Method and system for evaluating health status of aluminum processing equipment

Abstract

The invention belongs to the technical field of distributed assessment, and particularly relates to an aluminum processing equipment health status assessment method and system.The method comprises the steps that cross-plant equipment attributes and operation information are combined with a Merkel tree and an assessment algorithm, and a distributed fault assessment forest is constructed; secondly, carrying out distributed edge interaction state evaluation by utilizing the tree and a preset cross-plant real-time monitoring network, and obtaining evaluation results of single equipment and process associated equipment; and 3, according to the results and the root cause traceability early warning rule and strategy generation library, carrying out fault traceability, generating and sharing a maintenance strategy, meanwhile, visualizing the whole traceability and sharing process by means of a traceability early warning visualization mapping space, and finally forming a visualization fault assessment early warning forest. Accurate evaluation of the state of the cross-factory aluminum processing equipment, fault traceability and effective generation and sharing of a maintenance strategy are realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of distributed evaluation, and particularly relates to a method and system for evaluating the health status of aluminum processing equipment. Background Art

[0002] In the process of the aluminum processing industry moving towards intelligent and large-scale development, the importance of cross-plant equipment health status evaluation has become increasingly prominent. There are numerous equipment in each plant area, and the operating conditions are complex. Comprehensively and accurately grasping the equipment status is of great significance for ensuring production continuity and improving product quality. However, there are defects in the existing technologies. On the one hand, the phenomenon of data islands is serious, and it is difficult to share data among different plant areas, making it impossible to integrate multi-source information to build a comprehensive model. On the other hand, in the centralized learning mode, the centralized transmission of data is prone to privacy leakage risks, and in the face of the heterogeneity of equipment in different plant areas, traditional methods are difficult to effectively integrate data and accurately evaluate. These problems restrict the accuracy and security of cross-plant aluminum processing equipment status evaluation. Therefore, the present invention provides a method and system for evaluating the health status of aluminum processing equipment. Summary of the Invention

[0003] Aiming at the deficiencies of the existing technologies, the present invention proposes a method and system for evaluating the health status of aluminum processing equipment. The method first combines cross-plant equipment attributes and operation information with Merkle trees and evaluation algorithms to construct a distributed fault evaluation forest. Secondly, using the distributed fault evaluation forest and a preset cross-plant real-time monitoring network, it conducts distributed edge interaction status evaluation to obtain evaluation results of single equipment and process-related equipment. Thirdly, based on these results and the root cause tracing warning rule and strategy generation library, it conducts fault tracing, generates and shares maintenance strategies. At the same time, with the help of the tracing warning visualization mapping space, it visualizes the entire tracing and sharing process, and finally forms a visualized fault evaluation warning forest to achieve accurate evaluation of the status of cross-plant aluminum processing equipment, fault tracing, and effective generation and sharing of maintenance strategies.

[0004] To achieve the above object, the present invention provides the following technical solutions: A method for evaluating the health status of aluminum processing equipment, comprising: Based on cross-plant equipment attributes and operation information, combining with Merkle trees and evaluation algorithms, obtaining a distributed fault evaluation forest; Based on the distributed fault evaluation forest, combining with a preset cross-plant real-time monitoring network to conduct distributed edge interaction status evaluation, obtaining single equipment evaluation results and process-related equipment evaluation results; Based on the single equipment evaluation results and process-related equipment evaluation results, combining with the root cause tracing warning rule and strategy generation library configured by the distributed fault evaluation forest, conducting fault tracing and generating and sharing maintenance strategies, and combining with the tracing warning visualization mapping space to visualize the tracing and sharing process, obtaining a visualized fault evaluation warning forest.

[0005] Specifically, the distributed fault evaluation forest includes M regional distributed fault evaluation trees; each of the regional distributed fault evaluation trees includes N distributed fault evaluation sub-trees; the construction process of the regional distributed fault evaluation tree includes: Based on the monitoring device location points, device physical attributes, single-state operation information, and data flow relationships among all devices within a single plant area, the monitoring nodes and directed link relationships of the real-time monitoring subnet corresponding to the single plant area are constructed. Based on each monitoring node in the real-time monitoring subnet, an evaluation node corresponding to each device within the current plant area is mapped one by one. Based on the data flow relationships among devices, the associated operation information of the production processes corresponding to different devices, and historical device associated fault information, through an association algorithm combined with the Bayesian algorithm, the fault causal association degree and corresponding confidence level among devices are obtained. Based on the reciprocals of the fault causal association degrees among devices, the directed causal connections of each device are constructed, and the directed fault causal association connections among devices are visually mapped using the corresponding confidence level and the visual mapping constructed in the preset HSV space to obtain the visual directed causal connections. Based on the evaluation nodes corresponding to each device and the visual directed causal connections among devices, combined with the blockchain, a horizontal production process association evaluation chain based on production process association is obtained.

[0006] Specifically, the construction process of the regional distributed fault evaluation tree further includes: Taking the evaluation node corresponding to each device as the generation root node of each distributed fault evaluation sub-tree, based on the device status information and interaction information obtained in real time by the monitoring node corresponding to the generation root node, through the Merkle tree algorithm built into the monitoring node, a time information hash block sequence corresponding to each generation root node is obtained. Each time information hash block corresponding to each generation root node is used to store the device operation status information, physical attribute information, interaction information, and corresponding characteristic hash values of each monitoring node within a preset operation time interval. Based on the time information hash block sequence corresponding to each generation root node, through the evaluation algorithm built into the generation root node, the device status evaluation score of each generation root node under each time information hash block is obtained. The device status evaluation score of each generation root node under each time information hash block is combined with a preset fault discrimination score threshold and a fault type library to obtain whether there is a fault in the corresponding device under the current time information hash block. If it exists, obtain the corresponding fault level and fault type under the current time information hash block, and use the corresponding fault level and fault type under the current time information hash block to construct the abnormal state node of the corresponding distributed fault assessment subtree under the current time information hash block. If it does not exist, construct a normal state node; Repeat the state node construction process to obtain the abnormal state node set and normal state node set of each generated root node under different time information hash blocks.

[0007] Specifically, the state node is used to store the operating state and characteristic hash value corresponding to each time information hash block of the corresponding device; the operating state includes the normal state, the fault state and the corresponding fault timestamp; the fault state includes the fault type and the corresponding abnormal degree score and level; the normal state is represented by 0; the fault state is represented by the corresponding fault type and fault level value together; Specifically, the construction process of the regional distributed fault assessment tree further includes: If there is only one type of fault in the current abnormal state node set, use the generated root node as the initial abnormal state node, and construct the longitudinal time connection between the adjacent two abnormal state nodes corresponding to each generated root node through the total length of the operating time interval and the corresponding characteristic hash value of the normal state nodes included between the adjacent two abnormal state nodes; If there is more than one type of fault in the current abnormal state node set, construct the longitudinal time connection between the adjacent abnormal state nodes corresponding to the new fault type according to the total length of the operating time interval and the corresponding characteristic hash value of all the abnormal state nodes and normal state nodes corresponding to the non-new fault types between the first abnormal state node corresponding to the new fault type and the generated root node.

[0008] Specifically, the construction process of the regional distributed fault assessment tree further includes: When the current fault type appears again, construct the abnormal state node corresponding to the fault type and the corresponding longitudinal time connection under the first abnormal state node; When a new fault type appears again, repeat the process of generating the longitudinal time connection for more than one type of fault, and generate new abnormal state nodes and longitudinal time connections under the current generated root node; Construct the corresponding distributed fault assessment subtree under the current generated root node based on the abnormal state nodes corresponding to different fault types and the corresponding longitudinal time connections; Construct a regional distributed fault assessment tree through a distributed framework based on the distributed fault assessment subtrees corresponding to different generated root nodes, the horizontal production process association assessment chains corresponding to different time information hash blocks, and the verification hashes between the state nodes with interaction relationships between different distributed fault assessment subtrees.

[0009] Specifically, the construction process of the regional distributed fault evaluation tree further includes: Based on the regional distributed fault evaluation tree, through the real-time time information hash blocks obtained by different monitoring nodes in the current plant area, combined with the preset confidence threshold and fault discrimination score threshold corresponding to the fault causal correlation degree between devices and the fault type library for training, the trained regional distributed fault evaluation tree of the current plant area is obtained; When a new distributed fault evaluation subtree corresponding to a device appears, repeat the above process to update and train the current regional distributed fault evaluation tree.

[0010] Specifically, the construction process of the distributed fault evaluation forest includes: Store the feature hash values and operating status values in all state nodes in the regional distributed fault evaluation tree corresponding to each plant area and the verification hashes between state nodes with interaction relationships between different distributed fault evaluation subtrees into the interactive storage block constructed by combining the root node generated corresponding to each regional distributed fault evaluation tree and the blockchain. At the same time, based on the interactive storage block corresponding to each regional distributed fault evaluation tree, construct a block directed connection according to the production process flow interaction relationship between different plant areas; Based on the feature hash values and verification hashes stored in the interactive storage block corresponding to each regional distributed fault evaluation tree, obtain the cross-regional traceability verification proof between the interactive storage blocks with interaction relationships through the zero-knowledge proof algorithm; Map the cross-regional traceability verification proof into the block directed connection between the corresponding interactive storage blocks to construct a cross-regional directed block traceability verification chain.

[0011] Specifically, the construction process of the distributed fault evaluation forest further includes: Take each regional distributed fault evaluation tree as a federated interactive edge node, and at the same time take the cross-regional directed block traceability verification chain as the cross-regional secure traceability verification path and map the visualization in the visual directed causal connection into the corresponding block directed connection; Based on the federated interactive edge node, the cross-regional secure traceability verification path, the preset verification attack library, the preset traceability accuracy threshold and the security verification threshold, and combine the federated algorithm for training to obtain the trained distributed fault evaluation forest.

[0012] Specifically, the construction process of the visual fault evaluation and warning forest includes: Based on the trained distributed fault evaluation forest combined with the real-time monitoring network, monitor and evaluate all plant area devices in real time to obtain the regional distributed fault evaluation tree with abnormal status; Based on the horizontal production process association evaluation chain and the distributed fault evaluation subtree in the area-distributed fault evaluation tree with the current abnormal state, trace the root cause of the abnormality inside the plant area and the associated abnormal nodes, and according to the corresponding fault causal association degree and the corresponding confidence level between the root cause of the abnormality and the associated abnormal nodes, combine the preset warning levels to conduct traceability warnings of corresponding levels, and visualize the traceability process according to the visual directed causal connection; At the same time, based on the trained distributed fault evaluation forest, use the cross-region directed block traceability verification chain to conduct cross-plant area safety traceability for the distributed fault evaluation tree with the current abnormal state and the associated distributed fault evaluation trees with fault causal relationships, and utilize the block directed connection combined with visual mapping, the horizontal production process association evaluation chain and the distributed fault evaluation subtree in the associated distributed fault evaluation tree to conduct cross-plant area associated fault equipment traceability and visualization of the traceability process, and use the fault causal association degree and the corresponding confidence level corresponding to each connection relationship in the corresponding traceability path to combine the preset warning levels to conduct cross-plant area difference warnings, and obtain a visual fault evaluation warning forest; The warning levels include a main warning level and an associated warning level; the main warning level corresponds to the main fault equipment monitored at the current moment; the associated warning level corresponds to the associated fault equipment at the current moment; The main warning level conducts warning of the corresponding level through the fault level corresponding to the main fault equipment; the associated fault equipment conducts warning of the corresponding level through the fault level corresponding to the associated fault equipment and its fault causal association degree and confidence level with the main fault equipment.

[0013] An aluminum processing equipment health status evaluation system includes: an evaluation tree module, a monitoring and evaluation module, and an inference and mapping module; The evaluation tree module obtains a distributed fault evaluation tree based on cross-plant area equipment attributes and operation information, combining the Merkle tree and the evaluation algorithm; The monitoring and evaluation module conducts distributed edge interaction state evaluation based on the distributed fault evaluation tree in combination with a preset cross-plant area real-time monitoring network to obtain a single equipment evaluation result and a process-associated equipment evaluation result; The inference and mapping module conducts fault traceability and maintenance strategy generation and sharing based on the single equipment evaluation result and the process-associated equipment evaluation result in combination with the root cause traceability warning rule and strategy generation library configured by the distributed fault evaluation forest, and visualizes the traceability and sharing process in combination with the traceability warning visual mapping space to obtain a visual fault evaluation warning tree.

[0014] Compared with the prior art, the beneficial effects of the present invention are: In view of the deficiencies of the prior art, the present invention constructs a trusted storage structure for cross-plant equipment status data by means of a Merkle tree based on blockchain and a distributed evaluation algorithm, and realizes the verifiability and anti-tampering ability of the equipment operation status by using the time information hash block sequence and the feature hash value; combines the Bayesian association algorithm with the visualization mapping mechanism of the fault causal association degree to establish a horizontal production process association evaluation chain, effectively quantifies the fault propagation path and confidence level between equipment, and strengthens the causal reasoning accuracy of multi-equipment collaborative monitoring; integrates the regional distributed fault evaluation tree and the cross-regional directed block traceability verification chain through the federated learning framework to realize cross-plant collaborative modeling and security verification under data privacy protection, and uses zero-knowledge proof to ensure the trusted traceability of the cross-domain interaction process; based on the HSV space visualization mapping and multi-level early warning rules, dynamically maps the fault nodes, causal connections and traceability paths into a multi-dimensional visualization topology, and combines the longitudinal time connection and the spatio-temporal correlation analysis of the abnormal state node set to construct a spatio-temporal coupling characterization system for the fault evolution process, significantly improving the early identification of hidden faults and the cross-domain root cause location ability in cross-plant aluminum processing equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 FIG. is a flowchart of a method for evaluating the health status of an aluminum processing equipment according to Embodiment 1 of the present invention; Figure 2 FIG. is a module diagram of a system for evaluating the health status of an aluminum processing equipment according to Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Embodiment 1 Please refer to Figure 1 , an embodiment provided by the present invention: a method for evaluating the health status of an aluminum processing equipment, the steps include: A method for evaluating the health status of an aluminum processing equipment, including: S1. Based on the cross-plant equipment attributes and operation information, combined with the Merkle tree and the evaluation algorithm, obtain a distributed fault evaluation forest; S2. Based on the distributed fault evaluation forest, combined with a preset cross-plant real-time monitoring network, conduct a distributed edge interaction status evaluation to obtain a single equipment evaluation result and a process-related equipment evaluation result; S3. Based on the single equipment evaluation result and the process-related equipment evaluation result, combined with the root cause traceability early warning rule and strategy generation library configured by the distributed fault evaluation forest, conduct fault traceability and maintenance strategy generation and sharing, and visualize the traceability and sharing process through the traceability early warning visualization mapping space to obtain a visualized fault evaluation early warning forest.

[0017] Furthermore, the strategy generation library in this embodiment is obtained by constructing a deep text inference generation model and a knowledge graph based on historical fault data and maintenance strategies; Furthermore, the distributed fault assessment forest in this embodiment includes M regional distributed fault assessment trees; each of the regional distributed fault assessment trees includes N distributed fault assessment subtrees. Furthermore, the construction process of the regional distributed fault assessment tree in this embodiment includes: Based on the monitoring device location points, device physical attributes, single-state operation information, and data flow relationships among all devices within a single plant area, the monitoring nodes and directed link relationships of the real-time monitoring subnet corresponding to the single plant area are constructed. Furthermore, the specific variable information corresponding to the monitoring device location points, device physical attributes, single-state operation information, and data flow relationships among all devices in this embodiment includes: The monitoring device location points of all devices within a single plant area. For example, in a certain plant area, there are devices such as smelting furnaces, casting and rolling mills, and cold rolling mills. The smelting furnace is equipped with temperature sensors and pressure sensors; the casting and rolling mill is equipped with vibration sensors and speed sensors, etc. These location points are recorded in coordinate form, accurate to the specific installation location inside the device.

[0018] Device physical attributes, such as the model, size, material, rated power, etc. of the device. These attributes are used to describe the basic physical characteristics of the device and have important reference value for subsequent analysis of the device operation status.

[0019] Single-state operation information, such as the current temperature and pressure of the smelting furnace; the current vibration amplitude and running speed of the casting and rolling mill, etc. These information are the real-time operation parameters of the device at a certain moment.

[0020] The data flow relationships among all devices. For example, the molten aluminum produced by the smelting furnace is transported to the casting and rolling mill through a pipeline, which indicates that there is a data flow from the smelting furnace to the casting and rolling mill, and can be expressed as smelting furnace → casting and rolling mill.

[0021] Exemplarily, the detailed example process of constructing the monitoring nodes and directed link relationships of the real-time monitoring subnet in this embodiment includes: For each monitoring device location point, a monitoring node is created. For example, based on the temperature sensor location A1 of the smelting furnace, the monitoring node N_A1 is created, and this node contains sensor location information, device information to which it belongs, and an initial data structure for subsequent data storage.

[0022] According to the device physical attributes, relevant attribute labels are added to each monitoring node. For example, attributes such as device model and material are added to the monitoring node N_A1 to facilitate subsequent differentiation of the monitoring data characteristics of different devices.

[0023] Update the data values of the monitoring nodes in real time based on the single - status operation information. For example, when the temperature sensor detects a new temperature value T2, update the temperature data in the monitoring node N_A1 to T2.

[0024] Construct directed links according to the data flow relationship between devices. For example, for the data flow from the smelting furnace to the casting - rolling mill, create directed edges from the monitoring nodes related to the smelting furnace (such as N_A1, N_A2) to the monitoring nodes related to the casting - rolling mill (such as N_B1, N_B2) to form a directed link relationship. N_B1 is a monitoring node created based on the position B2 of the speed sensor, and the construction of other monitoring nodes will not be elaborated here. These directed edges not only represent the data flow but also can carry relevant information such as the data transmission rate and delay.

[0025] Furthermore, the output here in this embodiment includes: The set of monitoring nodes of the real - time monitoring subnet: It contains the monitoring nodes corresponding to all device monitoring points within a single factory area, such as {N_A1, N_A2, N_B1, N_B2,...}, and each node has detailed device location, attribute, and real - time data information.

[0026] The set of directed link relationships: The set of directed edges representing the data flow between devices, such as {(N_A1, N_B1), (N_A2, N_B2),...}, and each directed edge carries the necessary information related to data transmission.

[0027] Based on each monitoring node in the real - time monitoring subnet, map one - to - one to obtain the evaluation node corresponding to each device in the current factory area; Based on the data flow relationship between devices, the associated operation information of the production processes corresponding to different devices, and the historical device - associated fault information, through an association algorithm combined with the Bayesian algorithm, obtain the fault causal association degree and the corresponding confidence level between devices; Exemplarily, the detailed example process of calculating the fault causal association degree and confidence level in this embodiment includes: Obtain the data flow relationship between devices and the associated operation information of the production processes corresponding to different devices. For example, data flow information such as from the smelting furnace to the casting - rolling mill, and clarify the association path between devices; in this embodiment, the associated operation information of the production processes corresponding to different devices includes that there is a process association between the temperature and pressure of the smelting furnace and the aluminum liquid quality and casting - rolling speed of the casting - rolling mill. In this embodiment, when the temperature of the smelting furnace is too high, it may cause quality problems in the aluminum strips produced by the casting-rolling mill. These associated information includes the mutual influence relationship between process parameters, the description of the influence degree, and the historical equipment associated fault information. For example, it records the relevant information when the equipment failed in the past, such as the abnormal increase in the temperature of the smelting furnace at a certain moment, and then the aluminum strip breakage fault occurred in the casting-rolling mill, as well as data such as the time and frequency of the fault occurrence.

[0028] According to the historical equipment associated fault information, construct a prior probability model of the fault occurrence; for example, count the number of times of aluminum strip breakage faults in the casting-rolling mill after the abnormal increase in the temperature of the smelting furnace and the total number of abnormal increases in the temperature of the smelting furnace in the past period of time, and calculate the prior probability P τ (casting-rolling mill fault | abnormal temperature of the smelting furnace).

[0029] Utilize the data flow relationship between each equipment and the associated operation information of the production process to update the probability model of the fault occurrence. For example, considering the influence of factors such as pipeline pressure on the quality of molten aluminum during the transmission process of molten aluminum from the smelting furnace to the casting-rolling mill, and the influence of the quality of molten aluminum on the operating state of the casting-rolling mill, through the Bayesian formula, combined with new evidence, such as pipeline pressure data, update the probability of the casting-rolling mill having a fault.

[0030] For each pair of equipment, calculate the fault causal association degree between them. The fault causal association degree can be determined by calculating the probability change of one equipment's fault triggering another equipment's fault under various conditions. For example, when the smelting furnace has an abnormal temperature fault, the probability of the casting-rolling mill having a fault changes from P3 to P4, and the fault causal association degree can be defined as (P4 - P3) / P3.

[0031] At the same time, through the analysis of historical data and the model calculation results, determine the confidence level of the fault causal association degree. The confidence level reflects the reliability of the calculated fault causal association degree. For example, by counting the proportion of the calculated fault causal association degree that matches the actual situation in the past similar situations, determine the confidence level of the current calculation result.

[0032] Furthermore, the output results corresponding to this embodiment here include: the fault causal association degree matrix between each equipment and the fault causal association confidence level matrix between each equipment; among them, the fault causal association degree matrix between each equipment in this embodiment includes a two-dimensional matrix, the rows and columns of the matrix represent different equipment respectively, and the matrix elements are the fault causal association degrees between the corresponding equipment. For example, for equipment A and equipment B, the element [A, B] in the matrix represents the fault causal association degree of equipment A's fault triggering equipment B's fault; In this embodiment, the fault causal association confidence matrix between devices is also a two-dimensional matrix, corresponding to the fault causal association degree matrix, and the matrix elements are the confidence levels of the fault causal association degrees between the corresponding devices. For example, the element [A, B] in the matrix represents the confidence level of the fault causal association degree that the fault of device A causes the fault of device B.

[0033] Based on the reciprocals of the fault causal association degrees between devices, directed causal connections of each device are constructed, and using the corresponding confidence levels and the visualization mapping constructed by the preset HSV space, the directed fault causal association connections between devices are visually mapped to obtain visual directed causal connections. Exemplarily, the example process of constructing visual directed causal connections in this embodiment includes: Obtain the fault causal association degree matrix between devices; for example, the matrix describing the fault causal association degrees between devices obtained in the previous step, and the visualization mapping rules constructed by the preset HSV space, such as the HSV (Hue, Saturation, Value) space is used to map numerical information into visual colors. Furthermore, in this embodiment, it is set that the higher the fault causal association degree, the more the hue tends to red, and the saturation and value are determined according to the preset variation rules that are positively correlated with the hue to show the strength of the association degree.

[0034] For the fault causal association degree matrix between devices, calculate its reciprocal. Since the higher the fault causal association degree, the smaller its reciprocal, it is more convenient to perform numerical mapping in subsequent visualization.

[0035] According to the reciprocals of the fault causal association degrees, using the visualization mapping rules constructed by the preset HSV space, assign colors and other visualization attributes to the directed causal connections between each pair of devices. For example, for the connection between device A and device B with the reciprocal of the fault causal association degree being a1, according to the HSV mapping rules, set its hue to a value close to red, and determine the saturation and value according to specific rules. At the same time, set the line thickness according to the size of the reciprocal of the association degree. The smaller the reciprocal of the association degree (i.e., the higher the fault causal association degree), the thicker the line.

[0036] Use the corresponding confidence levels to further adjust the visualization attributes. For example, if the confidence level of the fault causal association degree between device A and device B is small, the transparency of the color of the connection line can be appropriately reduced to indicate that its reliability is high but not absolutely certain.

[0037] Further, the output in this embodiment includes: a visualized directed causal connection set, where the visualized directed causal connection set contains visual representations of all directed causal connections between devices, and each connection has visual attributes such as color, line thickness, and transparency determined according to the reciprocal of the fault causal correlation degree and confidence, intuitively showing the strength and reliability of the fault causal relationship between devices.

[0038] Based on the evaluation nodes corresponding to each device and the visualized directed causal connections between devices, combined with the blockchain, a horizontal production process correlation evaluation chain based on production process association is obtained; Taking the evaluation node corresponding to each device as the generation root node of each distributed fault evaluation subtree, based on the device status information and interaction information obtained in real time by the monitoring nodes corresponding to the generation root nodes, through the Merkle tree algorithm built into the monitoring nodes, a sequence of time information hash blocks corresponding to each generation root node is obtained; Exemplarily, the exemplary process for obtaining information related to the generation root node in this embodiment includes: Obtain a set of evaluation nodes corresponding to each device, where each evaluation node in the set corresponds to a specific device in the factory area, and includes various attributes and association information of the device, as well as the device status information and interaction information obtained in real time by the monitoring nodes. For example, the monitoring nodes obtain real-time temperature, pressure and other status information of the melting furnace, as well as data interaction information with other devices, such as the amount of molten aluminum transferred and the transfer time. These information are updated in real time to reflect the current operating conditions of the device.

[0039] For each evaluation node, through the associated monitoring node, device status information and interaction information are collected in real time. For example, for the generation root node corresponding to the melting furnace, the monitoring node continuously collects data from temperature sensors and pressure sensors, as well as data interaction information with the casting and rolling mill.

[0040] Using the Merkle tree algorithm built into the monitoring node, at a preset time interval, time information hash blocks are generated. Each time information hash block stores the device operation status information, physical attribute information, interaction information obtained by the monitoring node during this time period, and the feature hash value calculated from these information; for example, combine the temperature, pressure, molten aluminum transfer amount, etc. of the melting furnace within m minutes into a data block, calculate its SHA-256 hash value, and form the feature hash value in the time information hash block.

[0041] In chronological order, a sequence of time information hash blocks corresponding to each generation root node is generated. For example, starting from time 0, a time information hash block is generated every m minutes, and arranged in sequence to form a sequence {HB1, HB2, HB3,..., HBn}, where HBn represents the nth time information hash block.

[0042] Each time information hash block corresponding to each generated root node is used to store the device operation status information, physical attribute information, interaction information of each monitoring node within a preset operation time interval, and the corresponding feature hash values; Based on the time information hash block sequence corresponding to each generated root node, through the evaluation algorithm built into the generated root node, obtain the device status evaluation score of each generated root node under each time information hash block; The device status evaluation score of each generated root node under each time information hash block is combined with a preset fault discrimination score threshold and a fault type library to obtain whether there is a fault in the corresponding device under the current time information hash block; If there is, obtain the corresponding fault level and fault type under the current time information hash block, and use the corresponding fault level and fault type under the current time information hash block to construct an abnormal state node of the corresponding distributed fault evaluation subtree under the current time information hash block. If not, construct a normal state node; Repeat the state node construction process to obtain the abnormal state node set and normal state node set of each generated root node under different time information hash blocks; Exemplarily, the example process of fault judgment and state node construction in this embodiment includes: Obtain the device status evaluation score set and fault type library of each generated root node under each time information hash block, where the evaluation score set contains the set of device status evaluation scores for each time period; the fault type library contains information such as various possible device fault types, their corresponding feature descriptions, fault causes, etc. For example, the fault type library records "overtemperature fault", its feature is that the temperature exceeds the normal range to a certain extent, and the possible cause is a cooling system fault, etc.

[0043] Compare each device status evaluation score of each generated root node with a preset fault discrimination score threshold. For example, for the device status evaluation score Sn_melting of the generated root node of the smelting furnace, if Sn_melting is less than the preset fault discrimination threshold, it is determined that the device has a fault in this time period.

[0044] If it is determined that there is a fault, according to the equipment status evaluation score and the fault type library, determine the fault level and fault type. According to the fault judgment result, construct an abnormal state node. If the equipment is in a normal state (the equipment status evaluation score is greater than or equal to the fault discrimination score threshold), construct a normal state node. The node records the operating state as "0" (indicating the normal state), and obtains the characteristic hash value of the corresponding time information hash block. If the equipment has a fault, construct an abnormal state node. The node records the operating state as the fault type and fault level value, the fault timestamp, the abnormal degree score (such as mapped according to the equipment status evaluation score and the severity of the corresponding fault type in the fault type library), and the characteristic hash value of the corresponding time information hash block.

[0045] Repeat the above process to construct the corresponding abnormal state node set and normal state node set for each generated root node under different time information hash blocks.

[0046] Furthermore, the state nodes in this embodiment are used to store the operating state and the corresponding characteristic hash value of the corresponding equipment for each time information hash block; the operating state includes the normal state, the fault state and the corresponding fault timestamp; the fault state includes the fault type and the corresponding abnormal degree score and level; the normal state is represented by 0; the fault state is represented by the common value of the corresponding fault type and fault level; If there is only one fault type in the current abnormal state node set, use the generated root node as the initial abnormal state node, and construct the longitudinal time connection between the corresponding adjacent two abnormal state nodes for each generated root node through the total length of the operating time interval and the corresponding characteristic hash value of the normal state nodes included between the adjacent two abnormal state nodes. If there is more than one fault type in the current abnormal state node set, construct the longitudinal time connection between the adjacent abnormal state nodes corresponding to the new fault type according to the total length of the operating time interval and the corresponding characteristic hash value of all the abnormal state nodes and normal state nodes corresponding to the non-new fault types between the first abnormal state node corresponding to the new fault type and the generated root node.

[0047] When the current fault type appears again, construct the abnormal state node corresponding to the fault type and the corresponding longitudinal time connection under the first abnormal state node; When a new fault type appears again, repeat the process of generating the longitudinal time connection when there is more than one fault type, and generate new abnormal state nodes and longitudinal time connections under the current generated root node; Construct the corresponding distributed fault evaluation subtree under the current generated root node based on the abnormal state nodes corresponding to different fault types and the corresponding longitudinal time connections; Exemplarily, the example process of constructing the longitudinal time connection in this embodiment includes: Obtain all abnormal state nodes where the generated root node appears under different time information hash blocks and all normal state nodes where the generated root node appears under different time information hash blocks; Take the generated root node as the initial abnormal state node. For two adjacent abnormal state nodes, calculate the total length of the running time intervals corresponding to the normal state nodes included between them. For example, there are two normal state nodes, N1_normal_melting and N2_normal_melting, between the abnormal state nodes A1_melting and A2_melting. Record their running time intervals as [t1, t2] and [t3, t4] respectively, then the total length is (t2 - t1) + (t4 - t3).

[0048] Meanwhile, obtain the feature hash values corresponding to two adjacent abnormal state nodes. For example, the feature hash value of A1_melting is H1, and the feature hash value of A2_melting is H2.

[0049] Based on the above information, construct the vertical time connection between two adjacent abnormal state nodes corresponding to each generated root node. Record information such as the total length of the time interval, the feature hash value of the starting abnormal state node, and the feature hash value of the ending abnormal state node in the connection.

[0050] For the first abnormal state node corresponding to the new fault type, set it as A_new1, and calculate the total length of the running time intervals corresponding to all abnormal state nodes and normal state nodes included within the time interval between it and the generated root node. For example, before A_new1, there is A1_melting (belonging to the old fault type, obtaining the time length of the old fault type failure) and two normal state nodes, N1_normal_melting and N2_normal_melting, and calculate the total length of their running time intervals.

[0051] Based on the nodes corresponding to the new fault type and all abnormal state nodes corresponding to non - new fault types, as well as the total length of the running time intervals corresponding to the normal state nodes, and combining with the hash algorithm, obtain the feature hash value of the node corresponding to the new fault type. For example, the feature hash value of A_new1 is H_new1, the feature hash value of A1_melting is H1, and the feature hash value of N1_normal_melting is H_N1, etc.

[0052] Based on the nodes corresponding to the new fault type and all abnormal state nodes corresponding to non - new fault types, as well as the feature hash values of the normal state nodes, construct the vertical time connection between adjacent abnormal state nodes corresponding to the new fault type; record information such as the total length of the time interval and the feature hash values of the involved nodes in the connection.

[0053] When the current fault type appears again, construct the corresponding fault type abnormal state node and the corresponding vertical time connection under the first abnormal state node. For example, if the abnormal state node A3_melting of the same fault type as A1_melting appears again, construct a vertical time connection under A1_melting, and calculate the total length of the time interval between A1_melting and A3_melting and information such as the relevant feature hash value.

[0054] When a new type appears again, repeat the above process to generate a new abnormal state node and a vertical time connection at the currently generated root node. For example, if the abnormal state node A_new2 of a new fault type appears, calculate the total length of the time interval and the feature hash value between A_new2 and the previous relevant nodes, and construct a vertical time connection; in this embodiment, the previous relevant nodes include the generated root node, the existing abnormal state nodes, and the normal state nodes.

[0055] Furthermore, the set of vertical time connections corresponding to different fault types in this embodiment includes the set of vertical time connections corresponding to different fault types, and these connections reflect the sequence of occurrence of faults at different times in the device, the change of fault types, and the information of the normal operation time period.

[0056] Based on the verification hash between the state nodes with interaction relationships among the distributed fault assessment subtrees corresponding to different generated root nodes, the horizontal production process association assessment chains corresponding to different time information hash blocks, and different distributed fault assessment subtrees, construct a regional distributed fault assessment tree through a distributed framework.

[0057] Based on the regional distributed fault assessment tree, through the real-time time information hash blocks obtained by different monitoring nodes in the current plant area, combined with the preset confidence threshold and fault discrimination score threshold corresponding to the fault causal correlation degree between each device and the fault type library for training, obtain the trained regional distributed fault assessment tree corresponding to the current plant area; When a distributed fault assessment subtree corresponding to a new device appears, repeat the above process to update and train the current regional distributed fault assessment tree.

[0058] This process constructs a cross-level fault assessment system through multi-dimensional technology integration. First, a real-time monitoring subnet is established based on the physical attributes of the equipment and the relationship of data flow. The coordinate-level monitoring nodes and directed links are used to accurately map the equipment topology, and the holographic mapping of the physical space and the data space is realized by combining the dynamic update mechanism of the equipment status parameters. By fusing historical fault data and process-related information with the Bayesian algorithm, the fault causal correlation degree and confidence matrix between equipment are quantified, and a fault propagation model based on probability evolution is established, breaking through the limitation of isolated judgment of traditional threshold alarms. The visualization mapping rule that innovatively combines the reciprocal of the fault causal correlation degree with the HSV space is adopted to dynamically transform the association strength and confidence into a visual connection coupled with multiple parameters of hue-saturation-value, realizing the intuitive topological expression of complex causal relationship networks. With the help of blockchain technology, the evaluation nodes and causal connection information are used to construct a horizontal production process association evaluation chain. The Merkle tree is used to generate a sequence of time information hash blocks and embed the feature hash values to form a spatio-temporal fingerprint database of the operating state of the equipment throughout its life cycle. In the fault assessment link, an abnormal state node set is constructed through vertical time connection based on the distributed fault assessment subtree generated by the root node. The spatio-temporal correlation model of fault evolution is established by combining the length of the normal node operation interval and the feature hash value, effectively capturing the spatio-temporal coupling characteristics of intermittent faults and compound faults. When integrating multiple subtrees in the distributed framework, the verification hash is used to verify the cross-subtree interaction relationship, and the dynamic expansion of the regional assessment tree and cross-domain security verification are realized by combining the federated learning framework, forming a three-dimensional assessment system that combines fine-grained assessment at the equipment level and cross-plant collaborative traceability, significantly improving the early recognition accuracy of hidden faults and the multi-level root cause location efficiency in complex manufacturing systems.

[0059] Furthermore, the construction process of the distributed fault assessment forest in this embodiment includes: The feature hash values and operating state values of all state nodes in the regional distributed fault assessment tree corresponding to each plant area and the verification hashes between state nodes with interaction relationships between different distributed fault assessment subtrees are stored in the interactive storage block constructed by combining the corresponding generated root node of each regional distributed fault assessment tree with the blockchain. At the same time, based on the interactive storage block corresponding to each regional distributed fault assessment tree, a block directed connection is constructed according to the production process interaction relationship between different plant areas. Based on the feature hash values and verification hashes stored in the interactive storage block corresponding to each regional distributed fault assessment tree, through the zero-knowledge proof algorithm, a cross-regional traceability verification proof between interactive storage blocks with interaction relationships is obtained. The cross-regional traceability verification proof is mapped into the block directed connection between the corresponding interactive storage blocks to construct a cross-regional directed block traceability verification chain.

[0060] Taking each regional distributed fault evaluation tree as a federated interaction edge node, simultaneously using a cross-regional directed block traceability verification chain as a cross-regional security traceability verification path, and mapping the visualization mapping in the visual directed causal connection into the corresponding block directed connection; Based on the federated interaction edge node, the cross-regional security traceability verification path, the preset verification attack library, and the preset traceability accuracy threshold and security verification threshold, combined with the federated algorithm for training, a trained distributed fault evaluation forest is obtained.

[0061] Furthermore, the construction process of the visual fault evaluation and early warning forest in this embodiment includes: Based on the trained distributed fault evaluation forest and combined with the real-time monitoring network, all plant area equipment is monitored and evaluated in real time to obtain the regional distributed fault evaluation tree in an abnormal state. Based on the horizontal production process association evaluation chain and the distributed fault evaluation subtree in the currently abnormal regional distributed fault evaluation tree, trace the root cause of the abnormality and associated abnormal nodes within the plant area, and according to the corresponding fault causal association degree and confidence level between the root cause of the abnormality and the associated abnormal nodes, combined with the preset early warning level, conduct corresponding level traceability early warning, and visualize the traceability process according to the visual directed causal connection; At the same time, based on the trained distributed fault evaluation forest, through the cross-regional directed block traceability verification chain, perform cross-plant area security traceability on the distributed fault evaluation tree in the current abnormal state and the associated distributed fault evaluation trees with fault causal relationships, and utilize the block directed connection combined with the visual mapping and the horizontal production process association evaluation chain and the distributed fault evaluation subtree in the associated distributed fault evaluation tree to conduct cross-plant area associated fault equipment traceability and visualization of the traceability process, and use the fault causal association degree and confidence level corresponding to each connection relationship in the corresponding traceability path combined with the preset early warning level to conduct cross-plant area difference early warning, and obtain the visual fault evaluation and early warning forest; Furthermore, the early warning levels in this embodiment include a main early warning level and an associated early warning level; the main early warning level corresponds to the main fault equipment monitored at the current moment; the associated early warning level corresponds to the associated fault equipment at the current moment; Furthermore, the main early warning level in this embodiment conducts corresponding level early warning through the fault level corresponding to the main fault equipment; the associated fault equipment conducts corresponding level early warning through the fault level corresponding to the associated fault equipment and its fault causal association degree and confidence level with the main fault equipment. The main fault equipment in this embodiment is the equipment that directly triggers the abnormal alarm or the core influencing equipment in the fault causal chain.

[0062] Furthermore, in this embodiment, the maintenance strategy generation and sharing process is as follows: According to the functional attributes of the corresponding faulty device, the generated maintenance strategy is shared into the production root nodes corresponding to the devices with the same production function through the distributed fault evaluation forest.

[0063] This process deeply integrates blockchain technology with the distributed fault evaluation tree, combines the zero-knowledge proof algorithm to construct a cross-regional directed block traceability verification chain, and realizes multi-dimensional technological collaborative innovation. Specifically, by embedding the device status feature hash and the interactive verification hash into the interactive storage block of the blockchain, the immutability of the blockchain is used to ensure the traceability of the entire data life cycle; the zero-knowledge proof algorithm is used to generate a cross-regional traceability verification proof to achieve the trusted verification of the cross-plant interactive relationship on the premise of avoiding the exposure of sensitive data, effectively solving the contradiction between data privacy protection and verification effectiveness in traditional methods; the multi-plant distributed fault evaluation trees are fused through the federated learning framework, which not only preserves the data privacy of each region but also realizes the global optimization of model parameters, significantly improving the generalization ability of fault evaluation; the dynamic mapping mechanism of the visual directed causal connection and the block directed connection transforms the complex multi-level fault correlation relationship into an intuitive visual topology structure, greatly improving the fault location efficiency. In particular, through the joint constraint mechanism of the preset verification attack library and the double threshold, a dynamic security verification system is constructed during the federated training process, which not only enhances the model's defense capabilities against security threats such as data tampering and man-in-the-middle attacks but also realizes model self-optimization through the traceability accuracy feedback mechanism. The systematic integration of these technical means enables the constructed distributed fault evaluation forest to possess core capabilities such as cross-regional data security sharing, multi-source heterogeneous data fusion analysis, and complex fault correlation reasoning, providing trustworthy, efficient, and interpretable intelligent decision-making support for the collaborative operation and maintenance of large-scale industrial equipment groups.

[0064] Embodiment 2 Please refer to Figure 2 , another embodiment provided by the present invention: An aluminum processing equipment health status evaluation system, including: an evaluation tree module, a monitoring and evaluation module, and an inference and mapping module; The evaluation tree module obtains a distributed fault evaluation tree based on the cross-plant equipment attributes and operation information by combining the Merkle tree and the evaluation algorithm; The monitoring and evaluation module conducts a distributed edge interaction state evaluation based on the distributed fault evaluation tree in combination with a preset cross-plant real-time monitoring network to obtain a single device evaluation result and a process-related device evaluation result; The inference and mapping module conducts fault traceability and maintenance strategy generation and sharing based on the single device evaluation result and the process-related device evaluation result in combination with the root cause traceability early warning rules and strategy generation library configured by the distributed fault evaluation forest, and visualizes the traceability and sharing process in combination with the traceability early warning visual mapping space to obtain a visual fault evaluation early warning tree.

[0065] Embodiment 3 An electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, a method for evaluating the health state of an aluminum processing device is implemented.

[0066] A computer-readable storage medium stores computer instructions, and when the computer instructions run, a method for evaluating the health state of an aluminum processing device is executed.

[0067] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative rather than restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments without departing from the spirit and scope protected by the claims of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A method for evaluating the health status of aluminum processing equipment, characterized in that, Including: Based on the equipment attributes and operation information across factory areas, combined with the Merkle tree and the evaluation algorithm, a distributed fault evaluation forest is obtained; Based on the distributed fault evaluation forest and combined with a preset cross-factory real-time monitoring network, a distributed edge interaction status evaluation is carried out to obtain a single device evaluation result and a process-related device evaluation result; Based on the single device evaluation result and the process-related device evaluation result, combined with the root cause tracing warning rule and strategy generation library configured by the distributed fault evaluation forest, fault tracing and maintenance strategy generation and sharing are carried out, and the tracing and sharing process is visualized by combining with the tracing warning visualization mapping space to obtain a visualized fault evaluation warning forest.

2. The health status evaluation method of an aluminum processing device according to claim 1, characterized in that The distributed fault evaluation forest includes M regional distributed fault evaluation trees; The regional distributed fault evaluation tree includes N distributed fault evaluation subtrees; The construction process of the regional distributed fault evaluation tree includes: Based on the monitoring device location points, device physical attributes, single-state operation information configured for all devices within a single factory area, and the data flow relationship between each device, the monitoring nodes and the directed link relationship of the real-time monitoring subnet corresponding to the single factory area are constructed; Based on each monitoring node in the real-time monitoring subnet, an evaluation node corresponding to each device within the current factory area is mapped one by one; Based on the data flow relationship between each device, the production process-related operation information corresponding to different devices, and the historical device-related fault information, through an association algorithm combined with the Bayesian algorithm, the fault causal association degree and the corresponding confidence level between each device are obtained; Based on the reciprocal of the fault causal association degree between each device, a directed causal connection of each device is constructed, and the directed fault causal association connection between each device is visually mapped by using the corresponding confidence level and the visual mapping constructed by the preset HSV space to obtain a visualized directed causal connection; Based on the evaluation node corresponding to each device and the visualized directed causal connection between each device, combined with the blockchain, a horizontal production process-related evaluation chain based on the production process is obtained.

3. The health status assessment method of an aluminum processing device according to claim 2, characterized in that, The construction process of the regional distributed fault evaluation tree further includes: Taking the evaluation node corresponding to each device as the generation root node of each distributed fault evaluation subtree, based on the device state information and interaction information obtained in real time by the monitoring node corresponding to the generation root node, through the Merkle tree algorithm built in the monitoring node, a time information hash block sequence corresponding to each generation root node is obtained; Each time information hash block corresponding to each generation root node is used to store the device operation state information, physical attribute information, interaction information, and the corresponding characteristic hash value of each monitoring node within a preset operation time interval; Based on the time information hash block sequence corresponding to each generation root node, through the evaluation algorithm built in the generation root node, the device state evaluation score of each generation root node under each time information hash block is obtained; The device state evaluation score of each generation root node under each time information hash block is combined with a preset fault discrimination score threshold and a fault type library to obtain whether there is a fault in the corresponding device under the current time information hash block; If it exists, obtain the corresponding fault level and fault type under the current time information hash block, and use the corresponding fault level and fault type under the current time information hash block to construct the abnormal state nodes of the corresponding distributed fault evaluation subtree under the current time information hash block. If it does not exist, construct normal state nodes; Repeat the state node construction process to obtain the set of abnormal state nodes and the set of normal state nodes of each generated root node under different time information hash blocks.

4. The health status assessment method of an aluminum processing device according to claim 3, wherein, The state nodes are used to store the operating status and characteristic hash values corresponding to each time information hash block of the corresponding device; The operating status includes normal status, fault status and corresponding fault timestamps; The fault status includes fault type and corresponding abnormal degree score and level; The normal status is represented by 0; The fault status is jointly represented by the corresponding fault type and fault level value; The construction process of the regional distributed fault evaluation tree further includes: If there is only one type of fault in the current set of abnormal state nodes, use the generated root node as the initial abnormal state node, and construct the longitudinal time connection between the adjacent two abnormal state nodes corresponding to each generated root node through the total length of the operating time intervals and the corresponding characteristic hash values of the normal state nodes included between the adjacent two abnormal state nodes; If there is more than one type of fault in the current set of abnormal state nodes, construct the longitudinal time connection between the adjacent abnormal state nodes corresponding to the new type of fault according to the total length of the operating time intervals and the corresponding characteristic hash values of all the abnormal state nodes and normal state nodes corresponding to the non-new type of fault between the first abnormal state node corresponding to the new type of fault and the generated root node.

5. The health status evaluation method of an aluminum processing device according to claim 4, characterized in that The construction process of the regional distributed fault evaluation tree further includes: When the current type of fault appears again, construct the abnormal state nodes corresponding to the fault type and the corresponding longitudinal time connection under the first abnormal state node; When a new type of fault appears again, repeat the process of generating longitudinal time connections for more than one type of fault, and generate new abnormal state nodes and longitudinal time connections under the current generated root node; Construct the corresponding distributed fault evaluation subtree under the current generated root node based on the abnormal state nodes corresponding to different fault types and the corresponding longitudinal time connections; Construct a regional distributed fault evaluation tree through a distributed framework based on the distributed fault evaluation subtrees corresponding to different generated root nodes, the horizontal production process association evaluation chains corresponding to different time information hash blocks, and the verification hashes between the state nodes with interaction relationships between different distributed fault evaluation subtrees.

6. The health status evaluation method of an aluminum processing device according to claim 5, characterized in that, The construction process of the regional distributed fault evaluation tree further includes: Based on the regional distributed fault evaluation tree, through the real-time time information hash blocks obtained by different monitoring nodes in the current plant area, combined with the preset confidence threshold, fault discrimination score threshold and fault type library corresponding to the fault causal correlation degree between devices, train to obtain the regional distributed fault evaluation tree that has been trained in the current plant area; When a distributed fault evaluation subtree corresponding to a new device appears, repeat the above process to update and train the current regional distributed fault evaluation tree.

7. The health status evaluation method of an aluminum processing device according to claim 6, characterized in that The construction process of the distributed fault assessment forest includes: Store the feature hash values and operating status values in all state nodes of the regional distributed fault assessment tree corresponding to each plant area, and the verification hashes between state nodes with interaction relationships between different distributed fault assessment sub-trees, into the interactive storage block constructed by combining the root node generated corresponding to each regional distributed fault assessment tree with the blockchain. At the same time, based on the interactive storage block corresponding to each regional distributed fault assessment tree, construct a block directed connection according to the interactive relationship of the production process flows corresponding to different plant areas; Based on the feature hash values and verification hashes stored in the interactive storage block corresponding to each regional distributed fault assessment tree, obtain the cross-regional traceability verification proof between interactive storage blocks with interaction relationships through the zero-knowledge proof algorithm; Map the cross-regional traceability verification proof into the block directed connection between the corresponding interactive storage blocks to construct a cross-regional directed block traceability verification chain.

8. The health status evaluation method of an aluminum processing device according to claim 7, characterized in that, The construction process of the distributed fault assessment forest further includes: Take each regional distributed fault assessment tree as a federated interactive edge node, and at the same time take the cross-regional directed block traceability verification chain as the cross-regional security traceability verification path, and map the visualization mapping in the visual directed causal connection into the corresponding block directed connection; Based on the federated interactive edge node, the cross-regional security traceability verification path, the preset verification attack library, the preset traceability accuracy threshold and the security verification threshold, and combine with the federated algorithm for training to obtain the trained distributed fault assessment forest.

9. The method for evaluating the health status of an aluminum processing device according to claim 8, wherein, The construction process of the visual fault assessment and early warning forest includes: Based on the trained distributed fault assessment forest combined with the real-time monitoring network, monitor and evaluate all plant area equipment in real time to obtain the regional distributed fault assessment tree with abnormal status; Based on the horizontal production process association assessment chain and the distributed fault assessment sub-tree in the currently abnormal regional distributed fault assessment tree, trace the internal abnormal root causes and associated abnormal nodes in the plant area, and according to the corresponding fault causal association degree and the corresponding confidence level between the abnormal root causes and the associated abnormal nodes, combine with the preset early warning level for corresponding level traceability early warning, and visualize the traceability process according to the visual directed causal connection; At the same time, based on the trained distributed fault assessment forest, through the cross-regional directed block traceability verification chain, conduct cross-plant area security traceability for the distributed fault assessment tree with the current abnormal status and the associated distributed fault assessment trees with fault causal relationships, and use the block directed connection combined with the visualization mapping, the horizontal production process association assessment chain and the distributed fault assessment sub-tree in the associated distributed fault assessment tree to conduct cross-plant area associated fault equipment traceability and visualization of the traceability process, and use the fault causal association degree and the corresponding confidence level corresponding to each connection relationship in the corresponding traceability path to combine with the preset early warning level for cross-plant area difference early warning to obtain the visual fault assessment and early warning forest; The warning levels include a main warning level and an associated warning level; the main warning level corresponds to the main faulty device monitored at the current moment; the associated warning level corresponds to the associated faulty device at the current moment; The main warning level issues corresponding-level warnings through the fault level corresponding to the main faulty device; the associated faulty device issues corresponding-level warnings based on the fault level corresponding to the associated faulty device and the fault causal correlation degree and confidence between it and the main faulty device.

10. A health status evaluation system for aluminum processing equipment, which is used to implement the health status evaluation method for aluminum processing equipment described in any one of claims 1-9, characterized in that, It includes: An evaluation tree module, a monitoring and evaluation module, and an inference mapping module; The evaluation tree module obtains a distributed fault evaluation tree based on the equipment attributes and operation information across the plant areas, combined with the Merkle tree and the evaluation algorithm; The monitoring and evaluation module conducts a distributed edge interaction state evaluation based on the distributed fault evaluation tree and the preset real-time monitoring network across the plant areas to obtain the evaluation results of a single device and the evaluation results of process-related devices; The inference mapping module generates and shares fault traceability and maintenance strategies based on the evaluation results of a single device and the evaluation results of process-related devices, combined with the root cause traceability warning rule and policy generation library configured by the distributed fault evaluation forest, visualizes the traceability and sharing process in combination with the traceability warning visualization mapping space, and obtains a visualized fault evaluation warning tree.

Images