Big data analysis-combined forging and pressing operation state intelligent evaluation method and equipment

Abstract

The invention provides a forging and pressing operation state intelligent evaluation method and equipment combined with big data analysis, and relates to the technical field of forging and pressing evaluation, the method comprises the following steps: carrying out chain type carding on forging and pressing operation schemes of a forging press, establishing a forging and pressing operation chain, and carrying out multi-dimensional reference state deduction; acquiring a current forging and pressing monitoring flow in real time; according to the forging reference accompanying space, node association anomaly analysis and evaluation are carried out on the current forging monitoring flow; according to the current forging and pressing node, calling a fault causal ABF framework to carry out multi-element forging and pressing fault causal prediction; and performing node conduction forging and pressing fault tracing compensation on the current forging and pressing fault causal graph network according to the forging and pressing operation chain, and performing feedback adjustment on the forging and pressing operation chain according to the forging and pressing fault causal coupling graph network. The technical problem that in the prior art, the forging and pressing fault compensation is not accurate and the forging and pressing operation efficiency is further affected due to the fact that fault diagnosis is difficult to locate a root is solved, and the forging and pressing operation efficiency is improved through the forging and pressing fault tracing compensation.

Description

Technical Field

[0001] This application relates to the field of forging evaluation technology, and in particular to a method and equipment for intelligent evaluation of forging operation status combined with big data analysis. Background Technology

[0002] In traditional forging operations, equipment fault diagnosis usually relies on single-point monitoring and static threshold alarms. This fragmented monitoring method can only detect surface anomalies and cannot reveal the dynamic coupling relationship between parameters or the causal transmission path of the fault. This results in poor alarm interpretability, maintenance decisions are highly dependent on human experience, and fault compensation often only addresses surface phenomena without fundamentally solving the problem. This leads to equipment failure recurrence or production interruption, which seriously affects the continuity and efficiency of forging operations.

[0003] In summary, existing technologies suffer from the technical problem that fault diagnosis makes it difficult to pinpoint the root cause, leading to inaccurate fault compensation and further affecting the efficiency of forging operations. Summary of the Invention

[0004] The purpose of this application is to provide a method and equipment for intelligent evaluation of forging operation status that combines big data analysis, in order to solve the technical problem in the prior art that the root cause of fault diagnosis is difficult to locate, resulting in inaccurate fault compensation and further affecting the efficiency of forging operation.

[0005] In view of the above problems, this application provides a method and equipment for intelligent evaluation of forging operation status by combining big data analysis.

[0006] Firstly, this application provides an intelligent evaluation method for forging operation status combining big data analysis. This method is implemented using an intelligent evaluation device for forging operation status combining big data analysis. The method includes: chain-like analysis of the forging operation plan of the forging press to establish a forging operation chain; performing multi-dimensional baseline state deduction of the forging press based on the forging operation chain to establish a forging baseline accompanying space; and, when controlling the forging press to perform forging operations according to the forging operation chain, acquiring the current forging status in real time. The current forging monitoring flow corresponding to the pressing node is analyzed and evaluated based on the forging reference accompanying space to establish a current forging anomaly coupling matrix. Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging anomaly coupling matrix to establish a current forging fault causal graph network. Based on the forging operation chain, node-transmitted forging fault tracing compensation is performed on the current forging fault causal graph network to obtain a forging fault causal coupling graph network, and feedback adjustment is performed on the forging operation chain based on the forging fault causal coupling graph network.

[0007] Optionally, the forging operation plan is divided into nodes to determine the forging preheating plan, the forging impact plan, and the forging cooling plan; the forging preheating plan, the forging impact plan, and the forging cooling plan are chained together to generate the forging operation chain.

[0008] Optionally, a multi-dimensional reference state deduction is performed on the forging press according to the forging preheating scheme to generate a preheating reference accompanying space; a multi-dimensional reference state deduction is performed on the forging press according to the forging impact scheme to generate an impact reference accompanying space; a multi-dimensional reference state deduction is performed on the forging press according to the forging cooling scheme to generate a cooling reference accompanying space; and the forging reference accompanying space is generated by chaining the preheating reference accompanying space, the impact reference accompanying space, and the cooling reference accompanying space.

[0009] Optionally, interconnecting equipment of the same specification and model according to the forging press to obtain a cluster of forging press equipment; performing a large-scale data retrieval of normal equipment status samples on the cluster of forging press equipment according to the forging preheating scheme to obtain a preheating node equipment status registration set; performing central tendency analysis on the preheating node equipment status registration set to construct a preheating node equipment status benchmark matrix; performing a large-scale data retrieval and central tendency analysis of normal environmental status samples on the cluster of forging press equipment according to the forging preheating scheme to construct a preheating node environmental status benchmark matrix; performing a large-scale data retrieval and central tendency analysis of normal product status samples on the cluster of forging press equipment according to the forging preheating scheme to construct a preheating node product status benchmark matrix; integrating the preheating node equipment status benchmark matrix, the preheating node environmental status benchmark matrix, and the preheating node product status benchmark matrix to obtain the preheating benchmark accompanying space.

[0010] Optionally, the node association mapping activation of the forging reference accompanying space is performed according to the current forging node to obtain the current reference accompanying space, which includes the current equipment state reference matrix, the current environment state reference matrix, and the current product state reference matrix; multi-characteristic identification and cleaning is performed according to the current forging monitoring flow to establish the current equipment state monitoring matrix, the current environment state monitoring matrix, and the current product state monitoring matrix; anomaly identification and evaluation is performed on the current equipment state monitoring matrix according to the current equipment state reference matrix to obtain a first anomaly identification and evaluation matrix; anomaly identification and evaluation is performed on the current environment state monitoring matrix according to the current environment state reference matrix to obtain a second anomaly identification and evaluation matrix; anomaly identification and evaluation is performed on the current product state monitoring matrix according to the current product state reference matrix to obtain a third anomaly identification and evaluation matrix; anomaly attention allocation and aggregation are performed based on the first anomaly identification and evaluation matrix, the second anomaly identification and evaluation matrix, and the third anomaly identification and evaluation matrix to generate the current forging anomaly coupling matrix.

[0011] Optionally, based on the current forging node, fault type mining is performed on the forging press to obtain a current forging fault type matrix, which includes F forging fault type labels, where F is a positive integer greater than 1; based on the current forging fault type matrix, fault causal event retrieval is performed to obtain F forging fault causal event regions; based on the fault causal ABF architecture, multi-level causal reasoning learning is performed on the F forging fault causal event regions to establish F forging fault causal prediction branches; the current forging anomaly coupling matrix is ​​input into the F forging fault causal prediction branches to obtain F forging fault causal prediction paths; based on the F forging fault causal prediction paths, graph-networking is performed to generate the current forging fault causal graph network.

[0012] Optionally, the f-th forging failure causal event region is extracted based on the F forging failure causal event regions, where f is a positive integer and 1 ≤ f ≤ F; the fault causal ABF architecture is activated, which includes an ARIMA model, a BBN model, an FTA model, and a central server; the ARIMA model, BBN model, and FTA model are iteratively supervised and trained based on the f-th forging failure causal event region to establish a first forging failure causal prediction model, a second forging failure causal prediction model, and a third forging failure causal prediction model; loss characteristics are analyzed for the first, second, and third forging failure causal prediction models to establish a fault causal prediction mentor-student relationship; real-time monitoring logs of the central server are obtained, and security and privacy protection configurations are enhanced for the central server based on the real-time monitoring logs to establish a central trusted node; based on the fault causal prediction mentor-student relationship, the first, second, and third forging failure causal prediction models are distilled and strengthened based on the central trusted node to generate the f-th forging failure causal prediction branch.

[0013] Optionally, based on the current forging node, the node transmission characteristics of the forging operation chain are analyzed to determine the forging scheme of the transmission node; based on each forging fault causal prediction path in the current forging fault causal graph network, forging fault prediction is performed on the forging scheme of the transmission node to obtain F predicted transmission forging faults; based on the current forging fault causal graph network and the forging scheme of the transmission node, the accident triggering factors of the F predicted transmission forging faults are traced to obtain F transmission forging fault paths; based on the F transmission forging fault paths, the transmission forging fault characteristics of the current forging fault causal graph network are compensated to generate the forging fault causal coupling graph network.

[0014] Optionally, a forging fault alarm is generated based on the forging fault causal coupling graph.

[0015] Secondly, this application also provides an intelligent evaluation device for forging operation status combining big data analysis, used to execute the intelligent evaluation method for forging operation status combining big data analysis as described in the first aspect. The intelligent evaluation device for forging operation status combining big data analysis includes: a chain-based sorting module, used to sort the forging operation plan of the forging press in a chain, establish a forging operation chain, and perform multi-dimensional benchmark state deduction of the forging press based on the forging operation chain to establish a forging benchmark accompanying space; and a forging monitoring flow acquisition module, used to acquire the current forging monitoring flow corresponding to the current forging node in real time when controlling the forging press to perform forging operation according to the forging operation chain. The associated anomaly analysis and evaluation module is used to perform node-related anomaly analysis and evaluation on the current forging monitoring flow based on the forging reference accompanying space, and establish the current forging anomaly coupling matrix; the fault causality prediction module is used to perform multivariate forging fault causality prediction on the current forging anomaly coupling matrix based on the current forging node by calling the fault causality ABF architecture, and establish the current forging fault causality graph network; the fault tracing and compensation module is used to perform node-transmission forging fault tracing compensation on the current forging fault causality graph network based on the forging operation chain, obtain the forging fault causality coupling graph network, and perform feedback adjustment on the forging operation chain based on the forging fault causality coupling graph network.

[0016] One or more technical solutions provided in this application have at least the following beneficial effects:

[0017] By chain-like analysis of the forging operation scheme of the forging press, a forging operation chain is established. Based on the forging operation chain, a multi-dimensional baseline state deduction of the forging press is performed to establish a forging baseline accompanying space. When the forging press is controlled to perform forging operations according to the forging operation chain, the current forging monitoring flow corresponding to the current forging node is acquired in real time. Based on the forging baseline accompanying space, the node association anomaly analysis and evaluation of the current forging monitoring flow is performed to establish a current forging anomaly coupling matrix. Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging anomaly coupling matrix to establish a current forging fault causal graph network. Based on the forging operation chain, node-transmission forging fault tracing compensation is performed on the current forging fault causal graph network to obtain a forging fault causal coupling graph network, and the forging operation chain is adjusted based on the forging fault causal coupling graph network. In other words, by constructing a forging operation chain and performing multi-dimensional baseline state simulation, while acquiring the current forging monitoring flow, and through node association anomaly analysis and evaluation, a current forging anomaly coupling matrix is ​​established. The fault causal ABF architecture is then invoked to perform multivariate causal analysis, a detailed fault causal graph is established, the root cause of the fault is accurately identified, and fault tracing compensation is performed on the node transmission faults in the fault causal graph. Compensation is then performed from the root cause of the fault, improving the accuracy of forging fault compensation and thus enhancing forging operation efficiency.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the intelligent evaluation method for forging operation status based on big data analysis proposed in this application.

[0021] Figure 2 This is a schematic diagram of the intelligent evaluation device for forging operation status that combines big data analysis, as described in this application.

[0022] Figure labeling: Chain sorting module 11, forging monitoring flow acquisition module 12, correlation anomaly analysis and evaluation module 13, fault causal prediction module 14, fault tracing and compensation module 15. Detailed Implementation

[0023] This application provides an intelligent evaluation method and equipment for forging operation status that combines big data analysis. This solves the technical problem in existing technologies where difficulty in locating the root cause of faults leads to inaccurate fault compensation, further impacting forging efficiency. By constructing a forging operation chain and performing multi-dimensional baseline state simulation, while simultaneously acquiring the current forging monitoring flow, and through node-related anomaly analysis and evaluation, a current forging anomaly coupling matrix is ​​established. A fault causal analysis (ABF) architecture is invoked to perform multivariate causal analysis, creating a detailed fault causal network. This accurately identifies the root cause of the fault and provides fault tracing and compensation for node propagation in the fault causal network. Compensation is then performed at the fault root cause, improving the accuracy of forging fault compensation and thus enhancing forging operation efficiency.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides an intelligent evaluation method for forging operation status combining big data analysis. The method is executed by an intelligent evaluation device for forging operation status combining big data analysis, and specifically includes the following steps: The forging operation scheme of the forging press is sorted out in a chain, a forging operation chain is established, and a multi-dimensional benchmark state deduction of the forging press is performed based on the forging operation chain to establish a forging benchmark accompanying space.

[0026] Furthermore, this application also includes the following steps: dividing the forging operation scheme into nodes to determine the forging preheating scheme, the forging impact scheme, and the forging cooling scheme; and chaining the forging preheating scheme, the forging impact scheme, and the forging cooling scheme to generate the forging operation chain.

[0027] Specifically, the chain-like approach to forging operations ensures that each stage of the forging process is rationally arranged in terms of time and operational sequence. This chain-like approach breaks down the originally holistic production plan into an ordered sequence of steps based on its inherent physical stages and time order. The forging operation plan is divided into nodes, defining the forging preheating plan, forging impact plan, and forging cooling plan. The forging preheating plan heats the metal before forging begins, bringing it to a suitable temperature to help reduce cracks during processing and improve the metal's plasticity. The forging impact plan shapes the metal by applying impact force, typically involving rapid deformation at a specific temperature, using impact force to change the metal's shape. The forging cooling plan rapidly cools the metal after forging to ensure that the metal acquires the desired mechanical and physical properties during the cooling process.

[0028] The forging preheating, impact, and cooling processes are chained together. This means that preheating, impact, and cooling are sequentially linked into a continuous workflow chain, ensuring each stage works in tandem for a seamless forging process. For example, preheating is initiated by monitoring the die temperature using sensors T1 and T2. The preheating readiness signal is triggered when both T1 and T2 readings remain stable within the 308-312℃ range for 2 minutes. Upon receiving the preheating readiness signal, the impact is initiated: a preheated 1200℃ billet is fed in, and the first and second impacts are performed. Monitoring parameters include real-time force and displacement curves. The impact completion signal is triggered when the slider returns to the top dead center after the second impact. Upon receiving the impact completion signal, cooling is initiated: the robotic arm removes the workpiece, and a 120-second timer begins. Simultaneously, the die water cooling system is activated. The single-cycle completion signal is triggered when the workpiece cooling timer ends and the die surface temperature drops below 150℃, preparing for the next workpiece.

[0029] By using a chain-style forging operation plan, we can ensure smooth connection between each link, avoid production stagnation or repetitive operations caused by unreasonable process sequence, and thus improve overall operation efficiency.

[0030] Furthermore, this application also includes the following steps: performing multi-dimensional benchmark state deduction on the forging press according to the forging preheating scheme to generate a preheating benchmark accompanying space; performing multi-dimensional benchmark state deduction on the forging press according to the forging impact scheme to generate an impact benchmark accompanying space; performing multi-dimensional benchmark state deduction on the forging press according to the forging cooling scheme to generate a cooling benchmark accompanying space; and performing chain encapsulation based on the preheating benchmark accompanying space, the impact benchmark accompanying space, and the cooling benchmark accompanying space to generate the forging benchmark accompanying space.

[0031] Furthermore, this application also includes the following steps: interconnecting equipment of the same specification and model according to the forging press to obtain a cluster of forging press equipment; performing a large-scale data retrieval of normal equipment status samples on the cluster of forging press equipment according to the forging preheating scheme to obtain a preheating node equipment status registration set; performing central tendency analysis on the preheating node equipment status registration set to construct a preheating node equipment status benchmark matrix; performing a large-scale data retrieval and central tendency analysis of normal environmental status samples on the cluster of forging press equipment according to the forging preheating scheme to construct a preheating node environmental status benchmark matrix; performing a large-scale data retrieval and central tendency analysis of normal product status samples on the cluster of forging press equipment according to the forging preheating scheme to construct a preheating node product status benchmark matrix; integrating the preheating node equipment status benchmark matrix, the preheating node environmental status benchmark matrix, and the preheating node product status benchmark matrix to obtain the preheating benchmark accompanying space.

[0032] Specifically, forging equipment of the same specifications and model is interconnected to form a cluster of forging equipment. Based on the forging preheating scheme, historical data of the cluster of forging equipment is retrieved, extracting equipment status samples from the start to the end of preheating. This includes various parameters of the equipment under normal operating conditions, such as temperature, pressure, and vibration. For example, assuming a certain model of equipment has a temperature range of 800℃ to 1200℃, vibration data of 0.05mm / s to 0.1mm / s, and operating voltage of 380V during the preheating stage. The preheating node equipment status registration set is a set of equipment status data collected through big data retrieval of normal equipment status samples, forming the equipment status registration set for the preheating stage.

[0033] A central tendency analysis is performed on the preheating node equipment status registration set to identify its central or typical representative values ​​from a large amount of sample data. By calculating the mean, mode, or median, a standard benchmark for equipment status is constructed. This means finding the average level and normal fluctuation range of the preheating node equipment status registration set under normal conditions, such as the mean and standard deviation. The preheating node equipment status benchmark matrix is ​​a benchmark data matrix of equipment status obtained through central tendency analysis, reflecting the standard values ​​of various performance indicators of the equipment under normal operating conditions.

[0034] The analysis of environmental and product status is handled in the same way as the processing of equipment status. Based on the forging preheating scheme, a large-scale data retrieval of normal environmental status samples is performed on the forging equipment within the same cluster. This involves retrieving the environmental status from the start to the end of preheating, including environmental parameters under normal operating conditions, such as temperature and humidity. Similarly, central tendency analysis is performed on the retrieved normal environmental status samples to obtain the preheating node environmental status baseline matrix.

[0035] Similarly, based on the forging preheating scheme, a large data retrieval of normal product status samples is performed on the same cluster of forging equipment. That is, the product status is retrieved from the start to the end of the preheating period, including the normal status of different products during the preheating process, to ensure the quality and performance of the products.

[0036] By integrating the preheating node equipment state reference matrix, the preheating node environmental state reference matrix, and the preheating node product state reference matrix, a comprehensive preheating reference accompanying space is obtained, which serves as a reference standard for the preheating stage, ensuring that the forging press maintains a stable operating state during the preheating process.

[0037] The multi-dimensional baseline state deduction of the forging impact scheme and forging cooling scheme follows the same approach as the forging preheating scheme. Both methods retrieve the equipment state, environmental state, and product state under normal conditions and perform corresponding central tendency analysis to obtain the impact baseline associated space and the cooling baseline associated space. The impact baseline associated space is defined during the forging impact process, when the equipment deforms the metal through high-speed impact force. Different stages of the impact process (e.g., before, during, and after impact) have different requirements for the equipment and product state. The impact baseline associated space describes the standard range of equipment, environmental, and product states during this process. The cooling baseline associated space is the cooling stage after forging is completed and has a decisive influence on the mechanical properties of the metal. The cooling baseline associated space is used to deduce the equipment, environmental, and product states during the cooling process.

[0038] In summary, by extrapolating data from the preheating, impact, and cooling stages, preheating reference space, impact reference space, and cooling reference space are generated respectively, each containing standardized data of equipment, environment, and products at different stages.

[0039] The preheating reference space, impact reference space, and cooling reference space are chain-encapsulated, that is, the standard datasets of different stages are combined into a whole forging operation standard space, which includes standard data of equipment, environment, and product status at all stages of the entire forging process. For example, the forging of normal workpiece A must meet the following requirements: the die current and temperature meet the standards in the preheating stage; the pressure curve falls within the reference envelope in the impact stage, and the vibration does not exceed the standard; the temperature drop curve in the cooling stage conforms to the reference; and the die temperature and forging temperature transmitted between nodes are within the allowable range. Any deviation from the reference in any link or connection is considered abnormal. Through multi-dimensional reference state deduction and chain encapsulation, each stage of the forging process is standardized, ensuring that each link can operate according to the set standards.

[0040] When the forging press is controlled to perform forging operations according to the forging operation chain, the current forging monitoring stream corresponding to the current forging node is obtained in real time.

[0041] Specifically, throughout the forging process, each node in the forging operation chain has corresponding control strategies and parameters. These strategies are formulated based on specific operational requirements and executed by the automatic control system on the forging press. After each operation node is completed, it proceeds to the next node according to preset rules. Once a node is entered, the sensor group and data acquisition channel matching the preset monitoring scheme for that node are immediately activated. In other words, during the execution of each forging node, the forging press monitors the equipment status and operation process in real time through installed sensors, collecting real-time data to form the current forging monitoring stream corresponding to the current forging node. Real-time acquisition means the ability to quickly acquire various data streams from the forging press. For example, at the impact node, parameters such as impact force, vibration, and equipment pressure may be acquired in real time.

[0042] Data acquisition is not simply a collection process, but involves precise time alignment and encapsulation. Data from different physical interfaces is timestamped by a high-precision time server, ensuring that a peak value from a vibration sensor and a peak value from a pressure sensor can be accurately correlated to determine whether they occurred at the same time. Subsequently, these raw data packets, each with timestamps, node identifiers, and data values, are encapsulated in real time into a structured current forging and pressing monitoring stream.

[0043] When the workflow switches from one node to the next, the focus of data acquisition immediately shifts. For example, at the start of the impact node, sensors such as the main cylinder pressure, displacement encoder, and frame vibration accelerometer are boosted to their highest sampling frequencies, while preheating-related temperature sensors switch to low-frequency monitoring. This on-demand acquisition and dynamic focusing mode ensures the integrity of critical data while optimizing the utilization of network bandwidth and computing resources.

[0044] By acquiring real-time monitoring data streams at each stage of the forging process, real-time monitoring of the forging process can be achieved. The real-time data streams enable the forging press to automatically adjust based on the real-time monitoring information.

[0045] Based on the forging reference space, the current forging monitoring flow is analyzed and evaluated for node correlation anomalies, and a current forging anomaly coupling matrix is ​​established.

[0046] Furthermore, this application also includes the following steps: Activating the node association mapping of the forging reference accompanying space according to the current forging node to obtain the current reference accompanying space, wherein the current reference accompanying space includes the current equipment state reference matrix, the current environment state reference matrix, and the current product state reference matrix; performing multi-characteristic identification and cleaning according to the current forging monitoring flow to establish the current equipment state monitoring matrix, the current environment state monitoring matrix, and the current product state monitoring matrix; performing anomaly identification and evaluation on the current equipment state monitoring matrix according to the current equipment state reference matrix to obtain a first anomaly identification and evaluation matrix; performing anomaly identification and evaluation on the current environment state monitoring matrix according to the current environment state reference matrix to obtain a second anomaly identification and evaluation matrix; performing anomaly identification and evaluation on the current product state monitoring matrix according to the current product state reference matrix to obtain a third anomaly identification and evaluation matrix; and performing anomaly attention allocation and aggregation based on the first anomaly identification and evaluation matrix, the second anomaly identification and evaluation matrix, and the third anomaly identification and evaluation matrix to generate the current forging anomaly coupling matrix.

[0047] Specifically, based on the current forging node, data from the same node are associated with and activated in the forging reference accompanying space to obtain the current reference accompanying space, which includes the current equipment state reference matrix, the current environment state reference matrix, and the current product state reference matrix. The current reference accompanying space is a set of reference data for the current forging node, specifically deconstructed into reference matrices in three dimensions: equipment, environment, and product. It serves as a benchmark for determining whether the current state is normal.

[0048] The current forging monitoring stream is cleaned by multiple characteristics, including data alignment, invalid value removal, and noise filtering, to remove irrelevant data and noise, ensuring the accuracy and validity of the data. This results in the current equipment status monitoring matrix, the current environmental status monitoring matrix, and the current product status monitoring matrix, which represent the actual status of the forging press, environment, and product at the current operation node, covering features such as equipment temperature, pressure, vibration, product size, and surface quality.

[0049] Based on the current equipment status benchmark matrix, current environmental status benchmark matrix, and current product status benchmark matrix, anomaly identification and evaluation are performed on the current equipment status monitoring matrix, current environmental status monitoring matrix, and current product status monitoring matrix, respectively. This involves detecting deviations between the current monitoring data and standard benchmark values. For example, if the equipment temperature exceeds the predetermined normal range, or if the product dimensions deviate significantly, it will be marked as an anomaly. The first anomaly identification and evaluation matrix includes the actual value, benchmark expected value, deviation, and structured data based on statistical anomaly probability or deviation severity scores for each parameter related to the equipment status. The second identification and evaluation matrix includes the actual value, benchmark expected value, deviation, and structured data based on statistical anomaly probability or deviation severity scores for each parameter related to the environmental status. The third identification and evaluation matrix includes the actual value, benchmark expected value, deviation, and structured data based on statistical anomaly probability or deviation severity scores for each parameter related to the product status. Each matrix displays the degree of deviation from the standard value and the severity of the anomaly.

[0050] Anomaly attention is allocated based on the first, second, and third anomaly identification and evaluation matrices. This involves assigning weights to different state matrices according to the severity of each anomaly, aggregating them into a comprehensive forging anomaly coupling matrix. This matrix reflects the coupling relationship between equipment, environment, and product status, and how they collectively influence the forging process. The current forging anomaly coupling matrix specifically emphasizes the interrelated anomaly clusters most likely representing actual faults at the current node. Through anomaly identification and evaluation and the generated coupling matrix, anomalies in the forging process are accurately identified, helping operators take rapid remedial measures when anomalies occur. Timely identification and handling of anomalies avoids problems such as equipment overload or substandard product quality, thereby improving the stability and production efficiency of the forging process.

[0051] Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging anomaly coupling matrix, and a current forging fault causal graph network is established.

[0052] Furthermore, this application also includes the following steps: performing fault type mining on the forging press based on the current forging node to obtain a current forging fault type matrix, wherein the current forging fault type matrix includes F forging fault type labels, where F is a positive integer greater than 1; performing fault causal event retrieval based on the current forging fault type matrix to obtain F forging fault causal event regions; performing multi-level causal reasoning learning on the F forging fault causal event regions based on the fault causal ABF architecture to establish F forging fault causal prediction branches; inputting the current forging anomaly coupling matrix into the F forging fault causal prediction branches to obtain F forging fault causal prediction paths; and performing graph-networking based on the F forging fault causal prediction paths to generate the current forging fault causal graph network.

[0053] Furthermore, this application also includes the following steps: extracting the f-th forging failure causal event region based on the F forging failure causal event regions, where f is a positive integer, 1≤f≤F; activating the fault causal ABF architecture, which includes an ARIMA model, a BBN model, an FTA model, and a central server; performing iterative supervised training on the ARIMA model, the BBN model, and the FTA model based on the f-th forging failure causal event region to establish a first forging failure causal prediction model, a second forging failure causal prediction model, and a third forging failure causal prediction model; and performing the forging failure causal prediction model... The loss characteristics of the first model, the second model, and the third model for predicting the causal relationship of forging failures are analyzed to establish a teacher-student relationship for causal prediction of failures. Real-time monitoring logs of the central server are obtained, and security and privacy protection configurations of the central server are enhanced based on the real-time monitoring logs to establish a central trusted node. Based on the teacher-student relationship for causal prediction of failures, the first model, the second model, and the third model for predicting the causal relationship of forging failures are distilled and enhanced according to the central trusted node to generate the f-th branch for predicting the causal relationship of forging failures.

[0054] Specifically, based on the current forging node, the forging press undergoes fault type mining, identifying common fault types that may occur at this node, such as overheating, pressure runaway, excessive vibration, and temperature fluctuations. Each fault type represents a common fault mode. The forging fault type matrix is ​​a multi-dimensional matrix containing F distinct forging fault type labels. Each label represents a possible fault type. F is a positive integer greater than 1, representing different fault types. This process does not iterate through all possible faults but uses a rule engine to identify a list of faults relevant to the current forging node.

[0055] The current forging failure type matrix is ​​used for causal event retrieval, which involves searching for causal events related to different failure types. Each failure type may be caused by multiple factors, and causal event retrieval aims to identify these factors and their interrelationships to better understand the failure mechanism. For example, for bearing failures, the event area may contain reports of the past 100 bearing failures, each report containing a structured record of: grease particle size data from the months prior to failure, vibration spectrum changes from the week before failure, and the sudden temperature rise and torque drop at the moment of failure. The F forging failure causal event areas represent the complete chain of events and their statistical patterns for the F failure labels in the current forging failure type matrix, from initial causes to intermediate development processes and final manifestations.

[0056] A fault-cause (ABF) architecture is used to perform multi-level causal reasoning learning on F forging press fault causal event regions. Different models in the ABF architecture collaborate to learn each event region. When the ABF architecture is activated, multiple models are launched and connections between them are established. The input and output of each model are based on monitoring data and the characteristics of fault events. The ARIMA model processes time series data to analyze the operating trends of the forging press over a certain period. For example, it analyzes data on equipment temperature changes to predict whether the equipment temperature will exceed a critical value in the future. The BBN model evaluates the relationships between different fault types by building a Bayesian network. For example, the dependency between insufficient cooling water flow, equipment overheating, and fault occurrence. The FTA model constructs a fault tree to show how fault events hierarchically affect the stability of the entire system, helping to identify potential fault sources.

[0057] Specifically, the f-th causal event region for forging failures is randomly extracted from F causal event regions, representing the causal event data packet corresponding to a specific failure. The ARIMA, BBN, and FTA models are activated in parallel and fed with the same training set. The ARIMA model is trained to predict key monitoring parameters, such as vibration amplitude at a specific frequency, over time as the failure progresses. For example, it learns the typical curve of vibration characteristic value growth during the process of a bearing progressing from slight wear to severe spalling. The BBN model's structure is initially constructed by domain experts, but its core conditional probability table is learned through maximum likelihood estimation or Bayesian updates using training data. The FTA model's tree structure—top event, intermediate events, bottom events, and logic gates—is defined by experts. Its training focuses on validating and quantifying the prior probability of the bottom event and the confidence level of the entire chain using historical data. Iterative supervised training is performed on each model. During this process, the ARIMA, BBN, and FTA models receive data input, optimize parameters, and adjust models to improve the accuracy of failure prediction. After each model is trained, a forging failure causal prediction model will be generated. Based on historical and real-time data, this model can predict future failures and assess the risk and causes of failures. During this stage, the central server is responsible for coordinating and managing the operation of different models, ensuring the stability of data flow and processing.

[0058] Each model undergoes iterative supervised training, meaning its predictive ability is progressively optimized using a series of labeled data. The input data for ARIMA model training is time-series data, such as equipment temperature, pressure, and vibration. The data should include historical records indicating whether equipment malfunctions have occurred. Model fitting is performed based on historical data, estimating the autoregressive coefficients (AR), difference order (I), and moving average coefficients (MA). Predictions are made, and the differences between the predicted and actual results are compared to adjust the model parameters. The output is a prediction of the equipment's state at a future time, such as temperature or pressure. The input data for BBN model training consists of multiple variables, such as equipment state, environmental conditions, and operating modes. Each variable needs to be represented by a probability distribution, such as the probability of equipment overheating. A Bayesian network is built based on historical data, defining the relationships between nodes and edges. Through supervised learning, the network structure and conditional probability table are continuously optimized to improve the accuracy of inference. Based on the inference results, the output is the probability of equipment overheating, the likelihood of insufficient cooling water flow, etc. The input data for FTA model training consists of various events related to equipment failure and their probabilities, such as cooling system failure and equipment overload. Construct a fault tree, defining each fault event and its hierarchical relationship; calculate the probability of occurrence of each event based on historical data, and infer the most likely root cause of the fault. The output is a generated fault tree diagram, showing different fault events and their causal relationships.

[0059] During training, the output of each model serves as the input to the next model, forming a closed loop. For example, the ARIMA model predicts the future temperature of the device. If the predicted temperature exceeds a threshold, the output is passed to the BBN model to assess the likelihood of overheating. Based on the probability of overheating, the BBN model outputs the conditions for failure, which are then passed to the FTA model to help analyze whether overheating will lead to other system failures, such as cooling system failure.

[0060] After each model completes training, three fault causal prediction models are established based on the results of ARIMA, BBN, and FTA: The first forging fault causal prediction model is a device status prediction model generated based on ARIMA prediction results, determining whether a fault will occur. The second forging fault causal prediction model is based on BBN inference results, predicting the probability of faults such as overheating and insufficient cooling. The third forging fault causal prediction model is based on FTA analysis results, predicting the root cause of device faults. The three models, after iterative training, will generate specific fault prediction paths and output them to a central server. The central server is responsible for centrally storing, managing, and analyzing real-time monitoring data from different devices and sensors. Based on these results, it will make decisions and issue alarms or control commands for fault prevention or repair.

[0061] Loss characteristics of the three causal prediction models for forging failures (model 1, model 2, and model 3) were analyzed. This involved evaluating the performance of each model using independent validation sets and analyzing their respective prediction losses, i.e., errors. Models with superior performance and smaller errors were designated as teacher models, while those with slightly lower performance were considered student models. A teacher-student relationship for failure causal prediction was established by comparing the performance of different models. The prediction results provided by the teacher models will be used to guide the learning of the student models, thereby improving their performance.

[0062] Real-time monitoring logs from the central server are acquired, recording device status, environmental conditions, and operational data. After analyzing these logs, the security configuration is dynamically enhanced. Once it becomes a central trusted node, it oversees the distillation process. The softened probability outputs of the teacher model to a series of inputs, as a richer form of knowledge, are used to guide the retraining of the student model. The student model not only learns the original correct labels but also learns to imitate the teacher's more nuanced judgments. Through distillation, the performance of the student model is improved, and the three models become more consistent in their judgments. The central trusted node is a trusted core node established after the central server has undergone security hardening. Based on this real-time data, the central server's security and privacy protection configurations are enhanced to ensure data security and integrity. By strengthening security protection, risks such as data leakage and tampering are avoided.

[0063] Based on the teacher-student relationship in fault causal prediction, distillation enhancement is applied to the first, second, and third forging fault causal prediction models using a central trusted node. This distillation enhancement technique transfers knowledge from the teacher model to the student model, allowing them to learn from the teacher model's predictive capabilities and compensate for its original shortcomings. Distillation enhancement improves the prediction accuracy of the student model, ultimately generating a more accurate f-th forging fault causal prediction branch. Each branch represents a prediction path for different types of faults and helps identify potential fault sources and influencing paths.

[0064] Similarly, for the F-1 forging fault causal event regions (excluding the f-th forging fault causal event region), the above process is repeated to obtain the corresponding forging fault causal prediction branches, thus obtaining F forging fault causal prediction branches. The current forging anomaly coupling matrix is ​​used as input to these F forging fault causal prediction branches to obtain F forging fault causal prediction paths, which are all possible fault occurrence paths calculated using the fault causal ABF architecture.

[0065] Based on F causal prediction paths for forging failures, a graph network is constructed. Each node in the network represents a failure event, and each edge represents the causal relationship between events, forming the current causal graph network for forging failures. This network clearly defines the propagation paths and possible interactions of each failure event. For all F suspected failures, the current causal graph network merges, removes duplicates, and connects the F prediction paths based on their shared event nodes and logical relationships, forming a weighted, comprehensive causal relationship network containing multiple parallel or intersecting links. This reveals the competing failure hypotheses and their complete evolutionary logic behind all current anomalies.

[0066] By using the current forging fault causal network, operators can quickly identify potential fault chains and take corresponding preventative or remedial measures in advance. By establishing a forging fault type matrix, the fault types of the current equipment can be accurately identified, and their probability of occurrence can be predicted. Based on the multi-level causal reasoning of the fault causal (ABF) architecture, the potential causes and consequences of faults can be deeply explored, improving the accuracy of fault prediction and reducing false alarms and missed alarms. The generation of the fault causal network provides a clear view of the relationships between fault events and their propagation paths, helping operators quickly locate the fault source and take effective remedial measures.

[0067] Based on the forging operation chain, the current forging fault causal network is used to perform node propagation forging fault tracing compensation, thereby obtaining a forging fault causal coupling network, and the forging operation chain is adjusted based on the forging fault causal coupling network.

[0068] Furthermore, this application also includes the following steps: analyzing the node transmission characteristics of the forging operation chain based on the current forging node to determine the forging scheme of the transmission node; predicting forging faults for the forging scheme of the transmission node based on each forging fault causal prediction path in the current forging fault causal graph network to obtain F predicted transmission forging faults; tracing the accident triggering factors of the F predicted transmission forging faults based on the current forging fault causal graph network and the forging scheme of the transmission node to obtain F transmission forging fault paths; and compensating for transmission forging fault characteristics of the current forging fault causal graph network based on the F transmission forging fault paths to generate the forging fault causal coupling graph network.

[0069] Specifically, in the forging process chain, each node has unique operational characteristics, and these nodes have varying impacts on fault propagation. By analyzing the node transmission characteristics of the forging process chain based on the current forging node, the propagation path of the fault from one node to the next is identified. First, the state of the current forging node is identified, and how the fault characteristics of that node affect the normal operation of subsequent nodes is analyzed. The forging scheme for the transmission node identifies subsequent process nodes that may be affected by the current fault and their corresponding standard operating procedures.

[0070] For each path in the current causal network of forging failures, predict its impact on the transmission node scheme. Analyze the predicted paths to predict potential failures at the transmission nodes under the current scheme. Each path represents a propagation process from the fault source to the final impact. Treat each diagnosed fault (i.e., each predicted path in the causal network) as a cause, and combine this with the normal requirements of the forging scheme at the transmission nodes to infer new anomalies or fault types that this fault may cause in subsequent nodes, resulting in F predicted transmission forging failures.

[0071] Based on the current forging failure causal network and the forging scheme of transmission nodes, the triggering factors of F predicted transmitted forging failures are traced to identify the key factors leading to the failures. It is not only necessary to predict the impact, but also to identify the key trigger points leading to these transmitted failures. Simultaneously, it is necessary to examine whether there are factors in the transmission node scheme that exacerbate the problem, and to connect the current failure with the transmitted failure using a clear causal chain to form the transmitted forging failure path.

[0072] Based on F propagating forging fault paths, the current forging fault causal network is compensated for its propagating forging fault characteristics. This involves integrating the original current forging fault causal network with the F propagating forging fault paths, merging identical nodes, and connecting causal chains to form a coupled causal network for forging faults. Propagating forging fault characteristic compensation adds new nodes and edges to the current forging fault causal network to represent the propagation and impact range of the fault, thus forming a more comprehensive coupled causal network for forging faults that shows both the root cause of the current fault and its potential chain reactions.

[0073] For example, the current forging fault cause-effect graph shows the main fault as follows: partial blockage of the mold cooling water pipe, decreased cooling efficiency on the left side of the upper mold, leading to a temperature increase on the left side of the upper mold to 320℃ (reference temperature 310±5℃), resulting in increased flash on the left side during forging and a thinner workpiece. The fault characteristics are localized abnormal temperature in the mold, exhibiting both persistence and spatial locality. The proposed transmission node schemes are: 1. Next workpiece impact scheme, with a standard mold temperature of 310±5℃; 2. Mold life management scheme, requiring uniform temperature to prevent thermal stress cracking. The transmission prediction for the next workpiece impact scheme is as follows: the temperature on the left side of the mold remains excessively high; the metal flow on the left side is too good during the forging of the next workpiece; further increase in flash on the left side may lead to excessive metal filling of the cavity corners, forming creases. Triggering factor tracing: the current temperature deviation ΔT = +10℃, while the critical temperature deviation for filling the cavity corners of the next workpiece is ΔT_c = +8℃. Therefore, the current fault is the direct factor triggering the crease defect in the next workpiece. Predicted transmission faults: The temperature on the left side of the mold remains consistently high; uneven thermal expansion between the left and right halves of the mold; additional thermal stress is generated during mold closing; this accelerates the initiation of fatigue cracks at the parting surface. Triggering factor tracing: The thermal fatigue strength limit of the mold material corresponds to a temperature difference ΔT_max = 15℃. Currently, ΔT = 10℃ is close to the limit; continued operation will significantly shorten the mold's lifespan. Root cause chain: Blockage in the cooling water pipe causes the temperature on the left side of the upper mold to rise to 320℃, resulting in large flash and thinness on the left side of the current workpiece. Transmission chain 1: A sustained temperature of 320℃ on the left side of the upper mold leads to excessive metal fluidity on the left side of the next workpiece, potentially causing crease defects (65% probability). Transmission chain 2: A sustained temperature of 320℃ on the left side of the upper mold results in a 10℃ temperature difference between the left and right sides of the mold, creating thermal stress cycles and increasing the risk of cracks at the mold parting surface, with a projected lifespan reduction of 40%.

[0074] By analyzing each transmission node in detail, potential faults in the forging process can be accurately predicted, and corresponding countermeasures can be taken in a timely manner. Based on the characteristics of the transmission nodes and the fault prediction path, more targeted forging solutions can be generated, which can not only effectively mitigate the spread of faults when they occur, but also ensure the continuity of forging operations and production efficiency. By tracing the accident triggering factors, the root cause of the fault can be identified, thereby solving the problem at its source, rather than just dealing with the surface fault.

[0075] Furthermore, this application also includes the following step: generating a forging fault alarm based on the forging fault causal coupling diagram network.

[0076] Specifically, the forging fault causal coupling graph integrates fault information from multiple aspects, including equipment failure, environmental anomalies, and product status. By monitoring each node in the forging operation in real time, it quickly analyzes the fault propagation path in the graph and identifies possible fault development trends. First, it monitors the operating status of each node in the forging press and updates the status of each node in the causal graph in real time. If an anomaly occurs at a node, it automatically tracks the fault propagation path of that node and determines whether it will trigger failures in subsequent nodes.

[0077] Based on the generated fault causal coupling graph, different alarm trigger thresholds are set. For example, when the abnormal state of a node exceeds a predetermined threshold, such as equipment temperature exceeding the safety limit or pressure exceeding the operating range, the fault is considered to have a risk of escalation. Alarm trigger thresholds are typically set according to different production conditions and equipment parameters to ensure that the sensitivity of the alarms adapts to different operating conditions.

[0078] Once an abnormal state at a fault node is determined to have met the trigger conditions, the alarm system will promptly transmit fault information to operators, including the fault type, location, scope of impact, and emergency response measures. Notifications will be sent through multiple channels, including audible and visual alarms, SMS, email, and real-time notifications within the production management system, ensuring operators receive alerts promptly. For high-priority faults, multiple platforms will be linked to ensure a rapid response regardless of operator location. The generated alarms are not vague fault reports but rather action guidelines that include the root cause, precise location, quantified severity, predicted evolution, and specific recommendations. Timely warnings and guidance for operators to take action minimizes production interruptions due to equipment failures or anomalies, improving the stability and reliability of forging production.

[0079] In summary, the intelligent evaluation method for forging operation status combined with big data analysis provided in this application has the following beneficial effects: By chain-like analysis of the forging operation scheme of the forging press, a forging operation chain is established. Based on the forging operation chain, a multi-dimensional baseline state deduction of the forging press is performed to establish a forging baseline accompanying space. When the forging press is controlled to perform forging operations according to the forging operation chain, the current forging monitoring flow corresponding to the current forging node is acquired in real time. Based on the forging baseline accompanying space, the node association anomaly analysis and evaluation of the current forging monitoring flow is performed to establish a current forging anomaly coupling matrix. Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging anomaly coupling matrix to establish a current forging fault causal graph network. Based on the forging operation chain, node-transmission forging fault tracing compensation is performed on the current forging fault causal graph network to obtain a forging fault causal coupling graph network, and the forging operation chain is adjusted based on the forging fault causal coupling graph network. In other words, by constructing a forging operation chain and performing multi-dimensional baseline state simulation, while acquiring the current forging monitoring flow, and through node association anomaly analysis and evaluation, a current forging anomaly coupling matrix is ​​established. The fault causal ABF architecture is then invoked to perform multivariate causal analysis, a detailed fault causal graph is established, the root cause of the fault is accurately identified, and fault tracing compensation is performed on the node transmission faults in the fault causal graph. Compensation is then performed from the root cause of the fault, improving the accuracy of forging fault compensation and thus enhancing forging operation efficiency.

[0080] Example 2: Based on the same inventive concept as the intelligent evaluation method for forging operation status combined with big data analysis in Example 1, this application also provides an intelligent evaluation device for forging operation status combined with big data analysis. Please refer to the appendix. Figure 2 The intelligent evaluation device for forging operation status combined with big data analysis includes: The chain-based sorting module 11 is used to sort the forging operation plan of the forging press in a chain, establish a forging operation chain, and perform multi-dimensional benchmark state deduction of the forging press based on the forging operation chain to establish a forging benchmark accompanying space; the forging monitoring flow acquisition module 12 is used to acquire the current forging monitoring flow corresponding to the current forging node in real time when controlling the forging press to perform forging operation according to the forging operation chain; the association anomaly analysis and evaluation module 13 is used to perform node association anomaly analysis and evaluation of the current forging monitoring flow based on the forging benchmark accompanying space to establish a current forging anomaly coupling matrix; the fault causality prediction module 14 is used to perform multivariate forging fault causality prediction of the current forging anomaly coupling matrix based on the current forging node by calling the fault causality ABF architecture to establish a current forging fault causality graph network; the fault tracing and compensation module 15 is used to perform node-transmission forging fault tracing compensation of the current forging fault causality graph network based on the forging operation chain to obtain a forging fault causality coupling graph network, and perform feedback adjustment of the forging operation chain based on the forging fault causality coupling graph network.

[0081] Furthermore, the chain sorting module 11 in the intelligent evaluation device for forging operation status combined with big data analysis is also used to: divide the forging operation plan into nodes, determine the forging preheating plan, the forging impact plan, and the forging cooling plan; and chain the forging preheating plan, the forging impact plan, and the forging cooling plan to generate the forging operation chain.

[0082] Furthermore, the chain-like sorting module 11 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: performing multi-dimensional benchmark state deduction on the forging press according to the forging preheating scheme to generate a preheating benchmark accompanying space; performing multi-dimensional benchmark state deduction on the forging press according to the forging impact scheme to generate an impact benchmark accompanying space; performing multi-dimensional benchmark state deduction on the forging press according to the forging cooling scheme to generate a cooling benchmark accompanying space; and performing chain-like encapsulation of the preheating benchmark accompanying space, the impact benchmark accompanying space, and the cooling benchmark accompanying space to generate the forging benchmark accompanying space.

[0083] Furthermore, the chain-like sorting module 11 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: interconnecting equipment of the same specification and model according to the forging press to obtain a cluster of forging equipment; performing a big data retrieval of normal equipment status samples for the cluster of forging equipment according to the forging preheating scheme to obtain a preheating node equipment status registration set; performing central trend analysis on the preheating node equipment status registration set to construct a preheating node equipment status benchmark matrix; performing a big data retrieval and central trend analysis of normal environmental status samples for the cluster of forging equipment according to the forging preheating scheme to construct a preheating node environmental status benchmark matrix; performing a big data retrieval and central trend analysis of normal product status samples for the cluster of forging equipment according to the forging preheating scheme to construct a preheating node product status benchmark matrix; and integrating the preheating node equipment status benchmark matrix, the preheating node environmental status benchmark matrix, and the preheating node product status benchmark matrix to obtain the preheating benchmark accompanying space.

[0084] Furthermore, the associated anomaly analysis and evaluation module 13 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: performing node association mapping activation of the forging reference accompanying space according to the current forging node to obtain the current reference accompanying space, the current reference accompanying space including the current equipment status reference matrix, the current environment status reference matrix, and the current product status reference matrix; performing multi-characteristic identification and cleaning according to the current forging monitoring stream to establish the current equipment status monitoring matrix, the current environment status monitoring matrix, and the current product status monitoring matrix; performing anomaly identification and evaluation on the current equipment status monitoring matrix according to the current equipment status reference matrix to obtain a first anomaly identification and evaluation matrix; performing anomaly identification and evaluation on the current environment status monitoring matrix according to the current environment status reference matrix to obtain a second anomaly identification and evaluation matrix; performing anomaly identification and evaluation on the current product status monitoring matrix according to the current product status reference matrix to obtain a third anomaly identification and evaluation matrix; and performing anomaly attention allocation and aggregation according to the first anomaly identification and evaluation matrix, the second anomaly identification and evaluation matrix, and the third anomaly identification and evaluation matrix to generate the current forging anomaly coupling matrix.

[0085] Furthermore, the fault causal prediction module 14 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: performing fault type mining on the forging press based on the current forging node to obtain a current forging fault type matrix, wherein the current forging fault type matrix includes F forging fault type labels, where F is a positive integer greater than 1; performing fault causal event retrieval based on the current forging fault type matrix to obtain F forging fault causal event regions; performing multi-level causal reasoning learning on the F forging fault causal event regions based on the fault causal ABF architecture to establish F forging fault causal prediction branches; inputting the current forging anomaly coupling matrix into the F forging fault causal prediction branches to obtain F forging fault causal prediction paths; and performing graph-network sorting based on the F forging fault causal prediction paths to generate the current forging fault causal graph network.

[0086] Furthermore, the fault causal prediction module 14 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: extracting the f-th forging fault causal event region based on the F forging fault causal event regions, where f is a positive integer, 1≤f≤F; activating the fault causal ABF architecture, which includes an ARIMA model, a BBN model, an FTA model, and a central server; and performing iterative supervised training on the ARIMA model, the BBN model, and the FTA model based on the f-th forging fault causal event region to establish a first forging fault causal prediction model, a second forging fault causal prediction model, and a third forging fault causal prediction model. Three models are used: loss characteristics are analyzed for the first, second, and third models of forging fault causal prediction, and a fault causal prediction mentor-student relationship is established; real-time monitoring logs of the central server are obtained, and security and privacy protection configurations are enhanced for the central server based on the real-time monitoring logs to establish a central trusted node; based on the fault causal prediction mentor-student relationship, the first, second, and third models of forging fault causal prediction are distilled and enhanced according to the central trusted node to generate the f-th forging fault causal prediction branch.

[0087] Furthermore, the fault tracing and compensation module 15 in the intelligent evaluation device for forging operation status combined with big data analysis is also used for: analyzing the node transmission characteristics of the forging operation chain based on the current forging node to determine the forging scheme of the transmission node; predicting forging faults for the forging scheme of the transmission node according to each forging fault causal prediction path in the current forging fault causal graph network to obtain F predicted transmission forging faults; tracing the accident triggering factors of the F predicted transmission forging faults based on the current forging fault causal graph network and the transmission node forging scheme to obtain F transmission forging fault paths; and compensating for the transmission forging fault characteristics of the current forging fault causal graph network based on the F transmission forging fault paths to generate the forging fault causal coupling graph network.

[0088] Furthermore, the fault tracing and compensation module 15 in the intelligent evaluation device for forging operation status combined with big data analysis is also used to generate a forging fault alarm based on the forging fault causal coupling graph.

[0089] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1The intelligent evaluation method and specific examples of forging operation status combined with big data analysis in Example 1 are also applicable to the intelligent evaluation device of forging operation status combined with big data analysis in this embodiment. Through the foregoing detailed description of the intelligent evaluation method of forging operation status combined with big data analysis, those skilled in the art can clearly understand the intelligent evaluation device of forging operation status combined with big data analysis in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0090] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0091] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for intelligently evaluating the operating status of forging and pressing operations using big data analysis, characterized in that: include: The forging operation scheme of the forging press is sorted out in a chain, a forging operation chain is established, and a multi-dimensional benchmark state deduction of the forging press is performed based on the forging operation chain to establish the forging benchmark associated space. When controlling the forging press to perform forging operations according to the forging operation chain, the current forging monitoring stream corresponding to the current forging node is obtained in real time. Based on the forging reference associated space, the current forging monitoring flow is analyzed and evaluated for node correlation anomalies, and a current forging anomaly coupling matrix is ​​established. Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging abnormal coupling matrix, and a current forging fault causal graph network is established. Based on the forging operation chain, the current forging fault causal network is used to perform node propagation forging fault tracing compensation, thereby obtaining a forging fault causal coupling network, and the forging operation chain is adjusted based on the forging fault causal coupling network.

2. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, Based on the forging operation chain, a multi-dimensional baseline state deduction is performed on the forging press to establish the forging baseline associated space, including: Based on the forging preheating scheme, a multi-dimensional baseline state deduction of the forging press is performed to generate a preheating baseline associated space; Based on the forging impact scheme, a multi-dimensional benchmark state deduction of the forging press is performed to generate the impact benchmark associated space; Based on the forging cooling scheme, a multi-dimensional baseline state deduction of the forging press is performed to generate a cooling baseline associated space; The forging reference space is generated by chain-encapsulation based on the preheating reference space, the impact reference space, and the cooling reference space.

3. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 2, characterized in that, Based on the forging preheating scheme, a multi-dimensional baseline state deduction is performed on the forging press to generate a preheating baseline associated space, including: Interconnect the forging presses of the same specifications and models to obtain a cluster of forging presses; Based on the forging preheating scheme, a large data retrieval of normal equipment status samples is performed on the same cluster of forging equipment to obtain a preheating node equipment status registration set. Based on the preheating node equipment status registration set, a central trend analysis is performed to construct a preheating node equipment status baseline matrix; Based on the forging preheating scheme, a large-scale retrieval and central trend analysis of normal environmental state samples of the same cluster of forging equipment are performed to construct a benchmark matrix of environmental state of preheating nodes. Based on the forging preheating scheme, a big data retrieval and central trend analysis of normal product status samples are performed on the same cluster of forging equipment to construct a product status benchmark matrix for preheating nodes. By integrating the preheating node equipment status reference matrix, the preheating node environmental status reference matrix, and the preheating node product status reference matrix, the preheating reference accompanying space is obtained.

4. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, Based on the forging reference associated space, the current forging monitoring flow is analyzed and evaluated for node correlation anomalies, and a current forging anomaly coupling matrix is ​​established, including: Based on the current forging node, the node association mapping activation of the forging reference accompanying space is performed to obtain the current reference accompanying space, which includes the current equipment state reference matrix, the current environment state reference matrix, and the current product state reference matrix; Based on the current forging monitoring flow, perform multi-characteristic identification and cleaning, and establish a current equipment status monitoring matrix, a current environmental status monitoring matrix, and a current product status monitoring matrix; Based on the current device status baseline matrix, an anomaly identification and evaluation is performed on the current device status monitoring matrix to obtain a first anomaly identification and evaluation matrix; Based on the current environmental state baseline matrix, an anomaly identification and evaluation matrix is ​​performed on the current environmental state monitoring matrix to obtain a second anomaly identification and evaluation matrix. Based on the current product status baseline matrix, an anomaly identification and evaluation is performed on the current product status monitoring matrix to obtain a third anomaly identification and evaluation matrix. Anomaly attention allocation and aggregation are performed based on the first anomaly identification evaluation matrix, the second anomaly identification evaluation matrix, and the third anomaly identification evaluation matrix to generate the current forging anomaly coupling matrix.

5. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, Based on the current forging node, the fault causal ABF architecture is invoked to perform multivariate forging fault causal prediction on the current forging anomaly coupling matrix, and a current forging fault causal graph network is established, including: Based on the current forging node, the forging press is subjected to fault type mining to obtain the current forging fault type matrix. The current forging fault type matrix includes F forging fault type labels, where F is a positive integer greater than 1. Based on the current forging fault type matrix, fault causal event retrieval is performed to obtain F forging fault causal event regions; Based on the fault causal ABF architecture, multi-level causal reasoning learning is performed on the F forging fault causal event regions to establish F forging fault causal prediction branches. Input the current forging anomaly coupling matrix into the F forging fault causal prediction branches to obtain F forging fault causal prediction paths; Based on the F forging fault causal prediction paths, a graph network is generated to form the current forging fault causal graph network.

6. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 5, characterized in that, Based on the aforementioned fault causal ABF architecture, multi-level causal reasoning learning is performed on the F forging fault causal event regions to establish F forging fault causal prediction branches, including: Extract the f-th forging failure causal event region based on the F forging failure causal event regions, where f is a positive integer, 1≤f≤F; Activate the fault causal ABF architecture, which includes an ARIMA model, a BBN model, an FTA model, and a central server; Based on the f-th forging failure causal event region, the ARIMA model, BBN model and FTA model are iteratively supervised and trained to establish the first forging failure causal prediction model, the second forging failure causal prediction model and the third forging failure causal prediction model. Loss characteristics are analyzed for the first, second, and third models of forging fault causal prediction, and a teacher-student relationship for fault causal prediction is established. Obtain the real-time monitoring logs of the central server, and enhance the security and privacy protection configuration of the central server based on the real-time monitoring logs to establish a central trusted node; Based on the aforementioned fault causal prediction teacher-student relationship, the forging fault causal prediction first model, the forging fault causal prediction second model, and the forging fault causal prediction third model are distilled and strengthened according to the central trusted node to generate the f-th forging fault causal prediction branch.

7. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, Based on the forging operation chain, node propagation forging fault tracing compensation is performed on the current forging fault causal network to obtain a forging fault causal coupling network, including: Based on the current forging node, the node transmission characteristics of the forging operation chain are analyzed to determine the forging scheme of the transmission node; Based on each forging fault causal prediction path in the current forging fault causal graph network, forging fault prediction is performed on the forging scheme of the transmission node to obtain F predicted transmission forging faults. Based on the current forging fault cause-effect graph network and the forging scheme of the transmission node, the accident triggering factors of the F predicted forging faults are traced to obtain the F transmission forging fault paths. Based on the F forging fault paths, the current forging fault causal network is compensated for forging fault characteristics to generate the forging fault causal coupling network.

8. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, The forging operation plan of the forging press is analyzed in a chain, and a forging operation chain is established, including: The forging operation plan is divided into nodes to determine the forging preheating plan, forging impact plan, and forging cooling plan; The forging preheating scheme, the forging impact scheme, and the forging cooling scheme are chained together to generate the forging operation chain.

9. The intelligent evaluation method for forging operation status combined with big data analysis as described in claim 1, characterized in that, Based on the forging fault causal coupling diagram, a forging fault alarm is generated.

10. An intelligent evaluation device for forging operation status combining big data analysis, characterized in that: The step of implementing the intelligent evaluation method for forging operation status combining big data analysis as described in any one of claims 1 to 9, wherein the intelligent evaluation device for forging operation status combining big data analysis comprises: The chain sorting module is used to sort out the forging operation scheme of the forging press in a chain, establish the forging operation chain, and perform multi-dimensional benchmark state deduction of the forging press based on the forging operation chain to establish the forging benchmark accompanying space. The forging monitoring stream acquisition module is used to acquire the current forging monitoring stream corresponding to the current forging node in real time when controlling the forging press to perform forging operations according to the forging operation chain. The correlation anomaly analysis and evaluation module is used to perform node correlation anomaly analysis and evaluation on the current forging monitoring flow based on the forging reference accompanying space, and to establish the current forging anomaly coupling matrix. The fault causal prediction module is used to perform multivariate forging fault causal prediction on the current forging abnormal coupling matrix based on the current forging node and to establish the current forging fault causal graph network. The fault tracing and compensation module is used to perform node-based forging fault tracing and compensation based on the current forging fault causal network of the forging operation chain, obtain a forging fault causal coupling network, and perform feedback adjustment on the forging operation chain based on the forging fault causal coupling network.

Images