A method and device for intelligent operation and maintenance of equipment
By constructing an adaptive equipment health assessment model and a comprehensive decision-making mechanism, and combining multi-source data and lifecycle costs, the problems of model distortion and single maintenance strategies in equipment management systems are solved, and accurate identification of equipment health status and cost optimization are achieved.
Patent Information
- Application Number
- CN202611405636.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-10
- Publication Date
- 2026-11-10
AI Technical Summary
Existing equipment management systems, due to their use of fixed health assessment models and single maintenance strategies, are unable to accurately reflect the true degradation patterns of equipment. This leads to the gradual distortion of health assessment results over time, and the lack of deep coupling between physical condition and economic cost can easily result in over-maintenance or delayed updates.
By collecting multi-source data throughout the entire lifecycle of equipment, an adaptive health assessment model is constructed. Combining health index, fault risk value, and lifecycle cost, a comprehensive decision is generated. The model parameters are then corrected in reverse using the execution results of maintenance actions, forming an adaptive closed-loop mechanism.
It improves the accuracy of equipment health status identification, avoids over-maintenance or delayed updates, and achieves comprehensive cost optimization and scientific operation and maintenance decision-making throughout the equipment's entire life cycle.
Smart Images

Figure CN123070935A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of industrial equipment management technology, and more specifically, to a method and apparatus for intelligent operation and maintenance of equipment. Background Technology
[0002] Currently, in the field of industrial equipment operation and maintenance, existing equipment management systems typically employ fixed health assessment models and single maintenance strategy generation logic. However, due to the complex and variable operating environment of equipment and significant individual differences, static and fixed model parameters cannot accurately reflect the actual degradation patterns of equipment, leading to gradual distortion of health assessment results over time. Furthermore, existing decision-making mechanisms often rely solely on the physical state of the equipment or solely on economic costs for independent judgments, lacking a deep coupling between physical state evolution and economic cost constraints, which can easily result in problems such as over-maintenance or delayed updates. Summary of the Invention
[0003] To address the problems of static and rigid operation and maintenance models and single decision-making dimensions in existing technologies, which lead to the failure of maintenance strategies and the loss of cost control, this application proposes an intelligent operation and maintenance method and device to achieve adaptive evolution of the operation and maintenance model and optimization of the comprehensive cost of the entire equipment life cycle.
[0004] Firstly, a method for intelligent operation and maintenance of equipment is provided, which may include: Using the unique device identifier of the target device as the association primary key, collect and associate multi-source data at each stage of the device's entire lifecycle; A health assessment model is constructed based on the multi-source data. The health index of the equipment is calculated according to the evaluation indicators in the health assessment model. The failure risk value is predicted according to the health index and its changing trend. The remaining lifespan is predicted according to the cumulative operating information and health status of the equipment. Based on the health index, failure risk value, remaining lifespan, and lifecycle cost, a comprehensive decision including maintenance actions is generated. In response to the execution of the maintenance action, the status improvement data of the target device is obtained; Based on the improved status data, the parameters of the health assessment model are corrected, and the recommended priority of the maintenance actions is also adjusted.
[0005] In one possible implementation, multi-source data from all stages of the device's entire lifecycle is collected and correlated, including: Automatically collect equipment operating parameter data through sensors; Collect manual inspection data, maintenance work order data, and business data from ERP, MES, SCADA and EAM systems through mobile terminals or business system interfaces; The collected data is uniformly associated using the unique device identifier and timestamp as indexes.
[0006] In one possible implementation, a comprehensive decision is generated that includes operational actions, including: Determine the comprehensive cost of continued use and the comprehensive cost of replacement of the target equipment; When the overall cost of continued use is greater than the overall cost of renewal, and the remaining lifespan is less than a preset lifespan threshold, an overall decision including a device renewal action is generated. Otherwise, a comprehensive decision including maintenance or inspection actions is generated based on the health index.
[0007] In one possible implementation, the total cost of continued use includes the estimated maintenance and repair costs, the estimated operating energy costs, and the estimated downtime losses due to failures. The comprehensive cost of the upgrade includes the cost of purchasing new equipment, the cost of installing and commissioning new equipment, and the residual value of the original equipment after it is scrapped. The lifecycle cost includes procurement cost, installation and commissioning cost, operating cost, repair and maintenance cost, spare parts cost, scrapping and disposal cost, and downtime loss cost.
[0008] In one possible implementation, the health assessment model includes multiple assessment indicators and corresponding initial weights; Based on the improved condition data, the parameters of the health assessment model are corrected, including: Based on the aforementioned status improvement data, the correlation between each evaluation index and equipment failure is determined; Based on the correlation and the state change amount in the state improvement data, the initial weights of each evaluation index are dynamically adjusted to obtain updated weights.
[0009] In one possible implementation, the failure risk value is predicted as follows: based on the current health index, the trend of health index change, the rate of temperature change, the rate of vibration change, the historical failure frequency and the intensity of equipment use, the failure probability of the target equipment occurring within a set time window in the future is calculated, and then the failure probability is weighted according to the criticality of the target equipment and the degree of failure impact to obtain the risk value.
[0010] In one possible implementation, the recommended priority of the operational actions is modified, including: The actual improvement in status after the aforementioned maintenance actions were executed multiple times is statistically analyzed. When the actual improvement amount is lower than the preset improvement threshold for multiple consecutive times, the recommendation priority of the operation and maintenance action in the comprehensive decision-making is reduced. When the actual improvement in state is greater than or equal to the preset improvement threshold, the recommended priority is maintained or increased.
[0011] In one possible implementation, the evaluation metrics of the health assessment model include: Operating status indicators, including temperature, vibration, current, voltage, power, speed, and energy consumption; Reliability metrics include the number of failures, failure frequency, mean time between failures, and mean time to repair. Maintenance status indicators include cumulative number of repairs, time of the most recent repair, maintenance completion rate, and repair costs; Environmental and usage intensity indicators, including ambient temperature, ambient humidity, dust concentration, and average load rate.
[0012] Secondly, an intelligent equipment operation and maintenance device is provided, which may include: The acquisition unit is used to collect and associate multi-source data at each stage of the device's entire lifecycle, using the unique device identifier of the target device as the association primary key. The prediction unit is used to construct a health evaluation model based on the multi-source data, calculate the health index of the equipment according to the evaluation indicators in the health evaluation model, predict the failure risk value according to the health index and its changing trend, and predict the remaining lifespan according to the cumulative operating information and health status of the equipment. The generation unit is used to generate a comprehensive decision including operation and maintenance actions based on the health index, fault risk value R, remaining lifespan and life cycle cost. The acquisition unit is configured to acquire status improvement data of the target device in response to the execution of the maintenance action. The correction unit is used to correct the parameters of the health assessment model based on the status improvement data, and to correct the recommended priority of the operation and maintenance actions.
[0013] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.
[0014] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.
[0015] This application provides an intelligent equipment operation and maintenance method and device. By integrating multi-source data throughout the entire equipment lifecycle using a unique equipment identifier as the primary key, it breaks down data silos in the procurement, operation, and maintenance stages of traditional equipment management, providing a complete and accurate data foundation for health assessment and comprehensive decision-making. Based on this, it uses the status improvement data after the execution of maintenance actions to reverse-correct the parameters of the health assessment model and the recommended priority of maintenance actions, constructing an adaptive closed-loop mechanism of "status perception - decision execution - effect feedback - model evolution." This enables the model to continuously self-optimize based on the actual maintenance response of individual equipment, overcoming the defect of static models distorting their assessments over time, and significantly improving the accuracy of health status identification and the effectiveness of maintenance strategies. Furthermore, by deeply coupling physical dimension parameters such as health index and remaining lifespan with economic dimension parameters such as lifecycle cost to generate comprehensive decisions, it avoids over-maintenance caused by a single physical status orientation or safety risks caused by a single economic cost orientation, achieving optimization of the comprehensive cost throughout the equipment lifecycle and scientific operation and maintenance decision-making. Attached Figure Description
[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A flowchart illustrating an intelligent equipment operation and maintenance method provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of an intelligent equipment operation and maintenance device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but does not exclude other elements or objects. The terms "connection," "coupled," or "linked," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.
[0019] like Figure 1 As shown, this embodiment provides an intelligent equipment operation and maintenance method. This method relies on an intelligent equipment operation and maintenance system, which includes: an equipment identity and file management module 101, a lifecycle management module 102, a multi-source data acquisition module 103, a data governance and lifecycle database module 104, an equipment status feature construction module 105, an equipment health assessment module 106, a fault risk prediction module 107, a remaining lifespan prediction module 108, a lifecycle cost analysis module 109, an intelligent operation and maintenance decision-making module 110, a work order execution and feedback module 111, and a visualization management module 112. The method includes: Step S100: Using the unique device identifier of the target device as the association primary key, collect and associate multi-source data from each stage of the device's entire lifecycle.
[0020] After equipment enters the enterprise, a unique equipment code is first established as a unique equipment ID through the equipment identity and file management module 101. This unique equipment ID is generated by combining the enterprise code, plant code, equipment category code, production line code, and equipment serial number, for example, "EQ-01-PUMP-003-20260015", where "EQ" represents the equipment category, "01" represents the plant code, "PUMP" represents pump equipment, "003" represents the production line number, and "20260015" represents the serial number. For equipment with existing asset numbers, the system supports mapping the original asset number to this unique ID, and this ID remains unchanged during subsequent maintenance, parts replacement, and technical upgrades. After establishing the ID, the system creates a master file for the equipment containing fields such as equipment name, model, manufacturer, manufacturing date, purchase date, installation location, workshop, design life, rated power, rated speed, key components, and current operating status.
[0021] Subsequently, the multi-source data acquisition module 103 is used to collect data from all stages of the equipment's lifecycle, including procurement, acceptance, installation, commissioning, operation, inspection, maintenance, repair, modification, and even scrap assessment and decommissioning. This data includes operating parameters such as temperature, vibration, current, voltage, power, speed, and energy consumption automatically collected by sensors, as well as manual inspection records and maintenance work order details obtained through mobile terminals or business system interfaces, and business data from ERP (Enterprise Resource Planning) systems, MES (Manufacturing Execution System), SCADA (Supervisory Control and Data Acquisition) systems, and EAM (Enterprise Asset Management) systems.
[0022] The entire equipment lifecycle is divided into 15 consecutive stages: demand application, selection, procurement, delivery and acceptance, installation, commissioning, formal operation, daily inspection, maintenance, fault diagnosis and repair, parts replacement, technical transformation, performance verification, and scrap assessment and decommissioning. Each stage is associated with corresponding stage data (e.g., the procurement stage records the contract, price, and manufacturer) and clear stage completion conditions (e.g., the procurement stage is completed upon "delivery," and the acceptance stage is completed upon "acceptance passed"). When the system detects that the condition is met, it automatically advances the equipment to the next stage and generates a lifecycle event record. Each time a lifecycle event occurs, the system generates an event record containing an event ID, a unique equipment identifier, a timestamp, the current stage, the event type, the operator, the event content, and attached data. This record is stored hierarchically according to the sequence "equipment identifier → lifecycle stage → event → data → status change → maintenance result," thus forming a complete digital archive centered on the equipment.
[0023] The collected heterogeneous data undergoes cleaning, standardization, and time synchronization processing by the data governance and lifecycle database module 104. Due to the diverse sources and formats of the data, the data governance and lifecycle database module 104 requires in-depth data governance. First, the system standardizes all data, for example, unifying temperature to degrees Celsius (°C), pressure to megapascals (MPa), vibration to millimeters per second (mm / s), and time to UTC+8 standard time. Second, it handles abnormal data, removing obvious erroneous values exceeding physical limits (e.g., temperatures below -100°C), filling short-term missing data with interpolation from adjacent time points, and setting missing data markers for long-term missing data to avoid misleading the model. Third, it performs time alignment for data with different sampling frequencies, for example, mapping 1-second vibration data, 10-second temperature data, and 1-minute energy consumption data to a unified 1-minute time window, calculating statistical characteristics such as average, maximum, and root mean square values to form a state dataset with unified time granularity. Ultimately, all the processed data is linked to the same device entity using a unique device identifier, and a strict time-series chain is constructed using timestamps. This allows the operational status at any given moment to be traced back to the most recent maintenance, modification, or installation event, and linked backward to subsequent faults or performance changes. All data is uniformly stored in a lifecycle database using a unique device identifier and timestamp as indexes, forming a complete digital archive.
[0024] Step S200: Construct a health assessment model based on multi-source data, calculate the equipment's health index according to the evaluation indicators in the health assessment model, predict the failure risk value based on the health index and its changing trend, and predict the remaining lifespan based on the equipment's cumulative operating information and health status.
[0025] The equipment status feature construction module 105 extracts status features from the raw data for evaluation and prediction to construct a health evaluation model. The health evaluation model includes multiple evaluation indicators. Within a unified time window after data governance, the system converts the raw data into status features (i.e., evaluation indicators) for evaluation and prediction. Specifically: Temperature parameters are expressed as temperature deviation. ,in The current temperature of the device. This is the reference temperature for normal operation of the equipment, determined by the equipment manufacturer's technical specifications or historical normal operation data. This is the upper limit of the equipment's permissible temperature; exceeding this value poses a risk of damage to the equipment. Simultaneously calculate the rate of temperature change: Rate of change To determine the warming trend, among which, The rate of temperature change characterizes the rate at which temperature changes per unit time. For the current moment Temperature value; Before the current moment Temperature value at any given time; The time interval used to calculate the rate of change.
[0026] Vibration index uses the root mean square value of vibration. ,in, The root mean square value of vibration represents the energy intensity of the vibration signal and reflects the smoothness of the equipment's mechanical operation. The vibration amplitude (acceleration, velocity, or displacement) at the j-th sampling point, in units of . Consistent; This is to count the total number of vibration sampling points within the time window.
[0027] Current stability is achieved using the current variation coefficient. ,in, The coefficient of variation of current characterizes the relative degree of current fluctuation and reflects the stability of the electrical load of the equipment; The standard deviation of current data within the statistical time window; This represents the average value of the current data within the statistical time window.
[0028] Based on this, the equipment health evaluation module 106 calls the preset health evaluation model and calculates a quantitative health index (HI) by comprehensively considering multiple indicators such as operating status, reliability, maintenance status and environmental usage intensity.
[0029] Among them, operating status indicators include temperature, vibration, current, voltage, power, speed, and energy consumption, which are direct parameters characterizing the real-time physical health status of equipment. Temperature data is usually collected from key parts such as motor bearings, reducer housings, electrical cabinets, and hydraulic systems. Abnormal increases in temperature often indicate faults such as increased friction, cooling failure, or insulation aging. Vibration data includes vibration acceleration, velocity, displacement, and frequency characteristics, and is the most sensitive indicator for identifying imbalance, misalignment, bearing damage, and gear wear in rotating machinery. Current, voltage, and power data reflect the electrical load characteristics and energy conversion efficiency of the equipment. Speed data is used to verify whether the equipment is operating within its rated operating range. Energy consumption data is a macroscopic indicator for measuring the overall operating efficiency of the equipment.
[0030] Reliability metrics include the number of failures, failure frequency, mean time between failures (MTBF), and mean time to repair (MTBF). These are parameters that quantify the long-term operational stability of equipment from a statistical perspective. Failure frequency is defined as: Fault frequency is defined as (Number of failures / Cumulative operating time). Where F represents the failure frequency, measured in times per hour, representing the probability of a failure occurring per unit of operating time; The number of failures within the statistical period is expressed in times. This refers to the cumulative operating time of the equipment, expressed in hours. This metric eliminates the impact of variations in operating time on fault assessment. Mean Time Between Failures (MTBF) characterizes the average uptime between two consecutive failures of the equipment; Mean Time To Repair (MTTR) reflects the repairability of the equipment and the responsiveness of maintenance resources.
[0031] Maintenance status indicators include cumulative maintenance frequency, time to the most recent maintenance, maintenance completion rate, and maintenance cost, recording the intervention history and resource consumption throughout the equipment's entire lifecycle. The cumulative maintenance frequency needs to differentiate the weighting of preventative maintenance and fault repair; the time to the most recent maintenance is used to assess the timeliness of the current health assessment results; the maintenance completion rate measures the compliance of the established maintenance plan; and maintenance cost quantifies the economic cost of maintaining equipment operation.
[0032] Environmental and operational intensity indicators include ambient temperature, ambient humidity, dust concentration, and average load rate, which describe the external stress conditions and internal load levels of the equipment. Ambient temperature, ambient humidity, and dust concentration directly affect the equipment's heat dissipation efficiency, insulation performance, and mechanical wear rate; the average load rate reflects the ratio of the equipment's actual output power to its rated capacity.
[0033] Each evaluation indicator is standardized and scored from 0 to 100 points based on equipment-specific thresholds (preferably taken from equipment technical parameters or historical operating data). Specifically, the temperature score is linearly mapped to the following levels: normal range (90-100 points), slight deviation (80-90 points), proximity alarm (60-80 points), over-alarm (40-60 points), and over-protection (0-40 points); the vibration score is based on the normal threshold. (90~100 points), warning threshold (75~90 points), alarm threshold (50~75 points) and above (0~50 points) Divide into two groups, and simultaneously when the rate of change of vibration dV / dt > When the preset growth rate threshold is reached, the score is reduced even if the current value does not reach the alarm threshold. The fault frequency is calculated according to the above formula and compared with the historical average frequency. > ( The current failure frequency, If the historical average failure frequency is used, then the reliability is considered to have decreased.
[0034] The maintenance impact index categorizes maintenance into four types: preventative maintenance, general failure maintenance, major failure maintenance, and recurring failure maintenance, assigning them progressively increasing weighting coefficients. (That is, repeated fault repairs have the greatest negative impact on equipment health), calculate the comprehensive maintenance impact value. ,in These are the number of repairs for each type of service. This represents a maintenance impact index, a dimensionless value used to quantify the overall impact of maintenance history on the health status of equipment. Number of preventative maintenance operations; This refers to the number of repairs for common faults. Number of repairs for major malfunctions; This refers to the number of times a repeated fault can be repaired.
[0035] Finally, all standardized scores will be... The health index is obtained by weighted summation. , This represents the health index of the target device, a dimensionless value ranging from 0 to 100. This represents the standardized score of the i-th evaluation indicator, with a value ranging from 0 to 100. Let represent the weight corresponding to the i-th evaluation indicator; n represents the total number of evaluation indicators. The weights of each evaluation indicator satisfy the normalization constraint: the initial weights are determined by expert knowledge (e.g., temperature 0.15, vibration 0.20, current 0.10, energy consumption 0.10, failure frequency 0.15, maintenance status 0.10, utilization rate 0.05, environmental factors 0.15), and .
[0036] The system classifies health levels based on the Health Index (HI): 90-100 indicates a healthy state, 85-89 indicates good condition, 60-74 indicates a warning, and below 60 indicates a risk.
[0037] Simultaneously calculate the rate of change of the health index: ,in, It represents the rate of change of the health index, expressed in minutes per unit of time, and is used to quantify the speed at which the health status of equipment deteriorates. This represents the device health index at the current time t; express Equipment health index before the time period; This indicates the time interval used to calculate the rate of change. When HI is in a good range but... Even when the negative value is large, the system still determines that it needs to be closely monitored to avoid misjudgment based solely on the static value.
[0038] The health assessment model includes multiple evaluation indicators and corresponding initial weights. During system operation, the weights are adjusted based on historical failure samples. Indicators that have historically experienced frequent failures have higher weights, while those with lower correlation to failures have lower weights, forming a dynamic weighting mechanism of "initial expert weights → historical data correction → failure result verification → weight updates." This step breaks down the data silos between different stages in traditional management, providing a complete and consistent data foundation for subsequent accurate decision-making.
[0039] The prediction method for the failure risk value R is based on the current health index HI and the rate of change of the health index. Historical fault frequency F, temperature change rate Vibration change rate The failure probability P of the target equipment within a set time window is calculated by considering the impact of maintenance history (M), environmental impact (E), and equipment usage intensity (U). The failure probability is then weighted according to the criticality of the target equipment and the degree of failure impact (C) to obtain the risk value R.
[0040] The failure probability P is modeled using a multi-factor integrated approach: Where P represents the probability that the target device will fail within a set time window in the future, a dimensionless value ranging from 0 to 1, and f represents a multi-factor comprehensive function, which can be implemented using logistic regression, random forest, gradient boosting model or neural network model; For the current health index, The rate of change of the health index. Historical failure frequency , These are the rates of change of temperature and vibration, respectively. Due to historical impacts of maintenance, These are environmental parameters, including ambient temperature and humidity. To determine the load factor (e.g., average load rate), the system can employ logistic regression, random forest, or neural networks, taking statistical characteristics from the most recent 24 hours and 7 days, as well as lifecycle characteristics, as input, and outputting the probability of failure within a future time window. .
[0041] Building upon this, the system further introduces the failure impact level C to quantify the severity of equipment failure consequences. The value of C is dynamically determined based on the equipment's role and business attributes within the production line: for general auxiliary equipment (such as fans and lighting power supplies), C is set to 0.2~0.4; for important production equipment (such as conveying pumps and air compressors), C is set to 0.4~0.7; for critical production equipment (such as blast furnace blowers and rolling mill main motors), C is set to 0.7~0.9; for safety-critical equipment (equipment involving personal safety or environmental risks), C is set to 0.9~1.0. These values can be read from the "Criticality Level" field in the equipment file and can be manually adjusted according to actual production importance. Finally, the system calculates the risk value R: ,in This is a dimensionless value between 0 and 100, directly used for subsequent decision matrix determination. This risk value combines the probability of a failure occurring with the severity of its consequences, avoiding wasting maintenance resources by focusing solely on high-probability but low-impact failures, or causing significant losses by ignoring low-probability but catastrophic failures.
[0042] The remaining useful life (RUL) is predicted as follows: A basic RUL is obtained based on the design life and cumulative operating time of the target equipment. This basic RUL is then corrected using health correction factors, failure correction factors, maintenance correction factors, environmental correction factors, and load correction factors to predict the remaining useful life (RUL). RUL represents the projected remaining useful life of the equipment, in years (or hours). The design life of the equipment is given in years (or hours) and can be obtained from the equipment master file. The cumulative operating time of the equipment, expressed in years (or hours), is calculated from the equipment operation records. The correction factor K is decomposed as follows: ,in, Health correction factor, derived from the current health index HI; the higher the HI, the better. The closer to 1; Fault correction factor, determined based on historical fault frequency F; the higher the fault frequency, the greater the correction factor. The smaller; The maintenance correction factor is determined based on the historical impact M of maintenance; the greater the maintenance impact, the greater the correction factor. Adjustments should be made accordingly; Environmental correction factor, determined based on environmental parameter E (such as temperature, humidity, dust concentration), increases with severity of the environment. The smaller; The load correction factor is determined based on the average load rate; if the system is operating under overload for an extended period, then... Decrease.
[0043] To facilitate on-site personnel's intuitive understanding of equipment status, the system maps the Health Index (HI) to four levels: Healthy (90≤HI≤100), indicating that the equipment is in excellent operating condition and can be monitored normally; Good (85≤HI<90), indicating that the equipment has a slight degradation trend and the rate of change needs to be monitored; Warning (60≤HI<85), indicating that the equipment has experienced a significant performance decline, and the decision to increase inspections or arrange preventive maintenance should be made based on the rate of change; Risk (HI<60), indicating that the equipment has an extremely high risk of failure and immediate intervention is required.
[0044] Step S300: Based on the health index, failure risk value, remaining lifespan, and life cycle cost, generate a comprehensive decision that includes maintenance actions.
[0045] The intelligent operation and maintenance decision module 110 receives the health index HI from the equipment health evaluation module 106, the fault risk value R output by the fault risk prediction module 107, the remaining life (RUL) output by the remaining life prediction module 108, and the life cycle cost (LCC) calculated by the life cycle cost analysis module 109.
[0046] The life cycle cost (LC) includes at least the procurement cost. Installation and commissioning costs Operating energy consumption costs Repair and maintenance costs spare parts cost , disposal costs and downtime loss costs :
[0047] in, This represents the procurement cost, i.e., the expense of purchasing the equipment; This indicates the installation and commissioning costs, including the labor and material costs required for equipment installation, wiring, and commissioning. This represents the operating energy cost, which is the cost of electricity, fuel, and other energy consumed during equipment operation. It is collected in real time from the SCADA system and smart meters. It represents the cost of maintenance and repair, including the cost of spare parts and labor consumed in routine maintenance, fault repair and major overhaul, which is automatically collected from the work order settlement records of the EAM system. This refers to the cost of spare parts that are stockpiled to maintain the normal operation of the equipment. This indicates the cost of scrapping and disposal, namely the expenses required for equipment scrapping, dismantling, transportation, and environmental treatment; This represents the cost of downtime losses, calculated based on the duration of a single downtime and the downtime loss per unit time.
[0048] in, This represents the cost of downtime losses, expressed in yuan. Indicates the duration of a single downtime, in hours; This represents downtime loss per unit time, expressed in yuan / hour, and is linked to the output loss records of the MES system and the marginal contribution rate data of the finance department. This cost data is not only used for current update decisions but also serves as historical baseline data fed back into the cost prediction model, continuously improving the accuracy of subsequent predictions.
[0049] The intelligent operation and maintenance decision module 110 does not trigger alarms based on a single threshold, but deeply couples the physical dimension of health and lifespan status with the economic dimension of cost constraints.
[0050] Among them, the comprehensive cost will continue to be used. This refers to the total expected investment required to maintain the equipment in normal operation over a future pre-defined evaluation period.
[0051] In the formula, This indicates the estimated total cost of continued use of the equipment, expressed in yuan. This represents the estimated maintenance cost, which is predicted based on the equipment's current health index (HI), historical failure frequency (F), and component wear model. It covers both preventative maintenance costs and potential fault repair costs. This represents the estimated operating energy cost, calculated by analyzing the deviation between the current energy efficiency ratio and the rated energy efficiency of the equipment, combined with future production planning and scheduling. It reflects the additional energy consumption caused by equipment performance degradation. This represents the estimated cost of failures and downtime, calculated by multiplying the predicted failure probability P by the economic loss per unit of downtime. The result reflects the hidden risk costs brought about by the decline in equipment reliability.
[0052] Replacement cost refers to the net investment required to replace existing equipment and bring it to a usable state.
[0053] In the formula, This indicates the overall cost of equipment replacement; The cost of new equipment is indicated by the latest market price or supplier quote obtained through integration with the ERP system. This indicates the cost of installing and commissioning new equipment, estimated based on standard working hours and auxiliary material consumption. This indicates the residual value of the original equipment after it is scrapped, and is dynamically determined based on the current material weight of the equipment, the market conditions for second-hand goods, or the price quoted by recyclers.
[0054] After obtaining the above two costs, the intelligent operation and maintenance decision module 110 performs a dual condition judgment: only if Furthermore, the system determines that the equipment has entered the "economically obsolete" range only when the Remaining Life (RUL) output by the Remaining Life Prediction Module 108 is simultaneously lower than a preset lifespan threshold, thereby generating a comprehensive decision that includes equipment replacement actions. This dual constraint mechanism effectively avoids frequent replacement recommendations triggered by short-term maintenance cost fluctuations, while also preventing the equipment from being forcibly extended in situations where long-term operation is extremely uneconomical, even if it is not completely damaged. If the above replacement conditions are not met, the system reverts to the physical state-driven logic, generating a comprehensive decision that includes maintenance or inspection actions based on the range and trend of the health index, in order to maintain the equipment in its optimal operating state at the current stage of its life cycle.
[0055] Based on the health index, a comprehensive decision is generated that includes maintenance or inspection actions. This includes: when the health index is in a first preset range (90≤HI≤100), a normal operation decision is generated; when the health index is in a second preset range (85≤HI<90) and the rate of change of the health index... When the health index is below the change rate threshold, a decision is generated to increase the inspection frequency; when the health index is in the third preset range (60≤HI<85) or the change rate of the health index... When the health index is above the rate of change threshold, a preventive maintenance decision is generated; when the health index is in the fourth preset interval (HI<60), an emergency maintenance decision is generated; the numerical ranges of the first, second, third and fourth preset intervals decrease sequentially and do not overlap.
[0056] In the above-mentioned hierarchical decision-making logic, the rate of change of the health index This is the key criterion for distinguishing the decision path between the second and third preset intervals. The system introduces a rate of change threshold as an auxiliary judgment condition: when the health index is in the second preset interval, only when... Only when the rate of change is below (i.e., the absolute value of the negative value is greater than) a preset threshold is the equipment deemed to be rapidly deteriorating despite its current acceptable score, thus generating a decision to increase the frequency of inspections. If the rate of change does not exceed the threshold, it is considered normal fluctuation or slow aging, and routine inspections are maintained. When the health index falls into the third preset range, it indicates that the equipment risk has reached a level that cannot be ignored. Regardless of the rate of deterioration, a preventive maintenance decision must be generated directly to eliminate potential faults.
[0057] It should be noted that, to further improve the accuracy and robustness of trend judgment, this application introduces an adaptive time window mechanism when calculating the rate of change of the health index. Traditional fixed time window methods often struggle to accommodate the varying sensitivity requirements for trends at different health stages: a time window that is too long can smooth out early rapid deterioration signals, leading to missed detections; a time window that is too short is susceptible to sensor noise or transient operating conditions, resulting in false alarms. Therefore, this embodiment dynamically adjusts the time interval in the above formula based on the current health interval of the device. The value to be determined. Specifically, when the device is in the first preset range or the second preset range (i.e., healthy or good condition), the system will... By setting a relatively long time span (e.g., 7 days or 30 days), historical data is smoothed using a long window, effectively filtering out short-term random fluctuations caused by changes in environmental temperature and humidity, temporary load adjustments, etc., ensuring that only a continuous degradation trend will trigger an alert; and when the equipment enters the third or fourth preset interval (i.e., alert or high-risk state), the system automatically... Switching to a shorter time span (e.g., 24 hours or 4 hours) utilizes a shorter window to improve the sensitivity to sudden state changes, ensuring that the inflection point of accelerated fault deterioration can be captured as soon as possible. Furthermore, near the critical point of interval switching, the system can employ a weighted fusion strategy to simultaneously calculate the rate of change under both the longer and shorter time windows and perform a comprehensive evaluation, thus avoiding judgment jumps caused by switching to a single window.
[0058] It should be noted that, considering the complexity of industrial environments and the nonlinear characteristics of equipment degradation processes, future maintenance costs, energy consumption levels, and failure rates often exhibit significant uncertainty. Using only point estimates for cost comparisons can easily lead to misjudgments due to single instances of abnormal data or short-term fluctuations. Therefore, this embodiment further introduces an uncertainty quantification mechanism into cost prediction. Specifically, the system no longer outputs a single, definitive cost value, but instead uses Monte Carlo simulation based on the distribution characteristics of historical data to... , and Thousands of random sampling simulations were conducted on key parameters to construct a probability distribution function for the overall cost of continued use. Based on this, the system uses confidence intervals (such as the upper limit of cost at a 95% confidence level) as the decision-making basis, rather than a simple mean. For example, even if the mean of the cost of continued use is slightly lower than the replacement cost, if its 95% confidence upper limit significantly exceeds the replacement cost, indicating a very high risk of high maintenance expenditures in the future, the system will still tend to trigger an early update warning or recommend in-depth technical upgrades.
[0059] In step S400, in response to the execution of maintenance actions, the status improvement data of the target device is obtained.
[0060] Once the maintenance actions generated by the intelligent maintenance decision module 110 are converted into work orders and issued by the work order execution and feedback module 111, on-site personnel or automated systems will perform the corresponding repair, maintenance, or adjustment operations. The maintenance actions generated by the intelligent decision module are converted into work orders, which include fields such as work order number, unique equipment identifier, equipment name, risk level, abnormal indicators, abnormal time, recommended maintenance items, recommended completion time, required spare parts list, estimated working hours, estimated cost, and safety precautions.
[0061] After a work order is completed, the system not only records the execution result, but more importantly, it uses the multi-source data acquisition module 103 to collect the latest operating parameters of the equipment after the maintenance action is performed, and compares them with the status data before the execution, thereby quantifying the status improvement data. This status improvement data is not just a static text record in the maintenance log, but includes the change in health index before and after maintenance. Numerical feedback signals that can be calculated include the degree of recovery of key characteristic values (such as vibration amplitude and temperature deviation) and the extent of reduction in failure risk.
[0062] The maintenance actions generated by the intelligent operation and maintenance decision module 110 are converted into work orders. Each work order includes fields such as work order number, unique equipment identifier, equipment name, risk level, abnormal indicators, abnormal time, recommended repair items, recommended completion time, required spare parts list, estimated working hours, estimated cost, and safety precautions. After the work order is executed, the system re-collects equipment operating parameters and calculates the post-maintenance health index. Risk Value Remaining lifespan and lifecycle costs And compare it with the state before execution to obtain the state improvement amount (e.g. If the improvement amount after multiple executions of a certain type of maintenance action is consistently lower than the preset improvement threshold, the system automatically reduces the recommended priority of that action in subsequent decisions; otherwise, it maintains or increases it. Simultaneously, the updated health status and cost data are written back to the lifecycle database, forming a complete closed loop: equipment registration → data collection → status analysis → health assessment → prediction → decision-making → work order → result feedback → status update → model parameter update → next cycle.
[0063] Step S500: Based on the status improvement data, adjust the parameters of the health assessment model and adjust the recommended priority of operation and maintenance actions.
[0064] The system inputs the acquired status improvement data as feedback signals to the equipment health evaluation module 106 and the intelligent operation and maintenance decision module 110 to form a closed-loop control.
[0065] On the one hand, the health assessment model includes multiple evaluation indicators and corresponding initial weights; based on the condition improvement data, the parameters of the health assessment model are corrected, including: based on the condition improvement data, determining the correlation between each evaluation indicator and equipment failure. Based on correlation and state change in state improvement data The initial weights of each evaluation indicator are dynamically adjusted to obtain the updated weights. .
[0066] During the system initialization phase, the equipment health assessment module 106 sets a set of initial weights for each evaluation indicator in the health assessment model based on the technical manuals, industry standards, or domain expert experience provided by the equipment manufacturer. However, due to individual differences in equipment and the complexity of the operating environment, this set of static initial weights often fails to accurately reflect the true degradation pattern of the equipment over a long period. Therefore, this embodiment uses the acquired state improvement data as a feedback signal to adaptively correct the model parameters.
[0067] When determining the correlation between various evaluation indicators and equipment failures, the system does not simply employ linear statistical methods such as the Pearson correlation coefficient, because the failure mechanisms of industrial equipment often exhibit highly nonlinear and coupled characteristics. To more accurately capture the complex mapping relationship between evaluation indicators and the actual health status of equipment, this application employs interpretable artificial intelligence methods in calculating the correlation. Specifically, the system can use mutual information to quantify the degree of nonlinear dependence between indicators and failure states; or use grey relational analysis to handle the ranking of indicator correlations under conditions of small samples and limited information; furthermore, it can introduce the SHAP (SHapley Additive exPlanations) value method to accurately analyze the marginal contribution of each evaluation indicator to the current health score based on game theory principles. This correlation calculation method based on interpretable AI can effectively identify key weak features that are ignored in traditional linear analysis but actually have a strong indicative role for specific failure modes, thereby avoiding the model from getting trapped in local optima or overfitting.
[0068] After obtaining the correlation, the system combines the state change data in the state improvement data and dynamically adjusts the initial weights of each evaluation indicator using the following formula to obtain the updated weights:
[0069] In the formula, Indicates the first The updated weights of each evaluation indicator in this revision; Indicates the first The old weights of each evaluation indicator before this revision; Indicates the first The correlation between each evaluation metric and equipment failure or performance recovery is calculated using the aforementioned interpretable AI method; Indicates the first The change in the standardized score of each evaluation indicator before and after the execution of the operation and maintenance action, that is, the degree of recovery of the indicator in the status improvement data. The larger the absolute value, the more sensitive the indicator is to the operation and maintenance response. Indicates the total number of evaluation indicators; This represents the learning rate or correction step size factor, ranging from 0 to 1. It controls the magnitude of each correction to prevent the model from updating too quickly and becoming unstable. The physical meaning of this formula is: if a certain indicator is highly correlated with faults (… (Large), and significantly improved after this repair ( If a candidate is large, their voice in the overall evaluation should be enhanced; otherwise, it should be suppressed.
[0070] Furthermore, to avoid drastic fluctuations in model weights due to single abnormal maintenance records or occasional sensor malfunctions, this embodiment introduces a time-dimensional smoothing constraint into the weight adjustment mechanism. Specifically, the system sets up a historical memory mechanism based on a sliding window or an exponential decay factor. In one implementation, the formula above... The decision is not solely based on the most recent status improvement data, but rather on a weighted moving average of the correlation among the most recent N (e.g., N=10) valid maintenance events, with more recent events assigned a higher timeliness weight. In another implementation, the system automatically initiates a confidence check process after each weight update: only when the weight adjustment trend in the same direction remains consistent within M consecutive time windows is a change allowed. If the preset safety boundary threshold is exceeded, the system will either maintain the current weight or make only minor, tentative adjustments.
[0071] It should be noted that the modification of the health assessment model parameters in this application is not limited to adjusting the numerical values of the weights, but can also include the dynamic addition or subtraction of the evaluation index system or the adaptive switching of the model structure. For example, when the system detects repeatedly that the contribution of a certain type of sensor data to the improvement of the state approaches zero, it can automatically remove the index from the current evaluation set or perform dimensionality reduction; conversely, if a new abnormal pattern is found that is unrelated to the existing indicators, the automatic mining of new features can be triggered and incorporated into the evaluation system. Furthermore, for different degradation stages of equipment (such as the break-in period, the stable period, and the wear-out period), the system can adaptively switch between multiple preset sub-models, selecting the model structure that best matches the current physical state for calculation, rather than always using a single fixed model.
[0072] On the other hand, the recommended priority of operation and maintenance actions is revised, including: statistically analyzing the actual improvement in the status of operation and maintenance actions after multiple executions. When the actual state improves If the actual improvement is below the preset improvement threshold Th multiple times consecutively, the recommended priority of the maintenance action in the overall decision-making will be reduced; if the actual improvement is higher than or equal to the preset improvement threshold, the recommended priority will be maintained or increased.
[0073] Specifically, the intelligent operation and maintenance decision module 110 obtains the status improvement data after the execution of operation and maintenance actions through the work order execution and feedback module 111, and performs quantitative statistics on it to calculate the actual status improvement. This actual status improvement is not a qualitative description of the maintenance result, but a numerical measure based on the changes in key indicators before and after operation and maintenance. For example, for the operation and maintenance action of "replacing the bearing", the actual status improvement could be the difference between the health index HI before and after maintenance. It can also be the root mean square value of vibration. The system tracks the magnitude of the decrease or the degree of recovery from temperature deviation. It maintains an effect tracking queue in the lifecycle database for each type of operational action, recording the actual state improvement sequence after the action's N most recent executions, thus providing a statistical basis for subsequent priority determination.
[0074] In the judgment logic, the preset improvement threshold Th serves as a baseline for measuring the effectiveness of maintenance actions. This threshold is not fixed but can be configured differently based on equipment type, failure mode, and historical maintenance data. For example, for spindle maintenance of precision CNC machine tools, the preset improvement threshold might be set to an HI improvement of more than 15 points or a vibration decrease of more than 30%; while for tension adjustment of ordinary conveyor belts, the threshold might only be an HI improvement of 5 points. The initial threshold can be determined based on the maintenance effectiveness standards recommended by the equipment manufacturer or the average improvement level of similar equipment in the company's history, and is dynamically updated as maintenance data accumulates.
[0075] When the system detects the actual improvement in status of a certain maintenance action When the improvement amount is higher than or equal to the preset improvement threshold Th, it indicates that the action is effective in restoring the current equipment status. The system will maintain its current recommended priority, or appropriately increase its priority after the cumulative effective number reaches a certain amount, so that it will be recommended first in similar operating conditions in the future. Conversely, if the actual status improvement is lower than the preset improvement threshold multiple times consecutively, it indicates that the action may no longer be applicable to the current equipment degradation mode, or there may be systemic problems such as inadequate process execution. In this case, the system will automatically reduce the recommended priority of this maintenance action in the overall decision-making process to avoid resource waste and downtime losses caused by ineffective maintenance.
[0076] It should be noted that, to avoid the mistaken elimination of effective strategies due to single-time maintenance personnel errors or occasional interference, and to promptly identify truly failed maintenance solutions, this application introduces a tiered response strategy in the priority correction mechanism, rather than a simple binary upgrade / downgrade judgment. Specifically, the system defines different levels of response actions based on the number of consecutive times k that the actual improvement amount falls below a preset improvement threshold. In a preferred embodiment, when a maintenance action fails to meet the actual improvement target for the first time (k=1), the system only marks it as "observation status" and adds a verification prompt for the execution quality of the action when generating the next decision, but does not change its recommendation priority; when it fails to meet the target three times in a row (k=3), the system determines that the effectiveness of the action is questionable, reduces its recommendation priority by one level, and simultaneously triggers the search and recommendation of alternative maintenance solutions, such as replacing "cleaning the filter" with "replacing the filter" or upgrading "on-site dynamic balancing" to "return to factory fine balancing"; when it fails to meet the target five times or more in a row (k≥5), the system determines that the action has substantially failed under the current working conditions, automatically freezes it from the available maintenance action library, prohibits subsequent decisions from calling it again, and simultaneously generates a model retraining request, prompting algorithm engineers or domain experts to intervene and analyze the cause of failure, and use the accumulated negative sample data to perform in-depth optimization of the fault prediction model or decision rule base. This hierarchical and progressive negative feedback mechanism not only ensures the robustness of strategy evolution and prevents the model from overfitting to noise, but also ensures that it can quickly stop losses and iterate itself when faced with real failure modes, significantly improving the practical value and credibility of the intelligent operation and maintenance system in long-term operation.
[0077] Through the aforementioned dual correction mechanism, this embodiment constructs a complete technical closed loop of state perception, decision execution, effect feedback, and model evolution. The final result is: equipment documentation → data acquisition → state analysis → health assessment → fault prediction → lifespan prediction → cost analysis → maintenance / modification / upgrade decision → work order execution → maintenance result collection → health status update → model parameter update → entering the next lifecycle management cycle. This enables the operation and maintenance system to possess continuous learning capabilities similar to human experts, fundamentally solving the technical challenge of static models being unable to adapt to complex and changing operating conditions.
[0078] Furthermore, multi-source data from all stages of the equipment's lifecycle are collected and correlated, including: automatic collection of equipment operating parameter data through sensors; collection of manual inspection data, maintenance work order data, and business data from ERP, MES, SCADA, and EAM systems through mobile terminals or business system interfaces; and unified correlation of the collected data using unique equipment identifiers and timestamps as indexes.
[0079] In the specific implementation of multi-source data acquisition, for devices with sensor interfaces, the multi-source data acquisition module 103 achieves automatic acquisition of operating parameter data by deploying sensors in key locations. For example, temperature sensors are installed in locations such as motor bearings, reducer housings, electrical cabinets, and hydraulic systems to acquire temperature data T; vibration sensors are installed in motor bearing housings, pump bodies, or reducer housings to acquire vibration acceleration, velocity, displacement, and frequency characteristic values V; operating current I is acquired through current transformers or Hall effect sensors; operating voltage U is acquired through voltage acquisition modules; active power is acquired using smart meters or power calculation modules; and rotational speed N is acquired through encoders or controller communication interfaces.
[0080] For data that cannot be directly collected automatically, it is acquired through mobile terminals or business system interfaces. For example, inspection personnel scan equipment QR codes or RFID tags using mobile terminals, fill in the status of oil leaks, abnormal noises, temperature rises, etc., according to the inspection list automatically displayed by the system, and upload multimedia attachments; maintenance personnel enter fault details, maintenance measures, and cost information in the work order system; at the same time, the system regularly synchronizes procurement and spare parts data from the ERP system, production scheduling and output data from the MES system, real-time monitoring data from the SCADA system, and asset ledgers and maintenance plans from the EAM system through standard API interfaces, achieving comprehensive aggregation of cross-system data. The collected data is uniformly associated using unique equipment identifiers and timestamps as indexes.
[0081] Furthermore, a comprehensive decision including maintenance actions is generated, including: determining the comprehensive cost of continued use and the comprehensive cost of replacement of the target equipment; when the comprehensive cost of continued use is greater than the comprehensive cost of replacement and the remaining lifespan is lower than a preset lifespan threshold, a comprehensive decision including equipment replacement actions is generated; otherwise, a comprehensive decision including maintenance actions or inspection actions is generated based on the health index.
[0082] When generating comprehensive decisions, the intelligent operation and maintenance decision module 110 does not solely rely on the physical health status of the equipment, but incorporates economic dimensions into the decision-making closed loop. The system first calculates the comprehensive cost of continued use and the comprehensive cost of replacement of the target equipment at the current moment through the lifecycle cost analysis module 109. The comprehensive cost of continued use refers to the total expected investment required to maintain the equipment's normal operation over a preset evaluation period, and its calculation formula is as follows:
[0083] Accordingly, the total cost of replacement refers to the net investment required to replace existing equipment and bring it to a usable state, and its calculation formula is as follows:
[0084] The meaning and acquisition method of each parameter in the above two formulas have been described in detail in the above embodiments.
[0085] After obtaining the above two costs, the intelligent operation and maintenance decision module 110 performs a dual condition judgment: only if Furthermore, the system determines that the equipment has entered the "economically obsolete" range only when the Remaining Life (RUL) output by the Remaining Life Prediction Module 108 is simultaneously lower than a preset lifespan threshold, thereby generating a comprehensive decision that includes equipment replacement actions. This dual constraint mechanism effectively avoids frequent replacement recommendations triggered by short-term maintenance cost fluctuations, while also preventing the equipment from being forcibly extended in situations where long-term operation is extremely uneconomical, even if it is not completely damaged. If the above replacement conditions are not met, the system reverts to the physical state-driven logic, generating a comprehensive decision that includes maintenance or inspection actions based on the range and trend of the health index, in order to maintain the equipment in its optimal operating state at the current stage of its life cycle.
[0086] Furthermore, the total cost of continued use includes estimated repair and maintenance costs. Expected operating energy costs and estimated downtime costs due to malfunctions The overall cost of the upgrade includes the cost of purchasing new equipment. New equipment installation and commissioning costs and the residual value of the original equipment after scrapping The life cycle cost includes procurement cost, installation and commissioning cost, operating cost, maintenance cost, spare parts cost, disposal cost, and downtime loss cost.
[0087] Among them, the estimated maintenance cost in the comprehensive cost continues to be used to represent the estimated maintenance cost. This parameter is based on the current health index HI of the equipment, the historical failure frequency F, and the component wear model prediction, and covers the cost of preventive maintenance and potential failure repair. The estimated operating energy cost represents the estimated operating energy cost. It is calculated by analyzing the deviation between the current energy efficiency ratio and the rated energy efficiency of the equipment, combined with the future production plan scheduling, and reflects the additional energy consumption caused by the performance degradation of the equipment. This represents the estimated cost of failures and downtime losses, calculated by multiplying the predicted failure probability by the economic loss per unit of downtime. It reflects the hidden risk costs resulting from decreased equipment reliability.
[0088] The new equipment purchase cost in the updated comprehensive cost represents the cost of purchasing new equipment, which is obtained by connecting to the latest market price or supplier quotation in the ERP system; the new equipment installation and commissioning cost represents the cost of installing and commissioning new equipment, which is estimated based on standard working hours and auxiliary material consumption; the residual value of the original equipment after scrapping represents the residual value of the original equipment after scrapping, which is dynamically determined based on the current material weight of the equipment, the second-hand market conditions, or the price quoted by recyclers.
[0089] The meaning and acquisition methods of each cost in the Life Cycle Cost (LCC) have been described in detail in the foregoing embodiments. Operating costs and energy consumption data are collected in real time from the SCADA system and smart meters. Maintenance and repair costs and spare parts costs are automatically collected from the work order settlement records of the EAM system. Downtime loss costs are linked to the production loss records of the MES system and the marginal contribution rate data of the finance department, thereby ensuring the comprehensiveness of cost composition and the traceability of data.
[0090] Furthermore, the health assessment model includes multiple assessment indicators and corresponding initial weights; based on the state improvement data, the parameters of the health assessment model are corrected, including: determining the correlation between each assessment indicator and equipment failure based on the state improvement data; and dynamically adjusting the initial weights of each assessment indicator based on the correlation and the state change amount in the state improvement data to obtain updated weights.
[0091] Specifically, during the system initialization phase, the equipment health assessment module 106 sets a set of initial weights for each evaluation indicator in the health assessment model based on the technical manuals, industry standards, or domain expert experience provided by the equipment manufacturer. However, due to individual differences in equipment and the complexity of the operating environment, this set of static initial weights often fails to accurately reflect the true degradation pattern of the equipment over a long period. Therefore, this embodiment uses the acquired state improvement data as a feedback signal to adaptively correct the model parameters.
[0092] When determining the correlation between each evaluation indicator and equipment failure, the system does not simply employ linear statistical methods such as the Pearson correlation coefficient, because the failure mechanisms of industrial equipment often exhibit highly nonlinear and coupled characteristics. Specifically, to more accurately capture the complex mapping relationship between evaluation indicators and the actual health status of the equipment, this application employs interpretable artificial intelligence methods when calculating the correlation. Specifically, the system can use mutual information to quantify the degree of nonlinear dependence between indicators and failure states; or it can use grey relational analysis to handle the ranking of indicator correlations under conditions of small samples and limited information; furthermore, it can introduce the SHAP (SHapley Additive exPlanations) value method, based on game theory principles, to accurately analyze the marginal contribution of each evaluation indicator to the current health score result.
[0093] After obtaining the correlation, the system combines the state change data in the state improvement data and dynamically adjusts the initial weights of each evaluation indicator using the following formula to obtain the updated weights. (The formula has been described in detail above).
[0094] Furthermore, to avoid drastic fluctuations in model weights due to a single abnormal maintenance record or occasional sensor malfunction, this embodiment introduces a time-dimensional smoothing constraint into the weight adjustment mechanism. Specifically, the system sets up a historical memory mechanism based on a sliding window or an exponential decay factor. In one implementation, the correlation in the above formula is not determined solely by the most recent state improvement data, but rather by the weighted moving average of the correlation among the most recent N (e.g., N=10) valid maintenance events, with more recent events assigned higher time-related weights. In another implementation, after each weight update, the system automatically initiates a confidence check process: only when the weight adjustment trend in the same direction remains consistent within M consecutive time windows is a change allowed. If the preset safety boundary threshold is exceeded, the system will either maintain the current weight or make only minor, tentative adjustments.
[0095] Furthermore, the prediction method for the fault risk value is as follows: based on the current health index, the trend of health index changes, the rate of temperature change, the rate of vibration change, historical fault frequency, and equipment usage intensity, the probability of the target equipment failing within a future set time window is calculated. Then, the fault probability is weighted according to the criticality of the target equipment and the degree of impact of the fault to obtain the risk value. The fault probability P has been described in detail in the preceding embodiments. The system can use logistic regression, random forest, or neural networks, taking the statistical characteristics of the most recent 24 hours and 7 days, as well as lifecycle characteristics, as input, and outputting the fault probability P within the future time window.
[0096] Based on the failure probability P, the system further introduces the failure impact degree C to quantify the severity of equipment failure consequences. The value of C is dynamically determined according to the equipment's role and business attributes in the production line: for ordinary auxiliary equipment (such as ventilators and lighting power supplies), C is set to 0.2~0.4; for important production equipment (such as conveying pumps and air compressors), C is set to 0.4~0.7; for critical production equipment (such as blast furnace blowers and rolling mill main motors), C is set to 0.7~0.9; for safety-critical equipment (equipment involving personal safety or environmental risks), C is set to 0.9~1.0. The above values can be read from the "criticality level" field in the equipment file and can be manually adjusted according to the actual production importance. Finally, the system calculates the risk value R: R=P×C×100, where R is a dimensionless value between 0 and 100, directly used for subsequent decision matrix determination. This risk value combines the probability of a failure to the severity of its consequences, avoiding the waste of maintenance resources by focusing solely on high-probability but low-impact failures, or the significant losses caused by ignoring low-probability but catastrophic failures.
[0097] The remaining service life is predicted as follows: A basic remaining service life is obtained based on the design life and cumulative operating time of the target equipment. This basic remaining service life is then corrected using health correction factors, fault correction factors, maintenance correction factors, environmental correction factors, and load correction factors to predict the remaining service life. ,in, The meaning of each correction coefficient has been described in detail in the foregoing embodiments. After accumulating sufficient historical data, the system further uses a machine learning model to perform a secondary correction on the rule calculation results, forming a hybrid prediction of "rule base value + data-driven correction".
[0098] Lifecycle costs include at least procurement costs. Installation and commissioning costs Operating energy consumption costs Repair and maintenance costs spare parts cost , disposal costs and downtime loss costs :
[0099] The meaning and acquisition method of each cost, which is calculated based on the downtime of a single downtime and the downtime loss per unit time, have been described in detail in the foregoing embodiments.
[0100] The evaluation indicators of the health assessment model include: operational status indicators, including temperature, vibration, current, voltage, power, speed, and energy consumption; reliability indicators, including the number of failures, failure frequency, mean time between failures, and mean time to repair; maintenance status indicators, including the cumulative number of repairs, the time to the most recent repair, maintenance completion rate, and maintenance costs; and environmental and usage intensity indicators, including ambient temperature, ambient humidity, dust concentration, and average load rate. The specific meaning and scoring method of each indicator have been described in detail in the foregoing embodiments.
[0101] In real-world industrial scenarios, the four types of indicators mentioned above do not act in isolation but exhibit complex nonlinear coupling effects. If a single indicator is evaluated independently using only a fixed threshold, false alarms or missed alarms are easily caused by changes in operating conditions. Therefore, this application introduces a dynamic compensation mechanism for the coupling effects between indicators into the health evaluation model. Specifically, regarding the coupling between ambient temperature and vibration indicators, the system establishes a dynamic compensation model for vibration thresholds based on thermodynamic characteristics: when the ambient temperature exceeds the reference temperature (e.g., 40℃), considering the natural increase in vibration amplitude due to the decrease in metal material stiffness and lubricating oil viscosity, the system automatically raises the vibration alarm threshold or introduces a temperature-vibration joint scoring function to avoid misjudging normal thermally induced vibration increments as mechanical faults. Regarding the coupling between load rate and fault frequency, the system adopts a fault rate calculation method based on operating condition normalization: during high load rate operation (e.g., greater than 90% of rated power), the system divides the actual number of faults by the load weighting coefficient before calculating the fault frequency, thereby removing the inevitable risk increment brought about by high-intensity use, making the evaluation results more reflective of the inherent health of the equipment rather than simply the intensity of use. Furthermore, the system pre-sets corresponding cross-correction factors for the coupling relationships between humidity and insulation resistance, dust concentration and temperature rise rate, etc. This multi-dimensional coupling evaluation mechanism enables the health index to adaptively reflect the true health level of the equipment under complex and variable operating conditions, significantly improving the robustness and engineering practical value of intelligent operation and maintenance decisions.
[0102] Furthermore, the recommended priority of maintenance actions is revised, including: statistically analyzing the actual improvement in status after multiple executions of maintenance actions; reducing the recommended priority of maintenance actions in comprehensive decision-making when the actual improvement is lower than the preset improvement threshold for multiple consecutive times; and maintaining or increasing the recommended priority when the actual improvement is higher than or equal to the preset improvement threshold.
[0103] Specifically, the intelligent operation and maintenance decision module 110 obtains the status improvement data after the operation and maintenance actions are executed through the work order execution and feedback module 111, and performs quantitative statistics on it to calculate the actual status improvement. The system maintains an effect tracking queue for each type of operation and maintenance action in the lifecycle database, recording the actual state improvement sequence after the action has been executed the most recent N times, thus providing a statistical basis for subsequent priority determination.
[0104] In the judgment logic, the preset improvement threshold Th serves as a baseline for measuring the effectiveness of maintenance actions. This threshold can be configured differently based on equipment type, failure mode, and historical maintenance data.
[0105] When the system detects the actual improvement in status of a certain maintenance action When the value is ≥Th, it indicates that the action is effective in restoring the current device state, and the system will maintain its current recommended priority, or appropriately increase its priority after the cumulative effective times reach a certain number. Conversely, if the actual state improvement is lower than the preset improvement threshold multiple times consecutively, i.e. If the value is less than Th for k consecutive times (where k is the number of consecutive failures), it indicates that the action may no longer be applicable to the current equipment degradation mode. In this case, the system will automatically reduce the recommended priority of this maintenance action in the overall decision-making process.
[0106] To avoid the misuse of effective strategies due to operator errors or occasional interference, and to promptly identify truly failed maintenance solutions, this application introduces a tiered response strategy in its priority correction mechanism, rather than a simple binary upgrade / downgrade judgment. Specifically, the system defines different levels of response actions based on the number of consecutive times k that the actual improvement amount falls below a preset improvement threshold: When k=1, the system only marks it as "observational" and adds a verification prompt for the execution quality of the action in the next decision generation, but does not change its recommended priority; when k≥3, the system determines that the effectiveness of the action is questionable, lowers its recommended priority by one level, and simultaneously triggers the search and recommendation of alternative maintenance solutions; when k≥5, the system determines that the action has substantially failed under the current operating conditions, automatically freezes it from the available maintenance action library, prohibits subsequent decision-making from calling it again, and simultaneously generates a model retraining request, prompting algorithm engineers or domain experts to intervene and analyze the cause of failure, and to use the accumulated negative sample data to deeply optimize the fault prediction model or decision rule base. This hierarchical and progressive negative feedback mechanism not only ensures the robustness of policy evolution and prevents the model from overfitting to noise, but also ensures that it can quickly stop losses and iterate itself when faced with real failure modes.
[0107] In one example, the intelligent operation and maintenance application scenario of an industrial centrifugal pump throughout its entire lifecycle is used to fully verify the intelligent operation and maintenance method for equipment provided in this application. This scenario applies the abstract technical solutions described in the foregoing embodiments to a specific industrial site, demonstrating the closed-loop effect of the entire process from data collection, status assessment, hierarchical decision-making, feedback correction to cost-driven updates.
[0108] The target equipment is an industrial centrifugal pump, with a unique equipment identifier of EQ-01-PUMP-003-20260015. At the beginning of the equipment's lifecycle, the system establishes a digital file for the pump through the equipment identity and file management module 101, recording its purchase cost as 500,000 yuan, installation and commissioning cost as 50,000 yuan, design life as 10 years, rated power as 75kW, and rated speed as 1480r / min. Subsequently, the multi-source data acquisition module 103 deploys temperature sensors, vibration sensors, current sensors, and voltage acquisition modules at key locations on the pump, continuously collecting operating parameters at a minute-level frequency. Simultaneously, it synchronizes spare parts consumption records from the ERP system and production data from the MES system through the business system interface. All data is stored in the lifecycle database using EQ-01-PUMP-003-20260015 as the primary key, forming a complete digital foundation.
[0109] After three years of operation, the equipment has accumulated 18,000 hours of runtime, with 3 historical failures and 5 repairs, the most recent repair occurring 1,200 hours ago. At this point, the equipment health assessment module 106 calculates a current health index (HI) of 68 based on multi-dimensional evaluation indicators; the fault risk prediction module 107 calculates a fault risk value (R) of 52 based on health trends and maintenance history; and the remaining life prediction module 108, after considering health correction, fault correction, and environmental correction coefficients, predicts a remaining life (RUL) of 2.4 years. According to the described hierarchical decision-making logic, the health index of 68 falls within the second preset range (60 ≤ HI < 85), and the rate of change of the health index calculated by the system through an adaptive time window mechanism does not exceed the preset rate of change threshold, indicating that although the equipment has entered a warning state, the rate of deterioration is still controllable. Therefore, the intelligent operation and maintenance decision module 110 does not directly generate a maintenance work order, but instead generates a comprehensive decision to increase the frequency of inspections, requiring the original weekly inspections to be adjusted to daily inspections to more intensively capture potential fault symptoms and avoid resource waste caused by premature intervention.
[0110] As the equipment continued to operate, the vibration index showed a rapid deterioration trend. The equipment health assessment module 106 detected a continuous increase in the root mean square value of vibration, causing the health index HI to drop to 57 points, while the fault risk prediction module 107 output a fault risk value R that rose to 78. At this point, the health index had fallen into the fourth preset range (HI<60), and the fault risk had significantly increased. Based on the hierarchical decision-making logic, the intelligent operation and maintenance decision module 110 immediately generated a comprehensive decision including emergency repair actions, and automatically issued a repair work order through the work order execution and feedback module 111, clearly recommending the repair item as "bearing inspection and replacement," and associating the required spare parts model and safety precautions.
[0111] After the maintenance personnel replaced the bearing according to the work order, they reported the maintenance results via a mobile terminal. The multi-source data acquisition module 103 collected the equipment operating parameters again, showing that the vibration value had significantly decreased from 8.2 mm / s before the maintenance to 3.1 mm / s. The equipment health evaluation module 106 recalculated the health index HI, which recovered to 88 points. The system quantified the difference in state before and after the maintenance as state improvement data, confirming it as a valid maintenance event. Based on this, the model correction mechanism was triggered: on the one hand, the equipment health evaluation module 106 recalculated the correlation between the vibration index and the bearing failure based on the state improvement data, and appropriately increased the weight of the vibration index in subsequent evaluations according to the weight update formula; on the other hand, the intelligent operation and maintenance decision module 110 counted that the actual state improvement of the "bearing replacement" operation and maintenance action was higher than the preset improvement threshold, and thus maintained its recommended priority in the same type of failure mode. This process fully embodies the adaptive closed loop of "state perception - decision execution - effect feedback - model evolution".
[0112] As equipment enters the later stages of its service life, maintenance frequency and energy consumption levels continue to rise due to accelerated component aging. The lifecycle cost analysis module 109 provides real-time statistics on various expenditures and, based on Monte Carlo simulations, predicts the comprehensive cost of continued use for the next year. When the system determines Greater than the overall cost of replacement When the Remaining Life Prediction (RUL) output by the Remaining Life Prediction Module 108 is lower than the preset lifespan threshold, the Intelligent Operation and Maintenance Decision Module 110 no longer recommends repair or modification. Instead, it generates a comprehensive decision that includes equipment replacement actions, providing a scientific basis for enterprise asset replacement. The calculation formulas for continuing to use the comprehensive cost and the comprehensive replacement cost are as follows: In the formula, This indicates the estimated total cost of continuing to use the equipment; This indicates the estimated repair and maintenance costs, which are predicted based on the current health status and historical failure frequency. This indicates the projected operating energy cost, calculated based on the current energy efficiency deviation and production plan. This represents the estimated cost of failure and downtime losses, calculated by multiplying the failure probability by the downtime loss per unit time. In the formula, This indicates the overall cost of equipment replacement; This indicates the cost of purchasing new equipment; This indicates the cost of installing and commissioning new equipment; This indicates the residual value of the original equipment after it has been scrapped.
[0113] Corresponding to the above method, embodiments of this application also provide an intelligent equipment operation and maintenance device, such as... Figure 2 As shown, the device includes: The acquisition unit 210 is used to collect and associate multi-source data at each stage of the entire life cycle of the device, using the unique device identifier of the target device as the association primary key. The prediction unit 220 is used to construct a health evaluation model based on the multi-source data, calculate the health index of the equipment according to the evaluation indicators in the health evaluation model, predict the failure risk value according to the health index and its changing trend, and predict the remaining lifespan according to the cumulative operating information and health status of the equipment. The generation unit 230 is used to generate a comprehensive decision including operation and maintenance actions based on the health index, fault risk value R, remaining lifespan and life cycle cost. Acquisition unit 240 is used to acquire status improvement data of the target device in response to the execution of the operation and maintenance action; The correction unit 250 is used to correct the parameters of the health evaluation model based on the status improvement data, and to correct the recommended priority of the maintenance actions. The functions of each functional unit of the intelligent equipment maintenance device provided in the above embodiments of this application can be implemented through the above-described method steps. Therefore, the specific working process and beneficial effects of each unit in the intelligent equipment maintenance device provided in the embodiments of this application will not be repeated here.
[0114] This application also provides an electronic device, such as... Figure 3 As shown, it includes a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340.
[0115] Memory 330 is used to store computer programs; When the processor 310 executes the program stored in the memory 330, it performs the following steps: Using the unique device identifier of the target device as the association primary key, collect and associate multi-source data at each stage of the device's entire lifecycle; A health assessment model is constructed based on the multi-source data. The health index of the equipment is calculated according to the evaluation indicators in the health assessment model. The failure risk value is predicted according to the health index and its changing trend. The remaining lifespan is predicted according to the cumulative operating information and health status of the equipment. Based on the health index, failure risk value, remaining lifespan, and lifecycle cost, a comprehensive decision including maintenance actions is generated. In response to the execution of the maintenance action, the status improvement data of the target device is obtained; Based on the improved status data, the parameters of the health assessment model are corrected, and the recommended priority of the maintenance actions is also adjusted.
[0116] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.
[0117] The communication interface is used for communication between the aforementioned electronic devices and other devices.
[0118] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.
[0119] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0120] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 1 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.
[0121] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the intelligent operation and maintenance methods for devices described in the above embodiments.
[0122] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the intelligent operation and maintenance methods for devices described in the above embodiments.
[0123] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0124] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0125] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0127] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.
[0128] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.
Claims
1. A method for intelligent operation and maintenance of equipment, characterized in that, The method includes: Using the unique device identifier of the target device as the association primary key, collect and associate multi-source data at each stage of the device's entire lifecycle; A health assessment model is constructed based on the multi-source data. The health index of the equipment is calculated according to the evaluation indicators in the health assessment model. The failure risk value is predicted according to the health index and its changing trend. The remaining lifespan is predicted according to the cumulative operating information and health status of the equipment. Based on the health index, failure risk value, remaining lifespan, and lifecycle cost, a comprehensive decision including maintenance actions is generated. In response to the execution of the maintenance action, the status improvement data of the target device is obtained; Based on the improved status data, the parameters of the health assessment model are corrected, and the recommended priority of the maintenance actions is also adjusted.
2. The method according to claim 1, characterized in that, Collect and correlate multi-source data from all stages of the device's entire lifecycle, including: Automatically collect equipment operating parameter data through sensors; Collect manual inspection data, maintenance work order data, and business data from ERP, MES, SCADA and EAM systems through mobile terminals or business system interfaces; The collected data is uniformly associated using the unique device identifier and timestamp as indexes.
3. The method according to claim 1, characterized in that, Generate comprehensive decisions that include operational actions, including: Determine the comprehensive cost of continued use and the comprehensive cost of replacement of the target equipment; When the overall cost of continued use is greater than the overall cost of renewal, and the remaining lifespan is less than a preset lifespan threshold, an overall decision including a device renewal action is generated. Otherwise, a comprehensive decision including maintenance or inspection actions is generated based on the health index.
4. The method according to claim 3, characterized in that, The comprehensive cost of continued use includes the estimated maintenance and repair costs, the estimated operating energy consumption costs, and the estimated downtime loss costs. The comprehensive cost of the upgrade includes the cost of purchasing new equipment, the cost of installing and commissioning new equipment, and the residual value of the original equipment after it is scrapped. The lifecycle cost includes procurement cost, installation and commissioning cost, operating cost, repair and maintenance cost, spare parts cost, scrapping and disposal cost, and downtime loss cost.
5. The method according to claim 1, characterized in that, The health assessment model includes multiple assessment indicators and their corresponding initial weights; Based on the improved condition data, the parameters of the health assessment model are corrected, including: Based on the aforementioned status improvement data, the correlation between each evaluation index and equipment failure is determined; Based on the correlation and the state change amount in the state improvement data, the initial weights of each evaluation index are dynamically adjusted to obtain updated weights.
6. The method according to claim 1, characterized in that, The method for predicting the fault risk value is as follows: based on the current health index, the trend of health index change, the temperature change rate, the vibration change rate, the historical fault frequency and the equipment usage intensity, calculate the fault probability of the target equipment failing within a set time window in the future, and then weight the fault probability according to the criticality of the target equipment and the degree of impact of the fault to obtain the risk value.
7. The method according to claim 5, characterized in that, The recommended priority of the aforementioned operational actions has been revised, including: The actual improvement in status after the aforementioned maintenance actions were executed multiple times is statistically analyzed. When the actual improvement amount is lower than the preset improvement threshold for multiple consecutive times, the recommendation priority of the operation and maintenance action in the comprehensive decision-making is reduced. When the actual improvement in state is greater than or equal to the preset improvement threshold, the recommended priority is maintained or increased.
8. An intelligent equipment operation and maintenance device, characterized in that, The device includes: The acquisition unit is used to collect and associate multi-source data at each stage of the device's entire lifecycle, using the unique device identifier of the target device as the association primary key. The prediction unit is used to construct a health evaluation model based on the multi-source data, calculate the health index of the equipment according to the evaluation indicators in the health evaluation model, predict the failure risk value according to the health index and its changing trend, and predict the remaining lifespan according to the cumulative operating information and health status of the equipment. The generation unit is used to generate a comprehensive decision including operation and maintenance actions based on the health index, fault risk value R, remaining lifespan and life cycle cost. The acquisition unit is configured to acquire status improvement data of the target device in response to the execution of the maintenance action. The correction unit is used to correct the parameters of the health assessment model based on the status improvement data, and to correct the recommended priority of the operation and maintenance actions.
9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.
