Water reducing agent cooling equipment production management method and system

Abstract

The application discloses a water-reducing agent cooling equipment production management method and system, and relates to the technical field of production management. The method comprises the following steps: acquiring production event data of a water-reducing agent cooling equipment; performing format processing on the production event data to obtain formatted event data; constructing a workpiece production cycle event sequence according to the formatted event data; identifying abnormal events in the workpiece production cycle event sequence according to production process information and production standard information; performing reverse time tracing on the workpiece production cycle event sequence according to the abnormal events to identify multiple potential reasons for production efficiency reduction; and generating production strategy optimization suggestions according to the multiple potential reasons for production efficiency reduction. The application can generate production strategy optimization suggestions in combination with abnormal events and potential reasons for production efficiency reduction, so as to realize production management and improve production efficiency and equipment product quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production management, and particularly relates to a water reducing agent cooling equipment production management method and system. BACKGROUND

[0002] In the field of modern industrial manufacturing, it is crucial to maintain high efficiency and optimize resource allocation in the production process of water reducing agent cooling equipment. Existing systems compare benchmark information of production lines using fixed thresholds to identify production states. However, real production environments are complex and variable, such as replacing new materials leading to faster tool wear. Existing systems use old material characteristics for judgment, making it difficult to identify abnormal events, resulting in low production efficiency and low equipment product quality.

[0003] In summary, the technical problems in the related art need to be improved. SUMMARY

[0004] The main purpose of the embodiments of the present application is to provide a water reducing agent cooling equipment production management method and system, which can generate production strategy optimization suggestions by combining abnormal events and potential causes of production efficiency decline, to realize production management and improve production efficiency and equipment product quality.

[0005] In one aspect, the embodiments of the present application provide a water reducing agent cooling equipment production management method, comprising the following steps:

[0006] Obtain production event data of water reducing agent cooling equipment;

[0007] Format the production event data to obtain formatted event data;

[0008] According to the formatted event data, construct a workpiece production cycle event sequence;

[0009] According to the production process information and the production standard information, identify abnormal events in the workpiece production cycle event sequence;

[0010] According to the abnormal events, perform reverse time tracing on the workpiece production cycle event sequence to identify multiple potential causes of production efficiency decline;

[0011] According to the multiple potential causes of production efficiency decline, generate production strategy optimization suggestions.

[0012] In another aspect, the embodiments of the present application provide a water reducing agent cooling equipment production management system, comprising:

[0013] A data acquisition module for acquiring production event data of water reducing agent cooling equipment;

[0014] a data formatting module, configured to perform formatting processing on the production event data to obtain formatted event data;

[0015] an event sequence construction module, configured to construct a workpiece production cycle event sequence according to the formatted event data;

[0016] an abnormal event identification module, configured to identify an abnormal event in the workpiece production cycle event sequence according to production process information and production standard information;

[0017] a reverse time tracing module, configured to perform reverse time tracing on the workpiece production cycle event sequence according to the abnormal event to identify a plurality of production efficiency decline potential causes;

[0018] an optimization suggestion generation module, configured to generate a production strategy optimization suggestion according to the plurality of production efficiency decline potential causes.

[0019] The embodiments of the present application have at least the following beneficial effects: The embodiments of the present application first acquire production event data of a water reducing agent cooling device, perform formatting processing on the production event data to obtain formatted event data, then construct a workpiece production cycle event sequence according to the formatted event data, identify an abnormal event in the workpiece production cycle event sequence according to production process information and production standard information, perform reverse time tracing on the workpiece production cycle event sequence according to the abnormal event to identify a plurality of production efficiency decline potential causes, and finally generate a production strategy optimization suggestion according to the plurality of production efficiency decline potential causes, so that the production strategy optimization suggestion can be generated in combination with the abnormal event and the production efficiency decline potential causes to realize production management and improve production efficiency and equipment product quality.

[0020] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be achieved and obtained by means of the structures particularly pointed out in the description and the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced.

[0022] Figure 1 a flowchart of a water reducing agent cooling device production management method according to an embodiment of the present application;

[0023] Figure 2 a structural schematic diagram of a water reducing agent cooling device production management system according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0025] In the related art, in the modern industrial manufacturing field, especially in the production process of complex products such as water-reducing agent cooling equipment, it is crucial to maintain high efficiency and optimize resource allocation. The existing system compares the benchmark information of the production line using a fixed threshold to identify the production state, aiming to achieve optimal allocation of resources and maximize production efficiency. These benchmark information covers key content such as standard processing time of each process, performance parameters of various equipment, and pre-set material flow path. Under normal production conditions, this production line can maintain stable operation, and all resource scheduling and production efficiency management strictly follow the system's pre-set logic and parameters, ensuring the predictability and controllability of the production process. However, the real production environment often experiences some subtle changes that gradually erode the original production plan, leading to a slow decline in overall efficiency and low production efficiency.

[0026] For example, in order to improve the long-term durability of cooling coils in a specific corrosive environment, the relevant departments recently decided to introduce a new type of high-strength, corrosion-resistant alloy material for manufacturing key coil components in cooling equipment. Compared to the traditional alloy used in the past, this new material has improved hardness and toughness. When this new alloy material is precisely cut on a multi-axis numerical control cutting machine to form complex shapes, such as machining the spiral groove of the cooling coil or the special-shaped hole of the connecting flange, the actual wear rate of the cutting tool significantly accelerates. Specifically, the wear rate of the tool is about 15% faster than the standard wear model pre-set in the system. This difference is due to the physical properties of the new material, such as higher tensile strength and stronger resistance to plastic deformation, which result in increased cutting force and frictional heat on the tool during cutting, thereby accelerating the wear and breakage of the tool edge. Although this accelerated wear is an objective fact at the physical level, it is not severe enough to immediately cause equipment downtime, and the system's tool life prediction model has not been updated in time to reflect the processing characteristics of the new material, so this change has not been effectively identified by the system in the early stage.

[0027] Furthermore, despite the accelerated actual wear rate of the CNC cutting machine's tools, the system still uses a wear warning threshold set based on the characteristics of the old material. This means that when tool performance begins to decline due to wear, the monitoring system fails to trigger a timely warning signal for tool replacement, nor does it suggest adjusting machining parameters to adapt to the tool condition. Therefore, the CNC cutting machine continues to operate under suboptimal tool conditions for a period of time. This continuous suboptimal operation leads to slight but cumulative deviations in cutting accuracy. For example, the dimensions of the cut cooling coil assembly may slightly exceed tolerances, or its surface roughness may not meet design requirements. These deviations gradually become apparent with the accumulation of production batches, resulting in low product quality.

[0028] In the production line of water-reducing agent cooling equipment, when multiple situations occur simultaneously and interact with each other, such as changes in the processing characteristics of new materials, failure to update equipment monitoring logic in a timely manner, gradual accumulation of physical deviations between different processes, failure to synchronize resource scheduling information, and non-standard operations by on-site personnel, leading to a decline in overall production efficiency, it is necessary to understand the root cause analysis of the specific situation, accurately identify and measure the actual impact of each situation on the decline in production efficiency, determine the interaction between each situation, and generate targeted improvement measures.

[0029] The embodiments of this application will be explained in detail below with reference to the accompanying drawings:

[0030] Figure 1 This is an optional flowchart of a production management method for water-reducing agent cooling equipment provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0031] Step S101: Obtain production event data of the water-reducing agent cooling equipment;

[0032] Step S102: Format the production event data to obtain formatted event data;

[0033] Step S103: Construct a workpiece production cycle event sequence based on the formatted event data;

[0034] Step S104: Identify abnormal events in the workpiece production cycle event sequence based on production process information and production standard information;

[0035] Step S105: Based on the abnormal events, perform reverse time tracing of the workpiece production cycle event sequence to identify multiple potential causes of decreased production efficiency;

[0036] Step S106: Based on multiple potential causes of decreased production efficiency, generate production strategy optimization suggestions.

[0037] Steps S101 to S106 as shown in the embodiments of this application can combine abnormal events and potential causes of decreased production efficiency to generate production strategy optimization suggestions, thereby achieving production management and improving production efficiency and equipment product quality.

[0038] In some embodiments, steps S101-S106 may first acquire production event data of the water-reducing agent cooling equipment. Production event data is the starting point of the entire analysis process, and its comprehensiveness and accuracy directly affect the effectiveness of subsequent analyses. For example, various sensors, such as temperature sensors, pressure sensors, vibration sensors, and current sensors, can be deployed on the production line to collect equipment operating parameters in real time; information such as production orders, material batches, process flows, and personnel operations can be obtained by integrating a Manufacturing Execution System (MES) and an Enterprise Resource Planning (ERP) system; and information such as workpiece entry / exit times and quality inspection results can be recorded by manual input or scanning barcodes / QR codes. This data exists in raw, heterogeneous forms; for example, some data may be stored as CSV files, some recorded in a database, and some may be from real-time signal streams from a PLC.

[0039] The production event data is then formatted to obtain formatted event data. This process aims to transform heterogeneous raw data into a unified and standardized format, facilitating subsequent analysis. For example, all timestamps can be formatted uniformly, numerical values ​​in different units (such as Celsius and Fahrenheit for temperature) can be standardized to a single unit, missing values ​​can be marked or filled, and outliers can be identified and removed. Specifically, a series of data parsers and converters can be developed to adapt raw data from different sources and formats. For example, for process completion records from the MES system, fields such as workpiece ID (identifier), process name, start time, end time, and operator ID can be extracted and converted into structured JSON objects. For data streams from sensors, sampling and filtering can be performed, and the data can be associated with corresponding timestamps and device IDs. In essence, formatted event data refers to production event data that has been preprocessed to conform to a specific data structure and specification. Formatting aims to eliminate the heterogeneity, redundancy, and errors of the raw data, providing a unified and clean data source for subsequent data analysis and sequence construction.

[0040] Next, based on the formatted event data, a workpiece production cycle event sequence is constructed. A workpiece production cycle event sequence is a data structure that, centered on the workpiece, strings together all production events related to that workpiece in chronological order, forming a complete "production history." For example, for a cooling coil workpiece, its event sequence might include: material warehousing event, cutting start event, cutting end event, welding start event, welding end event, quality inspection event, assembly start event, assembly end event, finished product outbound event, etc. Each event contains information such as a timestamp, event type, involved equipment, operators, and key parameters. By traversing the formatted event data, all related events can be grouped according to the workpiece's unique identifier (such as serial number or batch number), and sorted according to the timestamps of the events, the workpiece production cycle event sequence can be obtained.

[0041] Based on production process information and production standard information, abnormal events in the workpiece production cycle event sequence are identified to detect deviations from preset specifications during production. For example, production process information might specify a standard processing time of 10 minutes for a certain process, and production standard information might set an allowable deviation range of ±1 minute. If the actual processing time for a workpiece is 15 minutes, this event will be identified as abnormal. During the identification process, actual parameters in the event sequence can be compared with production standard information, such as comparing actual processing time with standard processing time, actual energy consumption with standard energy consumption, and actual quality parameters with quality standards. Production process information can also be used to check whether the process sequence in the event sequence conforms to regulations and whether there are any skipped or repeated processes. Furthermore, thresholds can be set to monitor equipment operating parameters and material status in real time; once these exceed preset ranges, they are marked as abnormal events. It can be understood that production process information refers to normative documents or data describing the production process steps, process sequence, equipment configuration, and personnel responsibilities of water-reducing agent cooling equipment. Production standard information refers to the benchmark values ​​or ranges set for various indicators, parameters, and quality requirements in the production process. Examples include standard processing time, yield rate, and energy consumption limits. Abnormal events refer to events in the workpiece production cycle event sequence that deviate from production process information and production standard information, potentially leading to a decrease in production efficiency. These events may include equipment malfunctions, deviations in process parameters, material quality issues, and operational errors.

[0042] Based on abnormal events, the system performs reverse time tracing of the workpiece production cycle event sequence to identify multiple potential causes of decreased production efficiency. Starting from apparent abnormal events, the system can trace back along the timeline to explore the underlying causes of the anomalies. For example, if an abnormally long welding process is detected for a workpiece, the system can trace back to the cutting process preceding welding. By analyzing the event data of the cutting process, it might discover slight deviations in cutting precision, leading to uneven weld gaps, thus forcing the welding robot to spend extra time on adaptive adjustments. A rule-based inference engine can be used, pre-setting a series of causal chains. When an anomaly is detected, the corresponding tracing path is automatically triggered. For example, the rule could be defined as "If welding time is extended, check the precision data of the preceding cutting process."

[0043] Finally, based on multiple potential causes of decreased production efficiency, optimization suggestions for production strategies are generated, aiming to transform the analysis results into actionable improvement measures. For example, if "new materials causing accelerated tool wear" is identified as a potential cause, the system might suggest "updating the tool life prediction model," "adjusting the tool changeover cycle," and "optimizing cutting parameters." If "upstream process accuracy deviation" is a potential cause, suggestions might include "strengthening quality control in upstream processes" and "introducing online inspection equipment." Optimization suggestions can be generated by combining expert knowledge bases or machine learning models. Expert knowledge bases store typical solutions for different potential causes, while machine learning models can learn the impact of different optimization strategies on production efficiency based on historical data, thereby recommending the optimal strategy combination.

[0044] Through the above technical solution, this embodiment constructs a sequence of events in the workpiece production cycle, making the production process of each workpiece traceable and analyzable. When an abnormal event is identified, such as extended welding time, the system can use a reverse time tracing mechanism to trace back along the event sequence and accurately locate the precision deviation of the preceding cutting process that caused the welding problem. The vague phenomenon of "overall decline in production efficiency" can be decomposed into multiple specific and actionable potential causes. Furthermore, this embodiment can generate targeted production strategy optimization suggestions based on the identified multiple potential causes of declining production efficiency. For example, for the problem of accelerated tool wear, the system can suggest updating the tool life prediction model and adjusting the replacement cycle; for upstream precision deviations, it can suggest strengthening quality control or introducing online detection. This complete chain from problem identification to solution generation enables management to formulate and implement highly targeted optimization measures, thereby effectively solving deep-seated and intertwined production dilemmas and significantly improving the resource utilization and overall output of the water-reducing agent cooling equipment production line.

[0045] In some embodiments, step S103, constructing a workpiece production cycle event sequence based on formatted event data, may include, but is not limited to, the following steps:

[0046] The formatted event data is validated, including timestamp synchronization, workpiece identifier uniqueness, and data field integrity.

[0047] Based on the formatted event data after verification processing, construct the initial event sequence;

[0048] Perform sequence breakage analysis on the initial event sequence to identify the causes of sequence breakage;

[0049] Based on the cause of the sequence break, the initial event sequence is repaired. The repair process includes timestamp calibration, workpiece identity inference, and data field completion.

[0050] Conflict analysis is performed on the initial event sequence after the repair process to identify conflict events;

[0051] Conflict disambiguation is performed on conflicting events through cross-validation;

[0052] Based on the conflict events after conflict disambiguation, the initial event sequence after repair is updated to obtain the workpiece production cycle event sequence.

[0053] In some embodiments, formatted event data may be validated first to ensure its quality and consistency. This validation process includes timestamp synchronization verification, workpiece identifier uniqueness verification, and data field integrity verification. Timestamp synchronization verification checks whether the timestamps recorded by different devices or sensors are consistent to avoid event sequence errors due to time discrepancies. Workpiece identifier uniqueness verification ensures that each workpiece has a unique identifier throughout the production cycle, preventing data confusion. Data field integrity verification checks whether all necessary data fields are filled to avoid data omissions affecting subsequent analysis.

[0054] Then, based on the formatted event data after verification processing, an initial event sequence is constructed. This initial event sequence is a record of original production events arranged in chronological order, laying the foundation for subsequent analysis. Sequence breakage analysis is then performed on the initial event sequence to identify the causes of these breaks. In actual production processes, data acquisition or transmission may be interrupted, leading to breaks in the initial event sequence. Therefore, sequence breakage analysis is necessary to identify the causes of these breaks. For example, breaks may be caused by sensor malfunctions, network outages, or equipment downtime.

[0055] Next, based on the cause of the sequence break, the initial event sequence is repaired. This repair process includes timestamp calibration, workpiece identity inference, and data field completion. Timestamp calibration corrects time deviations caused by clock drift or synchronization issues; workpiece identity inference infers the correct workpiece identity using contextual information when the workpiece identifier is missing or ambiguous; and data field completion fills in missing data fields based on known information or preset rules.

[0056] Finally, conflict analysis is performed on the repaired initial event sequence to identify conflicting events. Conflicting events may still exist in the repaired initial event sequence, such as contradictory events recorded at the same time point or logical inconsistencies in event attributes. Therefore, conflict analysis is necessary to identify conflicting events. Simultaneously, conflict disambiguation is performed through cross-validation. Multiple data sources or different logical rules can be used to compare conflicting events to eliminate conflicts and determine the most reasonable and accurate event record. Based on the conflicting events after disambiguation, the repaired initial event sequence is updated to obtain an accurate, complete, and consistent workpiece production cycle event sequence.

[0057] Through the above technical solutions, this embodiment can significantly improve the data quality and reliability of the workpiece production cycle event sequence. Specifically, by verifying timestamp synchronization, workpiece identifier uniqueness, and data field integrity, analytical biases caused by data quality issues are effectively avoided. The identification and repair of sequence breaks ensure the continuity and integrity of production process events, preventing the loss of information in critical production stages. The analysis and disambiguation of conflicting events further improves the accuracy and consistency of the data, reducing the possibility of misjudgment. Ultimately, the constructed workpiece production cycle event sequence more accurately reflects the actual production situation, providing high-quality and highly reliable data support for subsequent anomaly identification, reverse time tracing, and the generation of production strategy optimization suggestions, thereby improving the effectiveness of the entire production management method and the accuracy of decision-making.

[0058] In some embodiments, step S104, identifying abnormal events in the workpiece production cycle event sequence based on production process information and production standard information, may include, but is not limited to, the following steps:

[0059] Acquire equipment operating parameters, material status, and environmental parameters;

[0060] Calculate the long-term trend based on equipment operating parameters, material status, and environmental parameters;

[0061] Determine the threshold for identifying abnormal events based on long-term trends;

[0062] Abnormal events are identified based on abnormal event identification thresholds, production process information, and production standard information.

[0063] In some embodiments, equipment operating parameters, material states, and environmental parameters can be acquired first. Equipment operating parameters refer to data reflecting various performance indicators of the water-reducing agent cooling equipment during the production process, such as equipment temperature, pressure, current, voltage, rotational speed, and vibration frequency. Material states refer to the physical and chemical properties of raw materials, semi-finished products, or finished products used in the production process, such as material composition, purity, viscosity, density, and particle size. Environmental parameters refer to the external conditions surrounding the production workshop or equipment, such as ambient temperature, humidity, dust concentration, and noise levels. These parameters can be acquired through sensors, monitoring systems, or manual recording, with the aim of providing a comprehensive data foundation for subsequent anomaly identification.

[0064] Then, based on equipment operating parameters, material conditions, and environmental parameters, long-term trends are calculated. Long-term trends refer to the statistical characteristics of these parameters over a period of time (e.g., hours, days, or weeks), such as their average value, fluctuation range, rate of change, or cumulative deviation. For example, historical data of these parameters can be analyzed using statistical methods such as moving averages, exponential smoothing, and regression analysis to reveal their regularity over time. The aim is to capture potential problems such as slow drift, periodic fluctuations, or gradual deterioration of parameters, rather than simply focusing on anomalies in instantaneous values.

[0065] Next, based on long-term trends, anomaly detection thresholds are determined. Anomaly detection thresholds are critical values ​​or ranges used to determine whether an event is abnormal. For example, these thresholds can be set based on the statistical distribution of long-term trends (such as multiples of the mean plus or minus the standard deviation), characteristics of historical anomalies, or expert experience. When the real-time value of a parameter or its trend exceeds the anomaly detection threshold, it is considered a potential anomaly. The aim is to establish an objective judgment standard to distinguish between normal fluctuations and genuine anomalies.

[0066] Finally, abnormal events are identified based on the abnormal event identification threshold, production process information, and production standard information. By comparing the parameters monitored in real time with these thresholds, processes, and standards, events that violate production specifications, exceed normal operating ranges, or may cause product quality problems can be accurately located. For example, if the temperature of a certain process is consistently higher than the upper limit of its long-term trend and exceeds the range allowed by production standards, it can be identified as an abnormal event.

[0067] Through the above technical solution, this embodiment enables refined identification of abnormal events during the production process of water-reducing agent cooling equipment. This embodiment considers not only the operating status of the equipment itself but also material characteristics and environmental factors, making the identification of abnormal events more comprehensive and accurate. By introducing the calculation of long-term trends and the determination of dynamic thresholds, the sensitivity and robustness of abnormal event identification are effectively improved. It can promptly detect potential risks of decreased production efficiency, providing a reliable basis for subsequent reverse-time tracing and production strategy optimization, thereby contributing to improving the level of intelligent production management and production efficiency.

[0068] In some embodiments, in step S105, the workpiece production cycle event sequence is traced back in reverse time based on the abnormal events to identify multiple potential causes of decreased production efficiency, which may include, but is not limited to, the following steps:

[0069] Based on the abnormal events, core input features and resource consumption are extracted from the workpiece production cycle event sequence. The core input features include weld gap change rate, local temperature gradient of material and predicted value of workpiece micro-warping. Resource consumption includes CPU cycle count and memory allocation.

[0070] Based on the core input features, the decision intent label is determined. The decision intent label includes weld strength, welding speed, heat input or dynamic deformation.

[0071] Adjust the complexity based on the parameters and assess the extension of the welding task cycle time.

[0072] A correlation analysis was performed on resource consumption, decision intent tags, and welding task cycle time extension to obtain the first correlation relationship;

[0073] A second association relationship is obtained by performing correlation analysis on the core input features, resource consumption, and decision intent labels;

[0074] Calculate the contribution of dynamic micro-warping based on the first and second association relationships;

[0075] Compare the cycle time extension of the welding task with the cycle time extension under the preset static deviation compensation model, and calculate the contribution of the cycle time extension.

[0076] The contribution of static machining deviations is evaluated based on the cutting accuracy.

[0077] Based on the contributions of dynamic micro-warping, cycle time extension, and static processing deviation, several potential causes of decreased production efficiency were identified.

[0078] In some embodiments, a decline in production efficiency is often the result of a complex interplay of multiple factors, such as fluctuations in equipment parameters, changes in material properties, environmental conditions, and operator behavior. Using only a macroscopic or single-dimensional reverse-time tracing method may fail to comprehensively and accurately reveal these deep-seated, interrelated underlying causes, leading to a lack of targeted and effective optimization of subsequent production strategies.

[0079] To this end, core input features and resource consumption can be extracted from the workpiece production cycle event sequence based on abnormal events. Core input features include weld gap change rate, local material temperature gradient, and predicted workpiece micro-warpage. These core input features are physical or geometric parameters that directly affect product quality and process stability. Resource consumption includes CPU cycle count and memory allocation, reflecting the computational load of the equipment or system during task execution. This data can be collected in real time through sensors integrated into the equipment, controller logs, or production management systems. Based on the core input features, decision intent labels are determined, including weld strength, welding speed, heat input, or dynamic deformation. The extracted core input features can be used to infer the goals that the current production process is focusing on or attempting to achieve. For example, when the weld gap change rate is large, the system may prioritize ensuring weld strength, adjust the welding speed to adapt to the change, or focus on potential dynamic deformation. These decision intent labels, such as weld strength, welding speed, heat input, or dynamic deformation, can be identified and classified using a pre-defined rule base or expert knowledge system.

[0080] Then, based on the parameter adjustment complexity, the welding task cycle time extension is evaluated to quantify the portion of the actual task completion time exceeding the standard time caused by anomalies or adjustments in the production process. Parameter adjustment complexity can serve as an important criterion for evaluating cycle time extension; for example, complex parameter adjustment processes typically require more time and resources, thus leading to cycle time extension. The welding task cycle time extension can be calculated by comparing the actual cycle time with the standard cycle time, combined with historical parameter adjustment data and the complexity model.

[0081] A correlation analysis is then performed on resource consumption, decision intent labels, and welding task cycle time extension to obtain the first correlation, aiming to establish a primary relationship between them. This analysis can reveal whether the increase in resource consumption is causally or correlated with a specific decision intent (e.g., increasing computational resources to maintain high weld strength) and the resulting cycle time extension. Simultaneously, a correlation analysis is performed on core input features, resource consumption, and decision intent labels to obtain the second correlation, aiming to establish a secondary relationship between them. This analysis provides a deeper understanding of how changes in core input features affect resource consumption and the decision intent adopted by the system. These correlation analyses can be achieved using statistical methods (such as correlation analysis and regression analysis) or data mining techniques (such as decision trees and neural networks).

[0082] Based on the first and second correlations, the contribution of dynamic micro-warping is calculated. This contribution quantifies the degree of impact of dynamic factors such as workpiece micro-warping on the decline in production efficiency. For example, by analyzing the correlation between predicted micro-warping values ​​and cycle time extension and resource consumption, the specific contribution ratio of micro-warping to the overall efficiency decline can be determined.

[0083] The cycle time extension for welding tasks is compared with that under a preset static deviation compensation model, and the contribution of the cycle time extension is calculated. The preset static deviation compensation model can predict the cycle time extension under normal static deviations (such as equipment wear and material tolerances) based on historical data or experience, thus distinguishing the impact of dynamic anomalies and static deviations on the cycle time. The preset static deviation compensation model can employ a deep neural network, which can be trained using a large amount of historical data to obtain the model.

[0084] The contribution of static machining deviations is assessed based on cutting accuracy. Cutting accuracy is a key indicator of upstream machining quality, and deviations can lead to decreased efficiency in subsequent welding or assembly processes. This contribution quantifies the degree of efficiency reduction caused by static machining issues such as insufficient cutting accuracy.

[0085] Finally, based on the contributions of dynamic micro-warping, cycle time extension, and static machining deviation, several potential causes of decreased production efficiency were identified. For example, if the contribution of dynamic micro-warping is high, "dynamic deformation caused by mismatched welding process parameters" may be identified as the main cause; if the contribution of cycle time extension is high, "production interruption caused by complex parameter adjustments" may be identified as the main cause; and if the contribution of static machining deviation is high, "insufficient precision in upstream cutting processes" may be identified as the main cause.

[0086] To illustrate this technical solution more clearly, a specific example is used below. Assume that at the welding station of the water-reducing agent cooling equipment, the system continuously monitors abnormal events through the data acquisition module, such as decreased welding quality or abnormally prolonged cycle time. At this time, the reverse-time tracing module is activated. First, core input features are extracted from the workpiece production cycle event sequence. For example, the weld gap change rate is detected to fluctuate from 0.5mm to 1.2mm, the local temperature gradient of the material increases from 5℃ / s to 15℃ / s, and the predicted value of workpiece micro-warping increases from 0.1mm to 0.3mm. Simultaneously, resource consumption is recorded; for example, the number of CPU cycles increases significantly during abnormal periods, and memory allocation also increases accordingly. Next, based on these core input features, the system determines a decision intent label. For example, due to the abnormal weld gap change rate and temperature gradient, the system judges that the current decision intent may be biased towards "maintaining weld strength" at the expense of welding speed, or that there is a risk of "dynamic deformation." Meanwhile, the complexity of parameter adjustments was assessed, revealing that welding parameters (such as current and voltage) underwent multiple complex adjustments to address the aforementioned anomalies, resulting in a 15% increase in the welding cycle time.

[0087] Subsequently, correlation analysis was conducted. The first correlation analysis revealed a strong correlation between high resource consumption and maintaining weld strength and cycle time extension. The second correlation analysis indicated that the weld gap change rate and local temperature gradient of the material were the main factors leading to increased resource consumption and dynamic deformation risk. Based on these correlations, the contribution of dynamic micro-warping was calculated, finding that workpiece micro-warping contributed 40% to the efficiency decline. Simultaneously, comparing the actual 15% cycle time extension with the 5% cycle time extension under the preset static deviation compensation model, the contribution of cycle time extension was calculated to be 30%. Finally, by analyzing cutting accuracy data, the contribution of static machining deviation was assessed at 20%. Combining these contributions, the system identified several potential causes of decreased production efficiency, including "excessive dynamic micro-warping," "complex adjustment of welding parameters leading to cycle time extension," and "insufficient accuracy of upstream cutting stations." This provides specific and quantitative basis for subsequent production strategy optimization; for example, it is recommended to optimize welding fixtures to reduce micro-warping, simplify parameter adjustment logic, and improve cutting processes to increase accuracy.

[0088] Through the above technical solution, this embodiment introduces multi-dimensional core input features and resource consumption, and combines them with decision intent tags for multi-level correlation analysis. This enables a more comprehensive and in-depth revelation of the potential factors leading to efficiency decline in the production process and their interaction mechanisms. For example, it can distinguish between efficiency decline caused by dynamic processes (such as micro-warpage) and efficiency decline caused by static processing deviations or parameter adjustment complexity, thus providing a more accurate and targeted basis for subsequent production strategy optimization. This refined cause identification capability significantly improves the scientificity and effectiveness of production management decisions, helping enterprises to solve production bottlenecks more efficiently and improve overall production efficiency and product quality.

[0089] In some embodiments, step S106, generating production strategy optimization suggestions based on multiple potential causes of decreased production efficiency, may include, but is not limited to, the following steps:

[0090] Step S201: Extract potential event data corresponding to each potential cause of decreased production efficiency from the workpiece production cycle event sequence;

[0091] Step S202: Calculate the contribution of efficiency decline to each potential event data;

[0092] Step S203: Based on the workpiece production cycle event sequence, identify the interactive event patterns among multiple potential causes of decreased production efficiency. The interactive event patterns include periodic micro-warping caused by the combined effect of material processing parameters and equipment component wear.

[0093] Step S204: Adjust the contribution of efficiency decline based on the interaction impact event pattern;

[0094] Step S205: Based on multiple revised contribution rates of efficiency decline and multiple potential causes of production efficiency decline, generate production strategy optimization suggestions.

[0095] In some embodiments, if the complex interactions between potential causes and their precise contribution to efficiency decline are not fully considered, the generated optimization suggestions may not achieve optimal results, or may even miss key efficiency bottlenecks. To address this, potential event data corresponding to each potential cause of production efficiency decline can be extracted from the workpiece production cycle event sequence. For each identified potential cause of production efficiency decline, event data directly or indirectly related to that potential cause can be screened and obtained from the complete workpiece production cycle event sequence. This potential event data forms the basis for subsequent analysis and quantification of contribution.

[0096] Then, the contribution of each potential event data point to the efficiency decline is calculated. Quantitative analysis can determine the specific impact of each potential event data point, or the potential cause it represents, on the overall decline in production efficiency. This contribution can be a percentage value or a relative weight, used to measure its importance in the efficiency decline.

[0097] Based on the workpiece production cycle event sequence, interactive event patterns among multiple potential causes of decreased production efficiency are identified. These interactive event patterns include periodic micro-warping caused by the combined effects of material processing parameters and equipment component wear. In-depth analysis of historical data and real-time event sequences reveals that different potential causes do not exist in isolation but may interact, reinforce, or cancel each other out, thus forming specific event patterns. For example, an interactive event pattern can be understood as periodic micro-warping caused by the combined effects of material processing parameters and equipment component wear. This means that a single change in material parameters or equipment wear may not be sufficient to cause a significant problem, but when both occur simultaneously and meet specific conditions, they can lead to periodic micro-warping of the workpiece, thereby affecting production efficiency.

[0098] The contribution of efficiency degradation is adjusted based on the interaction patterns of events. After identifying the interactions between potential causes, the previously calculated contribution of efficiency degradation can be adjusted. The initial calculations may not have taken into account these complex interactions, and the adjusted contribution will more accurately reflect the true impact of each potential cause in the actual production environment.

[0099] Finally, based on multiple revised contribution rates of efficiency decline and multiple potential causes of production efficiency decline, production strategy optimization suggestions are generated. Having obtained more accurate contribution rates of efficiency decline and fully understood the interactive effects between potential causes, the system can generate more targeted, comprehensive, and effective production strategy optimization suggestions. These suggestions not only address single problems but also take into account the synergistic effects of multiple factors, thereby achieving an overall improvement in production efficiency.

[0100] To illustrate this technical solution more clearly, a specific example is used below. Suppose that in the production process of a water-reducing agent cooling equipment, reverse time tracing identifies "weld gap change rate" and "local material temperature gradient" as two potential causes of decreased production efficiency. First, the system extracts potential event data related to the weld gap change rate and local material temperature gradient from the workpiece production cycle event sequence, such as specific welding parameter records and temperature sensor data. Next, the system calculates the efficiency reduction contribution of each of these two potential causes. For example, preliminary analysis might show that the weld gap change rate leads to a 15% efficiency reduction, while the local material temperature gradient leads to a 10% efficiency reduction. Subsequently, the system further identifies the interaction patterns between potential causes based on the workpiece production cycle event sequence. For example, analysis reveals that when the weld gap change rate exceeds a certain threshold, and equipment component wear (e.g., welding nozzle wear) is present simultaneously, it causes periodic micro-warping of the workpiece, significantly increasing the rework rate in subsequent processing. This "periodic micro-warping caused by the combined effect of weld gap change rate and equipment component wear" is the identified interaction pattern.

[0101] Based on this interactive event pattern, the system will revise the previously calculated contribution of efficiency degradation. For example, due to the discovery of periodic micro-warping, the actual contribution of the weld gap variation rate may be revised to 20%, while the contribution of equipment component wear is also taken into account. Finally, based on the revised contribution of efficiency degradation and a comprehensive understanding of these potential causes and their interactions, the system will generate production strategy optimization suggestions. These suggestions may include: adjusting welding parameters to strictly control the weld gap variation rate; implementing preventive maintenance plans, regularly inspecting and replacing welding nozzles to reduce equipment component wear; and fine-tuning the welding process under specific material processing conditions to avoid the occurrence of periodic micro-warping. In this way, the optimization suggestions can more comprehensively and accurately solve complex problems in actual production.

[0102] Through the above technical solution, this embodiment can significantly improve the accuracy and effectiveness of production strategy optimization suggestions. By introducing the identification of interactive event patterns and correcting the contribution of efficiency decline, this embodiment can gain a more comprehensive and in-depth understanding of the underlying causes of production efficiency decline, especially complex problems caused by the synergistic effects of multiple factors. For example, for complex problems such as periodic micro-warping caused by the combined effects of material processing parameters and equipment component wear, this embodiment can provide more targeted solutions, thereby avoiding poor optimization results due to one-sided analysis. Therefore, the generated production strategy optimization suggestions not only solve surface problems but also address the root causes, thereby achieving continuous optimization of production efficiency and stable operation of the production process.

[0103] In some embodiments, step S203, identifying interactive event patterns among multiple potential causes of decreased production efficiency based on the workpiece production cycle event sequence, may include, but is not limited to, the following steps:

[0104] Step S301: Obtain historical production data;

[0105] Step S302: Train a multilayer sensing network based on historical production data;

[0106] Step S303: Time-align the cutting station event data, welding station event data, and assembly station event data in the workpiece production cycle event sequence;

[0107] Step S304: Use a multilayer perceptron to extract features from the time-aligned cutting station event data, welding station event data, and assembly station event data to obtain the corresponding event features.

[0108] Step S305: Use a multilayer perceptron to identify the correlation of multiple event features to obtain the interactive influence event pattern.

[0109] In some embodiments, relying solely on simple rules or statistical analysis may fail to accurately capture the complex and nonlinear interactions in the production process, particularly potential correlations across workstations and time dimensions. This could lead to inaccurate corrections of efficiency degradation contributions, affecting the effectiveness of final production strategy optimization recommendations. To address this, historical production data can be acquired first. Various data related to the production of water-reducing agent cooling equipment over a past period can be collected from multiple data sources, such as production databases, sensor logs, and Manufacturing Execution Systems (MES). This includes data on equipment operating status, material batch information, process parameters, environmental conditions, and production event records for each workstation. This data is used to provide training samples for subsequent multilayer sensing networks.

[0110] Then, a multilayer perceptron is trained based on historical production data. A deep learning model with multiple layers of neurons can be constructed, enabling the network to learn and recognize complex patterns and relationships inherent in historical production data. During training, the multilayer perceptron is input into historical production data, and its internal weights and biases are adjusted through algorithms such as backpropagation, allowing it to effectively extract meaningful features from the input data and perform pattern recognition.

[0111] Next, time alignment is performed on the event data from the cutting, welding, and assembly stations within the workpiece production cycle event sequence. This allows event data generated at different points in time at different production stations (such as cutting, welding, and assembly) to be calibrated and synchronized according to a unified time benchmark. For example, events can be sorted by their timestamps, or interpolation and resampling methods can be used to ensure that data from different stations are comparable in the time dimension, facilitating subsequent joint analysis.

[0112] Multilayer perceptron (MPB) was used to extract features from time-aligned event data from cutting, welding, and assembly stations, yielding corresponding event features. The original time-aligned event data from each station can be used as input to the MPB. The network performs nonlinear transformations and abstractions on this raw data through its hidden layers, thereby extracting deep, high-dimensional features that represent the production status, process parameters, or potential problems of each station. These event features can more effectively capture key information in the data and filter out noise.

[0113] Finally, a multilayer perceptron (MLP) is used to identify the correlations between multiple event features, thus obtaining interactive event patterns. Event features extracted from different workstations can be used as input to the MLP. The network learns complex mapping relationships to analyze the interactions and dependencies between these features. For example, the network can identify how changes in a cutting parameter affect the final assembly quality through the welding process, or how a certain equipment wear pattern, together with a specific batch of materials, leads to periodic micro-warping. This allows for the discovery and quantification of complex patterns of interaction between different production stages and different potential causes, such as periodic micro-warping caused by the combined effects of material processing parameters and equipment component wear.

[0114] To illustrate this technical solution more clearly, a specific example is used below. Suppose that a water-reducing agent cooling equipment production line has frequently experienced workpiece micro-warping issues over the past year, leading to decreased efficiency in the assembly process. To identify the interactive event patterns causing this problem, historical production data, including data from cutting, welding, and assembly stations, was first acquired. This data encompassed equipment operating parameters, material batch information, ambient temperature, humidity, and detailed event records for each station. Subsequently, this historical production data was used to train a multilayer perceptron, enabling it to learn and identify potential patterns in the data. When analyzing the current workpiece production cycle event sequence, the event data from the cutting, welding, and assembly stations in the sequence are first precisely time-aligned to ensure that all relevant events are synchronized on the timeline.

[0115] Next, the time-aligned event data from each workstation is input into a pre-trained multilayer perceptron. The network extracts features from this data, obtaining a series of high-dimensional event features, such as vibration spectrum features during cutting, heat input distribution features during welding, and torque change features during assembly. Finally, the multilayer perceptron further performs correlation identification on these extracted event features. Through analysis, the network identifies an interactive event pattern: when a specific batch of material (with slightly deviated processing parameters) is processed at the cutting station, if a critical component (e.g., the welding torch nozzle) at the welding station is slightly worn, localized thermal stress concentration is more likely to occur during welding, leading to periodic workpiece micro-warping. This micro-warping manifests as the need for additional calibration time at the assembly station, thus extending the task cycle time. By identifying this interactive event pattern of "periodic micro-warping caused by the combined effect of material processing parameters and equipment component wear," targeted optimization suggestions can be proposed, such as strengthening quality inspection of specific batches of material and regularly inspecting and replacing critical worn components at the welding station, thereby effectively reducing the incidence of micro-warping and improving production efficiency.

[0116] Through the above technical solution, this embodiment can significantly improve the accuracy and depth of identifying the interaction patterns between potential causes of production efficiency decline. By introducing a multi-layer perceptron to learn and extract features from historical production data, and by performing time alignment on multi-station event data, the system can capture complex nonlinear correlations across workstations and time dimensions that are difficult to detect using traditional methods. This allows for more accurate quantification of the interactions between different potential causes, such as precisely identifying complex patterns like periodic micro-warping caused by the combined effects of material processing parameters and equipment component wear. This accurate identification of interaction patterns makes the correction of the contribution to efficiency decline more scientific and reasonable, thereby generating more targeted and effective production strategy optimization suggestions. This avoids poor optimization results or the introduction of new problems due to one-sided analysis, ultimately effectively improving the overall production efficiency and product quality of the water-reducing agent cooling equipment.

[0117] In some embodiments, in step S302, training the multilayer sensing network based on historical production data may include, but is not limited to, the following steps:

[0118] The quality of each historical event data in the historical production data is evaluated to obtain the corresponding event data quality level;

[0119] Adjust the learning weights of the multilayer sensing network for historical event data based on the quality level of the event data;

[0120] The multilayer perceptron is updated based on multiple learning weights.

[0121] In some embodiments, the quality of each historical event data point in the historical production data can be assessed first to obtain the corresponding event data quality level. The completeness, accuracy, consistency, and timeliness of each data point or record can be quantitatively analyzed. For example, the fill rate of data fields, the reasonable range of values, logical consistency with other related data, and the accuracy of the data collection timestamp can be evaluated. Thus, the corresponding event data quality level can be obtained. This level can be a discrete classification (e.g., "high," "medium," "low") or a continuous score value, used to characterize the reliability of the historical event data when training a multilayer perceptron.

[0122] Then, based on the quality level of the event data, the learning weights of the multilayer perceptron (MLP) for historical event data are adjusted. During model training, different importance can be assigned to data of different quality levels. Specifically, higher-quality historical event data can be given a larger learning weight, allowing it to play a greater role in updating model parameters; while lower-quality data can be given a smaller learning weight, or even excluded in some cases, to reduce its negative impact on model training. The aim is to guide the MLP to focus more on high-quality data, thereby learning more accurate and reliable event patterns.

[0123] The multilayer perceptron (MLP) is then updated based on multiple learned weights. During the iterative training of the LPP, historical event data with adjusted learned weights is used to calculate the loss function and gradient. For example, in the backpropagation algorithm, the contribution of each sample to the gradient is multiplied by its corresponding learned weight, thereby updating the network parameters. The goal is to ensure that the LPP can effectively extract features and patterns from high-quality data while reducing noise and bias introduced by low-quality data, ultimately forming a model with a higher ability to recognize interactive event patterns.

[0124] Through the above technical solution, this embodiment can significantly improve the accuracy and robustness of multilayer perceptrons in identifying interactive event patterns among multiple potential causes of production efficiency decline. Because the model can effectively distinguish and utilize high-quality historical data during training while reducing the negative impact of low-quality data, the trained model can more accurately capture complex and realistic event correlation patterns. This not only makes the generated production strategy optimization suggestions more targeted and effective but also improves the decision-making reliability of the entire production management system, thus providing a solid data foundation and model support for the continuous optimization of the production efficiency of water-reducing agent cooling equipment.

[0125] In some embodiments, in step S302, training the multilayer sensing network based on historical production data may include, but is not limited to, the following steps:

[0126] Based on the mapping relationship between event patterns and production efficiency decline in historical production data, the first weight of the multilayer sensing network is trained.

[0127] Identify the data type of each historical event in the historical production data;

[0128] Determine the forgetting factor for each data type;

[0129] The second weights of the multilayer perceptron are trained based on the forgetting factor corresponding to each data type.

[0130] In some embodiments, the multilayer sensing network (MLN) can be initially trained with weights based on the mapping relationship between event patterns and production efficiency decline in historical production data. Preliminary weight adjustments can be made to the MNN using known causal or correlational mappings between event patterns and production efficiency decline in historical production data. These mappings can be obtained through expert experience, historical statistical analysis, or a pre-established rule base, aiming to enable the network to learn the fundamental patterns of how different event patterns affect production efficiency. An event pattern refers to a recurring combination of events with a specific structure or sequence during the production process, such as equipment failure sequences, operational error patterns, or material defect types. Production efficiency decline refers to the negative impacts caused by these event patterns, such as prolonged production cycle time, decreased product quality, or increased resource consumption.

[0131] Then, identify the data type of each historical event in the historical production data. This can be categorized based on the data's source, format, or semantic content. For example, data types could include sensor data (such as temperature, pressure, vibration), equipment log data (such as operating status, error codes), operator input data, material batch information, environmental parameters, etc.

[0132] Next, determine the forgetting factor for each data type. The forgetting factor is a value between 0 and 1, used to quantify the timeliness and importance of that data type in the training of the multilayer perceptron. For example, for sensor data with high real-time requirements and frequent changes, a smaller forgetting factor can be set to ensure that the network can quickly respond to the latest data changes; while for data types that change slowly or serve as background information, a larger forgetting factor can be set to gradually reduce their impact on network weights. The forgetting factor can be set empirically based on factors such as the importance of the data type, update frequency, and direct impact on production efficiency, or dynamically adjusted through optimization algorithms.

[0133] Finally, based on the forgetting factor corresponding to each data type, a second weight training is performed on the multilayer perceptron. This training process can employ incremental learning or online learning, dynamically adjusting the learning weights for different data types using the forgetting factor. Specifically, data types with smaller forgetting factors (i.e., more important or more real-time data) are assigned higher learning weights during training, thus having a greater impact on the updating of network weights; conversely, data types with larger forgetting factors (i.e., less important or less timely data) are assigned lower learning weights.

[0134] Through the above technical solution, this embodiment can significantly improve the accuracy and robustness of multilayer sensing networks in identifying interactive event patterns among potential causes of declining production efficiency. Specifically, by assigning differentiated learning weights to different data types, the network can better focus on the most relevant and influential production data at present, avoiding identification bias caused by interference from old or secondary data. Therefore, this embodiment can more accurately reveal the hidden and dynamically changing interactive mechanisms in complex production processes, providing a solid data foundation for generating more targeted and timely production strategy optimization suggestions, thereby effectively improving the intelligence level and decision-making efficiency of water-reducing agent cooling equipment production management.

[0135] In some embodiments, after constructing the workpiece production cycle event sequence based on the formatted event data in step S103, the method may further include, but is not limited to, the following steps:

[0136] The cutting data during the cutting process is collected, including vibration waveforms and acoustic spectra.

[0137] Bandpass filtering is applied to the cutting data;

[0138] Fast Fourier Transform is performed on the bandpass filtered cutting data to extract frequency domain features, including energy distribution, peak frequency, and peak amplitude.

[0139] The frequency domain features are matched with the acoustic fingerprint database to identify the digital identifier of the workpiece;

[0140] Based on the workpiece's digital identifier, fracture location analysis is performed on the workpiece's production cycle event sequence to identify the fracture location of the event.

[0141] Based on the location of the fracture, information is supplemented into the sequence of events in the workpiece production cycle.

[0142] In some embodiments, during specific processing stages (such as cutting), the sequence of events in the workpiece production cycle may lack the ability to track individual workpieces in a refined manner, or there may be gaps or incomplete information at specific event nodes. This lack of detail may lead to insufficient accuracy in subsequent identification of abnormal events and tracing the causes of production efficiency decline, thereby affecting the effectiveness of production strategy optimization recommendations.

[0143] Therefore, cutting data can be collected first during the cutting process. Cutting data refers to various physical signals generated during the cutting of the workpiece. Cutting data includes vibration waveforms and acoustic spectra. Vibration waveforms reflect the dynamic interaction between the cutting tool and the workpiece, such as tool wear and chatter; acoustic spectra capture the acoustic characteristics generated during cutting, such as the frequency distribution and intensity of cutting sounds. These characteristics are closely related to the cutting state, material properties, and machining quality. By collecting this data, real-time and detailed physical information about the cutting process can be obtained.

[0144] Then, bandpass filtering is applied to the cutting data. The purpose is to remove irrelevant noise and interference signals from the cutting data, retaining only useful signals within a specific frequency range. For example, an appropriate passband can be set based on the operating frequency of the cutting equipment and the known noise frequency range to ensure the accuracy of subsequent analysis. A Fast Fourier Transform (FFT) is then performed on the bandpass-filtered cutting data to extract frequency domain features, including energy distribution, peak frequency, and peak amplitude. The Fast Fourier Transform (FFT) is an algorithm that converts time-domain signals into frequency-domain signals. Through the Fast Fourier Transform, the filtered vibration waveform and sound spectrum can be converted into their corresponding frequency domain representation. Frequency domain features can more clearly reveal the periodicity, resonance phenomena, and energy concentration regions in the cutting process. These features are important for distinguishing different cutting states or identifying specific workpieces.

[0145] The frequency domain features are then matched with the acoustic fingerprint database to identify the workpiece's digital identifier. The acoustic fingerprint database pre-stores unique frequency domain feature patterns corresponding to different workpieces or different cutting states. By matching the currently extracted frequency domain features with the fingerprints in the database, the unique digital identifier of the workpiece currently being processed can be accurately identified. This identifier helps track the production history and status of individual workpieces.

[0146] Based on the workpiece's digital identifier, a fracture location analysis is performed on the workpiece production cycle event sequence to identify the fracture location. Within the constructed workpiece production cycle event sequence, locations of potential discontinuities, missing data, or abnormal interruptions in the relevant event sequences of a specific workpiece can be located based on its digital identifier. For example, if the cutting event data of a workpiece is suddenly interrupted in the sequence, or deviates from the expected process, this can be identified as the fracture location.

[0147] Finally, information is supplemented into the workpiece production cycle event sequence based on the fracture location. After identifying the fracture location, detailed information obtained from the cutting data can be used to supplement the workpiece production cycle event sequence in a targeted manner. For example, missing cutting parameters, machining time, tool status, and other information can be added, or inaccurate event records can be corrected, thereby making the event sequence more complete and accurate, and providing a more reliable data foundation for subsequent production management and optimization.

[0148] To illustrate this technical solution more clearly, a specific example is used below. Assume a specific workpiece is undergoing CNC milling on a production line for a water-reducing agent cooling equipment. During the cutting process, vibration and acoustic sensors mounted on the machine tool spindle or workpiece fixture collect the vibration waveforms and sound spectrum generated by the cutting process in real time. For example, when the tool contacts the workpiece, specific vibration patterns and sounds are generated, and these signals are continuously recorded. Subsequently, this raw cutting data is transmitted to a data processing unit. First, bandpass filtering is performed to remove environmental noise and non-cutting-related vibrations from the equipment itself, retaining only the frequency components directly related to the cutting process. Next, a fast Fourier transform is performed on the filtered data to extract its frequency domain features, such as analyzing the energy distribution within a specific frequency range and identifying the main peak frequencies and their corresponding amplitudes. These frequency domain features form a unique "acoustic fingerprint." This fingerprint is then matched against a pre-established acoustic fingerprint database. This database stores typical acoustic fingerprints and their corresponding workpiece digital identifiers for different batches, materials, or designs of workpieces under standard cutting conditions. Through matching, the system successfully identified the unique digital identifier of the workpiece currently being processed, such as "workpiece A-001".

[0149] Based on the digital identifier "Workpiece A-001", the system analyzes the constructed sequence of production cycle events for the workpiece. Assuming that the cutting completion time record for "Workpiece A-001" is missing or discontinuous with the timestamps of subsequent assembly events, the system identifies a fracture location between the cutting station and the next station through fracture location analysis. At this point, the system supplements the production cycle event sequence for "Workpiece A-001" with detailed information obtained from the cutting data, such as precise cutting start time, cutting end time, and tool load changes. For example, it accurately records the actual cutting duration and whether abnormal vibrations occurred during the process (potentially indicating tool wear or material defects), inserting this supplementary information at the fracture location in the event sequence. Thus, the previously incomplete or ambiguous event sequence becomes more accurate and detailed, providing more reliable data support for subsequent production efficiency analysis and potential cause tracing.

[0150] Through the above technical solution, this embodiment can significantly improve the precision and data integrity of the workpiece production cycle event sequence. Especially in the critical cutting process, microscopic processing information can be obtained by real-time acquisition and analysis of vibration waveforms and acoustic spectra. This ability to accurately identify individual workpieces and analyze event fracture locations enables the production management system to promptly detect and correct data gaps or inconsistencies in the event sequence, thereby avoiding deviations in subsequent abnormal event identification and tracing the causes of production efficiency decline due to incomplete information. This not only improves the reliability of production event data but also provides a solid data foundation for generating more accurate production strategy optimization suggestions, further enhancing the intelligence level and decision-making accuracy of the water-reducing agent cooling equipment production management.

[0151] The beneficial effects of implementing the embodiments of the present invention include: First, the production event data of the water-reducing agent cooling equipment is obtained, and the production event data is formatted to obtain formatted event data. Then, based on the formatted event data, a workpiece production cycle event sequence is constructed. Next, based on the production process information and production standard information, abnormal events in the workpiece production cycle event sequence are identified. Based on the abnormal events, the workpiece production cycle event sequence is traced back in reverse time to identify multiple potential causes of decreased production efficiency. Finally, based on multiple potential causes of decreased production efficiency, production strategy optimization suggestions are generated. Thus, production strategy optimization suggestions can be generated by combining abnormal events and potential causes of decreased production efficiency to achieve production management and improve production efficiency and equipment product quality.

[0152] like Figure 2 As shown in the figure, this embodiment of the invention also provides a production management system for water-reducing agent cooling equipment, including:

[0153] Data acquisition module 401 is used to acquire production event data of the water-reducing agent cooling equipment;

[0154] The data formatting module 402 is used to format production event data to obtain formatted event data;

[0155] Event sequence construction module 403 is used to construct a workpiece production cycle event sequence based on formatted event data;

[0156] The abnormal event identification module 404 is used to identify abnormal events in the event sequence of the workpiece production cycle based on production process information and production standard information.

[0157] The reverse time tracing module 405 is used to perform reverse time tracing of the workpiece production cycle event sequence based on abnormal events, and identify multiple potential causes of decreased production efficiency.

[0158] The optimization suggestion generation module 406 is used to generate production strategy optimization suggestions based on multiple potential causes of decreased production efficiency.

[0159] The content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0160] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

Claims

1. A production management method for water-reducing agent cooling equipment, characterized in that, Includes the following steps: Acquire production event data for water-reducing agent cooling equipment; The production event data is formatted to obtain formatted event data; Based on the formatted event data, construct the workpiece production cycle event sequence; Based on production process information and production standard information, identify abnormal events in the event sequence of the workpiece production cycle; Based on the abnormal events, the production cycle event sequence of the workpiece is traced back in reverse time to identify multiple potential causes of decreased production efficiency; Based on the multiple potential causes of decreased production efficiency, production strategy optimization suggestions are generated. The step involves performing reverse time tracing of the workpiece production cycle event sequence based on the abnormal events to identify multiple potential causes of decreased production efficiency, including: Based on the abnormal events, core input features and resource consumption are extracted from the workpiece production cycle event sequence. The core input features include weld gap change rate, local temperature gradient of material, and predicted value of workpiece micro-warping. The resource consumption includes CPU cycle count and memory allocation. Based on the core input features, a decision intent label is determined, which includes weld strength, welding speed, heat input, or dynamic deformation. Adjust the complexity based on the parameters and assess the extension of the welding task cycle time. A correlation analysis was performed on the resource consumption, the decision intent label, and the welding task cycle time extension to obtain the first correlation relationship; A second correlation relationship is obtained by performing correlation analysis on the core input features, the resource consumption, and the decision intent label; Calculate the dynamic micro-warping contribution based on the first and second association relationships; The cycle time extension of the welding task is compared with the cycle time extension under the preset static deviation compensation model, and the contribution of the cycle time extension is calculated. The contribution of static machining deviations is evaluated based on the cutting accuracy. Based on the contribution of dynamic micro-warping, the contribution of cycle time extension, and the contribution of static processing deviation, the potential causes of the decline in production efficiency are identified.

2. The method according to claim 1, characterized in that, The step of constructing a workpiece production cycle event sequence based on the formatted event data includes: The formatted event data is validated, including timestamp synchronization validation, workpiece identifier uniqueness validation, and data field integrity validation. Based on the formatted event data after verification processing, construct the initial event sequence; Perform sequence breakage analysis on the initial event sequence to identify the causes of sequence breakage; Based on the cause of the sequence break, the initial event sequence is repaired, including timestamp calibration, workpiece identity inference, and data field completion. Conflict analysis is performed on the initial event sequence after the repair process to identify conflict events; The conflict events are disambiguated using cross-validation. Based on the conflict events after conflict disambiguation, the initial event sequence after repair processing is updated to obtain the workpiece production cycle event sequence.

3. The method according to claim 1, characterized in that, The step of identifying abnormal events in the workpiece production cycle event sequence based on production process information and production standard information includes: Acquire equipment operating parameters, material status, and environmental parameters; Calculate the long-term trend based on the equipment operating parameters, the material state, and the environmental parameters; Based on the aforementioned long-term trend, an abnormal event identification threshold is determined; The abnormal event is identified based on the abnormal event identification threshold, the production process information, and the production standard information.

4. The method according to claim 1, characterized in that, The step of generating production strategy optimization suggestions based on the multiple potential causes of decreased production efficiency includes: Extract potential event data corresponding to each potential cause of decreased production efficiency from the event sequence of the workpiece production cycle; Calculate the contribution of efficiency decline to each potential event data point; Based on the workpiece production cycle event sequence, an interactive event pattern among multiple potential causes of decreased production efficiency is identified, including periodic micro-warping caused by the combined effect of material processing parameters and equipment component wear. The contribution of the efficiency decline is adjusted based on the interaction impact event pattern. Based on the multiple revised contribution rates of efficiency decline and the multiple potential causes of production efficiency decline, the production strategy optimization suggestions are generated.

5. The method according to claim 4, characterized in that, The step of identifying interactive event patterns among multiple potential causes of decreased production efficiency based on the workpiece production cycle event sequence includes: Obtain historical production data; Train a multilayer sensing network based on the historical production data; Time alignment is performed on the cutting station event data, welding station event data, and assembly station event data in the workpiece production cycle event sequence; The multilayer sensing network is used to extract features from the time-aligned cutting station event data, welding station event data, and assembly station event data to obtain the corresponding event features. The interaction-influenced event pattern is obtained by associating and identifying multiple event features using the multilayer perceptual network.

6. The method according to claim 5, characterized in that, The step of training a multilayer perceptron based on the historical production data includes: The quality of each historical event data in the historical production data is evaluated to obtain the corresponding event data quality level; The learning weights of the multilayer sensing network for the historical event data are adjusted according to the event data quality level. The multilayer perceptron is updated based on the learning weights.

7. The method according to claim 5, characterized in that, The step of training a multilayer perceptron based on the historical production data includes: Based on the mapping relationship between event patterns and production efficiency decline in the historical production data, the first weight of the multilayer sensing network is trained. Identify the data type to which each historical event data belongs in the historical production data; Determine the forgetting factor for each data type; The second weights of the multilayer perceptron are trained based on the forgetting factor corresponding to each data type.

8. The method according to claim 1, characterized in that, After constructing the workpiece production cycle event sequence based on the formatted event data, the method further includes: The cutting data during the cutting process is collected, including vibration waveforms and acoustic spectrum. The cutting data is bandpass filtered; Fast Fourier Transform is performed on the bandpass filtered cutting data to extract frequency domain features, which include energy distribution, peak frequency, and peak amplitude. The frequency domain features are matched with the acoustic fingerprint database to identify the workpiece's digital identifier; Based on the workpiece's digital identifier, the fracture location is analyzed in the workpiece's production cycle event sequence to identify the event fracture location. Based on the location of the fracture, information is supplemented to the sequence of events in the production cycle of the workpiece.

9. A production management system for water-reducing agent cooling equipment, characterized in that, include: The data acquisition module is used to acquire production event data of the water-reducing agent cooling equipment; The data formatting module is used to format the production event data to obtain formatted event data; An event sequence construction module is used to construct a workpiece production cycle event sequence based on the formatted event data; An abnormal event identification module is used to identify abnormal events in the production cycle event sequence of the workpiece based on production process information and production standard information. The reverse time tracing module is used to perform reverse time tracing of the workpiece production cycle event sequence based on the abnormal event, and identify multiple potential causes of decreased production efficiency. The optimization suggestion generation module is used to generate production strategy optimization suggestions based on the multiple potential causes of decreased production efficiency. The reverse time tracing module is also used for: Based on the abnormal events, core input features and resource consumption are extracted from the workpiece production cycle event sequence. The core input features include weld gap change rate, local temperature gradient of material, and predicted value of workpiece micro-warping. The resource consumption includes CPU cycle count and memory allocation. Based on the core input features, a decision intent label is determined, which includes weld strength, welding speed, heat input, or dynamic deformation. Adjust the complexity based on the parameters and assess the extension of the welding task cycle time. A correlation analysis was performed on the resource consumption, the decision intent label, and the welding task cycle time extension to obtain the first correlation relationship; A second correlation relationship is obtained by performing correlation analysis on the core input features, the resource consumption, and the decision intent label; Calculate the dynamic micro-warping contribution based on the first and second association relationships; The cycle time extension of the welding task is compared with the cycle time extension under the preset static deviation compensation model, and the contribution of the cycle time extension is calculated. The contribution of static machining deviations is evaluated based on the cutting accuracy. Based on the contribution of dynamic micro-warping, the contribution of cycle time extension, and the contribution of static processing deviation, the potential causes of the decline in production efficiency are identified.

Images