An enterprise business report generation method and system based on large models

Abstract

The present invention relates to the technical field of report generation, and particularly to a method and system for generating enterprise business reports based on large models. The method includes the following steps: obtaining component operation status data and production plan data of industrial equipment; dividing the main-slave relationship of internal components of the equipment according to the component operation status data, designating the driving device as the main component, and designating the transmission component and the execution component as the slave components to construct a component transfer chain relationship diagram; monitoring the operation status parameters of the main components in the component transfer chain relationship diagram; extracting the main components whose fluctuation of the operation status parameters exceeds a preset threshold, and measuring the change of the response parameters of the connected slave components; generating abnormal transfer confirmation data according to the change of the response parameters. Based on the abnormal transfer characteristics between components, the present invention realizes early fault identification and propagation path tracking, and discovers the problem source before the fault expands.

Description

Technical Field

[0001] The present invention relates to the technical field of report generation, and particularly to a method and system for generating enterprise business reports based on large models. Background Art

[0002] Large models refer to deep learning models with huge parameter scales and complex computational structures, usually containing billions or even hundreds of billions of parameters. These models are based on the Transformer architecture and, through training with massive amounts of data, possess powerful language understanding and generation capabilities. Enterprise business reports are reports used for management and decision-making within an enterprise, generally including two major categories: financial reports and non-financial reports. Financial reports mainly reflect the financial position and operating results of an enterprise, such as balance sheets, income statements, and cash flow statements; non-financial reports cover the operating data, production data, sales data, etc. of an enterprise, and are used to evaluate the operating efficiency and business performance of the enterprise. The enterprise business report generation method refers to a series of processes and technical means to collect, organize, analyze the financial data, operating data, market data, etc. of an enterprise, and finally present them in the form of reports.

[0003] However, the enterprise business report generation method often has the following problems during equipment maintenance: Equipment maintenance is often a passive response. In most cases, problems can only be discovered after a failure occurs, resulting in unplanned downtime and high maintenance costs. Large industrial equipment (such as wind turbines and large metallurgical equipment) consists of tens of thousands of components, and the performance states of each component affect each other, forming complex fault evolution paths. Traditional maintenance reports are mostly based on single-index threshold monitoring and are difficult to identify system-level fault precursors. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a method and system for generating enterprise business reports based on large models to solve at least one of the above technical problems.

[0005] To achieve the above object, a method for generating enterprise business reports based on large models includes the following steps:

[0006] Step S1: Obtain the component operation status data and production plan data of industrial equipment;

[0007] Step S2: Divide the master-slave relationship of the internal components of the equipment according to the component operation status data, designate the driving device as the main component, and designate the transmission component and the execution component as the slave components, and construct a component transfer chain relationship diagram;

[0008] Step S3: Monitor the operating status parameters of the main components in the component transfer chain relationship diagram; extract the main components whose fluctuations in the operating status parameters exceed the preset threshold, and measure the changes in the response parameters of the connected slave components; generate abnormal transfer confirmation data based on the changes in the response parameters;

[0009] Step S4: Conduct a time interval pattern analysis based on a large model between the main component anomaly and the slave component anomaly according to the abnormal transfer confirmation data, and combine the preset historical failure data with the estimated running time required for the slave component to reach the maintenance critical state from the current state to obtain the component maintenance window data;

[0010] Step S5: Perform time matching between the component maintenance window data and the production plan data, and identify the production gap period or low-load period to obtain the maintenance time data; generate an equipment maintenance report including the component status signal light diagram, transfer chain map, maintenance time, and maintenance material list according to the maintenance time data and the component transfer chain relationship diagram.

[0011] The present invention realizes the intelligent optimization of equipment maintenance by obtaining the component operating status data and production plan data of industrial equipment. First, based on the component operating status data, the master-slave relationship of the internal components of the equipment is divided, the driving device is set as the main component, and the transmission component and the execution component are set as the slave components, thereby establishing the hierarchical structure and logical relationship between the components, and constructing the component transfer chain relationship diagram to accurately track the impact of the change in the operating status on the overall equipment. On this basis, monitor the operating status parameters of the main components, accurately identify the main components whose fluctuations exceed the preset threshold, and measure the changes in the response parameters of the connected slave components to ensure the accuracy of the abnormal transfer situation and generate abnormal transfer confirmation data, improving the control ability of the fault propagation path. Combining the time interval pattern analysis of the large model, calculate the impact degree of the main component anomaly on the slave components, and refer to the historical failure data to speculate the estimated running time required for the slave components to reach the maintenance critical state from the current state, ensuring that the maintenance strategy is forward-looking and feasible. Subsequently, perform time matching between the calculated maintenance window data and the production plan data, intelligently identify the production gap period or low-load period, minimize the impact of maintenance on production to the greatest extent, and improve equipment availability and production continuity. In addition, according to the maintenance time data and the component transfer chain relationship diagram, automatically generate an equipment maintenance report, including the component status signal light diagram, transfer chain map, maintenance time, and maintenance material list, ensuring that maintenance personnel can clearly master the equipment status, fault propagation path, and resources required for maintenance, improving the efficiency and accuracy of maintenance work. Overall, this method can optimize the maintenance timing while ensuring the normal operation of the equipment, reduce the downtime risk caused by sudden equipment failures, reduce unnecessary maintenance costs, and enhance the intelligent and visual level of equipment operation and maintenance management.

[0012] The present invention also provides an enterprise business report generation system based on a large model, which is used to execute the above-mentioned enterprise business report generation method based on a large model. The enterprise business report generation system based on a large model includes:

[0013] A data acquisition module, which is used to obtain the component operation status data and production plan data of industrial equipment;

[0014] A component relationship division module, which is used to divide the master-slave relationship of the internal components of the equipment according to the component operation status data, designate the driving device as the master component, designate the transmission component and the execution component as the slave components, and construct a component transfer chain relationship diagram;

[0015] An anomaly detection module, which is used to monitor the operation status parameters of the master components in the component transfer chain relationship diagram; extract the master components whose fluctuations of the operation status parameters exceed the preset threshold, and measure the changes in the response parameters of the connected slave components; generate anomaly transfer confirmation data according to the changes in the response parameters;

[0016] A maintenance window analysis module, which is used to perform a time interval pattern analysis based on a large model between the master component anomaly and the slave component anomaly according to the anomaly transfer confirmation data, and combine the preset historical failure data with the estimated running time required for the slave components to reach the maintenance critical state from the current state, to obtain the component maintenance window data;

[0017] A maintenance plan generation module, which is used to perform time matching between the component maintenance window data and the production plan data, and identify the production gap period or low load period to obtain the maintenance time data; generate an equipment maintenance report including a component status signal light diagram, a transfer chain map, the maintenance time, and a maintenance material list according to the maintenance time data and the component transfer chain relationship diagram.

[0018] The present invention accurately obtains the component operation status data and production plan data of industrial equipment through the data acquisition module, providing a reliable raw data basis for subsequent analysis and decision-making. The component relationship division module analyzes the operation status data of each component inside the equipment, reasonably sets the driving device as the main component, and divides the transmission and execution components into subordinate components, forming a component transfer chain relationship diagram. This operation helps to clarify the dependency relationships between components, thereby improving the accuracy of fault detection. The anomaly detection module monitors the operation status of the main component, promptly identifies the main component whose operation parameters fluctuate beyond the threshold, analyzes the changes in the response parameters of the connected subordinate components, and generates anomaly transfer confirmation data, providing a basis for fault location and diagnosis. The maintenance window analysis module analyzes the anomaly transfer time interval pattern between the main and subordinate components through a large model based on the anomaly transfer confirmation data, and combines historical fault data to accurately predict the time required for the subordinate component to reach the maintenance critical state, providing an accurate window period for equipment maintenance and reducing the risk of premature or late maintenance. The maintenance plan generation module further docks the maintenance window data with the production plan data to ensure that maintenance activities can be carried out during production breaks or low-load periods, avoiding interference with production. At the same time, the equipment maintenance report generated based on the component transfer chain relationship diagram, including the component status signal light diagram, transfer chain map, maintenance time, and maintenance material list, not only intuitively presents the health status of the equipment but also provides specific work plans for maintenance personnel, making the maintenance process more efficient, accurate, controllable, and maximizing the operation benefits of the equipment. Description of the Drawings

[0019] Other features, objects, and advantages of the present invention will become more apparent by reading the detailed description of the non-limiting embodiments with reference to the following drawings:

[0020] Figure 1 It is a schematic flow chart of the steps of a method for generating an enterprise business report based on a large model of the present invention;

[0021] Figure 2 For Figure 1 a detailed step flow chart of step S1 in Detailed Embodiments

[0022] The technical method of the present invention will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative work belong to the protection scope of the present invention.

[0023] In addition, the attached drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0024] It should be understood that although terms such as "first" and "second" may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0025] To achieve the above object, please refer to Figures 1 to 2 , the present invention provides a method for generating enterprise business reports based on a large model, and the method includes the following steps:

[0026] Step S1: Obtain the component operation status data and production plan data of industrial equipment;

[0027] In the injection molding equipment monitoring system of the embodiment of the present invention, a PCB 352C33 vibration sensor is deployed to monitor the vibration characteristics of the main motor EM-001 (37kW), a Pt100 temperature sensor is used to measure the temperature, and a LEM HAS400-S current sensor is used to monitor the current; pressure and flow sensors are installed on the hydraulic pump HP-001 (11kW); a pneumatic pressure sensor is installed on the pneumatic valve PA-001. All sensors are connected to the edge server through RS485 communication, and the acquisition frequency is 1 - 60 seconds / time. The system applies a filtering algorithm to process the data, and at the same time extracts the 30-day production plan from the ERP system, including the equipment operation time (three-shift system), product batches (such as daily production of ABS bumpers, injection pressure 21MPa), load rate (recorded hourly), and planned downtime (such as Sunday maintenance). The system integrates the sensor data and the production plan to form a comprehensive data structure.

[0028] Step S2: Divide the master-slave relationship of the internal components of the equipment according to the component operation status data, designate the driving device as the master component, and designate the transmission component and the execution component as the slave components, and construct a component transfer chain relationship diagram;

[0029] In an embodiment of the present invention, the physical connection relationships of 94 components of an injection molding device are extracted from drawings and a BOM list, and the master-slave relationships are divided based on the principle of energy flow. The components with energy input are designated as master components: 5 electrical master components such as the main drive motor EM-001, 2 hydraulic master components such as the main hydraulic pump HP-001, and 3 pneumatic master components such as the pneumatic valve PA-001, totaling 10 master components. The transmission components directly connected to the master components (such as the coupling LC-200 and the reduction gearbox GB-001) are marked as 25 first-level slave components, and the execution components connected to the first-level slave components are marked as 59 second-level slave components. The influence coefficient is calculated by analyzing 3 months of historical data. For example, when the temperature of the main motor rises by 10°C, the temperature of the reduction gearbox rises by 6.7°C, and the influence coefficient is 0.67. The force-directed algorithm is used to construct a component transfer chain relationship diagram, with nodes representing components and connections representing influence relationships (the thickness of the lines represents the strength), and different types of components are distinguished by colors (electrical in blue, hydraulic in green, and pneumatic in yellow).

[0030] Step S3: Monitor the operating state parameters of the master components in the component transfer chain relationship diagram; extract the master components whose fluctuations in the operating state parameters exceed the preset threshold, and measure the changes in the response parameters of the connected slave components; generate abnormal transfer confirmation data based on the changes in the response parameters;

[0031] In an embodiment of the present invention, the monitoring priorities are set according to the energy input of the master components and the number of connected slave components: 5 high-priority components such as the main motor EM-001 (once every 5 seconds), 2 medium-priority components such as the hydraulic pump HP-001 (once every 15 seconds), and 3 low-priority components such as the pneumatic valve PA-001 (once every 30 seconds). The data is smoothed by a sliding window (window of 10 points, weight of the latest point is 0.3). When the temperature of the main motor reaches 80°C (exceeding the standard by 15°C) and the vibration reaches 1.5 mm / s, or the pressure of the hydraulic pump drops to 18.5 MPa (1.5 MPa lower than the rated value), the system marks it as abnormal and records the timestamp. Subsequently, high-frequency sampling is performed on the slave components connected to the abnormal master components (once every 0.5 seconds), and the response characteristics of the slave components 30 seconds before the abnormality, 0 - 60 seconds after the abnormality, and 60 - 120 seconds after the abnormality are analyzed. For example, the temperature of the reduction gearbox rises from 52.3°C to 61.5°C, with a rising rate of 5.5°C per minute. The system determines the response delay time, peak response, and steady-state characteristics, matches them with the fault mode, and generates abnormal transfer confirmation data.

[0032] Step S4: Perform a time interval pattern analysis based on a large model between the master component abnormality and the slave component abnormality according to the abnormal transfer confirmation data, and combine the preset historical fault data with the estimated running time required for the slave component to reach the maintenance critical state from the current state to obtain component maintenance window data;

[0033] In the embodiment of the present invention, a master-slave component abnormal relationship matrix is constructed to determine the abnormal transmission path (such as main motor → coupling → speed reducer → bearing). The transmission time interval is calculated (the speed reducer responds 15 seconds after the main motor is abnormal, and the bearing responds after another 7 seconds), and the relationship with the load is calculated. The time data is input into a large model of the Transformer architecture (8-layer encoder, 12 attention heads), 35 feature vectors are matched with historical faults, and the "motor bearing overheating" mode is identified (matching degree 83%). Considering the environmental factors (workshop temperature of 28°C), the probability is adjusted to 100%. The system analyzes similar cases, considering the current temperature of the speed reducer is 78°C and the vibration is 1.8 mm / s. Under the condition of 85% load, it is predicted that the critical value (temperature of 90°C or vibration of 2.5 mm / s) will be reached after 17.5 hours. Different intervention measures are simulated: reducing the load to 65% can extend it to 30 hours, and increasing cooling can extend it to 36 hours, generating component maintenance window data.

[0034] Step S5: Perform time matching on the component maintenance window data and the production plan data, identify the production gap period or low-load period, and obtain the maintenance time data; generate an equipment maintenance report including a component status signal light diagram, a transmission chain map, the maintenance time, and a maintenance material list according to the maintenance time data and the component transmission chain relationship diagram.

[0035] In the embodiment of the present invention, 17 components to be repaired are scored, considering the remaining time (weight 0.5), the parameter deviation degree (weight 0.3), and the influence range (weight 0.2). Three components such as the speed reducer GB-001 (8.7 points) are listed as urgent repair categories, eight components such as the hydraulic valve HV-004 (6.2 points) are listed as planned repair categories, and six components such as the cooling pump CP-002 (3.5 points) are listed as monitored repair categories. The two-week production plan is extracted from the ERP, and six production gap periods (shutdown > 4 hours) and four low-load periods (< 40% load, > 6 hours) are identified, with a total of 80 hours of maintenance time. The multi-objective optimization algorithm matches the components with time periods, considering time urgency, maintenance windows, production impacts, and resource conflicts, and arranges the urgent components in the recent gap periods. Analyze the maintenance dependency relationship, merge upstream and downstream related components (such as hydraulic pumps and control valves), and optimize the 17 tasks into 9 tasks. Generate a maintenance report, including a component status diagram, a transmission chain map, a maintenance schedule, and a material list.

[0036] Preferably, step S1 includes the following steps:

[0037] Step S11: Deploy an equipment status collector, where the equipment status collector includes a vibration sensor, a temperature sensor, and a current sensor; install the vibration sensor on the surface of the main drive component, install the temperature sensor at the parts of the components prone to heat generation, connect the current sensor to the power input end, and collect the original equipment operation data;

[0038] An embodiment of the present invention uses a rotary kiln in a cement plant (95 meters long, 5.6 meters in diameter, driven by a 2500 kW main motor) as the implementation scenario. Three types of sensors are deployed: PCB 352C33 high-frequency vibration sensors (sensitivity 100 mV / g, ±50 g range), installed on 8 key drive components such as the main motor and reducer, and fixed to the radial and axial positions of the bearing housing with screws; Pt100 platinum resistance temperature sensors (-40°C to 450°C, accuracy ±0.5°C), installed on 12 easily heated parts such as the motor winding and bearing housing; LEMHAS400-S Hall effect current sensors (0 - 400 A range, response <1 μs), connected to the power input of the main motor. The sensors are connected to the data acquisition box through 4 - 20 mA signals or RS485, and different sampling frequencies are set: vibration 1000 Hz, temperature 0.1 Hz, current 100 Hz. The collected data is transmitted to the edge computing server through industrial Ethernet to form an original data stream containing timestamps and device IDs, with a rate of approximately 20 MB / hour.

[0039] Step S12: Perform signal noise and abnormal interference value removal processing on the original device operation data to obtain device operation data;

[0040] The embodiment of the present invention performs differential signal processing on the original data of the rotary kiln: For vibration data, a Butterworth low-pass filter (cutoff frequency 500 Hz) is used to remove high-frequency interference, and then db4 wavelet transform decomposition is used to remove environmental vibration; for temperature data, median filtering (window of 5 points) is used to eliminate mutation points, and exponential weighted moving average (α = 0.2) is used to smooth the trend; for current data, a notch filter (center frequency 50 Hz, bandwidth 5 Hz) is used to eliminate power grid interference, and Kalman filtering is used to remove spikes. The system automatically detects sensor failures: Data with a deviation exceeding 3 times the standard deviation or outside the physical range (such as temperature > 500°C) for 5 consecutive cycles is marked and excluded, and at the same time, a check warning is triggered. The processed data is saved in a structured format, including component ID, parameter type, processed value, and credibility score, and the data availability rate is increased from 89% to 97%.

[0041] Step S13: Classify and organize the device operation data based on component types, and calculate the parameter standard range of each component under normal operating conditions to obtain component operation reference data;

[0042] In the embodiments of the present invention, the data after the rotary kiln cleaning is classified according to component types: electrical drive type (extracting current and temperature parameters), mechanical drive type (extracting vibration and temperature parameters), thermal engineering type (extracting temperature distribution parameters), and hydraulic type (extracting pressure and flow parameters). The normal parameter ranges are established using the historical data during the stable operation period of the equipment (continuous operation > 72 hours, load 80% - 90%): for continuous parameters, the mean μ and standard deviation σ are calculated by statistical methods, and the normal range is defined as [μ - 2σ, μ + 2σ]; for complex parameters, the feature space boundary is established by the PCA method; for discrete parameters, a state transition model is established. For example, the normal range of the main motor bearing temperature is 60 ± 8°C, and the vibration speed is 0.5 - 2.8 mm / s; the temperature of the reducer oil is 55 ± 5°C; the main vibration frequency of the gearbox should be at the meshing frequency (12.5 Hz) and its multiples, and the peak acceleration < 4g. These parameters are integrated into a component operation benchmark database, which includes the standard ranges of each component under different loads (60%, 75%, 90%, 100%).

[0043] Step S14: Compare the component operation benchmark data with the component operation state parameters monitored in real time by the equipment state collector, and mark the components with parameters deviating from the benchmark range to obtain the component operation state data;

[0044] In the embodiments of the present invention, the edge server compares the current rotary kiln equipment data with the benchmark database in real time. The system first identifies the current working conditions (load rate, ambient temperature, running time), retrieves the corresponding benchmark range, and calculates the parameter deviation degree: for continuous parameters, the standardized calculation method (value - mean) / (range width / 2) is used, and values outside [-1, 1] are regarded as exceeding the standard; for complex parameters, the feature vector distance method is used; for discrete parameters, the state transition time and steady-state value are compared. For example, for the main motor bearing temperature of 72°C, the deviation degree is calculated as (72 - 60) / 8 = 1.5, exceeding the normal upper limit; a strong peak of non-meshing frequency appears in the vibration spectrum of the reducer, and the feature vector distance is 0.42 (exceeding the threshold of 0.3). The system classifies according to the deviation degree: 0.8 - 1.2 is a yellow warning, 1.2 - 2.0 is an orange warning, and > 2.0 is a red warning. All deviation situations (component ID, parameter type, current value, standard range, deviation degree, warning level) are recorded as operation state data, updated every 10 seconds, stored in the time series database, and displayed on the monitoring large screen.

[0045] Step S15: Obtain the production plan data, including the planned running time of the equipment, the production load rate, and the planned shutdown maintenance time.

[0046] In the embodiment of the present invention, the production plan of the rotary kiln in the cement plant is obtained through the SAP ERP and MES integrated system. The main production plan for the next 30 days is extracted through the API interface, including cement varieties, output, batches, and time arrangements: the rotary kiln operates throughout the day from Monday to Friday, operates from 8:00 to 20:00 on Saturday, and is shut down for maintenance on Sunday; the load rate is recorded hourly for each period, with a normal production period of 90%-95% and a switching period of 60%-70%; the shutdown maintenance time includes 8 hours of routine maintenance every Sunday and a 24-hour major overhaul once a month (on Tuesday of the third week of the next month); details of each maintenance content, such as routine inspection of the kiln head seal and cleaning of the dust collector, and the major overhaul includes adjustment of the support rollers and inspection of the drive, etc. The data is stored in JSON format, including accurate timestamps, equipment IDs, plan types, load rates, etc. The system monitors the execution of the plan in real time and automatically updates when there are deviations between the actual production and the plan (such as early completion or temporary failures), ensuring that maintenance decisions are based on the latest operation plan.

[0047] In the present invention, by deploying a device status collector, comprehensive monitoring of the operating status of industrial equipment is achieved. Vibration sensors are installed on the surface of the main drive components to capture vibration characteristics during operation, temperature sensors are arranged at the parts of the components prone to heat generation to detect temperature changes in real time, and current sensors are connected to the power input end to monitor the current fluctuations of the equipment, so as to accurately obtain the original equipment operation data. Subsequently, by removing signal noise and abnormal interference values from the collected data, the accuracy and reliability of the data are ensured, avoiding misjudgment from affecting the analysis results. On this basis, the equipment operation data is classified and sorted, and the parameter standard range under the normal operating state is calculated according to the component type to establish the component operation reference data, providing an accurate reference standard. Combining the real-time monitoring results of the device status collector, the current component operation status parameters are compared with the reference data to quickly identify the components with parameter deviation ranges, ensuring that abnormal changes in the equipment operation status can be detected and marked in a timely manner, providing a basis for subsequent maintenance decisions. At the same time, production plan data is obtained, including the planned operation time of the equipment, production load rate, and planned shutdown maintenance time, providing a time management basis for equipment operation and maintenance, combining fault analysis with production scheduling, and achieving the optimal formulation of maintenance strategies, thereby improving the stability of the equipment and production continuity.

[0048] Preferably, step S2 includes the following steps:

[0049] Step S21: Obtain the equipment structure drawing data, and extract the physical connection relationship and energy transfer path information of the components from the equipment structure drawing data, so as to obtain the component physical topology structure data;

[0050] In an embodiment of the present invention, taking an injection molding device as an example, engineering drawings are extracted from the enterprise PLM system, including two-dimensional assembly drawings in AutoCAD DWG format and three-dimensional models in STEP format. Processing is carried out using a CAD parsing tool: a graphic recognition algorithm is used to recognize the component boundaries and connection relationships in DWG; topological analysis is performed on the STEP model to recognize the solid components and spatial position relationships; components are matched with the BOM through coding rules to determine the model and function; a physical connection map between components is established based on physical contact and functional connection. At the same time, the energy transfer path is analyzed, such as electric energy from the power supply through the control cabinet to the motor, through mechanical transmission to the hydraulic system, and finally driving the injection and clamping units. The system generates physical topology structure data including 152 component nodes and 237 connection edges. Each node contains component ID, type, position, etc., and each connection contains connection type, transfer direction, and physical distance. <{

[0051] Step S22: Identify the energy input components in the device from the component physical topology structure data and mark them as main components, thereby obtaining preliminary main component identification data, where the energy input components include motors, hydraulic pumps, and pneumatic devices;

[0052] In an embodiment of the present invention, the energy input components in the injection molding machine are automatically identified through an energy source identification algorithm. This algorithm checks the functional attribute tags of components, filters "driving sources" and "energy converters"; analyzes the nodes in the topology diagram with energy input but no upstream source; and verifies in combination with the equipment professional knowledge rule base. Three main types of energy input components are identified: electrical (1 main drive motor of 37 kW, 4 auxiliary motors of 2.2 kW, 6 servo motors); hydraulic (2 main hydraulic pumps of 21 MPa, 3 auxiliary hydraulic pumps of 14 MPa); pneumatic (4 pneumatic control valve groups of 0.8 MPa). Each component is assigned a unique identifier (such as EM-001, HP-002, PA-003) and key parameters (power, pressure, flow rate) are recorded, forming a preliminary main component identification data table containing 17 main component records.

[0053] Step S23: Determine the transmission components directly connected to the main components based on the energy flow direction in the preliminary main component identification data, and mark them as first-level slave components, thereby obtaining first-level slave component association data, where the transmission components include gearboxes, pulleys, and bearings;

[0054] In the embodiment of the present invention, based on the preliminary main component identification data, the system analyzes the energy flow direction to determine the primary slave components. For electrical main components, the transmission devices are identified along the mechanical connection path, such as the coupling LC-200, the reduction gearbox (10:1), and the pulley (3:1) connected to the main motor EM-001; for hydraulic main components, the valves are traced along the hydraulic pipeline, such as the 25 MPa direction control valve and the 200 L / min flow control valve connected to the main hydraulic pump HP-001; for pneumatic main components, the air circuit is analyzed to identify the distributors and amplifiers. Verification is carried out in combination with the physical position and function, and the connection relationship and characteristics are recorded, such as the main motor and the reduction gearbox are "mechanically directly connected" with an efficiency of 98% and a delay of <10 ms; the hydraulic pump and the remote control valve are "hydraulically pipeline connected" with an efficiency of 85% and a delay of 50 - 100 ms. The system generates the primary slave component association data including 42 transmission components.

[0055] Step S24: Determine the execution components connected to the primary slave components along the energy transfer path according to the primary slave component association data, and mark them as secondary slave components, so as to obtain the master-slave component hierarchical data, where the execution components include actuators, working heads, and functional components;

[0056] In the embodiment of the present invention, the secondary slave components are determined along the energy transfer path according to the primary slave component association data. For the electrical-mechanical path, the execution components connected to the reduction gearbox are identified, such as the injection screw and the platen drive system; for the hydraulic path, the hydraulic cylinders and actuators connected to the control valves are identified; for the pneumatic path, the cylinders and the ejector pin system connected to the pneumatic distributor are identified. The system records the specifications, positions, and functional parameters of the secondary slave components, such as the maximum force, stroke, and response characteristics of the injection actuator. The energy transfer characteristics are analyzed, such as the mechanical efficiency from the reduction gearbox to the injection unit is 92% and the delay is about 25 ms; the efficiency from the hydraulic valve to the hydraulic cylinder is 88% and the delay is about 75 ms. Finally, the master-slave component hierarchical data including the complete hierarchical relationship of the primary and secondary slave components is generated, with a hierarchical structure of a total of 94 key components.

[0057] Step S25: Analyze the correlation of the operating parameters among the components according to the component operating state data, calculate the influence degree of the change of the main component parameters on the slave component parameters, and determine the parameter correlation strength, so as to obtain the component parameter influence matrix;

[0058] In an embodiment of the present invention, the correlation of each component in the operation data of the injection molding machine for 30 days (about 720 hours) is analyzed. The time series data is divided into 10-minute windows (step size 1 minute), generating 43,200 analysis windows. The change points of the main component parameters are marked, such as the main motor current increasing from 35 A to 48 A, and the responses of the slave components are traced, such as the temperature of the reduction gearbox increasing from 45 °C to 53 °C. The parameter response characteristics are calculated through statistical analysis: the average delay time (the average response of the reduction gearbox temperature to the change in the motor current is 15 minutes), the response amplitude ratio (a 37% increase in the motor current causes a 17.8% increase in the reduction gearbox temperature), and the response consistency (the frequency of this relationship). The influence coefficient is calculated: the change amplitude of the slave component / the change amplitude of the main component × the inverse delay factor × the consistency score. For example, the influence coefficient of the main motor on the reduction gearbox temperature = 17.8% / 37% × (1 / (15 / 60 + 0.1)) × 0.85 = 0.67. A 94×94 component parameter influence matrix is formed, and the element values range from 0 to 1, representing the influence intensity.

[0059] Step S26: Construct a visual component transfer chain relationship diagram according to the master-slave component hierarchical data and the component parameter influence matrix, where the nodes of the component transfer chain relationship diagram represent each component, the connecting lines represent the influence relationships between components, and the line thickness represents the influence intensity.

[0060] In an embodiment of the present invention, a component transfer chain relationship diagram of the injection molding machine is constructed. In the graphics engine, 94 component nodes are arranged according to the physical location and hierarchical relationship: the main component is at the top, the first-level slave components are in the middle, and the second-level slave components are at the bottom. The size of the nodes reflects the importance. For example, the main motor EM-001 is 100 pixels, and the auxiliary valve PA-004 is 60 pixels. Connecting edges are added to represent the energy flow direction, and the line thickness is adjusted according to the influence coefficient: 0.7 - 1.0 is 5 pixels, 0.4 - 0.7 is 3 pixels, 0.2 - 0.4 is 1.5 pixels, and <0.2 is a dotted line of 0.5 pixels. For example, the connection line from the main motor to the reduction gearbox is 4.5 pixels (influence coefficient 0.67). Color coding is used to distinguish component types: electrical components are blue, hydraulic components are green, and pneumatic components are yellow; connection types: mechanical connections are black, hydraulic connections are green, pneumatic connections are yellow, and signal connections are red. Finally, an interactive chart in SVG or HTML5 format is generated, which supports zooming, viewing details, and filtering display, providing a visual tool for fault analysis.

[0061] The present invention analyzes the data of the device structure drawings, extracts the physical connection relationships and energy transfer path information of the components, and accurately obtains the physical topology of the components. Based on the topology, the energy input components in the device, such as motors, hydraulic pumps, and pneumatic devices, are identified and marked as main components to ensure a correct understanding of the energy transfer process of the device. On this basis, according to the energy flow direction, the transmission components directly connected to the main components, such as gearboxes, pulleys, and bearings, are identified and marked as primary slave components to make the relationship between the main and slave components clearer. Further along the energy transfer path, the execution components connected to the primary slave components are determined, including actuators, working heads, and functional components, and are marked as secondary slave components to form a complete hierarchical relationship of the main and slave components. Combining the component operation status data, a correlation analysis is performed on the operation parameters between the components, the influence degree of the change of the main component parameters on the slave components is calculated, the dependency relationship between the components is quantified, and a component parameter influence matrix is obtained. Finally, based on the hierarchical relationship of the main and slave components and the parameter influence matrix, a visual component transfer chain relationship diagram is constructed, with nodes representing each component, connection lines representing the influence relationships between the components, and the thickness of the lines reflecting the influence intensity, providing intuitive data support for device operation monitoring, fault tracing, and maintenance optimization, and improving the accuracy of device status analysis.

[0062] Preferably, step S25 includes the following steps:

[0063] Step S251: Extract the time series of the operation parameters of each component under different working conditions from the component operation status data, so as to obtain the historical change data of the component parameters;

[0064] In the embodiment of the present invention, the time series of the operation parameters of each component under different working conditions are extracted from the data acquisition platform of the injection molding equipment. The continuous production data of the most recent 30 days are selected, covering various working conditions with different product models and process parameters. For main components such as motors, high-frequency sampling at 1 Hz is used, and for parameters with slow changes such as temperature, low-frequency sampling at 0.1 Hz is used. The original data is exported, including the current (1 Hz, ±0.1 A), temperature (0.1 Hz, ±0.5 °C), and vibration (10 Hz, ±0.01 mm / s) of the main motor EM-**001**, the pressure (1 Hz, ±0.1 MPa) and flow rate (1 Hz, ±0.5%) of the main hydraulic pump HP-**001**, and the air pressure and flow rate of the main pneumatic valve PA-**001** and other parameters. The data is preprocessed: outliers are detected using the 3σ rule, missing values are filled using linear interpolation, and noise is removed using a Butterworth low-pass filter. The processed data is organized into a standard time series, including time stamps, component IDs, parameter types, parameter values, and working condition identifiers. For example, the current data of the main motor when producing ABS products and the mold temperature is 80 °C contains approximately 2.6 million sampling points, forming a complete historical change curve. Finally, a parameter historical change data set is generated for 94 key components, with a total volume of approximately 500 GB.

[0065] Step S252: Segment the data of the main component in the historical change data of component parameters, identify the time points when the parameters change significantly, and use the significantly changed time points as reference time nodes, so as to obtain the main component parameter change marking data;

[0066] In the embodiment of the present invention, the historical data of the main component parameters is segmented and the significant change points are identified. Taking the main motor EM-001 as an example, it is segmented by production batches first, and each batch contains a complete injection molding cycle (20 - 35 seconds). The CUSUM cumulative sum algorithm is used to detect the trend change points, the target average value is set as the average value of normal operating parameters, the sensitivity parameter is 0.5 times the standard deviation, and the threshold is 5 times the standard deviation. At the same time, the sliding window method (window 5 seconds) is used to calculate the parameter change rate. When the current change rate exceeds 1 A / second or the temperature change rate exceeds 0.5 °C / second, it is marked as a change point. The system calculates the percentage of the difference between the stable values before and after the change in the value before the change as the change amplitude. The average value of the 5 seconds before the change point is taken before the change, and the average value of the 5 - 10 seconds after the change point is taken after the change. The key change points with a current change amplitude exceeding 10% or a temperature change amplitude exceeding 5% are screened out. For example, in a certain injection molding cycle, it is detected that the current of the main motor rises from 32 A to 46 A at t = 128356 seconds (change amplitude 43.8%), which is recorded as a key reference time node. The system identifies approximately 12,800 significant change points in 30 days of data, forming the main component parameter change marking data set.

[0067] Step S253: Analyze the parameter changes of the slave component before and after the corresponding time points according to the main component parameter change marking data, and calculate the time delay and response amplitude between the parameter changes of the slave component and the main component parameter changes, so as to obtain the parameter response characteristic data between components;

[0068] In the embodiment of the present invention, based on the main component change marker data, the parameter response characteristics of the slave components are analyzed. For each main component change point, a list of directly connected slave components is extracted from the slave component transfer chain diagram, and an observation window (10 seconds before to 120 seconds after the main component change point) is defined. Search for the parameter change pattern of the slave components within the window, and use the pattern matching algorithm to detect the parameter trend change. When the parameter difference signs of three consecutive time points are the same and non-zero, it is determined as an effective trend. Calculate the time difference between the slave component change point and the main component change point as the delay time, and calculate the percentage of the parameter value difference before and after the slave component change (5 seconds before and 5 seconds after the change point) to the value before the change as the response amplitude. Record the response consistency index, that is, the percentage of the number of samples with detected response relationships to the total number of samples. For example, after the current of the main motor EM-001 increases from 32A to 46A (43.8%), the temperature of the gearbox GB-001 rises from 42°C to 48°C (14.3%) after 15 seconds, and the consistency of this response in the samples reaches 85%. The system generates approximately 25,000 main-slave component parameter response records in total, forming the parameter response characteristic data between components.

[0069] Step S254: Calculate the parameter influence coefficient between each pair of main-slave components according to the parameter response characteristic data between components, so as to obtain the component influence coefficient matrix, where the parameter influence coefficient = (the change amplitude of the slave component parameter / the change amplitude of the main component parameter) × 1 / delay time;

[0070] In the embodiment of the present invention, based on the parameter response characteristic data, the influence coefficient between the main-slave components is calculated and a matrix is constructed. The influence coefficient calculation considers three factors: the ratio of the change amplitude of the slave component parameter to the change amplitude of the main component (transfer efficiency), the inverse factor of the time delay (the shorter the delay, the more direct the influence), and the response consistency percentage (to avoid overestimating accidental associations). The system sets the time delay influence parameter to 0.05 and the parameter to prevent the denominator from being zero to 0.1. Filter the valid samples with a response consistency greater than 50%, and take the average value of multiple measurements to reduce random fluctuations. For example, the calculation of the influence coefficient of the main motor EM-001 on the gearbox GB-001: the temperature change of the gearbox is 12% ÷ the current change of the main motor is 35% × 1 / (15 seconds delay × 0.05 + 0.1) × 85% consistency ≈ 0.55. The system calculates the influence coefficient for each pair of 94 key components, forming a 94×9,4 matrix, where most of the elements are 0 (no direct influence between the two components), and the non-zero elements are concentrated between the connected component pairs, and the matrix sparsity is about 85%.

[0071] Step S255: Normalize the component influence coefficient matrix to obtain the standard matrix of the component influence coefficient;

[0072] In the embodiment of the present invention, the component influence coefficient matrix is normalized to make different types of parameters comparable. First, global maximum-minimum normalization is performed. Each influence coefficient is subtracted by the minimum value (0.05) of the non-zero elements of the matrix, and then divided by the difference between the maximum value (0.92) and the minimum value, mapping all coefficients to the range of 0-1. Then, correction coefficients are applied to different types of component pairs: the electrical-mechanical relationship is multiplied by 1.0, the electrical-hydraulic relationship is multiplied by 1.2 (the hydraulic response is slower and the influence is often underestimated), and the hydraulic-pneumatic relationship is multiplied by 1.1. Considering the differences in component importance, weights are assigned to each component according to its position and function in the transmission chain. For example, the weight of the main motor is 1.0, and the weight of the auxiliary cooling pump is 0.7. The influence coefficient is adjusted by multiplying the geometric mean of the weights of the two components to highlight the influence of key components. Finally, fine-tuning is performed to ensure that important physical associations are not underestimated due to numerical calculations, generating a standard matrix of component influence coefficients with a value range of 0-1.

[0073] Step S256: Screen the component pairs with influence coefficients greater than the threshold from the standard matrix of component influence coefficients according to the preset influence threshold, so as to obtain the component parameter influence matrix.

[0074] In the embodiment of the present invention, hierarchical thresholds are set based on the standard matrix of influence coefficients for screening: the threshold for the first-level component pair (the main component and the directly connected slave component) is 0.2, the threshold for the second-level component pair (separated by one intermediate component) is 0.3, and the threshold for the third-level and above component pairs is 0.4. The system determines the level relationship of each pair of components through the shortest path in the component transmission chain relationship diagram, and then compares it with the corresponding threshold. For example, the main motor EM-001 and the reduction gearbox GB-001 are a first-level component pair, and the standardized influence coefficient 0.78 is greater than the threshold 0.2, so it is retained in the final matrix; while the main motor and the distal actuator AC-008 are a third-level component pair, and the influence coefficient 0.35 is less than the threshold 0.4, so it is set to zero in the final matrix. The sparsity of the screened component parameter influence matrix is increased to about 95%, only retaining the most significant component influence relationships in the equipment. The final matrix is stored in the compressed sparse row format and exported as JSON or CSV format for easy use in other analysis systems.

[0075] The present invention provides a basis for subsequent analysis by extracting the time series of operating parameters of components under different working conditions and constructing the historical change data of component parameters. The parameter data of the main component is segmented to identify the time points at which the parameters change significantly, and these time points are used as reference nodes to accurately locate key state changes. Based on the parameter change marking data of the main component, the parameter changes of the slave component before and after the corresponding time point are analyzed, the response time delay and amplitude of the slave component to the change of the main component are calculated, the dynamic correlation relationship between components is quantified, and the parameter response characteristic data between components is obtained. On this basis, the parameter influence coefficient between the master and slave components is calculated to numerically measure the influence degree of the change of the main component on the slave component, and a component influence coefficient matrix is formed. Subsequently, the component influence coefficient matrix is normalized to enable comparison of data of different magnitudes on the same scale and improve the accuracy of calculation. Finally, according to a preset influence threshold, component pairs with influence coefficients exceeding the threshold are selected from the standardized influence coefficient matrix to generate a component parameter influence matrix, which accurately reflects the dependence relationship between the master and slave components and provides data support for equipment status monitoring, fault propagation analysis, and predictive maintenance.

[0076] Preferably, the operating state parameters of the main component in the monitoring component transfer chain relationship diagram described in step S3 include: [[ID=X]]

[0077] Determine the monitoring priority of the main component according to the component transfer chain relationship diagram, where the main component with an energy input exceeding 15 kW or the number of connected slave components exceeding 5 is set as high priority, the main component with an energy input between 5 kW and 15 kW or the number of connected slave components between 3 and 5 is set as medium priority, and the remaining main components are set as low priority;

[0078] Perform time-division polling sampling on each main component based on the monitoring priority of the main component to obtain the original data of the operating state of the main component;

[0079] Perform smoothing processing based on a sliding window on the original data of the operating state of the main component. The window size is the first 10 sampling values, and weighted moving average processing is performed to obtain the smoothed data of the operating state of the main component. The weighted moving average processing is specifically that the weight of the most recent time point is 0.3, and the sum of the weights of the remaining 9 historical points is 0.7;

[0080] When the temperature deviation of the main component of the motor type in the smoothed data of the operating state of the main component exceeds 15 °C, the vibration deviation exceeds 0.8 mm / s, or the current deviation exceeds 12%, or when the pressure deviation of the main component of the hydraulic pump type exceeds 1.5 MPa, the flow deviation exceeds 10%, or the noise deviation exceeds 8 dB, or when the air pressure deviation of the main component of the pneumatic device type exceeds 0.1 MPa, the flow deviation exceeds 15%, or the valve response time extends by more than 30 ms, record the time point and mark the corresponding main component as an abnormal state to obtain the abnormal timestamp data of the main component;

[0081] Extract the set of slave components directly connected to the abnormal master component from the component transfer chain relationship diagram based on the abnormal timestamp data of the master component, and obtain the relevant slave component list;

[0082] Merge the smoothed data of the master component running status, the abnormal timestamp data of the master component, and the relevant slave component list into the running status parameters of the master component.

[0083] Based on the component transfer chain relationship diagram of the injection molding equipment in the embodiments of the present invention, the system first scans all 17 identified main components and determines their monitoring priority levels according to the preset priority division rules. The specific operation method is as follows: The system first reads the energy input parameters and the number of connected slave components of each main component from the component attribute database. For example, the rated power of the main drive motor EM-001 is 37kW, and the number of directly connected slave components is 6 (including the coupling LC-200, the reduction gearbox GB-001, etc.); the rated power of the main hydraulic pump HP-001 is 11kW, and the number of directly connected slave components is 4 (including various hydraulic valves and pipeline components); the rated power of the auxiliary pneumatic control valve PA-003 is 2.2kW, and the number of directly connected slave components is 2. The system classifies the main components according to the energy input threshold and the number of connected slave component threshold: When the energy input of the main component exceeds 15kW or the number of connected slave components exceeds 5, it is marked as high priority. For example, a total of 5 components including the main drive motor EM-001 (37kW, 6 slave components) and the main load motor EM-002 (22kW, 5 slave components) are marked as high priority; when the energy input of the main component is between 5kW and 15kW or the number of connected slave components is between 3 and 5, it is marked as medium priority. For example, a total of 7 components including the main hydraulic pump HP-001 (11kW, 4 slave components) and the auxiliary drive motor EM-003 (7.5kW, 3 slave components) are marked as medium priority; the remaining main components are marked as low priority. For example, a total of 5 components including the auxiliary pneumatic control valve PA-003 (2.2kW, 2 slave components) and the small auxiliary motor EM-006 (1.1kW, 1 slave component) are marked as low priority. Based on the monitoring priority of the main components, the system sets different data sampling frequencies: The sampling frequency of the high-priority main components is set to 5 seconds per time, the sampling frequency of the medium-priority main components is set to 15 seconds per time, and the sampling frequency of the low-priority main components is set to 30 seconds per time. The system develops a polling scheduler to sample each main component orderly according to the preset sampling schedule, avoiding network congestion caused by sending data requests to multiple sensors simultaneously. For different types of main components, the system configures dedicated data acquisition programs: The main components of the motor type mainly collect temperature, current, and vibration parameters, the main components of the hydraulic pump type mainly collect pressure, flow rate, and oil temperature parameters, and the main components of the pneumatic device type mainly collect air pressure, flow rate, and response time parameters. The system also dynamically adjusts the sampling frequency according to the operating state of the components. When the parameters of the main component are close to the warning value (such as the motor temperature rising to 85% of the rated temperature), the sampling frequency is automatically increased. For periods with obvious changes in working conditions (such as the injection stage and the holding pressure stage during the injection molding process), the system temporarily increases the sampling frequency of all main components to capture possible instantaneous abnormalities in the key process. For example, during the injection stage (lasting about 5 seconds) of a certain injection molding cycle, the system increases the sampling frequency of the main drive motor EM-001 from 5 seconds per time to 1 time per second.Through this time-division polling sampling mechanism, the system collected approximately 286,000 pieces of raw data on the operating status of the main components during the three-shift production period. For the acquired raw data, the system applied the sliding window smoothing processing technology to eliminate short-term fluctuations and random noise. The system created a data buffer for each parameter type of each main component, with a buffer size of 10 sampling points. For example, the temperature parameter buffer of the main drive motor EM-001 contained the temperature values of the most recent 10 samplings [65.2°C, 65.5°C, 66.1°C, 66.3°C, 66.8°C, 67.2°C, 67.5°C, 68.1°C, 68.4°C, 68.7°C] at a certain moment. The system applied the weighted moving average algorithm to the data in the buffer. The latest data point was given a weight of 0.3, and the remaining 9 historical data points shared a weight of 0.7 (each approximately 0.078). Specifically in the calculation, the system summed the data points after multiplying them by the corresponding weights. For example, the weighted average of the above temperature data was 67.4°C. For different types of parameters, the system adopted different smoothing processing technologies: for vibration data, in addition to the weighted moving average, frequency domain filtering technology was also applied to remove specific frequency noise; for parameters with obvious periodicity, such as the pressure change in the injection molding cycle, the system adopted a phase-aware smoothing algorithm to retain key periodic features. The system would also detect and process abnormal outliers. When the deviation of a data point from the weighted average exceeded a preset threshold (such as 3 times the standard deviation), this point would be excluded from the smoothing calculation to avoid the influence of sensor failures or interference signals. Based on the smoothed data of the main component operating status, the system identified and marked abnormal states of key parameters. The system read the normal operating range and abnormal thresholds of the parameters of each main component from the component configuration database: for motor-type main components, the standard operating temperature range was the rated temperature ±10°C (the rated temperature of the main drive motor EM-001 was 55°C, and the normal range was 45 - 65°C), the standard vibration range was 0.1 - 0.6 mm / s, and the rated current fluctuation range was ±8%; for hydraulic pump-type main components, the standard pressure fluctuation range was ±5% of the rated pressure (the rated pressure of the main hydraulic pump HP-001 was 21 MPa, and the normal range was 19.95 - 22.05 MPa); for pneumatic device-type main components, the standard air pressure fluctuation range was ±7% of the rated air pressure (the rated air pressure of the main pneumatic valve PA-001 was 0.8 MPa, and the normal range was 0.744 - 0.856 MPa).When the parameter deviates from the normal range by more than the preset threshold, the system marks the corresponding abnormal state: when the temperature of the main drive motor EM-001 reaches 80°C (deviating from the rated temperature by more than 15°C), the vibration value reaches 1.5 mm / s (exceeding the normal upper limit of 0.8 mm / s), or the current deviation reaches 15% (exceeding the normal fluctuation range of 12%), the system marks it as a motor abnormality; when the pressure of the main hydraulic pump HP-001 drops to 18.5 MPa (deviating from the rated pressure by more than 1.5 MPa) or the flow rate drops to 85% of the rated value (deviating by more than 10%), the system marks it as a hydraulic pump abnormality; when the air pressure of the pneumatic valve PA-001 drops to 0.65 MPa (deviating from the rated air pressure by more than 0.1 MPa) or the valve response time extends to 60 ms (exceeding the normal response time of 30 ms), the system marks it as a pneumatic equipment abnormality. The system records the accurate timestamp of the abnormality occurrence, the type of abnormality, the abnormal parameter value, and the deviation degree, forming the main component abnormality timestamp data. For the identified abnormal main components, the system extracts the slave components directly connected to these abnormal main components using the component transfer chain relationship diagram. First, the system reads the main component abnormality timestamp data to obtain all the abnormal main component IDs detected currently; then, queries the pre-constructed component transfer chain relationship diagram and executes the graph traversal algorithm to find all directly connected slave component nodes starting from the abnormal nodes; finally, filters those component pairs with an influence coefficient exceeding 0.4 in the component parameter influence matrix to form a list of related slave components, including information such as slave component ID, type, function description, connection relationship with the main component, and influence coefficient. The system finally integrates the smooth data of the main component operating state, the abnormality timestamp data, and the list of related slave components to form a structured set of main component operating state parameters, including a basic information part, a status data part, and an associated information part, providing comprehensive data support for subsequent abnormality transfer analysis and maintenance decision-making.

[0084] The present invention determines the monitoring priority of the main component through the component transfer chain relationship diagram, enabling the rational allocation of monitoring resources. The main components with an energy input exceeding 15 kW or more than 5 connected slave components are set as high priority to ensure real-time monitoring of critical components; the main components with an energy input between 5 kW and 15 kW or the number of connected slave components between 3 and 5 are set as medium priority to achieve balanced monitoring; the remaining main components are set as low priority, thereby optimizing the sampling frequency. Based on the monitoring priority, time-division polling sampling is adopted to improve the monitoring efficiency and obtain the original data of the main component operation status. By using the smoothing process of the sliding window and the weighted moving average method, the stability and anti-interference ability of the data are improved, and the influence of short-term fluctuations on the monitoring results is reduced. When the deviation of the main component operation status exceeds the set threshold, the abnormal timestamp is recorded in real time and the abnormal status is marked, making the abnormal detection more accurate and reliable. Subsequently, based on the abnormal timestamp data, the set of directly connected slave components is extracted from the component transfer chain relationship diagram to quickly locate the potentially affected slave components and improve the fault tracing ability. Finally, the smoothed data of the main component operation status, the abnormal timestamp data, and the relevant slave component list are integrated to form the complete operation status parameters of the main component, providing accurate data support for subsequent equipment health assessment, fault analysis, and predictive maintenance.

[0085] Preferably, extracting the main components whose fluctuation of the operation status parameters exceeds the preset threshold in step S3 and measuring the change of the response parameters of the connected slave components includes:

[0086] Calculating the main component fluctuation amplitude of the main component operation status parameters;

[0087] Comparing the main component fluctuation amplitude data with the preset fluctuation threshold to obtain the over-threshold main component identification data, where the specific preset fluctuation threshold is that the temperature fluctuation threshold of the main component of the motor type is 8%, and the rotational speed fluctuation threshold is 5%; the pressure fluctuation threshold of the main component of the hydraulic pump type is 12%; the air pressure fluctuation threshold of the main component of the pneumatic device type is 15%;

[0088] Screening the main components in the over-threshold main component identification data based on the physical connection distance and energy transfer efficiency according to the relevant slave component list to obtain the list of slave components to be measured;

[0089] Performing high-frequency sampling on the slave components in the list of slave components to be measured to obtain the original response data of the slave components;

[0090] Performing segmented processing on the original response data of the slave components before, during, and after the fluctuation, and calculating the average value, standard deviation, and change trend of each time period, respectively recording the temperature change slope, vibration spectrum change, and pressure fluctuation characteristics to obtain the response characteristic data of the slave components;

[0091] Perform a parameter change pattern analysis on the component response characteristic data, identify the response delay time, peak response, and steady-state response, and calculate the response decay rate and resonant frequency, so as to obtain the response parameter change situation.

[0092] In the operation monitoring of the injection molding equipment in the embodiments of the present invention, the system first calculates the fluctuation amplitude of the operation state parameters of each main component to identify abnormal parameter changes. The system sets an observation window (5 minutes) for different types of main components, and pays attention to the temperature, speed, and current parameters of motor components, the pressure, flow rate, and efficiency parameters of hydraulic pump components, and the air pressure, flow rate, and response time parameters of pneumatic device components. The maximum and minimum values of the parameters are recorded within the observation window. For example, the maximum temperature of the main drive motor EM-001 is 78°C, and the minimum temperature is 72°C (the standard operating temperature is 70°C). The system calculates the fluctuation amplitude, that is, the percentage of the difference between the maximum and minimum values in the standard operating value: the temperature fluctuation amplitude of the main motor EM-001 is (78 - 72) / 70×100% = 8.57%; the speed fluctuation amplitude is (1580 - 1490) / 1500×100% = 6%; the pressure fluctuation amplitude of the hydraulic pump HP-001 is (22.8 - 19.2) / 21×100% = 17.14%; the air pressure fluctuation amplitude of the pneumatic valve PA-001 is (0.92 - 0.76) / 0.8×100% = 20%. The system analyzes the fluctuation frequency characteristics through fast Fourier transform. For example, the main frequency of the main motor temperature fluctuation is 0.033 Hz (about one cycle every 30 seconds). The system compares the fluctuation amplitude data with the preset threshold, and reads the fluctuation threshold standards of different types of main components from the configuration database: the temperature fluctuation threshold of motor components is 8%, and the speed fluctuation threshold is 5%; the pressure fluctuation threshold of hydraulic pump components is 12%, and the flow rate fluctuation threshold is 10%; the air pressure fluctuation threshold of pneumatic device components is 15%, and the flow rate fluctuation threshold is 12%. When the actual fluctuation amplitude exceeds the threshold, it is marked as a fluctuation anomaly: the main motor temperature fluctuation of 8.57% > 8% is marked as a temperature fluctuation anomaly; the main motor speed fluctuation of 6% > 5% is marked as a speed fluctuation anomaly; the hydraulic pump pressure fluctuation of 17.14% > 12% is marked as a pressure fluctuation anomaly; the pneumatic valve air pressure fluctuation of 20% > 15% is marked as an air pressure fluctuation anomaly. The system records the anomaly type, fluctuation amplitude, degree of exceeding the threshold, and duration of each abnormal component, and classifies them according to the severity level: exceeding the threshold by 0 - 20% is a mild fluctuation anomaly, by 20 - 50% is a moderate fluctuation anomaly, and above 50% is a severe fluctuation anomaly. The system integrates the information of all fluctuation abnormal components into the data of the main components exceeding the threshold. Next, the system screens the slave components that may be affected, and extracts the slave components directly connected to each main component exceeding the threshold from the transmission chain relationship diagram of the slave components. For example, the coupling LC-200, gearbox GB-001, etc. connected to the main motor EM-001. The system conducts a preliminary screening based on the physical connection distance, and gives priority to the slave components closer to the main component. For example, the distance between the coupling LC-200 and the main motor EM-001 is 0.15 meters, and the distance between the gearbox GB-001 and the main motor is 0.3 meters, both of which are less than the threshold of 50 centimeters. The system also considers the energy transfer efficiency. For example, the energy transfer efficiency from the main motor to the coupling is 98%, and the efficiency to the gearbox is 95%, both of which are higher than the threshold of 65%.The system further considers the parameter influence coefficients, such as the temperature influence coefficient of the main motor on the reduction gearbox being 0.67 and that on the coupling being 0.82, both of which are higher than the threshold value of 0.3. Considering these factors comprehensively, the system generates a list of slave components to be tested with a priority ranking. The system conducts high-frequency sampling (once every 0.5 seconds) on the slave components to be tested and configures the monitoring parameters for different types of components: for mechanical transmission components such as the reduction gearbox GB-001, monitor temperature (accuracy ±0.2 °C), vibration (accuracy ±0.05 mm / s), and noise (accuracy ±1 dB); for hydraulic components such as the control valve HV-004, monitor pressure (accuracy ±0.05 MPa), flow rate (accuracy ±0.5%), and oil temperature (accuracy ±0.5 °C). The system continuously monitors for 120 seconds, synchronously records the ambient temperature (22 - 28 °C) and the equipment load status, and adopts a redundant sensor design to ensure data accuracy. The system processes the original response data of the slave components in segments, divided into the pre-fluctuation segment (30 seconds before to 0 seconds before the main component is abnormal), the mid-fluctuation segment (0 seconds to 60 seconds after the abnormality), and the post-fluctuation segment (60 seconds to 120 seconds after the abnormality). Calculate the statistical characteristics for each time period. For example, the average temperature of the reduction gearbox GB-001 in the pre-fluctuation segment is 52.3 °C (standard deviation 0.5 °C), in the mid-fluctuation segment is 57.8 °C (standard deviation 2.1 °C), and in the post-fluctuation segment is 61.5 °C (standard deviation 0.7 °C). The system calculates the temperature change slope. For example, the temperature change slope of the reduction gearbox in the mid-fluctuation segment is 5.5 °C / minute; analyzes the change in the vibration spectrum. For example, a new frequency component of 42 Hz appears in the reduction gearbox in the mid-fluctuation segment; analyzes the pressure fluctuation characteristics. For example, the pressure fluctuation range of the hydraulic control valve increases in the mid-fluctuation segment; conducts a fitting analysis on the change trends of each parameter to determine the change pattern. Finally, the system conducts an analysis of the parameter change pattern to determine the response delay time. For example, the temperature of the reduction gearbox starts to rise 15 seconds after the main motor is abnormal; identifies the peak response characteristics. For example, the highest temperature of the reduction gearbox reaches 63.2 °C, with a change amplitude of 10.9 °C; analyzes the steady-state response characteristics. For example, the temperature of the reduction gearbox stabilizes at 62.5 ± 0.5 °C; calculates the response decay rate (0.82) and the resonant frequency (0.05 Hz); identifies the abnormal transmission mode and path. For example, the abnormality is transmitted from the main motor to the coupling (8 seconds), then to the reduction gearbox (15 seconds), and finally to the bearing (22 seconds), confirming that the abnormality propagates step by step along the energy transmission chain. These analysis results form a complete situation of the response parameter changes, providing data support for subsequent abnormal transmission confirmation and fault diagnosis.

[0093] The present invention accurately evaluates the stability of the main component by calculating the fluctuation amplitude of the operating state parameters of the main component, compares the calculation result with a preset fluctuation threshold to identify the main components exceeding the threshold. Exclusive fluctuation thresholds are set for different types of main components, such as a temperature fluctuation of 8% and a rotational speed fluctuation of 5% for motor-type main components, a pressure fluctuation of 12% for hydraulic pump-type main components, and a pneumatic pressure fluctuation of 15% for pneumatic device-type main components, to ensure the pertinence and accuracy of anomaly detection. For the main components exceeding the threshold, screening is performed based on the physical connection distance and energy transfer efficiency to reduce the interference of irrelevant components and ensure the rationality of the analysis object, and then a list of slave components to be measured is obtained. High-frequency sampling is performed on the slave components to be measured to obtain accurate response data, and the data is segmented to extract key parameters in three time periods before, during, and after the fluctuation, including the average value, standard deviation, and change trend, so as to obtain the detailed response characteristics of the slave components. By calculating the temperature change slope, vibration spectrum change, and pressure fluctuation characteristics, the understanding of the operating state is further improved. Subsequently, based on the analysis of the parameter change pattern, the response delay time, peak response, and steady-state response of the slave components are identified to quantify the response characteristics of the system. At the same time, the response decay rate and resonance frequency are calculated to evaluate the dynamic adaptability of the slave components and the stability of the system. Finally, this method forms the change situation of the response parameters of the slave components, providing data support for accurately judging the abnormal transmission path, optimizing the equipment monitoring strategy, and improving the accuracy of predictive maintenance.

[0094] Preferably, generating the abnormal transmission confirmation data according to the change situation of the response parameters in step S3 includes:

[0095] Establishing a device abnormal mode recognition database;

[0096] Performing matching calculations based on vector similarity on the change situation of the response parameters according to the device abnormal mode recognition database to obtain fault mode matching degree data;

[0097] Screening potential fault types with a matching degree exceeding 75% from the fault mode matching degree data to obtain fault type probability distribution data;

[0098] Calculating the deviation rate between the actual response delay time and the preset theoretical transmission time of the slave component response parameters, and performing amplitude decay coefficient and spectral feature offset analysis to obtain abnormal transmission confirmation probability data;

[0099] Performing weighted fusion on the fault type probability distribution data and the abnormal transmission confirmation probability data, and calculating a credibility score. When the score exceeds the preset threshold of 0.8, it is confirmed as an effective abnormal transmission to obtain the abnormal transmission confirmation result;

[0100] Structurally integrate the main component information, slave component information, fault type probability, transmission characteristics, and credibility score according to the abnormal transmission confirmation result, so as to obtain the abnormal transmission confirmation data.

[0101] In the intelligent maintenance system of the injection molding equipment in the embodiments of the present invention, first, a database for identifying abnormal modes of the equipment is established as the knowledge basis for abnormal diagnosis. The system extracts the fault repair data of the past three years from the enterprise equipment maintenance records, covering 583 fault cases; three senior equipment maintenance experts are hired to classify the faults into 12 main fault categories and 48 typical fault modes; for each fault mode, a set of characteristic parameters is sorted out, including the abnormal parameter characteristics of the main components (such as the temperature rise rate, vibration spectrum characteristics, etc.) and the response characteristics of the slave components (such as the response delay time, peak response ratio, etc.); a characteristic vector with 20 - 30 dimensions is established for each fault mode; the development speed, severity, and recommended treatment measures of each fault mode are recorded; finally, a hybrid architecture database with self-learning ability is formed. Based on this database, the system performs a fault mode matching analysis on the current response parameter changes. First, the current observed temperature abnormality of the main motor EM-001 and the response of the slave components are integrated into a query characteristic vector, including about 25 characteristic dimensions such as the temperature rise rate (2.8 °C / minute), vibration frequency (35 Hz), current fluctuation (12%), and the response delay of the reduction gearbox (15 seconds), temperature rise (10.9 °C), etc.; the cosine similarity algorithm is used to calculate the similarity between the current characteristics and each fault mode in the database, and the similarity score is calculated after normalizing the characteristic vector; different weights are set for different characteristic dimensions, such as the weight of the temperature rise rate is 1.5, and the weight of the vibration frequency characteristic is 1.2; adaptive adjustment is made considering background information such as the equipment operation time and environmental temperature; finally, sorting is performed according to the scores, and the similarity, the number of matching characteristics, and the degree of key characteristic matching of each fault mode are recorded to form the fault mode matching degree data. The similarity between the current case and the "overheating of the motor bearing" mode is 0.83 (83%), the similarity with the "initial stage of motor winding short circuit" is 0.65 (65%), and the similarity with the "insufficient lubrication of the reduction gearbox" is 0.59 (59%). The system sets the screening threshold of the fault mode matching degree to 75%, and extracts the fault modes that meet the conditions from the matching results. In the current case, only the "overheating of the motor bearing" (83%) exceeds the threshold; adjustment is made in combination with the equipment historical fault records. This motor has experienced overheating caused by 2 "bearing overheats" and 1 "fan failure" in the past two years, so a weight of 0.15 is added; considering seasonal factors, the incidence rate of this fault is 30% higher in the high-temperature environment (28 °C) in summer than in winter, and another weight of 0.1 is added; the probability distribution is recalculated, and the final probability of the "overheating of the motor bearing" is 83%×(1 + 0.15 + 0.1)=103.75%, recorded as 100%; other fault modes with matching degrees lower than the threshold are used as secondary references, such as the "initial stage of motor winding short circuit" (15%) and the "insufficient lubrication of the reduction gearbox" (10%). The system verifies the physical rationality of the abnormal transmission, extracts the theoretical transmission time parameters between components from the equipment physical model, compares the actual response delay of the slave components with the theoretical time, and calculates the deviation rate.If the actual response delay from the main motor to the reduction gearbox is 15 seconds, while the theoretical time is 8 - 12 milliseconds and the deviation rate is approximately 124,900%, it indicates that this is caused by physical processes such as heat accumulation; analyzing the amplitude decay coefficient, the temperature rise of the main motor is 15°C, the temperature rise of the reduction gearbox is 10.9°C, and the coefficient is 0.73, which is within the reasonable range of 0.6 - 0.9; analyzing the spectral feature offset, the vibration of the main motor is 35 Hz, and the reduction gearbox is 34 Hz, with the offset within the reasonable range of ±5 Hz; comprehensively calculating the physical credibility is 0.88 (88%), indicating that the abnormal transmission is physically reasonable. The system sets the probability weight of the fault type to 0.6 and the probability weight of the abnormal transmission confirmation to 0.4, and calculates the comprehensive credibility score. For the transmission relationship where "motor bearing overheating" causes the abnormal temperature of the reduction gearbox, the comprehensive score is 100%×0.6 + 88%×0.4 = 95.2%, which is much higher than the preset threshold of 0.8 (80%), so it is confirmed as an effective abnormal transmission; comparing the scores of different transmission paths to determine the main and secondary transmission paths. Finally, the system integrates the main component information (EM - 001, main drive motor, temperature 80°C, exceeding the standard by 15°C, lasting for 45 minutes), relevant slave component information (GB - 001, LC - 200, etc., including response parameters, values, time), the probability of the fault type ("motor bearing overheating" 100%, "initial stage of motor winding short - circuit" 15%, "insufficient lubrication of the reduction gearbox" 10%), transmission characteristics (transmission path, speed about 8 cm / s, temperature transmission decay coefficient 0.73, heat conduction - type transmission), and the comprehensive credibility score (95.2%); calculates the fault severity level (level 3, moderate fault) and the development expectation (expected to develop into a level 4 serious fault after 12 hours), generates treatment suggestions and maintenance window suggestions, forms structured abnormal transmission confirmation data, and provides comprehensive support for maintenance decision - making.

[0102] The present invention constructs a database for identifying abnormal patterns of equipment, systematically stores historical failure patterns of various types of equipment, and provides basic data support for subsequent matching calculations. For the changes in response parameters of components, a vector similarity calculation method is used for matching analysis to quantify the matching degree of failure patterns, and potential failure types with a matching degree exceeding 75% are screened to form failure type probability distribution data, thereby improving the accuracy of failure identification. On this basis, by calculating the deviation rate between the actual response delay time and the theoretical transmission time, and combining the amplitude attenuation coefficient and spectral feature offset analysis, the characteristics of abnormal signal transmission between components are evaluated, and then the probability of abnormal transmission confirmation is quantified. Subsequently, the failure type probability distribution data and the abnormal transmission confirmation probability data are weighted and fused, and a credibility score is calculated to comprehensively measure the effectiveness of abnormal transmission. A credibility threshold of 0.8 is set. When the score exceeds this threshold, it can be determined that the abnormal transmission has been effectively confirmed, thereby improving the reliability of diagnosis. Finally, the method structurally integrates the master component information, slave component information, failure type probability, transmission characteristics, and credibility score to generate complete abnormal transmission confirmation data, providing a traceable and analyzable basis for equipment health management, and at the same time providing high-quality data support for accurate early warning and maintenance optimization.

[0103] Preferably, step S4 includes the following steps:

[0104] Step S41: Establish a master-slave component abnormal relationship matrix based on the abnormal transmission confirmation data, identify the main path and diffusion mode of abnormal transmission, and obtain abnormal propagation chain data;

[0105] In the operation monitoring of a large injection molding machine in an embodiment of the present invention, the system first extracts all abnormal transfer events recorded within the most recent 90 days from the abnormal transfer confirmation database, including 28 abnormal events of main components such as 1 37kW main drive motor, 4 2.2kW auxiliary motors, 6 servo motors, 2 hydraulic pumps, and 4 pneumatic control valve groups, and 63 response events of slave components. The system uses these data to construct a 94×94 abnormal relationship matrix of main and slave components (including all involved main and slave components), and the element value in the matrix represents the abnormal transfer confirmation probability between corresponding components. For example, the abnormal transfer confirmation probability between the main drive motor EM-001 and the speed reducer GB-001 is 0.87, indicating that when the main motor has an abnormality, there is an 87% probability that the speed reducer will subsequently have an abnormality. Based on this matrix, the system applies the path analysis algorithm in graph theory, specifically using the weighted breadth-first search method, to identify the main paths of abnormal transfer, such as the transfer path of abnormality from the main drive motor (EM-001) → coupling (LC-200) → speed reducer (GB-001) → bearing (B-001), and the abnormal confirmation probabilities are 0.87, 0.76, and 0.68 in sequence, forming the main transfer path; at the same time, the diffusion modes are identified, including the step-by-step attenuation mode (such as the gradually weakening influence of the main motor abnormality) and the branch diffusion mode (such as the abnormality of the hydraulic pump HP-001 affecting multiple hydraulic control valves simultaneously). The system integrates these identified paths, influence intensities, and diffusion modes into an abnormal propagation chain data structure, including information such as the abnormal source component, the sequence of affected components, the influence probabilities at all levels, and the diffusion characteristics, providing a basis for subsequent time analysis.

[0106] Step S42: Statistically analyze the abnormal transfer time intervals between connected components in the abnormal propagation chain data, calculate the average delay time and fluctuation range between different component pairs, and obtain the abnormal transfer time data between components;

[0107] In the embodiments of the present invention, for the abnormal propagation chain data, the system conducts in-depth analysis from the time dimension, focusing on the time characteristics of the abnormal transfer between the injection molding equipment components. Specifically in terms of operation, the system first extracts the timestamps of abnormal events between each pair of connected components from the historical monitoring data of the injection molding equipment. For example, the main drive motor EM-001 had a temperature abnormality at 10:23:45.32 on June 15, 2023, and the connected reduction gearbox GB-001 had a temperature increase abnormality at 10:24:12.58 on June 15, 2023. The calculated time interval is 27.26 seconds. The system statistically analyzes the historical abnormal events (a total of 47 times) of the same pair of components (such as EM-001 and GB-001), calculates the average delay time (28.35 seconds in this example) and its standard deviation (5.12 seconds), and thus determines the fluctuation range to be from 18.11 seconds to 38.59 seconds (average value ± 2 times the standard deviation). The system further analyzes the relationship between the delay time and the equipment operating conditions and finds that under high load conditions (such as when injecting high-viscosity materials), the average delay time is shortened to 22.65 seconds, while under low load conditions, it is extended to 34.82 seconds, indicating that the load has a significant impact on the abnormal transfer speed. In addition, the system also examines the environmental temperature factor and finds that for every 5°C increase in the environmental temperature, the average delay time decreases by approximately 2.3 seconds, indicating that temperature promotes abnormal transfer. Finally, the system establishes a complete abnormal transfer time data model for each pair of connected components in the abnormal propagation chain, including parameters such as average delay time, fluctuation range, sensitivity of influencing factors, and condition correlation, providing accurate time feature data for subsequent pattern learning and prediction.

[0108] Step S43: Input the abnormal transfer time data between components into the large model for time series pattern learning, identify typical abnormal development laws including periodic patterns, gradual change patterns, and jump patterns, and obtain the component abnormal time series feature data;

[0109] In the system of the injection molding equipment in the embodiments of the present invention, the calculated abnormal transfer time data between components is input into a large model based on the Transformer architecture for time series pattern learning. This model analyzes and identifies time patterns through the self-attention mechanism. In the specific implementation process, first, the abnormal transfer time data accumulated on the injection molding equipment is grouped by component pairs to construct a feature vector sequence containing a total of 968 abnormal transfer events within the past 12 months. 35 key features are extracted for each abnormal transfer event, including: main component type (such as 37kW main drive motor), slave component type (such as reduction gearbox), transfer time interval, main component abnormal intensity (such as temperature overrun degree), slave component response intensity, injection molding process parameters (such as injection pressure, holding pressure time, injection speed), mold temperature, ambient temperature, running duration, etc. After the data is organized into a standardized feature matrix, it is input into the large model. The model parameters include: 8-layer Transformer encoder, 12 attention heads in each layer, hidden layer dimension 768, training batch size 32, learning rate set to 0.0001, and trained using the Adam optimizer for 100 epochs. Through model analysis, the system successfully identifies three typical abnormal development patterns in the injection molding equipment: periodic pattern (such as the hydraulic pump HP-001 regularly shows pressure fluctuation abnormalities after running continuously for 48 hours, with obvious periodic characteristics), gradual change pattern (such as the temperature of bearing B-001 slowly rises at an average rate of 0.5°C per hour, showing a linear growth trend), and jump pattern (such as after the reduction gearbox GB-001 bears an instantaneous overload, the vibration value suddenly increases by 65% and stabilizes at a new level). The recognition accuracy of the model for these patterns reaches 92.3%, far higher than 74.5% of traditional statistical methods. The system integrates the identified pattern features, development laws, and applicable conditions into component abnormal time series feature data, including information such as pattern type, key parameters, fitting equation, prediction reliability, and applicable conditions.

[0110] Step S44: Extract the development trajectory of the fault from the preset historical fault database according to the component abnormal time series feature data, and match the evolution process of the fault from the initial abnormality to the maintenance critical state, so as to obtain the fault evolution reference data;

[0111] In the embodiment of the present invention, based on the characteristic data of the component abnormal time series, the system of the injection molding equipment then performs matching analysis with a preset historical fault database. This database contains 2,835 equipment fault cases recorded by the injection molding equipment manufacturer in the past 10 years. Each case details complete information such as the faulty component, initial abnormal manifestation, abnormal development process, repair measures, and replacement records. The system uses a similarity matching algorithm to compare the currently identified abnormal time series characteristics with historical cases and calculates the cosine similarity between the feature vectors. For example, when it is currently detected that the temperature of the main drive motor EM-001 shows a gradual increase pattern, with an initial temperature of 65°C and a rising rate of 0.8°C per hour, and there is an obvious peak in the vibration spectrum at 180 Hz, the system matches these characteristics with the cases in the historical database and finds a historical case (case ID: FL20180329-EM016) with a similarity of 0.87. This case records the complete development process of a motor of the same model from the start of the abnormality to the complete failure of the bearing, including the temperature starting from 67°C, rising at a rate of 0.9°C per hour, and reaching the critical value of 132°C after 72 hours, at which point the bearing fails. The system extracts the complete fault development trajectories of this case and other 5 cases with a similarity exceeding 0.75, analyzes the parameter change trends, acceleration decay points, and critical state characteristics, and based on the commonalities and differences of these cases, establishes a fault evolution reference model for the current abnormality. This model records the possible evolution paths from the current state (such as the bearing temperature of 65°C) to the maintenance critical state (such as the temperature exceeding the rated value by 40°C and reaching 95°C), including key stage points, parameter change rates, and state transition characteristics, forming complete fault evolution reference data.

[0112] Step S45: Calculate the remaining time required for the component to reach the maintenance critical value according to the fault evolution reference data to obtain the component maintenance window data, where the maintenance critical value for motor components is that the temperature exceeds the rated value by 40°C or the vibration amplitude exceeds 2.5 mm / s, the maintenance critical value for hydraulic pump components is that the pressure fluctuation exceeds the rated value by 25% or the internal leakage rate exceeds 12%, and the maintenance critical value for pneumatic device components is that the air pressure drops exceed the rated value by 30% or the response time extends by more than 150 ms.

[0113] In an embodiment of the present invention, based on the fault evolution reference data, the system of the injection molding equipment begins to accurately calculate the remaining time required for the component to reach the maintenance critical value. Taking the detected abnormal speed reducer GB-001 as an example, its current temperature is 78°C (28°C higher than the rated temperature of 50°C), and the vibration value is 1.8 mm / s. The system first determines the maintenance critical value of this component: as a motor component, its maintenance critical value is that the temperature exceeds the rated value by 40°C (i.e., reaches 90°C) or the vibration amplitude exceeds 2.5 mm / s. According to the prediction model of the fault evolution reference data, under the current equipment load conditions (injection pressure 21 MPa, injection speed 85 mm / s, ambient temperature 28°C), the temperature of the speed reducer is expected to rise at an average rate of 0.6°C per hour, and it takes (90 - 78) / 0.6 = 20 hours to reach the temperature critical value; at the same time, the vibration value increases at a rate of 0.04 mm / s per hour, and it takes (2.5 - 1.8) / 0.04 = 17.5 hours to reach the vibration critical value. The system takes the shorter of the two times, 17.5 hours, as the predicted remaining operating time, and based on historical reliability analysis, calculates the predicted confidence interval to be 17.5 ± 2.3 hours (95% confidence level). The system further analyzes and finds that if the equipment load is reduced to 65% (such as reducing the injection pressure to 15 MPa), the temperature rise rate can be reduced to 0.4°C per hour, and the remaining operating time can be extended to 30 hours; if additional cooling measures are implemented (such as adding a cooling fan), the temperature rise can be further suppressed, and the remaining time can be extended to 36 hours. The system comprehensively considers these factors and generates component maintenance window data, including component ID, current status value, critical value, predicted remaining time (basic situation and intervention situation), prediction reliability, and maintenance urgency, providing a decision-making basis for the maintenance plan of the injection molding equipment.

[0114] The present invention constructs a master-slave component anomaly relationship matrix to systematically identify the main paths and diffusion patterns of anomaly transmission, thereby forming anomaly propagation chain data, making the tracking of the anomaly development process more intuitive and quantifiable. On this basis, statistical analysis is carried out on the component anomaly transmission time intervals in the anomaly propagation chain data, and the average delay time and fluctuation range between each pair of components are calculated to accurately characterize the timing characteristics of anomaly transmission and improve the predictability of the anomaly development process. Subsequently, the component-to-component anomaly transmission time data is input into a large model for time series pattern learning to automatically identify the periodic, gradual, or jumpy patterns existing in the anomaly evolution process, extract key time series features, and provide in-depth data support for subsequent trend prediction and decision-making support. Combining with the historical fault database, the complete evolution trajectory of the fault from the initial anomaly to the maintenance critical state is further matched to ensure the accuracy of the anomaly development law and provide reference data based on real fault cases. Based on the fault evolution reference data, the remaining time required for the component to reach the maintenance critical value is calculated to form component maintenance window data, providing a scientific basis for equipment maintenance. Among them, specific maintenance critical standards are set for different types of components, such as the temperature and vibration amplitude of motor components, the pressure fluctuation and leakage rate of hydraulic pump components, and the air pressure drop and response time change of pneumatic device components, so as to implement different differential maintenance strategies for different components. This method improves the accuracy of equipment anomaly prediction as a whole, makes the maintenance plan more targeted, avoids resource waste or equipment damage caused by premature or late maintenance, and ultimately enhances the stability and reliability of equipment operation.

[0115] Preferably, step S5 includes the following steps:

[0116] Step S51: Classify the component maintenance window data by type, and classify the components into emergency repair type, planned repair type, and monitoring repair type based on the maintenance urgency to obtain maintenance priority data;

[0117] In the embodiment of the present invention, an urgency assessment is carried out on 17 components to be repaired identified from 94 components. A scoring model is established considering three key factors: the estimated remaining operating time (weight 0.5, 10 points for < 24 hours, 6 points for 24 - 72 hours, 3 points for 72 - 168 hours, 1 point for > 168 hours), the degree of deviation of abnormal parameters (weight 0.3, 10 points for deviation > 50%, 7 points for 30 - 50%, 4 points for 10 - 30%, 1 point for < 10%), and the scope of influence of component failure (weight 0.2, 10 points for affecting the whole machine, 6 points for affecting a single module, 3 points for affecting performance, 1 point for affecting accuracy). After the system calculates the comprehensive score of each component, it classifies them according to intervals: ≥ 8 points are for emergency repair category (such as the speed reducer GB - 001 with a score of 8.7) with a total of 3; 5 - 8 points are for planned repair category (such as the hydraulic control valve HV - 004 with a score of 6.2) with a total of 8; < 5 points are for monitoring and repair category (such as the cooling pump CP - 002 with a score of 3.5) with a total of 6. The system assigns a priority number (P1 / P2 / P3) and a sorting position to each component, and constructs a maintenance priority data structure. Step S52: Extract the planned operating time, production load rate, and planned downtime maintenance time of the equipment from the production plan data, establish an equipment operation time axis, and obtain an equipment utilization time axis;

[0118] In the embodiment of the present invention, the production plan data for the next two weeks is extracted from the enterprise production management system, and the detailed production plan (15 - minute granularity) of the equipment SJJ - 350 is exported in JSON format through the API interface, including the daily operation time arrangement (three - shift system on weekdays, two - shift system on weekends), the load rate at each time period (such as 95% load from 08:00 to 12:00 on Monday), the planned downtime maintenance time, and the detailed information of each production batch (such as producing automotive bumpers on June 18th, injection pressure 21 MPa, 3200 pieces). The system arranges this information in chronological order to construct an equipment utilization time axis of 1344 time points (14 days × 24 hours × 4 15 - minute intervals). Each point contains a time stamp, an operating state (1 / 0), a load rate, and a special mark, providing a basis for subsequent identification of maintenance time periods.

[0119] Step S53: Identify the production idle period and low - load period according to the equipment utilization time axis to obtain the data of available maintenance time periods, where the production idle period is defined as a period with continuous downtime exceeding 4 hours, and the low - load period is defined as a period with a production load rate lower than 40% and a continuous duration exceeding 6 hours;

[0120] In the embodiment of the present invention, the scanning device utilizes the time axis to automatically identify suitable maintenance periods. First, a search algorithm is applied to find "production idle periods" (periods of continuous downtime > 4 hours), and the continuous time points with a running state of 0 are grouped to calculate the duration. Six production idle periods are identified (such as a 16-hour downtime from 06:00 to 22:00 on Sunday), with a total of 52 hours of maintenance time. Secondly, "low-load periods" (periods with a load rate < 40% and a duration > 6 hours) are searched for, and four low-load periods are identified (such as producing low-precision components from 22:00 on Monday to 04:00 on Tuesday, with a load rate of 35%), with a total of 28 hours of potential maintenance time. The system integrates these 10 maintainable periods into a data set, including type, time, duration, average load rate, and characteristic description, providing a total of 80 hours of potential maintenance time.

[0121] Step S54: Match and analyze the maintenance priority data with the maintainable period data to obtain maintenance time data;

[0122] In the embodiment of the present invention, a multi-objective optimization algorithm is developed for matching maintenance priorities and periods, considering four factors: the matching degree of time urgency (such as the remaining running time of the speed reducer is 17.5 hours, and the nearest period starts after 12 hours, with a score of 85), the adequacy of the maintenance window (such as the control valve requires 3.5 hours of maintenance, and a certain period is 16 hours, with a score of 100), the production impact degree (the idle period is scored 0, and the low-load period is scored 10 - 40 according to the load rate), and the resource conflict degree (ranging from 0 points for no conflict to 50 points for serious conflict). The system establishes a 17×10 matching score matrix, applies the Hungarian algorithm to solve the optimal match, and assigns the most suitable maintenance period. For example, the speed reducer GB-001 is arranged in the idle period from 00:00 to 05:00 on Wednesday, and the control valve HV-004 is arranged in the downtime period from 06:00 to 22:00 on Sunday.

[0123] Step S55: Analyze the dependency relationship between the maintenance components based on the maintenance time data and the component transfer chain relationship diagram. When multiple components have an upstream and downstream relationship in the transfer chain, their maintenance times are merged and arranged to obtain optimized maintenance scheduling data;

[0124] In the embodiment of the present invention, the dependency relationship between maintenance components is analyzed based on the component transfer chain relationship diagram, the upstream and downstream relationships of the components to be maintained are extracted from the graph data structure, and an associated component set is established. For example, there is an upstream and downstream relationship between the control valve HV-004 and the hydraulic pump HP-001. The system uses a graph traversal algorithm to search for two-level associated components, and at the same time analyzes the maintenance disassembly and assembly sequence, and finds that there is an overlap in the disassembly of some components (for example, the same housing needs to be disassembled for maintaining the control valve and the distributor). When the disassembly and assembly overlap rate > 30%, it is determined as a highly associated task. The system optimizes the scheduling, combines the upstream and downstream components (such as HP-001 and HV-004) and arranges them in the same time period (06:00-22:00 on Sunday), reducing the maintenance time by 25%, and ensuring that the upstream components are maintained first or simultaneously. Finally, 17 maintenance tasks are combined into 9 centralized tasks, significantly improving the efficiency.

[0125] Step S56: Calculate the estimated execution time and required resources for each maintenance task for the optimized maintenance scheduling data, and perform enterprise resource constraints to obtain maintenance execution feasibility data;

[0126] In the embodiment of the present invention, the execution details and resource requirements of the maintenance tasks are calculated. The standard working hours, skill requirements, and spare part list of each component are extracted from the maintenance knowledge base. For example, the maintenance of the hydraulic system set requires 12.5 hours, 2 hydraulic professionals, 1 mechanical professional, and spare parts such as sealing rings and sensors. The system considers the complexity adjustment factor (the coefficient for the equipment running for 3.5 years is 1.15, the coefficient for not having a major overhaul for 8 months is 0.9, and the coefficient for moderate abnormality is 1.1), calculates the adjusted execution time to be 14.3 hours, and checks the availability of resources. It is found that there is only 1 hydraulic professional on Sunday (2 are required as standard), which will cause the working hours to increase by 40%, and there is only 1 pressure sensor in stock (2 are required), and the procurement cycle is 5 days. Based on this, feasibility data (score 70 points, there are risks but it is basically feasible) and deployment suggestions are generated.

[0127] Step S57: Generate an equipment maintenance report including a component status signal light diagram, a transfer chain map, a maintenance schedule, and a maintenance material list according to the maintenance execution feasibility data and the component transfer chain relationship diagram.

[0128] In the embodiment of the present invention, an equipment maintenance report is generated, which includes four parts: a component status signal light diagram (marking the status of each component with red, yellow, blue, and green on the 3D model); a transfer chain map (a network diagram based on the force-directed algorithm, with nodes as components and connections representing influence relationships); a maintenance schedule (in the form of a Gantt chart, with the horizontal axis as a 14-day time line and the vertical axis as 9 maintenance tasks, and the color blocks representing the maintenance time periods); a maintenance material list (listing spare parts, tools, and consumables by task group, including codes, specifications, quantities, inventory status, procurement cycle, and cost). The report is in HTML5 format, supports viewing and interactive functions on multiple devices, and is automatically distributed to relevant personnel to ensure the coordinated progress of maintenance work.

[0129] Based on the maintenance window data of components, the present invention classifies components by type and divides them into emergency maintenance category, planned maintenance category, and monitoring maintenance category according to the urgency of maintenance, making the maintenance priority clearer and helping to optimize the allocation of maintenance resources. Subsequently, the planned operation time, production load rate, and planned downtime maintenance time of the equipment are extracted from the production plan data to establish a complete equipment utilization timeline, providing data support for the selection of maintenance timing. On this basis, the production gap period and low load period are identified to form data on repairable time periods, so as to arrange maintenance tasks without affecting production efficiency and improve equipment availability. Then, the maintenance priority data is matched and analyzed with the repairable time period data to ensure that high-priority maintenance tasks can be executed in the most appropriate time period, avoiding production interruption or equipment damage caused by improper maintenance arrangements. Regarding the relevance between maintenance tasks, this method combines the component transfer chain relationship diagram to comprehensively analyze components with upstream and downstream dependencies and merge and optimize their maintenance times, reducing the resource consumption and production losses caused by multiple downtime maintenances. By further calculating the estimated execution time and required resources of each maintenance task and combining with the enterprise resource constraints, the feasibility of the maintenance plan in actual operation is ensured, avoiding maintenance delays caused by insufficient manpower, materials, or equipment. Finally, a complete equipment maintenance report is generated based on all the analysis results, including component status signal light diagrams, transfer chain maps, maintenance schedules, and maintenance material lists, making the maintenance management more intuitive and executable and ensuring the reliability of equipment operation and the accuracy of maintenance decisions.

[0130] The present invention also provides an enterprise business report generation system based on a large model for executing the above-mentioned enterprise business report generation method based on a large model. The enterprise business report generation system based on a large model includes:

[0131] A data acquisition module for obtaining the component operation status data and production plan data of industrial equipment;

[0132] A component relationship division module for dividing the master-slave relationship of the internal components of the equipment according to the component operation status data, designating the driving device as the main component, and designating the transmission component and the execution component as the slave components to construct a component transfer chain relationship diagram;

[0133] An anomaly detection module for monitoring the operation status parameters of the main components in the component transfer chain relationship diagram; extracting the main components whose fluctuations of the operation status parameters exceed the preset threshold and measuring the change of the response parameters of the connected slave components; generating anomaly transfer confirmation data according to the change of the response parameters;

[0134] A maintenance window analysis module, configured to perform a time interval pattern analysis based on a large model between the main component exception and the slave component exception according to the exception transfer confirmation data, and combine the preset historical failure data with the estimated running time required for the slave component current state to reach the maintenance critical state, so as to obtain the component maintenance window data;

[0135] A maintenance plan generation module, configured to perform time matching on the component maintenance window data and the production plan data, and identify the production gap period or the low load period to obtain the maintenance time data; generate an equipment maintenance report including the component status signal lamp diagram, the transfer chain map, the maintenance time, and the maintenance material list according to the maintenance time data and the component transfer chain relationship diagram.

[0136] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0137] The above are only the specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A method for generating enterprise business reports based on large models, characterized in that, It includes the following steps: Step S1: Obtain the component operation status data and production plan data of industrial equipment; Step S2: Divide the internal components of the equipment into master-slave relationships according to the component operation status data, designate the driving device as the master component, and designate the transmission component and the execution component as the slave components, and construct a component transfer chain relationship diagram; Step S3: Monitor the operation status parameters of the master components in the component transfer chain relationship diagram; extract the master components whose fluctuations of the operation status parameters exceed the preset threshold, and measure the changes in the response parameters of the connected slave components; generate abnormal transfer confirmation data according to the changes in the response parameters. Specifically, monitoring the operation status parameters of the master components in the component transfer chain relationship diagram is as follows: Determine the master component monitoring priority according to the component transfer chain relationship diagram. Among them, the master components with energy input exceeding 15kW or the number of connected slave components exceeding 5 are set as high priority, the master components with energy input between 5kW - 15kW or the number of connected slave components between 3 - 5 are set as medium priority, and the remaining master components are set as low priority; Based on the master component monitoring priority, conduct time-division polling sampling on each master component to obtain the original master component operation status data; Perform smoothing processing based on a sliding window on the original master component operation status data, with the window size being the first 10 sampling values, and perform weighted moving average processing to obtain the smoothed master component operation status data. Specifically, the weight of the most recent time point in the weighted moving average processing is 0.3, and the sum of the weights of the remaining 9 historical points is 0.7; When the temperature deviation of the motor-type master component in the smoothed master component operation status data exceeds 15°C, the vibration deviation exceeds 0.8mm / s, or the current deviation exceeds 12%, or when the pressure deviation of the hydraulic pump-type master component exceeds 1.5MPa, the flow deviation exceeds 10%, or the noise deviation exceeds 8dB, or when the air pressure deviation of the pneumatic device-type master component exceeds 0.1MPa, the flow deviation exceeds 15%, or the valve response time extends by more than 30ms, record the time point and mark the corresponding master component as an abnormal state to obtain the master component abnormal timestamp data; Extract the set of slave components directly connected to the abnormal master component from the component transfer chain relationship diagram according to the master component abnormal timestamp data to obtain a list of relevant slave components; Merge the smoothed master component operation status data, the master component abnormal timestamp data, and the list of relevant slave components into the operation status parameters of the master component; Specifically, extracting the master components whose fluctuations of the operation status parameters exceed the preset threshold and measuring the changes in the response parameters of the connected slave components is as follows: Calculate the fluctuation amplitude of the master component operation status parameters; Compare the master component fluctuation amplitude data with the preset fluctuation threshold to obtain the master component identification data exceeding the threshold. Specifically, the preset fluctuation threshold is that the temperature fluctuation threshold of the motor-type master component is 8%, and the rotational speed fluctuation threshold is 5%; the pressure fluctuation threshold of the hydraulic pump-type master component is 12%; the air pressure fluctuation threshold of the pneumatic device-type master component is 15%; According to the list of relevant slave components, screen the master components in the master component identification data exceeding the threshold based on the physical connection distance and energy transfer efficiency to obtain a list of slave components to be tested; Perform high-frequency sampling on the slave components in the list of components to be measured to obtain the original slave component response data; Perform segmented processing on the original slave component response data before, during, and after fluctuations, and calculate the average value, standard deviation, and change trend of each time period, respectively record the temperature change slope, vibration spectrum change, and pressure fluctuation characteristics to obtain the slave component response characteristic data; Perform parameter change pattern analysis on the slave component response characteristic data, identify the response delay time, peak response, and steady-state response, and calculate the response decay rate and resonance frequency to obtain the response parameter change situation; Step S4: Perform a time interval pattern analysis based on a large model between the master component anomaly and the slave component anomaly according to the anomaly transfer confirmation data, and combine the preset historical failure data with the estimated running time required for the slave component current state to reach the maintenance critical state to obtain the component maintenance window data; Step S5: Perform time matching between the component maintenance window data and the production plan data, and identify the production gap period or low load period to obtain the maintenance time data; Generate an equipment maintenance report including the component status signal light diagram, transfer chain map, maintenance time, and maintenance material list based on the maintenance time data and the component transfer chain relationship diagram.

2. The method for generating enterprise business reports based on large models according to claim 1, wherein Step S1 includes the following steps: Step S11: Deploy an equipment status collector, where the equipment status collector includes a vibration sensor, a temperature sensor, and a current sensor; install the vibration sensor on the surface of the main drive component, install the temperature sensor at the location of the heat-generating component, and connect the current sensor to the power input terminal to collect the original equipment operation data; Step S12: Perform signal noise and abnormal interference value removal processing on the original equipment operation data to obtain the equipment operation data; Step S13: Classify and organize the equipment operation data based on the component type, and calculate the parameter standard range of each component in the normal operation state to obtain the component operation reference data; Step S14: Compare the component operation reference data with the component operation state parameters monitored by the equipment status collector in real time, and mark the components with parameters deviating from the reference range to obtain the component operation state data; Step S15: Obtain the production plan data, including the equipment planned operation time, production load rate, and planned shutdown maintenance time.

3. The method for generating an enterprise business report based on a large model according to claim 2, wherein Step S2 includes the following steps: Step S21: Obtain the equipment structure drawing data, and extract the physical connection relationship and energy transfer path information of the components from the equipment structure drawing data to obtain the component physical topology structure data; Step S22: Identify the energy input components in the equipment for the component physical topology structure data, and mark them as the master components to obtain the preliminary master component identification data, where the energy input components include motors, hydraulic pumps, and pneumatic devices; Step S23: Determine the transmission components directly connected to the master component based on the energy flow direction in the preliminary master component identification data, and mark them as the first-level slave components to obtain the first-level slave component association data, where the transmission components include gearboxes, pulleys, and bearings; Step S24: Determine the execution components connected to the primary slave components along the energy transfer path according to the primary slave component association data, and mark them as secondary slave components, so as to obtain the master-slave component hierarchical data, where the execution components include actuators, working heads, and functional components; Step S25: Conduct an analysis of the correlation of operating parameters among components based on the component operating status data, calculate the degree of influence of the change in the parameters of the master component on the parameters of the slave components, and determine the parameter correlation strength, so as to obtain the component parameter influence matrix; Step S26: Construct a visual component transfer chain relationship diagram based on the master-slave component hierarchical data and the component parameter influence matrix, where the nodes of the component transfer chain relationship diagram represent each component, the connection lines represent the influence relationships among components, and the thickness of the lines represents the influence strength.

4. The method for generating an enterprise business report based on a large model according to claim 3, wherein Step S25 includes the following steps: Step S251: Extract the time series of the operating parameters of each component under different working conditions from the component operating status data, so as to obtain the historical change data of the component parameters; Step S252: Segment the data of the master component in the historical change data of the component parameters, and identify the time points when the parameters change significantly. Take the significantly changed time points as the reference time nodes, so as to obtain the master component parameter change marking data; Step S253: Analyze the parameter changes of the slave components before and after the corresponding time points according to the master component parameter change marking data, and calculate the time delay and response amplitude between the parameter changes of the slave components and the parameter changes of the master component, so as to obtain the parameter response characteristic data among components; Step S254: Calculate the parameter influence coefficients between each pair of master-slave components according to the parameter response characteristic data among components, so as to obtain the component influence coefficient matrix, where the parameter influence coefficient = (amplitude change of the slave component parameter / amplitude change of the master component parameter) × 1 / delay time; Step S255: Perform normalization processing on the component influence coefficient matrix, so as to obtain the component influence coefficient standard matrix; Step S256: Screen the component pairs with influence coefficients greater than the threshold according to the preset influence threshold for the component influence coefficient standard matrix, so as to obtain the component parameter influence matrix.

5. The method for generating an enterprise business report based on a large model according to claim 1, wherein The generation of abnormal transfer confirmation data according to the change situation of the response parameters in Step S3 includes: Establish an equipment abnormal mode recognition database; Perform a matching calculation based on vector similarity on the change situation of the response parameters according to the equipment abnormal mode recognition database to obtain the fault mode matching degree data; Screen the potential fault types with a matching degree exceeding 75% from the fault mode matching degree data to obtain the fault type probability distribution data; Calculate the deviation rate between the actual response delay time and the preset theoretical transfer time of the slave component response parameters, and conduct an analysis of the amplitude attenuation coefficient and the spectral feature offset to obtain the abnormal transfer confirmation probability data; Perform weighted fusion on the fault type probability distribution data and the abnormal transfer confirmation probability data, and calculate the credibility score. When the score exceeds the preset threshold of 0.8, it is confirmed as an effective abnormal transfer to obtain the abnormal transfer confirmation result; Structurally integrate the main component information, slave component information, fault type probability, transfer characteristics, and credibility score according to the abnormal transfer confirmation result to obtain the abnormal transfer confirmation data.

6. The method for generating an enterprise business report based on a large model according to claim 5, wherein Step S4 includes the following steps: Step S41: Establish a master-slave component abnormal relationship matrix based on the abnormal transfer confirmation data, identify the main path and diffusion mode of abnormal transfer, and obtain the abnormal propagation chain data; Step S42: Conduct a statistical analysis of the abnormal transfer time intervals of the connected components in the abnormal propagation chain data, calculate the average delay time and fluctuation range between different component pairs, and obtain the inter-component abnormal transfer time data; Step S43: Input the inter-component abnormal transfer time data into a large model for time series pattern learning, identify typical abnormal development laws including periodic patterns, gradual change patterns, and jump patterns, and obtain the component abnormal time series feature data; Step S44: Extract the development trajectory of the fault from the preset historical fault database according to the component abnormal time series feature data, and match the evolution process of the fault from the initial abnormality to the maintenance critical state to obtain the fault evolution reference data; Step S45: Calculate the remaining time required for the slave component to reach the maintenance critical value according to the fault evolution reference data to obtain the component maintenance window data, where the maintenance critical value of motor components is that the temperature exceeds the rated value by 40°C or the vibration amplitude exceeds 2.5 mm / s, the maintenance critical value of hydraulic pump components is that the pressure fluctuation exceeds the rated value by 25% or the internal leakage rate exceeds 12%, and the maintenance critical value of pneumatic device components is that the air pressure drops exceed the rated value by 30% or the response time extension exceeds 150 ms.

7. The method for generating an enterprise business report based on a large model according to claim 6, wherein Step S5 includes the following steps: Step S51: Classify the component maintenance window data by type, and classify the components into emergency repair type, planned repair type, and monitoring repair type based on the maintenance urgency to obtain the maintenance priority data; Step S52: Extract the equipment planned operation time, production load rate, and planned shutdown maintenance time from the production plan data, establish an equipment operation time axis, and obtain the equipment utilization time axis; Step S53: Identify the production gap period and low load period according to the equipment utilization time axis to obtain the repairable time period data, where the production gap period is defined as a period with continuous shutdown time exceeding 4 hours, and the low load period is defined as a period with a production load rate lower than 40% and a duration exceeding 6 hours; Step S54: Conduct a matching analysis of the maintenance priority data and the repairable time period data to obtain the maintenance time data; Step S55: Conduct a dependency analysis of the maintenance components according to the maintenance time data and the component transfer chain relationship diagram. When multiple components have an upstream and downstream relationship in the transfer chain, arrange their maintenance times together to obtain the optimized maintenance scheduling data; Step S56: Calculate the estimated execution time and required resources for each maintenance task for the optimized maintenance scheduling data, and perform enterprise resource constraints to obtain the maintenance execution feasibility data; Step S57: Generate an equipment maintenance report including component status signal lamp diagrams, transfer chain maps, maintenance schedules, and maintenance material lists according to the maintenance execution feasibility data and the component transfer chain relationship diagram.

8. An enterprise business report generation system based on a large model, characterized in that, For implementing the enterprise business report generation method based on a large model as described in claim 1, the enterprise business report generation system based on a large model includes: A data collection module for obtaining the component operation status data and production plan data of industrial equipment; A component relationship division module for dividing the master-slave relationship of the internal components of the equipment according to the component operation status data, designating the driving device as the master component, and designating the transmission component and the execution component as the slave components, and constructing a component transfer chain relationship diagram; An anomaly detection module for monitoring the operation status parameters of the master components in the component transfer chain relationship diagram; extracting the master components whose fluctuations in the operation status parameters exceed a preset threshold, and measuring the changes in the response parameters of the connected slave components; generating anomaly transfer confirmation data according to the changes in the response parameters; A maintenance window analysis module for performing a time interval pattern analysis based on a large model between the master component anomaly and the slave component anomaly according to the anomaly transfer confirmation data, and combining the preset historical failure data with the estimated operation time required for the slave components to reach the maintenance critical state from the current state to obtain the component maintenance window data; A maintenance plan generation module for matching the component maintenance window data with the production plan data in terms of time, and identifying the production gap period or low load period to obtain the maintenance time data; generating an equipment maintenance report including a component status signal light diagram, a transfer chain map, the maintenance time, and a maintenance material list according to the maintenance time data and the component transfer chain relationship diagram.

Images