Flour processing equipment fault prediction method and system based on big data

Abstract

The invention belongs to the technical field of fault prediction, and particularly relates to a flour processing equipment fault prediction method and system based on big data, and the method comprises the steps: obtaining the historical operation data and fault information of equipment, and constructing an initial data set; dividing the fault information into a first-level type and a second-level type, and generating a corresponding list; real-time operation parameters are collected, abnormal fluctuation points are identified, and alarm time nodes are generated; judging the effectiveness of the alarm signal in combination with a set time threshold value; and outputting a prediction result according to the effective alarm signal quantity. The system comprises an information acquisition module, a fault classification module, an operation monitoring module, an alarm analysis module and an output response module. According to the method, real-time monitoring and fault trend prediction of the equipment operation state can be realized, the intelligence and preventability of equipment operation and maintenance are improved, the sudden shutdown risk is reduced, and the method has relatively high practical value and popularization prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault prediction, and in particular relates to a flour processing equipment fault prediction method and system based on big data. Background Art

[0002] Flour processing, a fundamental component of food manufacturing, relies heavily on the continuous and stable operation of machinery, including multiple key units such as cleaners, grinders, sifters, and conveyors. Due to the long-term, high-speed operation of these machines, they are susceptible to factors such as variability in raw material properties, component wear, and electrical failures, leading to reduced efficiency, fluctuating product quality, and even unexpected downtime.

[0003] Traditional flour processing equipment maintenance relies primarily on periodic inspections and operator judgment. This approach is not only inefficient but also makes it difficult to detect potential faults in a timely manner, leading to delayed maintenance, sudden failures, and wasted resources. In recent years, the development of sensor technology and the Industrial Internet has enabled the real-time collection of large amounts of data during equipment operation, including parameters such as current, voltage, vibration, temperature, noise, and equipment operating status.

[0004] Some studies have attempted to introduce data analysis and fault diagnosis technologies, such as rule-based diagnostic systems, vibration feature analysis methods, or expert systems. Although these technologies have improved detection efficiency to a certain extent, they still have the following shortcomings: First, the data processing capability is limited, making it difficult to cope with the multi-source, heterogeneous, and highly volatile data characteristics of the flour processing process; second, there is a lack of high-precision modeling methods for typical fault types, resulting in low prediction accuracy; third, it is difficult to achieve real-time, dynamic fault warning and trend tracking in scenarios where multiple devices work together.

[0005] Therefore, it is urgent to propose a flour processing equipment fault prediction method and system based on big data, which can make full use of equipment operation data, build an accurate and real-time prediction model, and realize dynamic evaluation of the operating status of key equipment and early warning of failure risks. Summary of the Invention

[0006] In view of the above problems, the object of the present invention is to propose: A flour processing equipment fault prediction method based on big data includes the following steps: S1. Obtaining fault information: Obtain and summarize historical operating data, fault occurrence time, and fault type of flour processing equipment to form an initial data set; S2. Classification of fault information: Classify fault information into primary and secondary fault types based on the fault type. Primary fault types are faults that directly cause loss of equipment function, while secondary fault types are indirect faults caused by equipment abnormalities. S3. Construct the corresponding list: record the historical occurrence time corresponding to the first-level fault type as the reference time to construct the first-level fault list; record the occurrence time of the second-level fault type as the reference time to construct the second-level fault list; S4. Obtain monitoring data: Receive the real-time operating parameters of the equipment and extract the time point when the parameters change significantly as the monitoring starting point; combine the first-level fault list and the second-level fault list to generate the alarm time node, and establish the alarm history list in chronological order; S5. Determine the alarm signal: Set the time threshold interval for the warning time, and combine the alarm history list to determine the alarm signal in the current operation cycle; if the warning time falls within the time threshold interval, it is determined to be a valid alarm signal; if it exceeds the threshold interval, it is output as an invalid alarm signal; S6. Submit prediction results: Output prediction information based on the number of alarms and judgment results within the operation cycle; if there is a valid alarm signal, output equipment failure warning; if all alarm signals are invalid or no alarm signal is detected, output normal operation status.

[0007] In a preferred technical solution, in the step S1, the operating data in the initial data set is extracted, grouped, aggregated and stored according to different types, and the operating data includes real-time parameters of the equipment structure or components such as temperature, current, pressure, and speed.

[0008] In a preferred technical solution, in step S2, the structure identifier involved in each piece of fault information is compared with the device identifier of the abnormal operating parameter, and the fault type is determined in combination with the corresponding structure maintenance time.

[0009] In a preferred technical solution, in the S3 step, the time nodes of occurrence of various types of faults are extracted as a benchmark, a certain time length is calculated forward as a sampling interval, and the abnormal points within the interval are summarized into a first-level or second-level fault list; the first-level fault list also records the maintenance working hours information of the structure.

[0010] In a preferred technical solution, in the S4 step, the fault time node that is ranked later and has monitoring parameters is used as the sampling base point to determine the corresponding monitoring starting point; a sampling interval is set between the starting point and the current time, and the fluctuation points in the operating data within the interval are extracted as monitoring nodes; after unified conversion, the alarm time nodes are generated and classified in chronological order to form an alarm history list; wherein, the conversion parameter is the average maintenance working hours of the equipment structure.

[0011] In a preferred technical solution, in the step S5, before determining a valid alarm signal, a time threshold needs to be set as a reference, and the corresponding warning time point is obtained by subtracting the threshold from the alarm time point.

[0012] In a preferred technical solution, in step S6, if the number of valid alarm signals within a unit operating cycle exceeds the set alarm threshold, the equipment fault warning information is output and the current operation is interrupted; if the number of valid alarm signals does not exceed the threshold, it is determined that the equipment has not failed and the system operation status is normal.

[0013] The present invention also proposes a flour processing equipment fault prediction system based on big data, which is used to implement the method described above. The system includes: Information collection module, used to obtain historical fault data of equipment; The fault classification module is used to classify faults into primary and secondary faults according to the fault type and generate corresponding lists; Run monitoring module to extract real-time monitoring data and generate alarm signals; Alarm analysis module, used to determine whether the alarm signal is valid; The output response module is used to output fault warning information or normal operation status according to the judgment results.

[0014] Beneficial effects Compared with the existing technology, the flour processing equipment fault prediction method and system provided by the present invention based on big data has the following beneficial effects: 1. Significantly Improved Prediction Accuracy and Reliability: This invention classifies faults and constructs a fault list containing historical evolution patterns, providing a precise knowledge base for prediction. The dynamic time threshold interval judgment mechanism introduced effectively filters out noise signals generated by normal operating fluctuations or transient interference, ensuring that each output warning is based on precursors that closely match historical faults. This fundamentally addresses the high false alarm rate of existing technologies and ensures high confidence in prediction results.

[0015] 2. Implementing refined, hierarchical early warning and maintenance management: This invention clearly categorizes faults into primary faults (those that directly impact production) and incipient secondary faults, giving early warning information clear priorities. This hierarchical management mechanism enables the operations and maintenance team to differentiate responses and optimize resource allocation based on the severity of the warnings, transforming maintenance strategies from a crude, one-size-fits-all approach to a precise and efficient, hierarchical approach, significantly improving operations and maintenance efficiency.

[0016] 3. Enhanced real-time fault prediction and dynamic response capabilities: The method and system of this invention are designed specifically for continuous production scenarios. Through real-time parameter monitoring and dynamic updates of alarm history lists, the system continuously tracks the evolution of equipment health. Once the accumulation of valid alarm signals within a unit cycle exceeds a threshold, the system immediately triggers an early warning, buying valuable response time to prevent catastrophic failures and truly achieving the transition from "post-event repair" to "pre-event prediction."

[0017] 4. Improve the scientific and intelligent nature of maintenance decisions: This invention not only predicts faults but also provides rich decision-making support information. In the preferred solution, the first-level fault list integrates the "average maintenance man-hours" information of associated components. The output response module provides this information along with early warnings to the maintenance team. This enables the team to scientifically formulate maintenance plans and prepare spare parts, significantly reducing maintenance costs and unplanned downtime, and achieving data-driven intelligent maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Schematic diagram of the method flow of the present invention; Figure 2 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0019] In order to deepen the understanding of the present invention, the present invention will be further described in detail below with reference to the examples. The examples are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0020] Example 1 according to Figure 1 As shown, the flour processing equipment targeted in this embodiment is an automated flour processing production line with cleaning, crushing, screening, and conveying functions. The key equipment involved includes a raw material cleaning machine, a grinding mill, grading and screening equipment, and a horizontal screw conveyor. The production line is equipped with multiple sensors for collecting real-time parameters such as temperature, current, pressure, speed, and vibration.

[0021] S1. Obtain fault information: The core of this step is to comprehensively build a complete initial data set to support subsequent data comparison and fault trend judgment.

[0022] First, we retrieved the past two years of operational data from the workshop management system and the equipment's PLC control ports. The data format was standardized as a timestamp + parameter value structure. Simultaneously, we used the fault maintenance record system to extract information such as the time of occurrence of all fault events, the equipment involved, fault description, treatment measures, and whether it affected overall line operations. This data was then cross-organized using Excel and database formats to form a complete initial data set.

[0023] To improve data utilization, this step further categorizes and organizes operating data by device number and creates multidimensional parameter groups based on parameter types (such as temperature, current, pressure, and speed). This dataset not only covers state changes before and after a fault occurs, but also covers the stable range during normal operation, facilitating the subsequent identification of abnormal fluctuations. All data is uniformly stored in a time series database in units of seconds or minutes for subsequent access.

[0024] S2. Classification fault information: After obtaining complete fault records, fault types are categorized into primary and secondary faults based on their impact. Primary faults refer to major issues such as motor burnout, drive chain breakage, and spindle jamming that directly cause equipment downtime. Secondary faults include belt misalignment, temperature anomalies, and current fluctuations, which may not immediately cause downtime but may exhibit abnormal symptoms. Each fault record includes fields such as the device number, structural location, abnormality manifestation, and resolution time, which are then mapped to specific device structures through system matching.

[0025] The structural identifier is then compared with equipment parameters that have exhibited abnormal fluctuations in historical operating data. If a structure's abnormal operating parameters closely overlap with fault records, and if there is an overlap between the structural maintenance time and the fault handling time, the record can be confirmed as a primary or secondary fault, laying an accurate foundation for subsequent statistical analysis.

[0026] S3. Build the corresponding list: To facilitate subsequent alarm assessment, a timeline list must be created for each type of fault. In this step, the timeline of the first-level fault is used as the primary occurrence timeline. The sampling period is extended from 5 to 60 minutes forward, and fluctuations in all relevant parameters within this period are extracted to construct a timeline list for the first-level fault. Similarly, a 30-minute sampling period is used as a reference for the second-level fault list. Each list item includes the timestamp, fault type, corresponding device, affected parameters, and the last maintenance hours for the structure.

[0027] The first-level fault list also includes a working hours field, which records the cumulative operating hours from the time the equipment was installed or last overhauled until the current fault occurred. This field is used to dynamically adjust the weight when setting alarm thresholds, improving the accuracy of judgments.

[0028] S4. Obtain monitoring data: During normal equipment operation, the system continuously receives real-time data from each monitoring point, including but not limited to current, speed, vibration amplitude, temperature changes, etc. The data collection frequency is uniformly set to one record every 30 seconds, and real-time filtering is performed to eliminate occasional interference points.

[0029] When the system detects a significant change in a parameter within a short period of time (e.g., vibration exceeding a set upper limit, current jump exceeding 20%), it automatically records the time of the change as the monitoring starting point. At this point, the system calls the previously constructed primary and secondary fault lists and searches for fault time records adjacent to the starting point on the timeline. If they fall within the same time period, an alarm time node is generated and added to the alarm history list.

[0030] The system uses the fault time node with the lowest ranking (i.e., the most recent major fault) as the sampling base point and sets the new sampling interval from that time point to the current operating period. Within this interval, fluctuation points of all parameters are extracted and, after applying rule-based transformations (such as normalization and peak adjustment), form standard alarm time nodes, which are then archived in chronological order.

[0031] S5. Determine the alarm signal: The key to this step is determining whether the alarm signal has predictive significance. To do this, first establish a set of reference threshold ranges. For example, if the warning time for a certain fault type typically occurs between 10 and 40 minutes before the actual fault, then this range is used as the effective warning threshold for a Level 1 fault. The threshold for a Level 2 fault can be relaxed to 560 minutes.

[0032] When a new alarm signal is added to the list, the system subtracts its warning time from the fault time to calculate the lead time and determines whether this lead time falls within the threshold range. If it does, it is considered a valid alarm signal; if the lead time is too short or too long, it is considered an invalid alarm signal to avoid misjudgment.

[0033] Before making a judgment, the system subtracts the average maintenance hours from the alarm time to improve the specificity of the judgment. For example, if a motor typically experiences abnormal speed after 800 hours of operation, if the operating hours corresponding to the alarm time are close to this value, the judgment result is more likely to be a valid alarm.

[0034] S6. Submit prediction results: At the end of each complete operating cycle (e.g., 24 hours, one shift, or 10,000 data points), all alarm records within that cycle are compiled and the number of valid alarm signals is counted. If the number of valid alarms exceeds a set threshold (e.g., three), the system automatically issues an equipment failure warning and pushes it to the operator terminal and maintenance management platform.

[0035] If only sporadic invalid alarm signals are detected during the current cycle, or the system detects no parameter fluctuations, the system is deemed to be operating normally and no intervention is required. A preferred solution can also combine alarm trend changes over the past three cycles to create a weighted curve. If valid alarms continue to increase over the past three cycles, a "potential warning" alert can be issued, even if the threshold has not been exceeded in the current cycle, to increase awareness of the device.

[0036] In summary, this example systematically processes historical operating data, fault type classification, real-time parameter collection, alarm assessment, and result output for flour processing equipment, achieving early prediction and dynamic analysis of equipment failures. By introducing optimization strategies such as maintenance hours, sampling intervals, and dynamic thresholds, the accuracy and application flexibility of prediction results are significantly improved, avoiding the misjudgments and missed detections that traditionally rely on manual experience, demonstrating its practicality and potential for widespread adoption.

[0037] Example 2 To further illustrate the implementation effect of the present invention, an implementation example of a flour processing equipment fault prediction method based on an actual industrial workshop background is provided below.

[0038] This method combines historical data with real-time operating status to analyze fault trends of multiple flour processing equipment running in parallel, aiming to achieve early warning and equipment protection.

[0039] The flour processing line includes a purifier, a grinding mill, a wind screen, a feed elevator, and several conveying devices. Each device is equipped with a temperature sensor, a current transformer, a pressure probe, and a speed encoder, and is connected to the production line's central control system. To construct an implementable fault prediction method, the present invention proceeds through the following steps: S1. Obtain fault information: During this step, the operations department first exported the flour mill's maintenance records for the past three years, including the time, nature, equipment number, and duration of each piece of equipment failure. Furthermore, they obtained full-cycle operational data from the automation platform, including minute-by-minute data on temperature, current, bearing vibration, belt tension, and other parameters.

[0040] To facilitate subsequent analysis, technicians uniformly processed all data. Fault records were converted into structured data tables, and operating parameters were organized by timestamp, forming a multidimensional dataset with the device number as the primary key. By aggregating data from different devices, we can initially observe the changing trends of various parameters before and after the fault and determine which characteristic values ​​have significant predictive value.

[0041] S2, Classification fault information: On this production line, technicians divide faults into two types according to their impact: Level 1 and Level 2. Level 1 faults refer to major equipment problems that affect the entire line's shutdown, such as main motor burnout, grinding roller jamming, etc.; while Level 2 faults are mostly problems that have not directly caused shutdowns but have already shown signs of risks, such as excessively rapid bearing temperature rise, abnormal current jitter in a short period of time, etc.

[0042] The system automatically identifies each fault message and its associated equipment structure module, pairing it with the time point at which operational data fluctuations occurred. If a structure's operating parameters exhibit repeated anomalies before a fault (e.g., high current three or more times within a short period of time), the fault is classified as predictable and begins further processing.

[0043] S3. Build the corresponding list: For predicted and identifiable fault events, the system extracts their occurrence time and constructs a first-level fault time list and a second-level fault time list based on the fault type. Each record contains the following fields: device number, fault time, abnormal parameter type, parameter average value, maximum deviation value, and operating time before the fault.

[0044] For example, if the main shaft of a powder cleaning machine breaks at 14:33 on May 15, 2024, and its current has five abnormal spikes between 14:18 and 14:32 on the same day, the system will automatically use "14:33" as the base time and push it forward 15 minutes as the sampling interval, recording all abnormal changes in this interval and summarizing them as an item in the first-level fault list.

[0045] To further enhance accuracy, the system has added an "operating hours" field to the first-level fault list, which indicates the total operating time of the equipment since its last overhaul. This metric helps dynamically adjust the scope of future alarm judgments, making fault predictions more targeted.

[0046] S4. Obtain monitoring data: During this step, the system is in a continuous monitoring state. All device operating parameters are refreshed every 10 seconds and processed uniformly by the master control server. If the system detects that a parameter (such as temperature, current, or speed) fluctuates beyond a set range (e.g., ±15%) within a short period of time, the current time is immediately recorded as a suspected abnormality.

[0047] At this point, the system initiates an alarm search mechanism, searching forward from the current anomaly to see if there's a similar trend with the primary or secondary fault time list. If the fluctuation point is highly consistent with the historical distribution of a particular type of fault (e.g., similar deviation magnitude and duration), the current time point is designated as the "initial alarm time node."

[0048] In practice, the system also incorporates a multi-point simultaneous judgment mechanism. For example, if a grinding mill exhibits three abnormal indicators simultaneously within 10 minutes—rising temperature, increasing current, and increasing vibration amplitude—that period is designated as a high-risk period, and all fluctuation points are uniformly marked in the alarm history list.

[0049] S5. Determine the alarm signal: To prevent false alarms and oversensitivity, the system activates the "warning effectiveness judgment" logic after identifying the alarm time node. First, a reasonable time threshold interval is set for each type of fault, for example, the warning range for a level 1 fault is 10 minutes, and for a level 2 fault it is 60 minutes.

[0050] The system then compares the time of the alarm node with the possible future fault time and calculates the warning lead time. If this value falls within the corresponding threshold range, the alarm signal is considered a "valid alarm"; otherwise, it is marked as invalid.

[0051] It's worth noting that the present invention preferably uses the equipment's current operating hours as one of the weights for judgment. For example, a slight fluctuation in the operating time of a recently replaced bearing might be considered normal; whereas, if the same structure has been operating beyond its normal maintenance interval, a fluctuation of the same magnitude is more likely to be considered a valid alarm.

[0052] S6. Submit prediction results: After the equipment has run for a full cycle (e.g., an 8-hour shift or 24 hours a day), the system categorizes and counts all alarm signals. If the number of valid alarm signals within a certain period reaches a set threshold (e.g., three consecutive fluctuations within the valid range), a fault warning is automatically triggered, prompting the on-duty personnel to conduct an inspection.

[0053] In the control system, prediction results are not only displayed as text prompts on the control interface but can also be output to the intelligent maintenance platform for generating maintenance task orders. If there are few valid alarm signals or all are judged to be invalid, the system will automatically output a "normal operation" result without manual intervention.

[0054] To enhance stability, the system also supports a rolling warning mechanism, that is, when the weighted average of the alarm scores of three consecutive cycles exceeds the set threshold, an early warning can be issued even if the signal of the current cycle is weak, to prevent short-term misjudgment due to trend changes.

[0055] In summary, this implementation approach, through comprehensive data collection, fault classification, feature analysis, real-time comparison, and dynamic feedback, has established a fault prediction method applicable to various types of flour processing equipment. This method not only identifies potential equipment problems at an early stage but also makes reasonable judgments based on the equipment's operating time, its fluctuation history, and the interaction of multiple parameters. This significantly improves prediction accuracy and applicability, demonstrating excellent practical results.

[0056] Example 3 like Figure 2As shown, this embodiment proposes a big data-based flour processing equipment fault prediction system for implementing the aforementioned big data-based flour processing equipment fault prediction method. The system can collect, identify, compare, and analyze multi-dimensional parameters during the operation of flour processing equipment, providing real-time early warning and dynamic response to potential equipment failures. The system has a clear structure and comprehensive functionality, primarily comprising the following modules: Information Collection Module: This module collects operating parameters and historical fault data from flour processing equipment and organizes them into a structured dataset. This historical fault data includes the time, type, structural location, and corresponding maintenance records for each device. Operating parameters include temperature, current, vibration frequency, rotational speed, air pressure, and voltage. This data is transmitted to the central processing unit in real time via various industrial sensors and bound to the device identifier using a timestamp alignment mechanism to form a unified initial dataset. This module also features data cleaning, automatically removing outliers and signal interference to ensure the accuracy and stability of subsequent analysis.

[0057] Fault Classification Module: This module is used to intelligently classify historical fault information and distinguish between primary and secondary faults. A primary fault refers to a serious problem that directly causes equipment interruption or entire line shutdown, such as a broken spindle or a burned-out motor. A secondary fault refers to a system operating abnormality that does not cause an immediate shutdown, such as abnormal current, increased vibration, or continuously rising temperatures. The module analyzes the correlation between the equipment structure identifiers and abnormal operating parameter trends in the fault records to identify the time points at which various faults occurred and the scope of their possible impact, and then generates a primary and secondary fault list. These two lists will serve as one of the bases for judging the effectiveness of the alarm and will participate in the screening and comparison of alarm signals in subsequent steps.

[0058] Operation Monitoring Module: This module is responsible for real-time monitoring and data extraction of equipment operating status. The system collects real-time operating parameters from each monitoring point at a fixed frequency (e.g., every 30 seconds) and calculates the rate of change, fluctuation amplitude, and the coordinated changes between parameters in real time. If a parameter value fluctuates significantly within a short period of time (e.g., exceeding a preset floating threshold), the system automatically records the time of the fluctuation and activates the alarm recognition logic. The module also includes a built-in sliding window mechanism to determine whether parameter anomalies persist, further preventing false alarms caused by occasional interference.

[0059] Alarm Analysis Module: This module is used to determine the validity of initially identified alarm signals. The module first compares the generated alarm time node with the historical time intervals recorded in the primary and secondary fault lists to determine whether there is a potential correspondence. If the alarm signal's time point falls within the valid warning time interval (e.g., 10 to 40 minutes in advance) and has a high similarity with the historical fault parameter distribution, the signal is considered a "valid alarm signal." Otherwise, the signal is classified as an "invalid alarm signal" and does not participate in the final prediction results.

[0060] The module also incorporates equipment maintenance information, such as work hours and repair frequency, to determine whether a structure is in a "high-risk operating period." For example, if a grinding mill's main shaft has operated for more than 1,000 hours, even slight fluctuations in its vibration signal can be considered a valid warning, and the system will automatically assign a higher alarm weight.

[0061] Output Response Module: This module generates predictive feedback based on the alarm analysis results. If the cumulative number of valid alarm signals within an operating cycle exceeds a preset threshold (e.g., three consecutive times), the system will output an "Equipment Failure Warning" alert. This alert is displayed in a pop-up window on the local control terminal and simultaneously sent to the maintenance management platform and the maintenance personnel's mobile device for timely intervention.

[0062] If no valid alarms are generated during the operation cycle, or if all alarms are deemed invalid, the system outputs a "Device Operating Normally" status and records it in the operation log. The module also supports the generation of periodic operation reports, including fault trend graphs, parameter fluctuation records, warning count statistics, and fault type distribution, for analysis by the operation and maintenance team.

[0063] In summary, the big data-based flour processing equipment fault prediction system provided in this embodiment combines historical data, real-time monitoring, and multi-dimensional parameter comparison to implement a closed-loop control mechanism from information collection to prediction output. This not only effectively improves the accuracy and response speed of fault warnings, but also reduces human misjudgment, enhances the scientific and intelligent level of equipment operation and maintenance, and has strong industrial application value and widespread application prospects.

[0064] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A flour processing equipment fault prediction method based on big data, characterized in that: The following steps are involved: S1. Obtain and summarize historical operating data, fault occurrence time, and fault type of flour processing equipment to form an initial data set; S2. According to the different fault types, they are divided into primary fault types and secondary fault types. Primary fault types are faults that directly cause loss of equipment function, and secondary fault types are indirect faults caused by equipment abnormalities. S3. Record the historical occurrence time corresponding to the first-level fault type as the reference time and construct a first-level fault list; Record the occurrence time of the secondary fault type as the reference time and construct a secondary fault list; S4. Receive the real-time operating parameters of the equipment and extract the time point when the parameters change significantly as the monitoring starting point; Combine the first-level fault list and the second-level fault list to generate the alarm time node, and establish the alarm history list in chronological order; S5. Set the time threshold interval for the warning time, and judge the alarm signal in the current operation cycle in combination with the alarm history list; If the warning time falls within the time threshold interval, it is determined to be a valid alarm signal; If it exceeds the threshold range, the output is an invalid alarm signal; S6. Output prediction information based on the number of alarms and the judgment results within the operation cycle; if there is a valid alarm signal, output equipment failure warning; If all alarm signals are invalid or no alarm signal is detected, the output is in normal operation state.

2. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In the step S1, the operating data in the initial data set is extracted, grouped, aggregated and stored according to different types. The operating data includes real-time parameters of the device structure or components such as temperature, current, pressure, and rotation speed.

3. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In the step S2, the structure identifier involved in each piece of fault information is compared with the device identifier of the abnormal operating parameter, and the fault type is determined in combination with the corresponding structure maintenance time.

4. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In the step S3, the time nodes of each type of fault occurrence are extracted as a benchmark, a certain time length is calculated forward as a sampling interval, and the abnormal points within the interval are summarized into a first-level or second-level fault list; the first-level fault list also records the maintenance working hours information of the structure.

5. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In the step S4, a fault time node that is ranked lower and has monitoring parameters is used as a sampling base point to determine the corresponding monitoring starting point; a sampling interval is set between the starting point and the current time, and fluctuation points in the operating data within the interval are extracted as monitoring nodes; alarm time nodes are generated after unified conversion, and are classified in chronological order to form an alarm history list; wherein, the conversion parameter is the average maintenance working hours of the equipment structure.

6. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In the step S5, before determining a valid alarm signal, a time threshold needs to be set as a reference, and the corresponding warning time point is obtained by subtracting the threshold from the alarm time point.

7. The method for predicting flour processing equipment failure based on big data according to claim 1, characterized in that: In step S6, if the number of valid alarm signals within the unit operation cycle exceeds the set alarm threshold, the equipment failure warning information is output and the current operation is interrupted; if the number of valid alarm signals does not exceed the threshold, it is determined that the equipment has not failed and the system operation status is normal.

8. A flour processing equipment fault prediction system based on big data, used to implement the method according to any one of claims 1 to 7, characterized in that: The system comprises: Information collection module, used to obtain historical fault data of equipment; The fault classification module is used to classify faults into primary and secondary faults according to the fault type and generate corresponding lists; Run monitoring module to extract real-time monitoring data and generate alarm signals; Alarm analysis module, used to determine whether the alarm signal is valid; The output response module is used to output fault warning information or normal operation status according to the judgment results.

Images