve the accuracy of root cause localization, the system needs to obtain the specific processing time period of the current material batch on that upstream processing equipment. Since the system has established a binding relationship between the identification of the processing equipment and the identification of the material batch, and this binding relationship is continuously recorded and updated throughout the entire material flow process, the flow trajectory data of any equipment stores the timestamp information of each material batch entering and leaving the equipment.

[0101] In this embodiment of the application, by accurately acquiring the processing time period, the system can further refine the retrieval scope of subsequent alarm data from the entire historical record of the upstream equipment to the time window directly related to the batch of materials, thereby avoiding including irrelevant alarms generated by the equipment in other time periods in the analysis scope, thus improving the accuracy and efficiency of root cause tracing.

[0102] S42. Extract alarm data located within the historical processing time period.

[0103] Specifically, based on the historical processing time periods of the material batches at the upstream processing equipment, the system performs precise time-dimensional filtering on the historical alarm records of the upstream processing equipment. Using the start and end times of the historical processing time periods as search boundaries, the system performs time interval queries in the alarm database of the upstream processing equipment, extracting alarm data whose occurrence time falls within that historical processing time period, and using this as a candidate data set for subsequent root cause analysis.

[0104] Because upstream processing equipment may generate a large number of alarm records during daily operation, and these alarms involve processing of different time periods and different material batches, including all of them in the analysis without screening would introduce a large amount of noise data unrelated to the current fault, interfering with the root cause determination results. Therefore, this method precisely trims the alarm data by defining a specific historical processing time period, retaining only alarm events that overlap with the current material batch in the time dimension. That is, only alarms that occurred during the upstream equipment's processing of this batch of materials will be included in the candidate range, thereby focusing the focus of subsequent root cause analysis from all historical alarms of the equipment to a limited set of alarm records directly related to this material batch.

[0105] S43. Based on the process knowledge model, perform causal analysis on the alarm data and identify the upstream processing equipment corresponding to the alarm data that has a causal relationship with the valid alarm events of the current processing equipment as the root cause equipment.

[0106] Specifically, based on the candidate alarm data extracted from the historical processing time period, a preset process knowledge model is invoked to perform causal reasoning, identify alarm events that have a causal relationship with the current valid alarm events from the candidate alarm data, and then pinpoint the root cause device.

[0107] In some embodiments, the specific process of causal analysis is as follows: the system extracts each candidate alarm data and analyzes its alarm type, alarm content, and alarm occurrence time one by one. For each candidate alarm data, the system, based on the process dependency relationship between processes defined in the process knowledge model, determines whether the upstream equipment anomaly reflected by the candidate alarm data can substantially affect the processing quality or equipment status of the downstream current processing equipment through material flow and process transfer chain, and ultimately trigger the current valid alarm event.

[0108] Once the system identifies candidate alarm data with a causal relationship, it determines the upstream processing equipment corresponding to that candidate alarm data as the root cause device that triggered the current valid alarm event. If there are multiple candidate alarm data with a causal relationship, the system can further determine the highest priority primary cause device as the root cause device based on the time sequence of alarm occurrence, alarm severity, and the strength of the causal relationship with the current valid alarm event, and mark other related devices as related influencing devices.

[0109] In some embodiments, such as Figure 2 As shown, after generating and outputting alarm information, the method also includes:

[0110] S6. Based on the process knowledge model, determine the downstream processing equipment of the current processing equipment.

[0111] Specifically, after completing root cause localization, the system further utilizes the material flow relationships defined in the process knowledge model to assess the potential downstream impacts of the fault. Starting with the processing equipment that currently generates a valid alarm event, the system follows the forward material flow path defined in the process knowledge model to query one or more downstream processing equipment that receive the same batch of materials for the next processing step after the current processing equipment. It then obtains information such as the equipment number, the process node to which the downstream processing equipment belongs, and its current operating status.

[0112] Simultaneously, based on the process dependencies defined in the process knowledge model, and combined with the alarm type and severity of the currently valid alarm events, the system predicts the potential impact on downstream processing equipment. For example, if the current alarm type involves machining dimensional deviations, the process knowledge model can infer that these dimensional deviations will directly affect the clamping accuracy and machining quality of downstream equipment, thereby predicting a high risk of cascading anomalies in downstream equipment.

[0113] By proactively identifying and predicting the impact on downstream processing equipment through this step, maintenance personnel can not only know the root cause of the fault, but also predict the potential downstream impact of the fault in advance, thereby taking preventive intervention measures before the actual spread of the fault's impact.

[0114] S7. Based on the valid alarm events and the flow information of material batches, predict the impact range of the valid alarm events on downstream processing equipment, and add the impact range to the alarm information.

[0115] Specifically, based on the fact that S has already determined the downstream processing equipment of the current processing equipment, the system further predicts the propagation trend of the fault. The system first obtains the current flow information of the material batch based on the established binding relationship, including whether the material batch has entered the downstream processing equipment, the expected time node for entering the downstream processing equipment, and the type of processing operation to be executed in the downstream processing equipment.

[0116] Subsequently, the system combines the alarm type of valid alarm events, the degree of deviation of abnormal state data, and the fault propagation rules between adjacent processes defined in the process knowledge model to predict the potential impact of valid alarm events on downstream processing equipment, so as to obtain prediction results.

[0117] Furthermore, the system integrates the above prediction results into impact range information, which includes at least the equipment number of the potentially affected downstream processing equipment, the corresponding process node, the possible anomaly type, the probability level of impact, and recommended preventive intervention measures. Finally, the system adds this impact range information to the generated alarm information as supplementary content, thus realizing a shift from passive response to proactive early warning.

[0118] In some embodiments, generating and outputting alarm information specifically includes:

[0119] S51. Obtain the base of the material processing process flow diagram.

[0120] Specifically, the system pre-builds and stores a process flow diagram of a monocrystalline silicon processing line. This process flow diagram graphically describes the complete processing flow from raw material input to finished product output, arranging each process node according to the actual production process sequence, and labeling the corresponding processing equipment and its identification at each process node. In this embodiment, the process flow diagram not only includes the static positional relationship of each process node and equipment node, but also visually shows the material flow path between each piece of equipment in the form of connecting lines, that is, the complete path of material flowing from the processing equipment of the cutting process to the processing equipment of the squaring process and the grinding process.

[0121] S52. Map the alarm content of the current processing equipment and the information of the root cause equipment to the corresponding process node on the process flow chart base.

[0122] Specifically, based on the acquired process flow diagram base, the system further performs a graphical mapping operation of alarm information. First, the system queries the process flow diagram base for the process node to which the current processing equipment belongs, based on the equipment identification identifier. Then, it generates an alarm notification marker at the corresponding position of that process node, enabling maintenance personnel to immediately identify the alarm location on the flow diagram. In this embodiment, the alarm notification marker includes core alarm information such as the alarm type of the current processing equipment, the specific value of the abnormal status data, and the time of alarm occurrence.

[0123] Subsequently, based on the identified root cause device's equipment information, the system queries the process node to which the root cause device belongs in the process flow diagram base, and generates a root cause identifier at that process node. The root cause identifier includes information such as the root cause device's equipment number, the process node to which it belongs, the historical alarm type, and the time when the alarm occurred.

[0124] S53. Generate and output a panoramic alarm view of the process. The panoramic alarm view of the process can visualize the propagation path and impact range of the fault.

[0125] Specifically, based on the completed mapping of alarm information and root cause information, the analysis results of the aforementioned steps are integrated to generate a unified visual alarm view. In this view, the system uses the process flow diagram as a background, and marks the alarm content of the current processing equipment and the information of the root cause equipment at the corresponding process nodes. It also uses graphical elements to visualize the propagation path of the fault, thereby intuitively showing the complete link of the fault from the root cause equipment to the current processing equipment along the process route.

[0126] In some embodiments, after real-time acquisition of the current processing equipment status data, the method further includes:

[0127] S32. When the duration or percentage of the duration of the abnormal signal is less than a preset threshold, the abnormal signal is determined to be an interference signal and the interference signal is filtered out.

[0128] Specifically, this step, together with step S3, constitutes the complete judgment closed loop of the sliding window debouncing algorithm. After the system completes the statistics of the duration or duration ratio of the abnormal signal within the preset sliding time window, if the calculation results show that the duration does not reach the preset time length threshold, or the duration ratio is less than the preset proportion threshold, it indicates that the abnormal signal only appears briefly or occurs intermittently within the observation window, and its persistence and stability are insufficient to constitute a reliable basis for judging the actual operating status of the equipment.

[0129] In this situation, the system determines that the abnormal signal is not caused by a real equipment malfunction or process anomaly, but rather by interference signals caused by external random factors such as the strong electromagnetic environment in the workshop, momentary sensor disturbances, or brief fluctuations in the data acquisition link. Based on this determination, the system performs a filtering operation on the abnormal signal, that is, it does not generate an alarm event, does not trigger subsequent cross-process traceability procedures, and does not record the abnormal signal as valid alarm data.

[0130] A second aspect of this application provides an intelligent fault alarm and analysis device, comprising:

[0131] The data acquisition module 101 is used to collect the status data of the current processing equipment in real time and bind the identity identifier of the current processing equipment with the identity identifier of the material batch;

[0132] The judgment module 201 is used to determine whether the status data of the current processing equipment is an abnormal signal. If the status data of the current processing equipment is an abnormal signal, it counts the duration percentage or frequency of the abnormal signal within a preset sliding time window. It is also used to compare the duration percentage or frequency of the abnormal signal with a preset threshold. When the duration percentage or frequency of the abnormal signal is greater than or equal to the preset threshold, a valid alarm event is generated.

[0133] The filtering module 301 is used to respond to valid alarm events, determine the upstream processing equipment of the current processing equipment based on a preset process knowledge model, and trace and query the historical alarm data of the material batch in the upstream processing equipment according to the identification of the material batch.

[0134] The output module 401 is used to determine the root cause device that triggered the valid alarm event based on historical alarm data; and to generate and output alarm information, which includes the alarm content of the current processing equipment and the device information of the root cause device.

[0135] Optionally, the judgment module 201 is specifically used for:

[0136] The length of the sliding time window and the preset threshold are adaptively configured according to the type of abnormal signal;

[0137] Within the sliding time window, the duration of the abnormal signal and the ratio of the duration to the length of the sliding time window are calculated in real time to obtain the duration percentage.

[0138] A valid alarm event is generated when the duration percentage is greater than or equal to a preset threshold.

[0139] Optionally, the preset process knowledge model includes process nodes and equipment nodes in the material processing process, as well as the material flow relationship and process dependency relationship between process nodes and equipment nodes.

[0140] Optionally, output module 401 is specifically used for:

[0141] Obtain the historical processing time period of the material batch on the upstream processing equipment;

[0142] Extract alarm data located within the historical processing time period;

[0143] Based on the process knowledge model, causal analysis is performed on the alarm data, and the upstream processing equipment corresponding to the alarm data that has a causal relationship with the valid alarm events of the current processing equipment is identified as the root cause equipment.

[0144] Optionally, the device also includes a prediction module 501, which is used to determine the downstream processing equipment of the current processing equipment based on the process knowledge model; and to predict the impact range of the valid alarm event on the downstream processing equipment based on the valid alarm event and the flow information of the material batch, and add the impact range to the alarm information.

[0145] Optionally, output module 401 is specifically used for:

[0146] Obtain the base of the material processing technology flow chart;

[0147] Map the alarm content of the current processing equipment and the information of the root cause equipment to the corresponding process node on the process flow chart base.

[0148] Generate and output a panoramic alarm view of the process, which can visualize the propagation path and impact range of the fault.

[0149] Optionally, the judgment module 201 is also used for:

[0150] When the duration or percentage of the duration of an abnormal signal is less than a preset threshold, the abnormal signal is determined to be an interference signal and the interference signal is filtered out.

[0151] A third aspect of this application provides an electronic device, comprising:

[0152] At least one processor; and,

[0153] A memory that is communicatively connected to at least one processor; wherein,

[0154] The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the intelligent fault alarm and analysis method provided in any of the first aspects of this application.

[0155] The fourth aspect of this application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the intelligent fault alarm and analysis method provided in any of the first aspects of this application.

[0156] A third aspect of this application provides a device for intelligent fault alarm and analysis, see [link to relevant documentation]. Figure 4 The figure illustrates a structural diagram of a device for intelligent fault alarm and analysis provided in an embodiment of this application, such as... Figure 4 As shown, the device includes a processor 401 and a memory 402:

[0157] The memory 401 is used to store computer programs and transmit the computer programs to the processor;

[0158] The processor 402 is used to execute the request control method described in the above embodiments according to the instructions in the computer program.

[0159] A fourth aspect of this application provides a computer-readable storage medium, see [link to previous document]. Figure 5 The figure illustrates a schematic diagram of a computer-readable storage medium provided in an embodiment of this application, such as... Figure 5 As shown, the computer-readable storage medium is used to store a computer program 501, which is used to execute the request control method described in the above embodiments.

[0160] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The device embodiments described above are merely illustrative. The units and modules described as separate components may or may not be physically separate. Furthermore, some or all of the units and modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0161] The above description is only a specific embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for intelligent fault alarm and analysis, characterized in that, include: Real-time collection of status data of the current processing equipment, and binding of the identification of the current processing equipment with the identification of the material batch; If the current status data of the processing equipment is an abnormal signal, the duration percentage or frequency of the abnormal signal is counted within a preset sliding time window; When the duration or frequency of the abnormal signal is greater than or equal to a preset threshold, a valid alarm event is generated. In response to the valid alarm event, the upstream processing equipment of the current processing equipment is determined based on a preset process knowledge model, and the historical alarm data of the material batch in the upstream processing equipment is traced and queried according to the identification of the material batch. Based on the historical alarm data, the root cause device that triggered the valid alarm event is identified; An alarm message is generated and output, which includes the alarm content of the current processing equipment and the equipment information of the root cause equipment.

2. The method according to claim 1, characterized in that, The step of calculating the duration percentage of the abnormal signal within a preset sliding time window specifically includes: The length of the sliding time window and the preset threshold are adaptively configured according to the type of the abnormal signal and / or the status data. Within the sliding time window, the duration of the abnormal signal and the ratio of the duration to the length of the sliding time window are calculated in real time to obtain the duration percentage.

3. The method according to claim 1, characterized in that, The preset process knowledge model includes process nodes and equipment nodes in the material processing process, as well as the material flow relationship and process dependency relationship between the process nodes and the equipment nodes.

4. The method according to claim 1, characterized in that, The step of determining the root cause device that triggered the valid alarm event based on the historical alarm data specifically includes: Obtain the historical processing time period of the material batch at the upstream processing equipment; Extract alarm data located within the historical processing time period; Based on the process knowledge model, causal analysis is performed on the alarm data, and the upstream processing equipment corresponding to the alarm data that has a causal relationship with the valid alarm event of the current processing equipment is identified as the root cause equipment.

5. The method according to claim 1, characterized in that, After generating and outputting alarm information, the method further includes: Based on the aforementioned process knowledge model, the downstream processing equipment of the current processing equipment is determined; Based on the valid alarm events and the flow information of the material batches, the impact range of the valid alarm events on the downstream processing equipment is predicted, and the impact range is added to the alarm information.

6. The method according to claim 1, characterized in that, Generate and output alarm information, specifically including: Obtain the base of the material processing technology flow chart; Map the alarm content of the current processing equipment and the information of the root cause equipment to the process node corresponding to the process flow chart base. Generate and output a panoramic alarm view of the process, which can visually display the propagation path and impact range of the fault.

7. The method according to claim 1, characterized in that, After real-time acquisition of the current status data of the processing equipment, the method further includes: When the duration or percentage of the duration of the abnormal signal is less than the preset threshold, the abnormal signal is determined to be an interference signal and the interference signal is filtered out.

8. An intelligent fault alarm and analysis device, characterized in that, include: The data acquisition module is used to collect the status data of the current processing equipment in real time and bind the identity identifier of the current processing equipment with the identity identifier of the material batch; The judgment module is used to determine whether the status data of the current processing equipment is an abnormal signal. If the status data of the current processing equipment is an abnormal signal, it counts the duration percentage or frequency of the abnormal signal within a preset sliding time window. It is also used to compare the duration percentage or frequency of the abnormal signal with a preset threshold. When the duration percentage or frequency of the abnormal signal is greater than or equal to the preset threshold, a valid alarm event is generated. The filtering module is used to respond to the valid alarm event, determine the upstream processing equipment of the current processing equipment based on a preset process knowledge model, and trace and query the historical alarm data of the material batch in the upstream processing equipment according to the identification of the material batch. The output module is used to determine the root cause device that triggered the valid alarm event based on the historical alarm data; and to generate and output alarm information, the alarm information including the alarm content of the current processing equipment and the device information of the root cause device.

9. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the intelligent fault alarm and analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the intelligent fault alarm and analysis method as described in any one of claims 1 to 7.

Images