Online monitoring method and system for fatigue state of forging press based on acoustic signal

Abstract

The application provides an online monitoring method and system for fatigue state of a forging press based on acoustic signals, and belongs to the technical field of online monitoring. Specifically, the method comprises the following steps: determining whether optimization processing of the monitoring processing method is needed according to the abnormality degree of the analysis and processing results of the fatigue state of different forging presses; determining the updating method of the optimization control target in the forging press according to the analysis and processing results of the fatigue state of different forging presses and the abnormality degree of the corresponding analysis and processing results; determining the coincidence degree of the forging press with suspected abnormal fatigue state according to the updating processing results of the optimization control target; and determining the monitoring and analysis method of the fatigue state of the forging press in combination with the matching degree of the optimization control target and the monitoring processing method of the acoustic signals of the forging press, thereby improving the reliability of the monitoring and analysis processing of the fatigue state.

Description

Technical Field

[0001] This invention belongs to the field of online monitoring technology, and particularly relates to an online monitoring method and system for fatigue state of forging presses based on acoustic signals. Background Technology

[0002] Each forging operation generates a broadband impact sound wave. Traditional condition monitoring relies on vibration acceleration sensors, which mainly analyze vibration characteristics from low to mid frequencies (0-10kHz) to identify and process fatigue conditions.

[0003] To address the aforementioned technical issues, the invention patent application CN202610195745.0, "A Method for Identifying and Monitoring Disconnection of a Scraper Conveyor Based on a Sound Array and Spatiotemporal Feature Network," employs adaptive spectral subtraction and improved beamforming for multi-channel denoising and focusing. It combines weighted loss function training with pruning quantization to adapt to edge devices, achieving millisecond-level identification, location, and closed-loop response from early signs of chain loosening to chain breakage. However, the following technical problems remain: In the process of evaluating and analyzing the fatigue state of forging presses based on acoustic signals, the reliability of the fatigue state evaluation model is easily affected by external acoustic signals. Therefore, it is an urgent technical problem to solve how to determine different monitoring and analysis methods for forging presses and how to dynamically update the monitoring and analysis methods using the monitoring and analysis results, so as to improve the reliability of fatigue state evaluation and analysis.

[0004] Specifically, this application provides a method and system for online monitoring of fatigue state of forging presses based on acoustic signals. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a method for online monitoring of fatigue state of forging presses based on acoustic signals, which includes: S1 determines the method for monitoring and processing the acoustic signals of the forging press based on the data of the forging press and the degree of correlation between the forging press and the forging data of different forging presses; S2 extracts and processes acoustic feature signals of different forging presses based on the monitoring and processing method, analyzes and processes the fatigue state of different forging presses based on the extraction results of the acoustic feature signals, determines the degree of abnormality of the analysis and processing results of the fatigue state of different forging presses based on the variation of the analysis and processing results of the fatigue state of different forging presses, and proceeds to the next step when it is determined that the monitoring and processing method needs to be optimized based on the degree of abnormality. S3 determines the method for updating the optimization control target in the forging press based on the analysis and processing results of the fatigue state of different forging presses and the degree of abnormality of the corresponding analysis and processing results; S4 determines the degree of overlap with the forging press with suspected abnormal fatigue state based on the updated processing result of the optimized control target, and determines the monitoring and analysis method of fatigue state of the forging press by combining the matching degree between the optimized control target and the monitoring and processing method of the acoustic signal of the forging press.

[0006] The beneficial effects of this invention are as follows: Based on the number of forging presses and the correlation of forging data between different forging presses, the need for analyzing and processing the acoustic signals of forging presses under interference conditions is determined. That is, the fewer the number of forging presses and the lower the correlation of forging data between them, the higher the need for analyzing and processing the acoustic signals of forging presses under interference conditions. The monitoring and processing method of the acoustic signals of forging presses is determined based on the need for analyzing and processing the acoustic signals of forging presses under interference conditions, thus laying the foundation for determining the reliability of the fatigue assessment model in identifying the aging state under interference conditions.

[0007] By optimizing the degree of overlap between the control target and the forging press with suspected fatigue anomalies, and the matching degree between the optimized control target and the acoustic signal monitoring and processing method of the forging press, the reliability of updating the monitoring and processing data of the optimized control target and the reliability of monitoring and analysis of the forging press with suspected fatigue anomalies are determined. Based on the reliability of updating the monitoring and processing data of the optimized control target and the reliability of monitoring and analysis of the forging press with suspected fatigue anomalies, the update processing requirements of the optimized control target are determined. Based on the update processing requirements of the optimized control target, the monitoring and analysis method of the fatigue state of the forging press is determined, thereby further improving the reliability of fatigue state assessment and analysis.

[0008] Furthermore, the forging press data includes the number of forging presses.

[0009] Furthermore, the correlation between the forging data of the forging press and different forging presses is determined based on the similarity of the number of forging operations between the forging press and different forging presses.

[0010] Furthermore, the method for determining the acoustic signal monitoring and processing method of the forging press is as follows: S11 uses the forging press data to determine the number of forging presses; S12 determines the similarity of the number of forging operations between different forging presses based on the correlation between the forging data of the forging press and different forging presses, and classifies the forging presses into different forging press groups based on the similarity. S13 determines the monitoring and processing method for the acoustic signal of the forging press based on the number of the forging presses and the number of forging presses in the forging press group to which the forging press is located.

[0011] Furthermore, it was determined that the monitoring and processing methods needed to be optimized, specifically including: S21. Based on the analysis and processing results of the fatigue state of the forging press, determine the average value of the fatigue state of the forging press in different analysis and processing processes, and take the average value of the fatigue state of the forging press in different analysis and processing processes as the benchmark value of the fatigue state of the forging press. S22 determines the variable processing process in the analysis and processing of the forging press based on the deviation between the fatigue state value and the benchmark value during the analysis and processing of the forging press, and determines the monitoring deviation risk coefficient of the forging press based on the proportion of variable processing processes in the analysis and processing of the forging press. S23 determines whether the monitoring and processing methods need to be optimized based on the monitoring deviation risk coefficient of different forging presses.

[0012] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described method for online monitoring of fatigue state of a forging press based on acoustic signals when running the computer program.

[0013] Other features and advantages will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0014] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0015] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart of an online monitoring method for fatigue state of forging presses based on acoustic signals; Figure 2 This is a flowchart illustrating the method for determining the monitoring and processing of acoustic signals from a forging press. Figure 3 It is a flowchart for determining the optimization process of the monitoring and processing methods that need to be implemented. Detailed Implementation

[0017] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that the invention will be thorough and complete, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore their detailed description will be omitted.

[0018] The terms “a,” “one,” “the,” and “the” are used to indicate the existence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended meaning of inclusion and that other elements / components / etc. may exist in addition to the listed elements / components / etc.

[0019] Example 1 To solve the above problems, according to one aspect of the present invention, such as Figure 1 As shown, an online monitoring method for fatigue state of forging presses based on acoustic signals is provided, specifically including: S1 determines the method for monitoring and processing the acoustic signals of the forging press based on the data of the forging press and the degree of correlation between the forging press and the forging data of different forging presses; Furthermore, the forging press data includes the number of forging presses.

[0020] Furthermore, the correlation between the forging data of the forging press and different forging presses is determined based on the similarity of the number of forging operations between the forging press and different forging presses.

[0021] Specifically, such as Figure 2 As shown, the method for determining the monitoring and processing method of the acoustic signal of the forging press is as follows: In this embodiment, the need for analyzing and processing the acoustic signals of the forging presses under interference conditions is determined based on the number of forging presses and the degree of correlation between the forging data of different forging presses. That is, the fewer the number of forging presses and the lower the degree of correlation between the forging data of other forging presses, the higher the need for analyzing and processing the acoustic signals of the forging presses under interference conditions. The monitoring and processing method of the acoustic signals of the forging presses is determined based on the need for analyzing and processing the acoustic signals of the forging presses under interference conditions, thereby laying the foundation for determining the reliability of the fatigue assessment model in identifying the aging state under interference conditions.

[0022] The core objective of this embodiment is to determine the monitoring and processing methods for the acoustic signals of each forging press based on forging press data and the correlation between forging data of different forging presses. This lays the foundation for determining the reliability of the fatigue assessment model in identifying aging states under interference conditions. The core logic is: by quantifying the number of forging presses and the data distribution characteristics within forging press groups, different acoustic signal monitoring trigger strategies are configured for different forging press groups. This allows for more frequent monitoring of forging presses with fewer machines and lower correlation, thereby achieving full identification of interference states under limited monitoring resources. The overall logic follows the process of "determining the number of forging presses → dividing groups according to the similarity of forging times → determining the monitoring method based on the number and group distribution."

[0023] S11 uses the forging press data to determine the number of forging presses; The forging press data refers to the basic information dataset of all forging presses covered by the current monitoring system, including basic attributes such as forging press number, model, commissioning time, and rated forging times. The number of forging presses is the total number of valid online forging presses in the dataset.

[0024] Suppose a forging press monitoring system covers multiple online forging presses. By reading the forging press data, the total number of currently active online forging presses can be determined.

[0025] This step provides a basic quantitative basis for subsequent group division and monitoring method determination. Its significance lies in ensuring that the configuration of the monitoring strategy is based on the actual online scale, avoiding statistical distortion caused by interference from offline devices, thereby ensuring the accuracy of subsequent decision-making logic.

[0026] S12 determines the similarity of the number of forging operations between different forging presses based on the correlation between the forging data of the forging press and different forging presses, and classifies the forging presses into different forging press groups based on the similarity. The correlation of the forging data refers to the consistency of the trend of the number of forgings of two forging presses in the same period. It is quantified by calculating the correlation coefficient or distance metric between the historical forging number sequences of different forging presses. The similarity of the number of forgings refers to the closeness of the cumulative number of forgings of different forging presses. The higher the similarity, the closer the working load characteristics of the two forging presses are. The forging press group refers to the set formed by dividing forging presses in the same forging number range into the same group. The forging presses in each group have similar cumulative forging number and fatigue characteristics.

[0027] Suppose a system has multiple forging presses. The system is divided into several intervals based on the number of forging operations (such as low load interval, medium load interval, high load interval, etc.). Multiple forging presses whose cumulative forging operations fall within the same interval are grouped into the same group, thus forming multiple forging press groups.

[0028] This step groups forging presses with similar fatigue characteristics together. Its significance lies in providing a structured basis for configuring subsequent differentiated monitoring strategies, ensuring that forging presses within the same group can refer to each other, and improving the reliability of horizontal comparisons in fatigue state analysis.

[0029] S13 determines the monitoring and processing method for the acoustic signal of the forging press based on the number of the forging presses and the number of forging presses in the forging press group to which the forging press is located.

[0030] The monitoring and processing method refers to the conditional strategy for triggering acoustic signal monitoring of the forging press, which determines under what circumstances acoustic signal acquisition and feature extraction will be initiated. Different monitoring and processing methods have different trigger sensitivities to adapt to monitoring needs under different scales and degrees of correlation.

[0031] Assuming there are multiple forging press groups in the system, different monitoring trigger conditions are configured for different groups based on whether the total number of forging presses exceeds a preset forging press number threshold and whether the number of forging presses in each group reaches a preset number threshold.

[0032] This step enables differentiated allocation of monitoring strategies. Its significance lies in ensuring monitoring coverage while giving higher trigger frequencies to a small number of forging press groups with low correlation, thereby improving the reliability of fatigue assessment model identification under disturbance conditions.

[0033] It should be noted that the forging presses are divided into different forging press groups, specifically including: Forging presses that fall within the same forging cycle range will be grouped into the same forging press group.

[0034] The forging frequency range refers to a continuous numerical segment pre-defined according to the cumulative forging frequency of the forging press. Each range corresponds to a load level, and forging presses falling into the same range have high homogeneity in terms of fatigue level and acoustic characteristics.

[0035] Assume the system is divided into several intervals based on the cumulative number of forging operations of the forging press. All forging presses whose cumulative forging operations fall within the same interval are assigned to the same forging press group, and forging presses from different intervals are not allowed to be mixed.

[0036] This step establishes an objective standard for group division. Its significance lies in using a quantifiable range of forging times as the basis for grouping, avoiding subjective biases in manual grouping, ensuring homogeneity within groups, and thus improving the effectiveness of horizontal comparison of fatigue states.

[0037] Specifically, based on the number of forging presses and the number of forging presses in the forging press group, a method for monitoring and processing the acoustic signals of the forging presses is determined, including: Case 1: If the number of forging presses is greater than the preset threshold for the number of forging presses, then based on the monitoring data of different forging presses, the degree of influence of the interference signal on the identification and processing of the aging state of the forging presses can be determined. Therefore, the monitoring and processing method for the acoustic signals of all forging presses is determined to be that when the amplitude characteristic value of the noise signal is greater than the preset threshold, the acoustic signal monitoring and processing of the forging presses is performed. The amplitude characteristic value of the noise signal refers to the quantitative index of the amplitude of environmental noise not generated by the forging operation itself during the acoustic signal acquisition process. When the value exceeds the preset threshold, it indicates that the current environmental interference is strong and targeted acoustic signal monitoring and processing needs to be initiated. The preset threshold for the number of forging presses refers to the critical number value for judging whether the scale of the forging presses is sufficient to support interference analysis.

[0038] Assuming that the total number of online forging presses in the system exceeds the preset threshold for the number of forging presses, it indicates that the monitoring data of different forging presses at the current scale are sufficient for mutual reference, and the influence of interference signals can be identified through horizontal comparison. Therefore, a strategy of "triggering monitoring when the amplitude characteristic value of noise signal is greater than the preset threshold" is uniformly configured for all forging presses.

[0039] In this case, the monitoring triggering conditions are relatively conservative. The significance of this is that when there are enough forging presses, the cross-sectional comparison of multiple machines can be used to avoid false triggering, reduce unnecessary monitoring costs, and at the same time ensure the ability to identify real interference.

[0040] Case 2: If the number of forging presses is not greater than a preset forging press number threshold, obtain the number of forging presses in the forging press group to which the forging press is located. If the number of forging presses in the forging press group to which the forging press is located is less than a preset number threshold, then determine that the acoustic signal monitoring and processing method of the forging press is to perform acoustic signal monitoring and processing of the forging press when the amplitude characteristic value of the noise signal is greater than a preset threshold or when the acoustic signal monitoring and processing of the forging press has not been performed within the most recent first preset time period. The first preset duration refers to the time-based monitoring interval configured for groups with insufficient forging presses, ensuring that even if the noise signal amplitude does not exceed the threshold, a monitoring can be forcibly triggered within this duration to prevent long-term monitoring gaps; the preset quantity threshold refers to the critical value for judging whether the number of forging presses in a single group is sufficient.

[0041] Assuming that the total number of online forging presses in the system does not exceed the preset threshold for the number of forging presses, and the number of forging presses in a certain group is less than the preset threshold, then the group is configured with a strategy of "triggering when the noise signal amplitude characteristic value is greater than the preset threshold or when it has not been monitored within the most recent first preset time period", which adds a time dimension trigger guarantee compared to case 1.

[0042] The introduction of a time-based fallback mechanism in this situation is significant because it provides additional protection for small groups, preventing the early changes in fatigue state from being missed due to noise signals not reaching the threshold for an extended period, and improving the monitoring reliability in small-scale scenarios.

[0043] Case 3: The forging press group that performs acoustic signal monitoring when the amplitude characteristic value of the noise signal is greater than a preset threshold or when the acoustic signal monitoring of the forging press has not been performed within the most recent first preset time period is designated as a reliable monitoring and analysis group. It is determined whether the number of reliable monitoring and analysis groups exceeds a preset reliable group number threshold. If so, the acoustic signal monitoring and processing method for the remaining forging press groups is determined to be: when the amplitude characteristic value of the noise signal is greater than a preset threshold, the acoustic signal monitoring and processing of the forging press is performed. If not, the acoustic signal monitoring and processing method for the remaining forging press groups is determined to be: when the amplitude characteristic value of the noise signal is greater than a preset threshold or when the acoustic signal monitoring and processing of the forging press has not been performed within the most recent second preset time period, the acoustic signal monitoring and processing of the forging press is performed.

[0044] The reliable monitoring and analysis group refers to the forging press group that has been assigned to the monitoring method described in Case 2 (including the first preset duration fallback strategy). This group has a high monitoring coverage and frequency, which can provide a reference for the strategy configuration of the remaining groups. The second preset duration is an extension of the first preset duration, which is used to further strengthen the time fallback guarantee for the remaining groups when the number of reliable monitoring groups is insufficient. It can be understood that the first preset duration is shorter than the second preset duration.

[0045] Assuming that multiple reliable monitoring and analysis groups have been formed after allocation in scenario 2, if the number of reliable monitoring and analysis groups exceeds the preset threshold for the number of reliable groups, it indicates that there are enough high-frequency monitoring reference groups, and the remaining groups can adopt a relatively conservative "noise amplitude threshold triggering" strategy. If the number of reliable monitoring and analysis groups does not reach the threshold, an enhanced strategy of "noise amplitude threshold triggering or second preset duration fallback" is configured for the remaining groups to compensate for the lack of high-frequency reference groups.

[0046] This approach enables adaptive strategy allocation for the remaining groups. Its significance lies in dynamically adjusting the trigger sensitivity of the remaining groups based on the coverage of reliable monitoring groups, avoiding information blind spots in the overall monitoring system due to insufficient high-frequency reference groups, thereby improving the systematicness and adaptability of the monitoring method configuration.

[0047] Furthermore, fatigue state analysis and processing of different forging presses are carried out, specifically including: The extracted acoustic feature signal of the forging press after denoising is output to the fatigue state analysis model, and the fatigue state value of the forging press is obtained based on the output of the fatigue state analysis model.

[0048] The denoised acoustic feature signal refers to the acoustic feature components related to the fatigue state of the forging press that are retained after noise reduction processing, including vibration frequency features, amplitude envelope features, impact features, etc.; the fatigue state analysis model refers to a pre-trained machine learning or signal processing model used to map acoustic features to fatigue state values; the fatigue state value refers to the numerical value output by the model that quantitatively represents the current fatigue degree of the forging press, and the higher the value, the more severe the fatigue degree.

[0049] Suppose that after a forging press triggers acoustic signal monitoring, it completes noise reduction and extracts acoustic feature signals. These feature signals are then input into a fatigue state analysis model to obtain the fatigue state value of the forging press in this analysis.

[0050] This step transforms acoustic signals into quantifiable fatigue assessment results. Its significance lies in achieving an objective numerical characterization of the fatigue level of the forging press through modeling, providing a standardized data foundation for subsequent deviation risk analysis and monitoring method optimization.

[0051] In this embodiment, the system monitors a total of 20 online forging presses (M1~M20), and the preset threshold for the number of forging presses is 15.

[0052] S11 execution: By reading the forging press data, it is confirmed that the number of currently effective online forging presses is 20, which is greater than the preset forging press number threshold of 15.

[0053] S12 Execution: Based on the cumulative number of forging operations of each forging press, groups are divided according to the following intervals: Forging press group A (low load range, cumulative forging times 0~50000 times): M1, M2, M3, M4, M5, a total of 5 units; Forging press group B (medium and low load range, cumulative forging times 50,001~100,000): M6, M7, M8, M9, M10, a total of 5 units; Forging press group C (medium-high load range, cumulative forging times 100,001~150,000 times): M11, M12, M13, M14, M15, a total of 5 units; Forging press group D (high load range, cumulative forging times of more than 150,001 times): M16, M17, M18, M19, M20, a total of 5 units.

[0054] S13 Execution (Case 1 Trigger): Since the total number of forging presses is 20, which is greater than the preset threshold of 15 forging presses, Case 1 is directly triggered. The acoustic signal monitoring and processing method for all forging presses (M1~M20) is determined as follows: when the amplitude characteristic value of the noise signal is greater than the preset threshold, the acoustic signal monitoring and processing of the forging press is performed.

[0055] S2 extracts and processes acoustic feature signals of different forging presses based on the monitoring and processing method, analyzes and processes the fatigue state of different forging presses based on the extraction results of the acoustic feature signals, determines the degree of abnormality of the analysis and processing results of the fatigue state of different forging presses based on the variation of the analysis and processing results of the fatigue state of different forging presses, and proceeds to the next step when it is determined that the monitoring and processing method needs to be optimized based on the degree of abnormality. Specifically, such as Figure 3 As shown, the monitoring and processing methods that need to be optimized are identified, specifically including: In this embodiment, based on the changes in the fatigue state analysis results of the forging press, the stability of the evaluation results of the fatigue state analysis model for different forging presses under severe interference is determined. Based on the stability of the evaluation results of the fatigue state analysis model in different forging presses, it is determined whether the monitoring and processing method needs to be optimized. This lays the foundation for ensuring the reliability of the update processing of the fatigue state analysis model and the reliability of the fatigue state evaluation and analysis processing when the stability is poor.

[0056] The core objective of this embodiment is to extract acoustic characteristic signals from each forging press based on the monitoring and processing method determined in S1, complete fatigue state analysis, and quantify the monitoring deviation risk based on the changes in the analysis results, thereby determining whether the monitoring and processing method needs to be optimized. Its core logic is: by calculating the historical average fatigue state value of each forging press as a benchmark value, identifying the deviation processing processes, determining the monitoring deviation risk coefficient based on the proportion of these deviation processing processes, and then determining whether optimization is needed through multi-level judgment. The overall logic follows the process of "extracting acoustic features → fatigue state analysis → determining benchmark value → statistical analysis of variation processes → calculating monitoring deviation risk coefficient → determining whether optimization is needed".

[0057] S21. Based on the analysis and processing results of the fatigue state of the forging press, determine the average value of the fatigue state of the forging press in different analysis and processing processes, and take the average value of the fatigue state of the forging press in different analysis and processing processes as the benchmark value of the fatigue state of the forging press. The analysis and processing process refers to a complete fatigue state analysis process completed after each triggering of acoustic signal monitoring and processing, and each analysis generates a fatigue state value; the average value of the fatigue state value refers to the arithmetic mean of the fatigue state values ​​obtained by a certain forging press in several historical analysis and processing processes; the benchmark value refers to the fatigue state stability reference level with the above average value as a reference, representing the typical fatigue degree of the forging press under normal interference level.

[0058] Suppose that a forging press has generated fatigue state values ​​in multiple past analysis and processing processes. Calculate the arithmetic mean of these fatigue state values ​​and use this mean as the benchmark value for the forging press, which will be used as a reference for subsequent deviation judgment.

[0059] This step establishes a personalized fatigue state benchmark for each forging press. Its significance lies in eliminating the influence of occasional fluctuations through historical averages, forming a stable reference level, thereby making subsequent deviation judgments more objective and accurate, and avoiding the impact of single extreme values ​​on the reliability of overall deviation analysis.

[0060] S22 determines the variable processing process in the analysis and processing of the forging press based on the deviation between the fatigue state value and the benchmark value during the analysis and processing of the forging press, and determines the monitoring deviation risk coefficient of the forging press based on the proportion of variable processing processes in the analysis and processing of the forging press. The aforementioned variation processing process refers to the analysis and processing process where the deviation rate from the benchmark value of the forging press is greater than a preset deviation rate threshold. The deviation rate is calculated as: |Current fatigue state value - Benchmark value| ÷ Benchmark value. The percentage of variation processing processes refers to the proportion of variation processing processes to the total number of analysis and processing processes. The monitoring deviation risk coefficient is a risk indicator quantified by the percentage of variation processing processes, representing the degree of interference affecting the current monitoring system. The larger the coefficient, the more frequent the fluctuations in the monitoring results and the more severe the interference.

[0061] Suppose that a forging press has completed multiple analysis and processing processes in its history. In some of these processes, the deviation rate between the fatigue state value and the benchmark value exceeds a preset deviation rate threshold. These processes are identified as variable processes. The monitoring deviation risk coefficient of the forging press is obtained by dividing the number of variable processes by the total number of analysis and processing processes.

[0062] This step transforms the fluctuation of fatigue state into a quantifiable risk coefficient. Its significance lies in providing objective numerical basis for whether to trigger the optimization of monitoring methods, avoiding misjudgments based on a single deviation, and thus improving the accuracy of optimization trigger decisions.

[0063] S23 determines whether the monitoring and processing methods need to be optimized based on the monitoring deviation risk coefficient of different forging presses.

[0064] The optimization of the monitoring and processing method refers to adjusting the triggering conditions or monitoring frequency based on the current acoustic signal monitoring triggering strategy, so as to reduce the impact of interference signals on the stability of fatigue state analysis results and ensure the reliability of fatigue state analysis model update processing.

[0065] Assuming there are multiple forging presses in the system, by comparing the monitoring deviation risk coefficient of each forging press with the preset deviation coefficient threshold, and by statistically analyzing the proportion of forging presses with monitoring risks, it can be determined whether there is a systemic deviation risk in the current overall monitoring system, and based on this, it can be determined whether to initiate the optimization of the monitoring and processing method.

[0066] This step enables multi-level quantitative judgment of optimization needs. Its significance lies in preventing over-optimization due to the abnormality of individual forging presses, and also preventing delays in optimization due to the underestimation of overall risk, thereby achieving a balance between resource utilization efficiency and monitoring reliability.

[0067] Specifically, the variation processing process in the analysis and processing of the forging press is the analysis and processing process when the deviation rate from the reference value of the forging press is greater than a preset deviation rate threshold.

[0068] The deviation rate threshold refers to the critical proportion by which the result of a certain analysis and processing process deviates significantly from the benchmark. When the deviation rate between a certain fatigue state value and the benchmark value exceeds the threshold, the process is recorded as a change process.

[0069] Assuming a preset deviation rate threshold is a certain percentage, if the deviation rate of the fatigue state value obtained by a certain forging press in a certain analysis process from the benchmark value exceeds the threshold, then the analysis process is identified as a change process and is counted in the change process count of the forging press.

[0070] This step clarifies the criteria for identifying the change process. Its significance lies in ensuring the consistency of the identification criteria for the change process through a unified deviation rate threshold, and avoiding the lack of comparability of risk coefficients due to inconsistent deviation judgment criteria between different forging presses.

[0071] Understandably, based on the monitoring deviation risk coefficient of different forging presses, it is determined whether the monitoring and processing methods need to be optimized, specifically including: Scenario 1: If there is a forging press with a monitoring deviation risk coefficient greater than the preset deviation coefficient threshold, then it is determined that the monitoring and processing method needs to be optimized. The preset deviation coefficient threshold refers to the critical value for determining whether the monitoring deviation risk of a forging press has reached the point where optimization must be triggered. When the monitoring deviation risk coefficient of any forging press exceeds this threshold, it indicates that the monitoring results of the forging press have been affected by interference to a degree that cannot be ignored, and optimization must be initiated immediately.

[0072] If the monitoring deviation risk coefficient of one forging press in the system exceeds the preset deviation coefficient threshold, then regardless of the risk coefficient levels of other forging presses, it is directly determined that the monitoring and processing methods need to be optimized.

[0073] This situation establishes the highest priority optimization triggering rule, which is significant because it implements zero tolerance for serious interference risks and ensures that any high risk from a single forging press can trigger a timely optimization response at the system level, preventing the overall assessment reliability from being affected by the downplaying of local risks.

[0074] Case 2: If there are no forging presses with a monitoring deviation risk coefficient greater than the preset deviation coefficient threshold, forging presses with a monitoring deviation risk coefficient greater than the preset coefficient threshold are designated as monitoring risk forging presses. If the proportion of monitoring risk forging presses in the forging presses is greater than the preset proportion threshold, then it is determined that the monitoring processing method needs to be optimized. The monitored risk forging press refers to a forging press whose monitoring deviation risk coefficient reaches a preset coefficient threshold but has not yet exceeded the preset deviation coefficient threshold, and is in the medium risk range; the preset proportion threshold refers to the critical proportion for judging whether the monitored risk forging press has constituted a systemic risk in the whole.

[0075] Assuming that the monitoring deviation risk coefficients of all forging presses do not exceed the preset deviation coefficient threshold, but the monitoring deviation risk coefficients of some forging presses are above the preset coefficient threshold, these forging presses are identified as monitoring risk forging presses; if the number of monitoring risk forging presses accounts for more than the preset proportion threshold of all forging presses, then it is determined that optimization processing is required.

[0076] This situation triggers optimization from the perspective of overall risk proportion. Its significance lies in the fact that when multiple forging presses are simultaneously at medium deviation risk, the overall monitoring system has shown a systematic deviation tendency. At this time, starting optimization can prevent medium risk from accumulating and evolving into severe risk.

[0077] Scenario 3: If the proportion of the monitored risk forging press in the total number of forging presses is not greater than the preset proportion threshold, the monitoring deviation risk value is determined by the proportion of the monitored risk forging press in all forging presses and the average value of the monitoring deviation risk coefficient of different forging presses. It is then determined whether the monitoring deviation risk value is greater than the preset risk threshold. If yes, it is determined that the monitoring processing method needs to be optimized. If no, it is determined that the monitoring processing method does not need to be optimized.

[0078] The monitoring deviation risk value refers to a composite risk index that comprehensively considers the proportion of monitoring risk forging presses and the average monitoring deviation risk coefficient of all forging presses. Its calculation formula is: Monitoring deviation risk value = Proportion of monitoring risk forging presses × Average monitoring deviation risk coefficient. The preset risk threshold refers to the critical value for judging whether the overall risk level needs to be triggered for optimization.

[0079] For example, assuming that the proportion of monitoring risk forging presses does not exceed the preset proportion threshold, the monitoring deviation risk value needs to be further calculated. If the proportion of monitoring risk forging presses is 0.20 and the average monitoring deviation risk coefficient of all forging presses is 0.35, then the monitoring deviation risk value = 0.20 × 0.35 = 0.070. If the preset risk threshold is 0.060, then 0.070 > 0.060, and it is determined that the monitoring processing method needs to be optimized.

[0080] This situation achieves a refined judgment of the overall risk level through synthetic risk indicators. Its significance lies in avoiding misjudgments based solely on the proportion or mean. By multiplying the two, the intensity and breadth of risk are comprehensively measured, thus providing a final, accurate screening of potential systemic risks when neither situation 1 nor situation 2 is triggered.

[0081] S3 determines the method for updating the optimization control target in the forging press based on the analysis and processing results of the fatigue state of different forging presses and the degree of abnormality of the corresponding analysis and processing results; Furthermore, the method for determining the update method of the optimization control target in the forging press is as follows: In this embodiment, based on the analysis and processing results of the fatigue state of the forging press and the degree of abnormality of the corresponding analysis and processing results, the number of forging presses with suspected fatigue state abnormalities and the reliability of the monitoring and analysis processing of forging presses with suspected fatigue state abnormalities are determined. Using the number of forging presses with suspected fatigue state abnormalities and the reliability of the monitoring and analysis processing of forging presses with suspected fatigue state abnormalities, the update processing requirements of the fatigue state assessment model are determined. Based on the update processing requirements, the update method of the optimization control target in the forging press is determined. Thus, while ensuring the reliable monitoring of the forging press, the update processing requirements of the fatigue state assessment model are also guaranteed through the update of training data.

[0082] The core objective of this embodiment is to identify a group of forging presses at fatigue risk based on the fatigue state analysis results and their degree of anomaly, and to determine the optimal control objective and its update method through multi-level judgment logic, providing reliable training data for updating the fatigue state assessment model. Its core logic is as follows: identifying forging presses whose fatigue state benchmark value exceeds the threshold as fatigue risk forging presses; combining their monitoring deviation risk coefficient; judging whether different update methods are triggered layer by layer; and achieving a refined determination of the optimal control objective update method through the calculation of comprehensive risk weight values ​​and monitoring analysis demand factors. The overall logic follows the process of "identifying fatigue risk forging presses → judging their proportion → analyzing monitoring deviation risk → calculating comprehensive risk weight values ​​→ calculating monitoring analysis demand factors → determining update methods".

[0083] S31 uses the analysis and processing results of the fatigue state of the forging press to determine the benchmark value of the fatigue state value of the forging press, and forge presses with benchmark values ​​greater than the preset fatigue state threshold are designated as fatigue risk forging presses. The fatigue risk forging press refers to a forging press whose baseline value exceeds the preset fatigue state threshold. The fatigue level of such forging presses has reached a level that requires special attention and priority inclusion in model update training. The preset fatigue state threshold is a critical baseline value for judging whether a forging press has entered the high-risk fatigue range. Exceeding this threshold indicates that the fatigue state of the forging press is at a high level and there is a significant risk of failure.

[0084] Assuming there are multiple forging presses in the system, and the reference values ​​for each forging press have been calculated in S21, forging presses whose reference values ​​exceed the preset fatigue state threshold are identified as fatigue risk forging presses, and their number and proportion in all forging presses are counted.

[0085] This step establishes the core focus of fatigue state analysis. Its significance lies in focusing limited optimization and control resources on the forging press group with the highest fatigue risk, avoiding unnecessary optimization of low-risk forging presses, thereby improving resource allocation efficiency.

[0086] Furthermore, the above steps include the following: Determine whether the proportion of fatigue risk forging presses in the forging presses is greater than the preset risk forging press proportion threshold. If so, determine that the method for updating the optimization control target in the forging press is to take the forging press with the monitoring deviation risk coefficient less than the preset risk coefficient threshold as the optimization control target, thereby determining under what noise conditions the forging press is prone to monitoring deviation, and thus realizing the update processing of the fatigue state analysis model. If not, proceed to step S32. The preset risk forging press proportion threshold refers to the critical proportion for judging whether the fatigue risk forging press group has occupied a large proportion of all forging presses. When the fatigue risk forging press proportion exceeds this threshold, it indicates that the fatigue problem of the overall equipment group is relatively common. Forging presses with lower monitoring deviation risk coefficients (i.e., more reliable monitoring results) should be selected as the optimization control target. The above control targets have enough reliable monitoring data, which can help to identify and process acoustic features with high interference risk. The preset risk coefficient threshold refers to the upper limit of monitoring deviation risk for screening reliable optimization control targets.

[0087] If the proportion of fatigue risk forging presses to all forging presses exceeds the preset threshold for the proportion of risk forging presses, then forging presses with a monitoring deviation risk coefficient less than the preset risk coefficient threshold are selected as the optimization control target to ensure that the reliability of the updated identification of training data used for model updates is high.

[0088] This judgment establishes a rapid screening path in scenarios where fatigue risk is widely distributed. Its significance lies in the fact that when the overall fatigue level of the equipment group is high, there is no need to consider other factors, thereby improving the efficiency of model update processing.

[0089] S32 determines the monitoring deviation risk coefficient of the fatigue risk forging press based on the degree of abnormality of the analysis and processing results corresponding to the fatigue risk forging press; The degree of abnormality in the analysis and processing results refers to the severity and frequency of the fatigue state value deviating from the benchmark value during the historical analysis and processing of the fatigue risk forging press, which is quantified by the monitoring deviation risk coefficient. The calculation method of the monitoring deviation risk coefficient here is the same as in S22, that is, the number of variable processing processes ÷ the total number of analysis and processing processes.

[0090] Assuming that the proportion of fatigue risk forging presses determined by S31 does not exceed the preset threshold for the proportion of risk forging presses, the monitoring deviation risk coefficient of each fatigue risk forging press is calculated for subsequent S321 judgment.

[0091] This step, after narrowing down the scope to fatigue risk forging presses, further quantifies the monitoring reliability of each fatigue risk forging press. Its significance lies in accurately identifying which fatigue risk forging presses have higher interference levels in scenarios where fatigue risk distribution is relatively concentrated, thus providing a basis for subsequent differentiated optimization of control target selection.

[0092] It should be noted that the above steps include the following: S321 Determine whether there is a fatigue risk forging press with a monitoring deviation risk coefficient greater than the preset deviation coefficient threshold. If yes, determine that the method for updating the optimization control target in the forging press is to take the forging press with a monitoring deviation risk coefficient less than the preset risk coefficient threshold as the optimization control target, thereby determining under what noise conditions the forging press is prone to monitoring deviation, and thus realizing the update processing of the fatigue state analysis model. If not, proceed to step S322. The preset deviation coefficient threshold is applied here to a subset of fatigue risk forging presses. When there is a forging press in the fatigue risk forging press whose monitoring deviation risk coefficient exceeds the threshold, it indicates that the monitoring results of some fatigue risk forging presses are seriously unreliable, and it is necessary to select a forging press with better monitoring stability from all forging presses as the optimization control target.

[0093] If, in the fatigue risk forging press, the monitoring deviation risk coefficient of one forging press exceeds the preset deviation coefficient threshold, then the forging press whose monitoring deviation risk coefficient is less than the preset risk coefficient threshold among all forging presses is taken as the optimization control target, thereby improving the update efficiency of training data.

[0094] S322 determines the comprehensive risk weight value of the fatigue risk forging press by multiplying the benchmark value of different fatigue risk forging presses with the monitoring deviation risk coefficient. It then determines whether the sum of the comprehensive risk weight values ​​of different fatigue risk forging presses is less than a preset risk weight threshold. If so, the method for updating the optimization control target in the forging press is to take the forging press with a monitoring deviation risk coefficient less than a preset risk coefficient threshold and a benchmark value above a preset fatigue state threshold as the optimization control target. This determines under what noise conditions the forging press is prone to monitoring deviation, thereby updating the fatigue state analysis model. If not, proceed to step S33. The comprehensive risk weight value refers to the product of the benchmark value of a fatigue risk forging press and its monitoring deviation risk coefficient. This indicator reflects both the fatigue level of the equipment (benchmark value) and the monitoring instability level (monitoring deviation risk coefficient), and comprehensively reflects the risk priority of the equipment. The preset risk weight threshold refers to the critical value for judging whether the sum of the comprehensive risk weight values ​​of all fatigue risk forging presses reaches the requirements of high-level optimization processing.

[0095] The calculation formula is: Comprehensive risk weight value (single unit) = Benchmark value × Monitoring deviation risk coefficient. The sum of comprehensive risk weight values ​​= ∑ (Comprehensive risk weight value of each fatigue risk forging press). Assuming that the monitoring deviation risk coefficient of the fatigue risk forging press exceeds the preset deviation coefficient threshold, calculate the comprehensive risk weight value for each fatigue risk forging press and sum them. If the sum of the comprehensive risk weight values ​​is less than the preset risk weight threshold, then the forging press whose monitoring deviation risk coefficient is less than the preset risk coefficient threshold and whose benchmark value is not lower than the preset fatigue state threshold is taken as the optimization control target.

[0096] This step introduces a joint screening mechanism of fatigue level and monitoring stability by comprehensively considering risk weight values. Its significance lies in simultaneously considering two dimensions: equipment risk level (benchmark value) and monitoring reliability (deviation risk coefficient). It selects forging presses with high fatigue level and relatively stable monitoring as the optimization control target, which on the one hand ensures the quality of updated data, and on the other hand achieves reliable monitoring of forging presses with high fatigue level.

[0097] S33 uses the baseline value of the fatigue risk forging press and the monitoring deviation risk coefficient of different fatigue risk forging presses to determine the update method of the optimized control target in the forging press.

[0098] Specifically, in the above steps, the monitoring and analysis requirement factor of the forging press is determined by the average comprehensive risk weight value of different fatigue risk forging presses and the proportion of fatigue risk forging presses in the forging press. It is then determined whether the monitoring and analysis requirement factor of the forging press is greater than a preset factor threshold. If so, the method for updating the optimal control target of the forging press is to take the forging press with a monitoring deviation risk coefficient less than a preset risk coefficient threshold as the optimal control target, thereby determining under what noise conditions the forging press is prone to monitoring deviation, and thus updating the fatigue state analysis model. If not, the method for updating the optimal control target of the forging press is to take the forging press with a monitoring deviation risk coefficient less than a preset risk coefficient threshold and a benchmark value above a second preset fatigue state threshold (less than the preset fatigue state threshold) as the optimal control target, thereby determining under what noise conditions the forging press is prone to monitoring deviation, and thus updating the fatigue state analysis model.

[0099] The monitoring and analysis demand factor refers to a composite index of the average comprehensive risk weight value of fatigue risk forging presses and the proportion of fatigue risk forging presses, used to measure the urgency of updating the optimization control target of the current equipment group as a whole; its calculation formula is: Monitoring and analysis demand factor = average comprehensive risk weight value × proportion of fatigue risk forging presses; the second preset fatigue state threshold is less than the preset fatigue state threshold, used to appropriately expand the screening range of optimization control targets when the monitoring and analysis demand is low.

[0100] Assuming the average comprehensive risk weight value of fatigue risk forging presses is a certain value, and the proportion of fatigue risk forging presses is a certain percentage, the two are multiplied to obtain the monitoring and analysis requirement factor. If this factor is greater than the preset factor threshold, then the forging press with a monitoring deviation risk coefficient less than the preset risk coefficient threshold is taken as the optimization control target; if it is not greater than the preset factor threshold, then the forging press with a monitoring deviation risk coefficient less than the preset risk coefficient threshold and a benchmark value not lower than the second preset fatigue state threshold is taken as the optimization control target.

[0101] This step achieves dynamic adjustment of the selection range of optimized control targets under high / low urgency scenarios by monitoring and analyzing demand factors. Its significance lies in selecting the widest range of reliable targets when the urgency is high to ensure the sufficiency of data for model updates, and introducing a second fatigue state threshold as an additional screening condition when the urgency is low, reducing the difficulty of data analysis while ensuring the reliability of monitoring forging presses with high fatigue states.

[0102] Furthermore, if the forging press is an optimized control target, the fatigue state of the forging press is analyzed and processed using a fatigue state analysis model according to a preset time period and the acoustic signal monitoring and processing method, thereby determining under what noise conditions the forging press is prone to monitoring deviation.

[0103] The preset time period refers to the time interval for periodically analyzing the fatigue state of the optimized control target. By periodically analyzing and accumulating fatigue state data of the optimized control target under different noise environments, continuous high-quality training data is provided for updating the fatigue state assessment model.

[0104] Assuming a forging press is identified as the target for optimized control, the system will periodically analyze and process its fatigue state according to a preset time period (e.g., every 24 hours) and the acoustic signal monitoring and processing method determined for the forging press, and continuously collect its monitoring data under different noise conditions.

[0105] This step transforms the optimization control objective into a continuous data acquisition task. Its significance lies in accumulating high-quality model training data through systematic periodic monitoring, enabling the fatigue state assessment model to continuously learn the fatigue characteristics of the forging press under various noise interference conditions, and continuously improve its identification accuracy and reliability under interference conditions.

[0106] S4 determines the degree of overlap with the forging press with suspected abnormal fatigue state based on the updated processing result of the optimized control target, and determines the monitoring and analysis method of fatigue state of the forging press by combining the matching degree between the optimized control target and the monitoring and processing method of the acoustic signal of the forging press.

[0107] Furthermore, the method for determining the fatigue state monitoring and analysis method of the forging press is as follows: In this example, the reliability of updating the monitoring and processing data of the optimized control target and the monitoring and analysis of the forging press with suspected fatigue anomalies are determined by optimizing the degree of overlap between the optimized control target and the forging press with suspected fatigue anomalies, as well as the degree of matching between the optimized control target and the monitoring and processing method of the acoustic signal of the forging press. Based on the reliability of updating the monitoring and processing data of the optimized control target and the monitoring and analysis of the forging press with suspected fatigue anomalies, the update processing requirements of the optimized control target are determined. Based on the update processing requirements of the optimized control target, the monitoring and analysis method of the fatigue state of the forging press is determined, thereby further improving the reliability of fatigue state assessment and analysis.

[0108] The core objective of this embodiment is to determine the final fatigue state monitoring and analysis method for forging presses based on the degree of overlap between the optimized control target update processing results and the fatigue risk forging presses, as well as the degree of matching between the optimized control targets and the acoustic signal monitoring and processing methods, so as to further improve the reliability of fatigue state assessment and analysis processing. Its core logic is: by analyzing the proportion of fatigue risk forging presses in the optimized control targets, it determines whether the fatigue risk forging presses are adequately monitored and covered; and by combining the fatigue state assessment and analysis process distribution data of the optimized control targets, it determines the triggering conditions for monitoring and analysis of non-optimized control target forging presses, thus achieving the optimal configuration of the monitoring and analysis methods for all forging presses. The overall logic follows the process of "judging fatigue risk coverage → judging the number of non-target risk forging presses → analyzing the distribution of the assessment process → determining the total number of control targets → outputting the monitoring and analysis method".

[0109] S41 determines the degree of overlap with the forging press with suspected abnormal fatigue state based on the updated processing result of the optimized control target, and determines the optimized control target in the fatigue risk forging press. The degree of overlap refers to the proportion of the number of intersections between the set of optimized control targets and the set of fatigue risk forging presses to the total number of fatigue risk forging presses. The higher the degree of overlap, the greater the proportion of fatigue risk forging presses included in the optimized control targets, and the more comprehensive the monitoring coverage of fatigue risk. The optimized control targets in the fatigue risk forging presses refer to the set of forging presses that simultaneously meet the fatigue risk forging press conditions and the optimized control target conditions.

[0110] Assuming there are 7 fatigue risk forging presses in total, and several other presses after screening for optimal control objectives, some of these presses are both fatigue risk forging presses and optimal control objectives. These presses are the optimal control objectives among the fatigue risk forging presses, and their proportion to the total number of fatigue risk forging presses is the degree of overlap.

[0111] This step quantifies the adequacy of monitoring coverage for high-fatigue-risk equipment by the degree of overlap, and its significance lies in ensuring that the equipment with the most severe fatigue conditions can be reliably monitored.

[0112] Specifically, the above steps include the following: S411 determines whether the proportion of the number of optimized control targets in the fatigue risk forging press is less than the preset control target proportion threshold. If so, the fatigue risk forging press is not reliably monitored. Therefore, in order to improve the update processing efficiency of the optimized control targets, the monitoring and analysis method of the fatigue state of the forging press is determined. That is, the monitoring and analysis method of the fatigue state of the forging press excluding the optimized targets is as follows: if the monitoring deviation risk coefficient in the most recent preset period is less than the preset risk coefficient threshold, or the number of fatigue state evaluation and analysis processes in the most recent preset period is less than the preset evaluation process number threshold, then the fatigue state of the forging press will be analyzed and processed according to the preset time period and the acoustic signal monitoring and processing method in the future preset time period. If not, proceed to step S412. The preset control target proportion threshold refers to the critical value for judging whether the coverage ratio of optimized control targets in fatigue risk forging presses is sufficient. When the proportion is lower than the threshold, it indicates that a large number of fatigue risk forging presses are not included in the optimized control targets, and an active triggering strategy needs to be adopted for other forging presses to improve the update efficiency of optimized control targets. The preset time period refers to the backtracking time window used to evaluate the most recent monitoring status. The preset evaluation process number threshold refers to the critical number of times to judge whether the recent fatigue status evaluation and analysis of a forging press is sufficient. The preset duration refers to the continuous execution time window after triggering active monitoring and analysis.

[0113] Assuming there are 7 fatigue risk forging presses, only 2 of which are included in the optimization control target, the proportion = 2 ÷ 7 ≈ 0.286. If the preset control target proportion threshold is 0.40, then 0.286 < 0.40, triggering the S411 "if" path. For the forging presses other than the optimization control target, configure the following: if the monitoring deviation risk coefficient is less than the preset risk coefficient threshold or the number of evaluation processes is less than the preset evaluation process number threshold in the most recent preset period, then fatigue state analysis will be performed periodically in the future preset period.

[0114] In this case, the significance of proactively increasing the monitoring frequency of forging presses with non-optimal control targets lies in improving the update efficiency of optimal control targets by monitoring other forging presses through proactive triggering strategies when fatigue risk coverage is insufficient.

[0115] S412 determines whether the number of fatigue risk forging presses that do not belong to the optimization control target is greater than the preset threshold for the number of risk forging presses. If yes, proceed to step S42. If no, determine the monitoring and analysis method for the fatigue state of the forging press. That is, the monitoring and analysis method for the fatigue state of forging presses other than the optimization target is that as long as the monitoring deviation risk coefficient in the most recent preset time period is less than the preset risk coefficient threshold, the fatigue state of the forging press will be analyzed and processed in the future preset time period according to the preset time cycle and the acoustic signal monitoring and processing method.

[0116] The preset threshold for the number of risk forging presses refers to the critical value for determining whether the number of uncovered fatigue risk forging presses is sufficient to enter a more complex analysis process. When the number of uncovered fatigue risk forging presses is small, a relatively simplified trigger condition (judged solely by the monitoring deviation risk coefficient) can be used to determine the monitoring and analysis method.

[0117] Assuming that the number of forging presses that are not part of the optimal control target is 2, which is less than the preset threshold of 3 for the number of risk forging presses, then the forging presses that are not part of the optimal control target will be directly configured: if the monitoring deviation risk coefficient is less than the preset risk coefficient threshold in the most recent preset period, then fatigue state analysis will be performed periodically in the future preset period, and the process will not enter S42.

[0118] This step determines whether to proceed to a more refined analysis process based on the number of uncovered fatigue risk forging presses. Its significance lies in adopting a simplification strategy for small-scale uncovered cases, reducing system computational overhead, and improving system operating efficiency while ensuring basic monitoring coverage.

[0119] S42 uses the degree of matching between the optimized control target and the acoustic signal monitoring and processing method of the forging press to determine the distribution data of the fatigue state evaluation and analysis process of the optimized control target; The matching degree refers to the consistency between the frequency of the actual acoustic signal monitoring triggered by the optimized control target and the trigger frequency of its configured monitoring and processing method. The higher the matching degree, the more complete the monitoring data of the optimized control target. The distribution data of the fatigue state assessment and analysis process refers to the time distribution characteristics between each fatigue state assessment and analysis process in the history of the optimized control target. The core indicator is the interval between adjacent assessment and analysis processes.

[0120] Suppose that a certain optimization control target undergoes fatigue state analysis according to a preset time period (30 days). The timestamp sequence of several actual evaluation and analysis processes is recorded. The interval between two adjacent evaluation processes is calculated to form the interval duration sequence and mean of the optimization control target.

[0121] This step quantifies the actual monitoring data accumulation status of the optimized control targets in the form of distributed data. Its significance lies in identifying which optimized control targets have insufficient monitoring coverage frequency (the average interval duration is too small, indicating frequent triggering), providing a basis for further screening of efficient data sources.

[0122] Furthermore, the above steps include the following: Based on the distribution data of the fatigue state evaluation and analysis process of the optimized control target, the average interval time between different adjacent evaluation and analysis processes is determined. It is then determined whether there is an optimized control target whose average interval time is less than a preset interval time threshold. If so, proceed to step S43. If not, determine the monitoring and analysis method for the fatigue state of the forging press. That is, the monitoring and analysis method for the fatigue state of the forging press other than the optimized target is as follows: if the monitoring deviation risk coefficient in the most recent preset time period is less than the preset risk coefficient threshold, or if the number of fatigue state evaluation and analysis processes in the most recent preset time period is less than the preset evaluation process number threshold, then the fatigue state of the forging press will be analyzed and processed in the future preset time period according to the preset time cycle and the acoustic signal monitoring and processing method. The preset interval duration threshold refers to the critical duration for judging whether the interval between adjacent evaluation and analysis processes is too short (i.e., whether the monitoring trigger is too frequent). If the average interval duration is less than the threshold, it indicates that the monitoring frequency of the optimized control target is too high, which may result in repeated data collection and wasted resources. When there is no such optimized control target, it indicates that the monitoring frequency of all optimized control targets is reasonable, and the standard two-condition triggering strategy can be used for non-optimized control targets.

[0123] Assuming the preset interval time threshold is 2 hours, and the average interval time of each optimized control target is more than 2 hours, then there is no optimized control target with an average interval time of less than 2 hours. For non-optimized control targets, the forging press is configured as follows: when the monitoring deviation risk coefficient is less than the preset risk coefficient threshold or the number of evaluation processes is less than the preset evaluation process number threshold in the most recent preset time period, fatigue state analysis is performed periodically in the future preset time period.

[0124] S43 uses the distribution data of the optimization control objectives in the fatigue risk forging press and the fatigue state evaluation and analysis process of different optimization control objectives to determine the monitoring and analysis method of the fatigue state of the forging press.

[0125] The selected control target refers to the optimized control target that meets the dual conditions of "optimized control target in fatigue risk forging press" and "average interval duration less than preset interval duration threshold", that is, a forging press that simultaneously has the characteristics of high fatigue risk and high frequency monitoring; the total number of control targets refers to the sum of the number of optimized control targets in fatigue risk forging press and the number of selected control targets, which reflects the current comprehensive monitoring coverage capability for high-risk forging press.

[0126] Assuming there are 3 optimized control targets in the fatigue risk forging press, and 2 of them have an average interval duration less than the preset interval duration threshold (overlapping with the screened control targets), the total number of control targets = 3 + 2 = 5 (if there is overlap, deduplication is performed; assuming no overlap, then it is 5). If the total number of control targets is 5, which is less than the preset control target number threshold of 6, then the forging press is configured with non-optimized control targets: when the monitoring deviation risk coefficient is less than the preset risk coefficient threshold or the number of evaluation processes is less than the preset evaluation process number threshold in the most recent preset period, fatigue state analysis will be performed periodically in the future preset period. If the total number of control targets is ≥ 6, then the forging press is configured with more stringent triggering conditions: when the monitoring deviation risk coefficient is less than the preset risk coefficient threshold in the most recent preset period or there is no fatigue state evaluation analysis process in the most recent preset period, fatigue state analysis will be performed periodically in the future preset period.

[0127] This step determines the final monitoring and analysis triggering strategy by controlling the total number of targets. Its significance lies in the fact that when the comprehensive monitoring coverage of high-risk forging presses is strong (total number of control targets ≥ threshold), stricter triggering conditions are adopted to avoid resource waste caused by over-monitoring; when the coverage is relatively insufficient (total number of control targets < threshold), a more lenient triggering strategy is adopted, which includes a fallback condition for the number of evaluation processes, to ensure the efficiency of updating the optimized control targets and the efficiency of updating the training data.

[0128] Example 2 In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described method for online monitoring of fatigue state of a forging press based on acoustic signals when running the computer program.

[0129] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0130] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0131] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for online monitoring of fatigue state of a forging press based on acoustic signals, characterized in that, Specifically, it includes: The method for monitoring and processing the acoustic signals of the forging press is determined based on the data of the forging press and the degree of correlation between the forging press and the forging data of different forging presses. Based on the monitoring and processing method, acoustic feature signals of different forging presses are extracted and processed. Based on the extraction results of the acoustic feature signals, fatigue state analysis of different forging presses is performed. Based on the variation of the analysis and processing results of fatigue state of different forging presses, the degree of abnormality of the analysis and processing results of fatigue state of different forging presses is determined. When it is determined that the monitoring and processing method needs to be optimized based on the degree of abnormality, the next step is performed. Based on the analysis and processing results of fatigue states of different forging presses and the degree of abnormality of the corresponding analysis and processing results, the method for updating the optimization control target in the forging press is determined. Based on the updated processing results of the optimized control target, the degree of overlap with the forging press with suspected abnormal fatigue state is determined. In combination with the matching degree between the optimized control target and the monitoring and processing method of the acoustic signal of the forging press, the monitoring and analysis method of the fatigue state of the forging press is determined.

2. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 1, characterized in that, The forging press data includes the number of forging presses.

3. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 1, characterized in that, The correlation between the forging data of the forging press and different forging presses is determined based on the similarity of the number of forging operations between the forging press and different forging presses.

4. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 1, characterized in that, The method for determining the acoustic signal monitoring and processing method of the forging press is as follows: Based on the forging press data, determine the number of forging presses; Based on the correlation between the forging data of different forging presses, the similarity of the number of forging operations between different forging presses is determined, and the forging presses are divided into different forging press groups based on the similarity. Based on the number of forging presses and the number of forging presses in the forging press group, a method for monitoring and processing the acoustic signals of the forging presses is determined.

5. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 4, characterized in that, The forging presses are divided into different forging press groups, specifically including: Forging presses that fall within the same forging cycle range will be grouped into the same forging press group.

6. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 4, characterized in that, Based on the number of forging presses and the number of forging presses in the forging press group, a method for monitoring and processing the acoustic signals of the forging presses is determined, specifically including: If the number of forging presses is greater than a preset threshold for the number of forging presses, the method for monitoring and processing the acoustic signals of all forging presses is to perform acoustic signal monitoring and processing when the amplitude characteristic value of the noise signal is greater than a preset threshold.

7. The online monitoring method for fatigue state of forging presses based on acoustic signals as described in claim 1, characterized in that, The fatigue state analysis and processing of different forging presses are carried out, specifically including: The extracted acoustic feature signal of the forging press after denoising is output to the fatigue state analysis model, and the fatigue state value of the forging press is obtained based on the output of the fatigue state analysis model.

8. The online monitoring method for fatigue state of forging press based on acoustic signals as described in claim 1, characterized in that, The method for determining the fatigue state monitoring and analysis method of the forging press is as follows: Based on the updated processing results of the optimized control objective, the degree of overlap with the forging press with suspected abnormal fatigue state is determined, and the optimized control objective in the fatigue risk forging press is determined. By utilizing the degree of matching between the optimized control target and the acoustic signal monitoring and processing method of the forging press, the distribution data of the fatigue state evaluation and analysis process of the optimized control target is determined; Based on the distribution data of the optimization control objectives in the fatigue risk forging press and the fatigue state evaluation and analysis process of different optimization control objectives, a monitoring and analysis method for the fatigue state of the forging press is determined.

9. The online monitoring method for fatigue state of forging press based on acoustic signals as described in claim 8, characterized in that, When the proportion of the number of optimized control targets in the fatigue risk forging press is less than the preset control target proportion threshold, the monitoring and analysis method for the fatigue state of the forging press is determined. That is, the monitoring and analysis method for the fatigue state of the forging press excluding optimized targets is as follows: if the monitoring deviation risk coefficient in the most recent preset time period is less than the preset risk coefficient threshold, or if the number of fatigue state evaluation and analysis processes in the most recent preset time period is less than the preset evaluation process number threshold, then the fatigue state of the forging press will be analyzed and processed in the future preset time period according to the preset time cycle and the acoustic signal monitoring and processing method.

10. A computer system, comprising: A memory and processor connected by communication, and a computer program stored in the memory and capable of running on the processor, characterized in that, when the processor runs the computer program, it executes the online monitoring method for fatigue state of a forging press based on acoustic signals as described in any one of claims 1-9.

Images