A comprehensive health evaluation method for coking key equipment

Abstract

The application discloses a kind of coking key equipment comprehensive health evaluation methods, comprising: setting two-level digital model in each equipment to be monitored structure;Real-time detection sample construction analysis data;Characteristic extraction is carried out to preprocessed signal;The same two-level index normalized eigenvalue monitored by different components and the same kind of sensor is introduced key component self-defined weight, and adaptive weight correction is introduced;The weighted average of normalized eigenvalue is calculated as calibration value;According to the calibration value and the subjective weight of the first index and the objective weight of the second index calculated, the equipment operation health condition is judged by comprehensive evaluation value.The application realizes the real-time monitoring of coking production line, health judgment, and solves the problems of cooperative monitoring and linkage control in coking production line, precise real-time monitoring under extreme working conditions, health monitoring of core process and equipment, etc., to ensure the safe and reliable operation of coking equipment, support preventive maintenance of equipment, and reduce the risk of unplanned shutdown of equipment.

Description

Technical Field

[0001] This invention relates to the field of equipment testing and fault diagnosis technology, and more specifically, to a comprehensive health evaluation method for key coking equipment. Background Technology

[0002] In the coking process, in addition to the main product coke, a large amount of carbon-containing byproducts such as coke oven gas, coal tar, and coking dust are generated, which also possess high energy and chemical utilization value. Against this backdrop, constructing a high-value utilization system across the entire chain of "gasification to syngas - methanol preparation - ethanol deep processing" using coking byproducts as raw materials has become the core mainstream direction for the green and low-carbon transformation and value-added extension of the industrial chain in the coking industry. Currently, the research and development of technologies in the coking product gasification and methanol preparation industrial chain focuses primarily on core process optimization, catalyst upgrading, and breakthroughs in equipment localization. However, research on process monitoring and intelligent control systems for the entire process still lags significantly behind the needs of industrial development, becoming a core bottleneck restricting the safe, stable, and efficient operation of the entire industrial chain. In this context, technical methods capable of accurately, in real-time, and comprehensively monitoring and managing equipment health to ensure safe and stable production and reduce operation and maintenance costs have become indispensable core technologies for modern coking enterprises. Summary of the Invention

[0003] Due to the aforementioned deficiencies in existing technologies, this invention provides a comprehensive health evaluation method for key coking equipment. By setting up a two-level digital model, different calculation weights are assigned to the characteristic values ​​of secondary indicators based on the type of different components and the different operating environments. The weighted average of the normalized characteristic values ​​is used as the calibration value. The health status of the equipment is judged based on the calibration value, the subjective weights of the primary indicators, and the objective weights of the secondary indicators. This enables real-time monitoring and health assessment of the coking production line, and solves problems such as collaborative monitoring and linkage control in the coking production line, accurate real-time monitoring under extreme operating conditions, and health monitoring of core processes and equipment. This ensures the safe and reliable operation of coking equipment, supports preventative maintenance, and reduces the risk of unplanned equipment downtime.

[0004] To achieve the above objectives, the present invention provides a comprehensive health evaluation method for key coking equipment, comprising the following steps:

[0005] The equipment is classified according to the type of equipment to be evaluated, and a digital model is set for the structure to be monitored in each piece of equipment. The digital model is divided into two levels. The first-level digital model is determined by industry-standard production experience and sets first-level indicators, which serve as the guide for the second-level digital model. The second-level digital model sets second-level indicators according to the equipment type, production function and / or operating environment, and sets corresponding detection sensors on the corresponding parts of the equipment based on the second-level digital model to obtain real-time detection data samples of each monitoring part of the component.

[0006] Based on real-time sensor detection samples, analytical data is constructed and preprocessed. This preprocessing removes invalid operational data, abnormal interference data from the field, and sensor data acquisition errors from the detection signals. Specifically, the signal preprocessing operation can be performed according to the sensor data acquisition timestamp, removing data outside the specified window, and removing data from the window within the specified window, specifically data from the 5-10 minutes prior to start-up and shutdown, thus filtering out invalid operational data from the detection signals. This process is repeated according to a 3-step process. The principle is to discard data that exceeds the data range to resolve abnormal interference data on site; the zero-point drift correction method is used to collect the zero-point offset value when the machine is stopped. As the original correction value collected by the operating status sensor, according to Correct sensor data acquisition errors; perform additional X / Y channel synchronization correction, X / Y channel sensitivity correction, X / Y direction polarity correction, phase delay correction, and order tracking when the rotation speed changes on the full vector spectrum vibration signal to keep the two channels synchronized and avoid inaccurate data due to asynchrony.

[0007] Data preprocessing can preserve the true signals under the effective operating conditions of the unit, providing reliable data for subsequent analysis and ensuring the accuracy of detection and analysis.

[0008] Feature extraction is performed on the preprocessed signal to extract the physical quantities corresponding to the secondary indicators required for analyzing the health status of the equipment, forming an analysis feature matrix. The analysis feature matrix is ​​then normalized to eliminate the influence between different dimensions. The preprocessed signal is also normalized to eliminate the influence of the original data dimensions.

[0009] Based on the importance of key components in production, the consequences of failure, historical failure probability, maintenance difficulty, and safety risks, custom weights are introduced for the normalized feature values ​​of the same secondary indicators monitored by similar sensors on different components of the equipment. In the formula n represents the number of sensors at the detection site; These are normalized monitoring values. The weighted mean calibration method assigns weights to the secondary indicators of sensors in different locations based on their installation importance (e.g., core rotating components > output shaft > housing). Due to the typical characteristics of coking equipment and the environmental conditions of coking production, some core and typical components have a high degree of impact on production, a high probability of failure, and a significant impact from downtime. To accurately amplify early fault characteristics, precisely locate weak points, and provide high-quality data for subsequent data processing, important components are assigned larger weights.

[0010] Furthermore, considering abnormal fluctuations during equipment operation, an adaptive weight correction is introduced to ensure that abnormal fluctuations in critical components can be detected promptly and are not averaged out by ordinary fluctuations. A weighted average of normalized eigenvalues ​​is calculated as the calibration value; the normalized calibration value reflects the degree of danger in equipment operation, with higher values ​​indicating poorer equipment health. More specifically, the normalization process transforms secondary indicators of different dimensions into dimensionless relative values, enabling comparison and summation of different indicators. Based on parameters such as real-time component risk values, risk growth rate, changes in the main vibration direction of the full vector spectrum, and abnormal changes in forward and reverse precession, the basic weights are dynamically adjusted to improve the ability to detect early faults.

[0011] For the calibration values ​​of primary and secondary indicators of equipment, the subjective weights corresponding to the primary indicators are calculated using the Fuzzy Analytic Hierarchy Process (FAHP), taking into account the fuzziness of expert engineering experience and judgment, and reducing the randomness of subjective weighting. The objective weights corresponding to the secondary indicators are calculated using the Conflicting Correlation Method (CRITIC), fully considering the comparative strength and conflict between indicators, and improving the rationality of objective weighting. Thus, the determination of the weights of primary and secondary indicators, through the introduction of FAHP and CRITIC methods, yields a reasonable comprehensive weight for both primary and secondary indicators. Linear weighting optimization is then applied to the primary and secondary weights of each piece of equipment, ensuring that the final weights used for health value calculations take into account both experience and data.

[0012] The component's comprehensive risk value is calculated based on the corrected calibration value of the monitoring data, as well as the weights of the primary and secondary indicators. This comprehensive risk value is then converted into a health score of 0-100. The calibration value is a risk-type calibration value; a higher value indicates a higher equipment risk. The health score is inversely related to the comprehensive risk value. The formula for calculating the health score is:

[0013] In the formula This is a health correction factor. When the calculated health value deviates significantly from the actual condition of the equipment, the factor can be adjusted (e.g., ...). (Take 1.05), recalculate health value; double summation term Calculate the comprehensive risk value of the k-th component, with a value ranging from [0,1]. A larger value indicates a higher risk for the equipment. Here, n is the number of primary indicators. Let i be the number of secondary indicators under the i-th primary indicator. The final comprehensive weight of the j-th secondary indicator is... This is the normalized calibration value of the index corresponding to the k-th component;

[0014] Determine the health status of the equipment based on its health values.

[0015] This invention utilizes sensors to detect digital signals and generate corresponding real-time monitoring signal models for equipment structures. By purifying, preprocessing, and normalizing the data signals, the health status of the equipment is calculated using a digital health assessment model. Based on the importance of key components in coking equipment during production, their failure consequences, historical failure probabilities, maintenance difficulty, and safety risks, this invention sets custom basic weights for the monitoring characteristics of different components. This ensures that abnormal characteristics of key components such as gearboxes, couplings, and motor input / output shafts are highlighted in the health assessment. Furthermore, the basic weights are dynamically adjusted based on parameters such as the component's real-time risk value, risk growth rate, change in the main vibration direction of the full vector spectrum, and abnormal changes in forward and reverse precession. When critical components exhibit early abnormal fluctuations, the weight of that component in the health assessment is automatically increased, thus preventing abnormal characteristics from being diluted by data from other normal components and improving the ability to detect early faults. Based on the status of each evaluation indicator, a real-time health assessment model for the equipment is determined. Using monitoring data from each equipment component and corresponding subjective and objective weights, the health score of the equipment is calculated, determining the overall health level of the equipment. Based on the system's determined health level, a direct understanding of the system's health status can be achieved, identifying specific faults or structures requiring maintenance, enabling proactive protection of the equipment system. Therefore, this invention can perform real-time monitoring and diagnosis of key coking equipment, automatically determining damaged parts and equipment health status, accurately assessing the current health level of the equipment, and achieving "pre-emptive maintenance," providing a guarantee for safe, energy-saving, economical, and green production. This invention, through its core advantages such as a detection model grading system, custom weights for key components, adaptive weight correction for abnormal fluctuations and weighted assignment of subjective and objective weights, and quantification of health values ​​on a percentage basis, is fully adaptable to the harsh operating conditions of high temperature, heavy load, and continuous operation of key equipment in coking production. It can accurately locate weak links, ensure the accuracy and objectivity of equipment health value evaluation, and realize the transformation from "post-event maintenance" to "pre-event maintenance".

[0016] Furthermore, the primary indicators include vibration, electrical, and process indicators. The secondary indicators for vibration include one or more of time domain, frequency domain, waveform, and demodulation indicators. The secondary indicators for process indicators include one or more of temperature, pressure, flow rate, thermal stress risk indicators, and dust wear risk indicators. The thermal stress risk indicators are determined by one or more of temperature deviation, temperature gradient, and temperature rise rate. The dust wear risk indicators are determined by one or more of dust concentration, filtration or sealing pressure difference, and high-frequency impact characteristics. The secondary indicators for electrical indicators include one or more of voltage, current, and power indicators. The sensors installed on each equipment component include one or more of velocity sensors, acceleration sensors, displacement sensors, temperature sensors, voltage sensors, current sensors, pressure sensors, flow sensors, and dust concentration sensors.

[0017] Furthermore, for the key monitoring sections of the rotating equipment, two orthogonal vibration sensors are arranged according to the full vector spectrum requirements to collect signals in the X and Y directions respectively. The full vector spectrum vibration signals are also preprocessed by X / Y channel synchronization correction, X / Y channel sensitivity correction, X / Y direction polarity correction, phase delay correction, and order tracking when the rotational speed changes. The preprocessing of the full vector spectrum vibration signals also includes dual-channel installation angle compensation and baseline drift compensation. Among them, dual-channel installation angle compensation is used to correct the two vibration signals that are not strictly orthogonal to orthogonal vibration signals, and baseline drift compensation is used to eliminate the effects of zero-point drift and temperature drift during long-term operation of the sensors. The full vector spectrum is obtained by constructing a complex vibration signal for each monitoring section, and the vibration characteristics corresponding to the monitoring section are extracted from the full vector spectrum.

[0018] By introducing full vector spectrum analysis into vibration detection signals, complex signals from two mutually perpendicular directions on the same monitoring section are fused to extract features such as full vector spectrum amplitude, principal vector, secondary vector, forward and reverse precession components, and principal vibration direction angle. Compared with single-direction spectrum analysis, this method can more comprehensively reflect the vibration intensity, vibration direction, and rotor operating state of rotating equipment. For vibration signals monitored using full vector spectrum technology, further channel sensitivity calibration, phase synchronization correction, and direction polarity correction are performed on the basis of the above preprocessing operations to avoid data inaccuracies caused by asynchrony.

[0019] Furthermore, the vibration characteristics include one or more of the following: total vector amplitude, principal vibration vector, secondary vibration vector, forward and reverse precession components, and principal vibration direction angle.

[0020] Furthermore, considering the inherent characteristics of vibration signals and the strong correlation between vibration and equipment failure, dynamic hazard values ​​are used to normalize vibration-related indicators. The calculation formula is as follows: ,in, Let j be the feature value of the j-th secondary index of the k-th component. These are reference values ​​for normal operation. These are industry-standard alarm values. This is the dynamic coefficient for operating conditions, determined based on equipment load, speed deviation, and temperature conditions, with a value range of 0.8 to 1.5; specifically, 0.8 to 0.95 for light load stable conditions; 0.95 to 1.05 for rated stable conditions; 1.15 to 1.30 for heavy load and high temperature conditions; and 1.30 to 1.50 for extreme heavy load or drastic fluctuation conditions. For other constraint-related indicators, an upper limit normalization method is used, and the calculation formula is as follows: ,in This represents the corresponding hazardous value for the indicator. In this way, it adapts to the differences in operating conditions under different loads / speeds, making the hazardous value more closely reflect the actual operating conditions of the equipment.

[0021] Furthermore, the process of obtaining the subjective weights of the primary indicators using the FAHP method includes the following steps:

[0022] Construct a hierarchical structure: with equipment health status assessment as the target layer, first-level indicators as the criterion layer, and second-level indicators as the indicator layer, forming a three-level hierarchical structure;

[0023] Constructing a fuzzy judgment matrix: For the first-level indicators of the criterion layer, based on the overall goal of equipment health status, multiple coking experts were invited to conduct pairwise comparisons of the relative importance between indicators using a fuzzy scale of 0.1 to 0.9, thus constructing a fuzzy judgment matrix. Where n is the number of primary indicators. The fuzzy scaling value represents two indicators of equal importance. A value of 0.5 indicates that the two indicators are equally important, while a larger value indicates that the former indicator is more important than the latter, and the following conditions must be met: When an expert's score deviates from the scores of other experts by more than a set threshold, the expert's weight can be reduced or a re-score can be required.

[0024] Calculating subjective weights: Normalize the fuzzy judgment matrix, sum the normalized columns, and then divide the summed vector by the matrix order n to obtain the subjective weight vector of the first-order index. The calculation formula is as follows:

[0025] The resulting weight vector satisfies the normalization constraint. ;

[0026] Consistency check: Calculate the consistency index of the judgment matrix, where is the largest eigenvalue of the judgment matrix; combine with the average consistency index RI to calculate the consistency ratio CR. When CR < 0.1, the judgment matrix passes the consistency check, and the obtained subjective weights of the primary indicators are valid; if it fails, the fuzzy judgment matrix is ​​readjusted until it passes the check; through the above process, the final weights of the primary indicators are finally obtained. ,Right now .

[0027] Furthermore, the process of obtaining the objective weights of secondary indicators using the CRITIC method includes the following steps:

[0028] Zero-value determination of calibration: When the calibration value of a secondary indicator is zero across all components, it is determined to be a normal indicator with all zeros. The information content of this indicator in the current period is zero, and it does not participate in the objective weight competition. When the calibration value of a secondary indicator is zero only in some components, it is determined to be a locally zero indicator. The low-risk meaning represented by the zero value is retained, and a very small positive number is added to the denominator of the coefficient of variation calculation. To avoid computational instability, if a column for a secondary indicator shows a value of 0, a minimum value of 0.01 is added to all values ​​in that column. The corrected data does not change the dispersion characteristics of the data or the correlation between indicators. The correction formula is: ,in Let j be the standard deviation of the j-th indicator. The average value of the j-th indicator;

[0029] Comparison strength of indicators: The coefficient of variation is used to characterize the comparison strength of indicators. A large coefficient of variation indicates that the indicator is more dispersed among different components, has a stronger ability to distinguish the health status of equipment, and contributes more to the health assessment.

[0030] Calculating the conflict between indicators: Based on the Pearson correlation coefficient, the conflict between secondary indicators belonging to the same primary indicator is calculated. A higher conflict rate indicates lower information overlap and higher independent information content between the indicator and other indicators. The conflict quantification value of the j-th secondary indicator is: ( ),in Let be the Pearson correlation coefficient between the j-th and h-th secondary indicators, and m be the number of secondary indicators under the primary indicator; The larger the value, the greater the conflict between this indicator and other indicators at the same level;

[0031] Calculate the comprehensive information content of the indicators ,in Let be the coefficient of variation of the j-th secondary indicator: taking into account the strength of comparison and conflict, calculate the comprehensive information content of the secondary indicator. The greater the comprehensive information content, the higher the importance of the indicator in the health assessment.

[0032] Calculating the objective weights of secondary indicators: The comprehensive information content of all secondary indicators belonging to the same primary indicator is normalized to obtain the corresponding weights of the secondary indicators. The calculation formula is as follows: ,in Let the objective weights of the j-th secondary indicators under the i-th primary indicator satisfy the normalization constraints. Through the above process, the final weights of the secondary indicators under each primary indicator are calculated. ,Right now .

[0033] Furthermore, based on the hierarchical weighting results, the final comprehensive weight of each secondary indicator in the entire evaluation system is calculated using the hierarchical multiplication method. When calculating the final comprehensive weight of the secondary indicators, a subjective and objective weight balancing distribution coefficient is introduced. ,in The combined secondary weight of the j-th secondary indicator under the i-th primary indicator is the weighted sum of the expert subjective secondary weight and the CRITIC objective secondary weight. The final comprehensive weight is the product of the primary indicator FAHP weight and the combined secondary weight. The formula for calculating the final comprehensive weight is: ,in Let be the final comprehensive weight of the j-th secondary indicator under the i-th primary indicator in the entire evaluation system; This represents the FAHP weight of the i-th primary indicator to which the secondary indicator belongs; Let be the expert subjective weight of the j-th secondary indicator under the i-th primary indicator; Let be the objective weight of the CRITIC for the j-th secondary indicator under the i-th primary indicator.

[0034] Furthermore, equipment health values ​​are determined according to the "weakest link" principle. The lowest score among all components in the equipment is selected as the component health value, and the lowest health value of a component is the equipment health value. To ensure continuous production and prevent production losses caused by potential equipment failures and downtime, the equipment evaluation adopts the weakest link principle to improve the reliability of equipment safety testing. If a sensor malfunction leads to an incorrect calibration value, resulting in a health value less than 0, the validity of the sensor data must be checked first to eliminate data errors before calculation. If the data is correct, the health level operation must be strictly followed.

[0035] Furthermore, before determining the equipment based on the weakest link principle, the health values ​​of each component are first filtered using a moving average, and the validity of single abnormal measurement points is judged. When a single measurement point experiences communication interruption, constant signal, over-range, zero-point drift anomaly, or isolated sudden change, the measurement point is masked and the component calibration value is recalculated. When the number of valid measurement points is lower than a set lower limit, a sensor fault alarm is output, instead of directly judging the entire machine as a shutdown state. Finally, the lowest score of each component in the equipment is selected as the component health value, and the lowest health value of the component is the equipment health value. The formula for calculating the smoothed health value after moving average filtering is as follows: ,in This represents the health value of the k-th component in the t-th evaluation period, L is the sliding window filter length, and q represents the time offset index within the sliding window, i.e., which evaluation period to trace back.

[0036] Furthermore, the equipment health level is divided into four levels: a health value in the range of (80, 100) is good, a health value in the range of (60, 80) is usable, a health value in the range of (40, 60) requires maintenance, and a health value ≤40 requires shutdown.

[0037] Compared with the prior art, the above invention has the following advantages or beneficial effects:

[0038] (1) This invention creates a multi-index evaluation model based on the type of equipment to be evaluated and monitored. The multi-index model consists of primary evaluation indicators and secondary evaluation indicators. Based on the digital model, corresponding sensors are set at each part of the equipment to obtain monitoring data of each secondary evaluation indicator in that component. Vibration signals of rotating equipment are collected according to the full vector spectrum arrangement requirements. Directionality, precession, and synthetic amplitude features are extracted through full vector spectrum fusion. Custom weights and adaptive component weights are constructed based on the importance of key components and real-time abnormal fluctuations. The weighted values ​​of each secondary indicator are obtained from the feature values ​​and weights, and then normalized to obtain the weighted average value of the secondary indicators as the calibration value. The weights of the primary and secondary indicators are calculated using the fuzzy hierarchical analysis method (FAHP) and the standard conflict correlation method (CRITIC). Finally, the health value of each component is calculated based on the calibration value of the secondary indicators of each component and the corresponding secondary weights and primary weights. Based on the health value of the components, the overall health status of the equipment is evaluated. This invention can combine equipment characteristics and production characteristics to detect the health status of equipment in real time, reflect the operating status of equipment, ensure stable production, reduce maintenance costs, and has high engineering value.

[0039] (2) In order to ensure continuous production and prevent production losses caused by the possibility of equipment failure and downtime, the present invention adopts the short board principle for equipment evaluation, that is, the lowest health value is selected as the health value of the equipment component and the equipment in the evaluation results, so as to control the fault within the most stable range and improve the reliability of equipment safety detection. In order to ensure the accuracy of the short board principle, before the short board principle is implemented, the component health value is averaged by sliding filter and the effectiveness judgment and shielding treatment are performed on the single abnormal measurement point.

[0040] (3) This invention, through its core advantages such as detection model grading system, full vector spectrum vibration feature extraction, key component custom weight, abnormal fluctuation adaptive weight correction and subjective and objective weight weight assignment, sensor installation position weight assignment, percentage health value quantification, and short board principle, is fully adaptable to the harsh operating conditions of high temperature, heavy load and continuous operation of key equipment in coking production. It can accurately locate weak links, ensure the accuracy and objectivity of equipment detection health value evaluation, and realize the transformation from "post-event maintenance" to "pre-event maintenance". Attached Figure Description

[0041] The invention, its features and advantages will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. Like reference numerals denote like parts throughout the drawings.

[0042] Figure 1 This is a flowchart of a comprehensive evaluation method for key coking equipment according to the present invention;

[0043] Figure 2This is a schematic diagram of the overall structure of the primary indicators of the equipment constructed in this invention;

[0044] Figure 3 This is a schematic diagram of the structure of the secondary indicators of the equipment constructed in this invention;

[0045] Figure 4 This is a schematic diagram of the monitoring indicators of a specific example device in this invention;

[0046] Figure 5 This is a schematic diagram of the monitoring part of a specific example device in this invention. Detailed Implementation

[0047] The structure of the present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the invention. The execution order of actions, steps, etc., in the apparatus and methods shown in the claims, specification, and drawings of the present invention can be implemented in any order unless a specific order is explicitly specified, and as long as the output of the preceding processing is not used in the subsequent processing.

[0048] In the following detailed description, numerous specific details are set forth to provide a more thorough understanding of the invention. It will be apparent to those skilled in the art that well-known algorithms or models are not shown in detail to avoid obscuring the spirit of the invention; and that the techniques not detailed in the following effect examples are readily available prior art.

[0049] The technical solution of the present invention will now be described in detail with reference to specific embodiments.

[0050] Example

[0051] See Figure 1 This embodiment discloses a comprehensive health evaluation method for key coking equipment. This example selects the hot air blower unit in the coking gasification zone as the evaluation equipment and conducts a comprehensive health evaluation of its operating status. The specific process is as follows:

[0052] Step 1: Create a multi-index evaluation model based on the digital model of the equipment to be evaluated.

[0053] See Figure 2 and Figure 3 Based on the unit's digital model and on-site monitoring conditions, data tags are set up according to four levels: equipment, component, measuring point, and measurement location. Starting from the free end of the drive motor and ending at the fan blade of the working machine, the digital model of the equipment is constructed. The digital model of the equipment includes basic information and a simplified structural diagram. The basic information is shown in Table 1, including but not limited to the unit code, rated speed, and bearing model.

[0054] Table 1 Basic Information of the Unit Digital Model

[0055]

[0056] This example selects the hot air blower unit in the coking area as the evaluation object. Its multi-index evaluation system is selected based on the digital model of the hot air blower unit, using sensitive characteristic parameters of common equipment faults as the selection criterion. For example... Figure 4 As shown, a multi-index evaluation system is constructed, comprising three primary indices: vibration, process, and electrical. The secondary indices for vibration include first-harmonic amplitude, second-harmonic amplitude, third-harmonic amplitude, effective filtered value, peak value, kurtosis, passband amplitude, and impact index. The secondary indices for process are temperature-induced thermal stress risk and dust abrasion risk. The secondary indices for electrical include three-phase stator current, harmonic distortion, three-phase incoming line voltage, and operating power.

[0057] See Figure 5 The unit includes a motor free end 5, a motor drive end 4, a motor 6, a bearing housing near the motor end 3, a bearing housing near the fan end 2, and a fan unit 1. A horizontal acceleration sensor Ha, a vertical acceleration sensor Va, and a temperature sensor Tt are installed at the bearing housing near the fan end 2, the bearing housing near the motor end 3, the motor drive end 4, and the motor free end 5. An A-phase current sensor Ia-i, a B-phase current sensor Ib-i, and a C-phase current sensor Ic-i are installed at motor 6. A temperature sensor Tt is installed at fan unit 1. Based on the equipment structure, a set of components to be evaluated is defined as follows: ,in This indicates the k-th part of the component being inspected. This indicates the number of parts that participated in the evaluation.

[0058] For vibration signals from rotating equipment, this invention, following the sensor arrangement requirements of full vector spectrum technology, sets up two mutually perpendicular radial vibration sensors on the same monitoring section to collect horizontal and vertical vibration signals respectively. For each component requiring full vector spectrum analysis... Vibration signals were collected in two orthogonal directions: . This represents the horizontal vibration signal of the k-th component. This represents the vertical vibration signal of the k-th component.

[0059] The arrangement of full-vector vibration sensors should meet the following requirements: two sensors are located at the same axial interface or as close as possible to the same monitoring section; the two measurement directions are perpendicular to each other; the two channels sample synchronously; the sensitivity, range and acquisition frequency of the two channels are consistent or have been calibrated; and the polarity and installation direction of the sensors are clearly recorded.

[0060] For rotating equipment requiring order analysis, speed signals are acquired simultaneously: This indicates the real-time rotational speed signal of the device.

[0061] Step 2: Based on the digital model of the equipment, install sensors for vibration, temperature, current, and voltage on the monitored components, and continuously collect N sets of sample data at equal intervals. The monitoring data is represented in matrix form. The original monitoring data matrix for the 2B-2332 hot air blower unit can then be constructed, in the following form: ;

[0062] In the formula The vibrations are, in order: horizontal vibration at the free end of the motor, vertical vibration at the free end of the motor, horizontal vibration at the drive end of the motor, vertical vibration at the drive end of the motor, horizontal vibration near the motor end of the bearing housing, vertical vibration near the motor end of the bearing housing, vertical vibration near the fan end of the bearing housing, and vertical vibration near the fan end of the bearing housing. The three-phase current signals are recorded sequentially; the temperature signals of each key component are recorded as follows: , , , ; This represents the number of sample groups obtained through continuous sampling at equal intervals.

[0063] For the original monitoring data matrix Preprocessing is performed. The preprocessing steps include: 1. Removing invalid data from the start-up and shutdown phases based on timestamps; 2. Based on 3... The principle is to eliminate sudden abnormal values ​​during operation; third, to correct the zero-point drift of the sensor and retain the true detection signal under the effective operating state of the equipment.

[0064] For vibration signals using full vector spectrum technology, channel sensitivity calibration, phase synchronization correction, and directional polarity correction are performed in data processing; further, dual-channel installation angle compensation and baseline drift compensation are performed.

[0065] For dual-channel installation angle compensation: Let the actual installation angle between the two vibration sensors be... The ideal included angle is The original measurement signals from the two sensors are respectively , .when When a non-orthogonal signal is converted to an orthogonal signal, the following formula is used: , ,in, , This is the orthogonal vibration signal after angle compensation. It is a very small positive number. In actual installation, when When the angle is within the specified range, it can be corrected through angle compensation; when the angle exceeds this range, the system prompts to recheck the sensor installation orientation.

[0066] For baseline drift compensation: Let the k-th measurement point... The baseline drift in the two directions are respectively , The baseline-compensated signal is then: , ,in, , For the k-th measuring point Sensitivity coefficient of two-channel sensor.

[0067] The baseline drift can be obtained through sliding estimation in either the shutdown or steady-state conditions, and the corresponding calculation formula is as follows: , ,in, This is the baseline update factor, and its value range is... ; , For the k-th measuring point under stable operating conditions or shutdown conditions The average value of the vibration signal obtained from the two elements.

[0068] For phase delay compensation: if If there is a data acquisition delay between the two channels, the delay amount is estimated based on the channel cross-correlation function: ,in, For the k-th measuring point The cross-correlation function of the channels. Based on... Align one of the signals.

[0069] The corrected vibration signal is filtered. For low-frequency faults such as imbalance, misalignment, and looseness, the 1X, 2X, and 3X frequency-related components are primarily retained. For gearbox faults, the gear meshing frequency and its harmonics and sidebands are primarily retained. For rolling bearing faults, the high-frequency impact components are primarily retained, and further envelope demodulation analysis can be performed. When there are significant fluctuations in equipment speed, the speed signal is used as a reference. By performing synchronous resampling or order tracking processing on the vibration signal, the vibration characteristics can be transformed from the frequency domain to the order domain for analysis.

[0070] Full vector spectrum processing is performed on the two orthogonal vibration signals of each monitoring section. For the k-th component, its horizontal and vertical vibration signals are constructed into a complex vibration signal. , where j is the imaginary unit.

[0071] Then, perform a Fourier transform on the complex vibration signal to obtain the full vector spectrum: When using order analysis, it can be represented as . Indicates frequency, Indicates order, This represents the corner sampling sequence.

[0072] After preprocessing, an effective data matrix is ​​obtained. Its form is:

[0073] For effective data matrix Secondary indicators were further extracted from the signals at each measurement point to construct a multi-source feature matrix. For example, secondary indicators for vibration signals include first harmonic amplitude, second harmonic amplitude, third harmonic amplitude, effective value of filter, peak value, kurtosis, passband amplitude, and impulse index; secondary indicators for electrical signals include three-phase stator current, harmonic distortion, three-phase incoming line voltage, and operating power.

[0074] Secondary indices for vibration are extracted based on the full vector spectrum results. The positive and negative frequency components in the full vector spectrum reflect the positive and negative precession characteristics of rotational vibration, respectively. Let the amplitudes of the positive and negative components of the k-th component at frequency f be:

[0075]

[0076]

[0077] in, The number of sampling points. Indicates the positive precession component. It represents the anti-precession component.

[0078] Further calculations are performed on the principal amplitude, secondary amplitude, and precession characteristics at this frequency. The principal amplitude can be expressed as: The amplitude of the secondary vibration vector can be expressed as: The precession ratio can be expressed as: .in, To prevent extremely small numbers with a denominator of zero, The time indicates that the current movement is dominant. This indicates that the anti-advancement movement is dominant. The larger the value, the more pronounced the precession characteristic in that direction.

[0079] For frequency-related faults, the full vector spectrum amplitudes of the k-th component (1X, 2X, 3X) are extracted. For gearbox faults, the gear meshing frequency is calculated based on the input shaft frequency, output shaft frequency, number of teeth on the driving gear, and number of teeth on the driven gear, thereby extracting the full vector spectrum amplitude of the gear meshing frequency and obtaining the sideband characteristics of the gear meshing frequency. For rolling bearing faults, after high-frequency bandpass filtering of the orthogonal vibration signal, an envelope analysis input signal can be constructed based on the full vector spectrum synthesized signal. Hilbert envelope demodulation is then performed on the envelope analysis input signal to obtain the envelope signal. Spectral analysis is then performed on the envelope signal to extract the amplitude characteristics near the fault characteristic frequencies of the bearing inner ring, outer ring, rolling elements, and cage.

[0080] Therefore, the secondary index of vibration has been expanded from the original unidirectional spectral characteristics to full vector spectrum fusion characteristics.

[0081] Considering the high temperature, dust, heavy load, and continuous operation environment of coking production sites, this invention focuses on adding thermal stress risk indicators and dust wear risk indicators to the secondary process indicators. These indicators are used to characterize the impact of high-temperature thermal deformation, uneven thermal expansion, coke powder particle wear, seal deterioration, and lubrication contamination on the health status of equipment.

[0082] The thermal stress risk index can be determined by temperature deviation, temperature gradient, and temperature rise rate. The relevant calculation formula is as follows: ,in The thermal stress risk index for the k-th component; The current temperature; The baseline temperature for health; Temperature alarm threshold; This refers to the temperature difference or temperature gradient between different locations of the same component. The allowable temperature difference threshold; The rate of temperature rise; The allowable temperature rise rate threshold; Let the weighting coefficients satisfy: .

[0083] Dust abrasion risk indicators can be determined by dust concentration, filtration or sealing pressure difference, and high-frequency impact characteristics. The relevant calculation formula is as follows: ,in This is the dust wear risk index for the k-th component; The concentration of dust or coke powder around the equipment; The dust concentration alarm threshold; This refers to the pressure difference of the filter, the pressure difference of the sealing air, or the pressure difference of the dust removal system. This is the differential pressure alarm threshold; Let be the high-frequency impact characteristic value of the k-th component; This is the threshold for high-frequency impact alarms; For the weighting coefficients, satisfying .

[0084] When no dust concentration sensor is installed on site, filter pressure difference, sealing air pressure attenuation, bearing high-frequency impact growth, and temperature rise can be used as indirect indicators of dust wear.

[0085] The full-vector-spectrum vibration characteristics, electrical characteristics, and technological characteristics are combined to form a multi-source feature matrix: Feature Matrix The structure is as follows.

[0086] Its form is: The first line This indicates the horizontal vibration of the motor's free end. 1x, 2x, and 3x represent the first, second, and third harmonic frequencies, respectively. b represents the passband amplitude, s represents the effective value of the filter, p represents the peak value, k represents the kurtosis value, and m represents the impact index.

[0087] Step 3: Analyze the feature matrix. After normalization, the normalized risk matrix is ​​obtained: For vibration-related indicators, dynamic hazard values ​​are used for normalization, and the calculation formula is as follows: ,in, Let j be the feature value of the j-th secondary index of the k-th component. These are reference values ​​for normal operation. These are industry-standard alarm values. This is the dynamic coefficient of the operating condition, which is determined based on the equipment load, speed deviation, and temperature conditions, and its value ranges from 0.8 to 1.5. Specifically, the value is 0.8 to 0.95 for light load stable operating conditions; 0.95 to 1.05 for rated stable operating conditions; 1.15 to 1.30 for heavy load and high temperature operating conditions; and 1.30 to 1.50 for extreme heavy load or drastic fluctuation operating conditions.

[0088] For other constraint-type indicators, the upper limit normalization method is used, and the calculation formula is as follows: ,in This represents the risk value corresponding to the indicator.

[0089] After normalization, each indicator is converted into a dimensionless risk quantity, with a larger value indicating a higher risk. Using the maximum threshold method, the maximum value of each normalized indicator is selected to reflect the health status of the monitored equipment components.

[0090] When monitoring equipment, if a certain secondary indicator has multiple measuring points distributed across different parts of the equipment, in order to focus on monitoring the core components and amplify the impact of weak points, it is necessary to perform weighted calculations based on the location weights of the measuring points to obtain the calibration value of the secondary indicator for that component.

[0091] In this embodiment, based on extensive past production experience, it is known that the core sensitive parts of the hot air blower unit are mainly concentrated at the free end and the drive end of the motor. From an engineering analysis perspective, the free end of the motor should be considered the core part in this example, followed by the drive end, and then the bearing housing near the fan end and the bearing housing near the motor end. Therefore, the calibration value of the j-th secondary index of the k-th component is obtained, calculated using the following reference formula: ,in This indicates the number of measurement points participating in the weighting of the k-th component under the j-th secondary indicator. This represents the weight of the k-th component under the j-th secondary indicator.

[0092] To enable the weights to adaptively adjust according to changes in equipment status, this invention further constructs an abnormal fluctuation correction factor. Let the comprehensive abnormal factor of the k-th component in the current evaluation period be... It is composed of the anomaly amplitude factor, the risk growth rate factor, and the full-vector directional anomaly factor, and the calculation reference formula is as follows: . For abnormal amplitude factors, For risk growth rate factor, The directional anomaly factor of the whole vector spectrum. This is the adjustment coefficient.

[0093] The abnormal amplitude factor is used to characterize the degree to which the current risk value exceeds the warning threshold. The calculation formula is as follows: In the formula The number of secondary indicators for the k-th component participating in the evaluation; This is the normalized risk value of the j-th indicator for the k-th component in the current cycle; For a very small positive number, we can take: ; Let j be the warning threshold for the j-th indicator, and its value range is: (Usually acceptable, general early warning indicators:) =0.6~0.7; Key safety indicators: 0.5~0.6; Severe risk indicator: =0.8 or higher (entering the alarm range).

[0094] The risk growth rate factor is used to characterize the growth trend of component risk within adjacent evaluation periods. The calculation reference formula is as follows: ,in This represents the average risk of the k-th component during the current evaluation period. This is the average risk value from the previous evaluation period; This is the normalized threshold for the risk growth rate. It can be determined based on the evaluation period: when the evaluation period is short, A value of 0.05 to 0.10 is acceptable; however, when the evaluation period is relatively long, A value of 0.10~0.30 is acceptable; for continuous operation of coking equipment, 0.10~0.20 is preferred.

[0095] The full-vector spectral directional anomaly factor is used to characterize the degree of deviation of directional characteristics such as the principal vibration direction and the ratio of forward and reverse precession relative to a healthy baseline. The calculation reference formula is as follows: ,in This is the current main vibration direction angle; The principal vibration direction angle under healthy baseline conditions; The current precession ratio; The precession ratio under healthy baseline conditions; The maximum permissible offset in the main vibration direction; This represents the maximum permissible offset of the precession ratio. The adjustment coefficient satisfies: and The recommended parameter range is: Desirable ; 1-2 are acceptable; Each can be set to 0.5, or adjusted according to the equipment's sensitivity to changes in direction or precession.

[0096] Based on the basic custom weights and the abnormal fluctuation correction factor, the adaptive weight of the k-th component is calculated using the following formula: ,in This represents the adaptive weight of the k-th component within the current evaluation period. Let k be the comprehensive anomaly factor of the k-th component within the current evaluation period. Define a custom weight for the k-th component; satisfying When a critical component exhibits an abnormal increase in amplitude, a greater risk growth rate, or a sudden change in the characteristics of its full vector spectrum, its corresponding... This increases the weight of the component in the overall machine health assessment, thus automatically increasing its weight.

[0097] If the k-th component contains multiple measurement points, then the measurement points within that component are first weighted at the measurement point level.

[0098] After introducing adaptive weights, the corrected calibration value of the j-th secondary index of the k-th component is: .in, Correct calibration values ​​for components after introducing custom and adaptive weights.

[0099] Then, based on the calculated secondary indicator calibration values ​​of each component, an indicator correction calibration value matrix is ​​constructed. .

[0100] Its form is: Among them, rows 1 to 4 can correspond to the free end of the motor, the drive end of the motor, the end of the bearing housing near the motor, and the end of the bearing housing near the fan, respectively; This indicates the calibration value of the first component on the first secondary indicator.

[0101] This step indicates that location weighting and value calibration are considered in data processing.

[0102] Based on the raw monitoring data and processed using the above algorithm, the calibration values ​​of secondary indicators for different monitoring locations of the unit were obtained. See Table 2. The four main monitoring locations of the unit exhibit vibration in both horizontal and vertical directions; the average of the horizontal and vertical calibration values ​​is presented here.

[0103] Table 2. Index calibration values ​​for four monitoring points of the unit.

[0104]

[0105] Step 4: Calculate the subjective weights of the primary indicators using the Fuzzy Hierarchical Analysis (FAHP). With equipment health status as the target layer and vibration, process, and electrical as the three primary indicators, the set of primary indicators is defined as: "Vibration," "Electrical," and "Process." Construct a fuzzy judgment matrix. The fuzzy judgment matrix is ​​as follows, and its form is: Its satisfaction When an expert's score deviates from other experts' scores by more than a set threshold, the expert's weight can be reduced or a re-score can be required.

[0106] This fuzzy judgment matrix was constructed by multiple experts in related fields using a fuzzy scale of 0.1-0.9 to pairwise assess the relative importance of the indicators. It meets the modeling requirements of the FAHP method and satisfies the realistic rationality of matching the current state of the device in this example. Therefore, it is legitimate as a primary subjective weight input matrix.

[0107] Calculating subjective weights: Normalize the fuzzy judgment matrix, sum the normalized columns, and then divide the summed vector by the matrix order n to obtain the subjective weight vector of the first-order index. The calculation formula is as follows:

[0108] The resulting weight vector satisfies the normalization constraint. .

[0109] Considering that the core components and anomalies of the hot air blower unit are concentrated in the bearing vibration and impact characteristics during operation, the vibration index is more important than the process index and electrical index in this implementation.

[0110] Referring to Table 3, after normalizing the judgment matrix according to the FAHP calculation process, the subjective weights of the first-level indicators can be obtained, among which vibration weight is the largest, followed by process weight, and then electrical weight.

[0111] Table 3 Weight values ​​of primary indicators

[0112]

[0113] After obtaining the weights of the primary indicators, a consistency check is performed on the fuzzy judgment matrix. Based on the above technical approach, the largest eigenvalue of the judgment matrix is... For a matrix n=3, RI=0.58 is used. The consistency ratio can be calculated using the following formula: The calculated CR < 0.1. Therefore, the judgment matrix satisfies the consistency test and can be used for subsequent weight calculation.

[0114] The objective weights of the corresponding secondary indicators are calculated using the standard conflict correlation (CRITIC) method. The component indicator calibration value matrix is ​​then used. (where p is the number of components participating in the evaluation;) The input is the number of secondary indicators under the i-th primary indicator.

[0115] Then, zero-value correction, comparison intensity calculation, indicator conflict calculation, and comprehensive information content calculation are performed on each secondary indicator in turn.

[0116] Zero-value correction can prevent numerical instability caused by a calibration value of 0 in subsequent calculations.

[0117] Let the correction calibration values ​​of the j-th secondary index on p components be listed as follows: If satisfied If the value is zero, it indicates that the j-th secondary indicator does not reflect any risk for any component within the current evaluation period, and is considered a normal indicator with all zeros. In this case, instead of assigning the same minimum positive number of 0.01 to this indicator, its comparative strength is set to: Its information content is: .

[0118] This indicator does not participate in the CRITIC objective weight competition for the current period, and the remaining non-zero effective indicators are re-normalized and weighted.

[0119] If satisfied and If the value is zero, it indicates that the j-th secondary indicator is a local zero indicator, meaning that some components are normal while others are at risk. In this case, the low-risk meaning represented by the zero value is retained, and instead of adding a fixed value to the entire column, a very small positive number is added only to the denominator of the coefficient of variation calculation. To avoid the calculation being unstable due to a mean of zero: ,in Let j be the standard deviation of the j-th indicator. Let be the average value of the j-th indicator.

[0120] If satisfied This indicates that the indicator shows almost no difference between different components, lacking sufficient distinguishing ability. .

[0121] If the information content of all secondary indicators under the same primary indicator is 0, that is... Then, the preset empirical weights or equal weights of the secondary indicators under this primary indicator are adopted: ,in, Let represent the number of secondary indicators under the i-th primary indicator.

[0122] After the correction is completed, calculate the standard deviation of the j-th secondary indicator. And calculate the arithmetic mean of the corrected calibrated values ​​of the j-th secondary index on the p-th component. The ratio of the standard deviation to the arithmetic mean of the standardized values. The coefficient of variation (CV) characterizes the comparative strength of this indicator across different components. A larger CV indicates a higher degree of dispersion of the indicator across different components, a stronger ability to distinguish the health status of equipment, and a greater contribution to health assessment.

[0123] Further calculate the conflict between secondary indicators under the same primary indicator. First, calculate the Pearson correlation coefficient between the j-th and h-th secondary indicators. Based on this, calculate the conflict quantification value of the j-th secondary indicator. .

[0124] After obtaining the standard deviation and conflict quantification values, the information content of the j-th secondary indicator is calculated. The formula for information content is as follows: . The comparative strength of the j-th secondary indicator; The conflict quantification value of the j-th secondary indicator: ( ),in Let be the Pearson correlation coefficient between the j-th and h-th secondary indicators, and m be the number of secondary indicators under the primary indicator; The higher the value, the greater the conflict between this indicator and other indicators at the same level.

[0125] The information content of all secondary indicators under the same primary indicator is normalized to obtain the objective weights of the secondary indicators within the group. The formula for calculating the objective weights is as follows: ,in Let the objective weights of the j-th secondary indicators under the i-th primary indicator satisfy the normalization constraints. (There are a total of 1,0 ... under the i-th primary indicator) (Two secondary indicators).

[0126] Step 5: After obtaining the objective weights of the secondary indicators, introduce the subjective and objective weight balance distribution coefficient, and multiply it hierarchically with the subjective weights of the primary indicators to obtain the final comprehensive weight.

[0127] Introducing a subjective and objective weighting balance coefficient: When historical samples are limited, equipment operating conditions are complex, or expert experience is highly reliable, A value of 0.5 to 0.7 is acceptable; when there are sufficient historical test samples, stable data, and clear equipment operating patterns, A value of 0.3 to 0.5 is acceptable; in this example, We set it to 0.5 to maintain a balance between the weight of expert experience and the weight of objective data.

[0128] Then, the combined secondary weight of the j-th secondary indicator under the i-th primary indicator is the weighted sum of the expert subjective secondary weight and the CRITIC objective secondary weight: ,in, Let be the expert subjective weight of the j-th secondary indicator under the i-th primary indicator; The larger the value, the more the evaluation relies on expert experience; conversely, the smaller the value, the more the evaluation relies on objective differences in the monitoring data.

[0129] The fusion of secondary weights should satisfy the following:

[0130] The final overall weight is: ,in Let be the final comprehensive weight of the j-th secondary indicator under the i-th primary indicator in the entire evaluation system.

[0131] The overall risk value is calculated based on the comprehensive weight and component calibration value, and then the component's health value is calculated. The formula for calculating the health value is:

[0132] In the formula This is a health correction factor. When the calculated health value deviates significantly from the actual condition of the equipment, the factor can be adjusted (e.g., ...). (Take 1.05), recalculate health value; double summation term Calculate the comprehensive risk value of the k-th component, with a value ranging from [0,1]. A larger value indicates a higher risk for the equipment. Here, n is the number of primary indicators. Let i be the number of secondary indicators under the i-th primary indicator. The final comprehensive weight of the j-th secondary indicator is... This is the normalized calibration value of the index corresponding to the k-th component.

[0133] Step 6: Finally, according to the principle of the weakest link, select the minimum health value among the health values ​​of each component as the overall health value of the machine. .

[0134] To avoid misjudging the overall health value due to momentary anomalies of a single sensor, communication failures, or measurement point drift, this invention performs sliding mean filtering on the component health values ​​before implementing the weakest link principle, and performs validity judgment and shielding processing on single abnormal measurement points.

[0135] Let the health value of the k-th component in the t-th period be... A sliding window of length L is used for filtering to obtain a smoothed health value.

[0136] The formula for smoothed health value is: ,in represents the health value of the k-th component in the t-th evaluation period; L is the sliding window filtering length, which can be 3 to 10 evaluation periods depending on the equipment sampling period and alarm response requirements; q represents the time offset sequence number within the sliding window, i.e., which evaluation period to trace back.

[0137] When it is necessary to highlight the most recent state, the weighted moving average can also be used. The formula for the weighted moving average is: ,in For sliding window permissions, the health value closer to the current moment has a higher weight. Set a validity flag for the k-th component and s-th measurement point: when... The measuring point is valid at that time. At that time, the measuring point is invalid or blocked.

[0138] When a measuring point meets any of the following conditions, it is judged as a suspected measuring point abnormality: (1) No data or communication interruption for multiple consecutive evaluation cycles; (2) The signal remains constant for a long time and the variance is less than the set threshold; (3) The signal exceeds the physical range of the sensor; (4) The zero drift exceeds the allowable range; (5) The sudden change of a single measuring point is obvious, but the adjacent measuring points or similar indicators of the same component do not show synchronous abnormalities.

[0139] After shielding, the calibration value of the j-th index of the k-th component is recalculated based on the effective measurement points. The calculation formula is as follows: ,in The number of measurement points for the k-th component; The weight of the measurement point of the s-th measurement point of the k-th component under the j-th index; This corresponds to the normalized risk value; It is a very small positive number.

[0140] When the number of valid measuring points is less than the set lower limit, the equipment is not directly shut down, but a sensor fault alarm is output: At this point, the health status of the component is marked as "data invalid or needs to be verified," and the system prompts you to check the corresponding sensor, wiring, and acquisition channel.

[0141] The final equipment health value is determined according to the principle of improving the weakest link: Among them, the component health values ​​used in the defect assessment should be the monitoring values ​​after sliding filtering and measurement point validity assessment.

[0142] To avoid single abnormal shutdowns, a continuous confirmation mechanism can be added, the formula of which can be written as: (and continuously satisfy C evaluation cycles), where The shutdown threshold is C, which can be 2 to 5.

[0143] The health level of the equipment unit is assessed based on the calculated equipment health values. The health value is determined according to the minimum value principle, and in this method, the equipment health level is divided into four levels: a health value in the range (80, 100) is good; a health value in the range (60, 80) is usable; a health value in the range (40, 60) requires maintenance; and a health value ≤ 40 requires shutdown.

[0144] In addition to displaying the unit's overall health value, the health system also provides the current health values ​​of other core components in the unit, allowing operators to assess the equipment's health status in advance based on these values.

[0145] This example uses the hot air unit in coking production as a practical case, following the above-mentioned equipment health assessment method. It includes the calibration values ​​of the secondary indicators of each core component of the unit calculated using the above method, the subjective weights of the primary indicators of each core component calculated using fuzzy hierarchical analysis (FAHP), and the objective weights of the secondary indicators of each core component calculated using standard conflict correlation (CRITIC).

[0146] In summary, analysis of monitoring data from the hot air unit in coking production using this comprehensive evaluation system shows that the health values ​​of the unit's core components are as shown in Table 4. Based on effective system monitoring and the minimum value principle, the system's health value is 60 points.

[0147] Table 4 Health Values ​​of Core Components of the Unit

[0148]

[0149] In summary, the comprehensive health evaluation method for key coking equipment proposed in this invention can combine equipment characteristics and production features to detect the health status of equipment in real time, reflect the operating status of equipment, ensure stable production, reduce maintenance costs, and has high engineering value.

[0150] Those skilled in the art should understand that variations can be implemented by combining existing technology with the above embodiments, which will not be elaborated here. Such variations do not affect the essence of the present invention, and will not be elaborated here either.

[0151] The preferred embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and the devices and structures not described in detail should be understood as being implemented in a conventional manner in the art. Any person skilled in the art can make many possible variations and modifications to the technical solutions of the present invention using the methods and techniques disclosed above, or modify them into equivalent embodiments with equivalent changes, without departing from the scope of the present invention. This does not affect the essential content of the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the present invention's technical solutions still fall within the protection scope of the present invention.

Claims

1. A comprehensive health evaluation method for key coking equipment, characterized in that, Includes the following steps: The equipment is classified according to the type of equipment to be evaluated, and a digital model is set up for the structure to be monitored in each piece of equipment. The digital model is divided into two levels. The first-level digital model is determined by industry-standard production experience and sets first-level indicators, which serve as the guide for the second-level digital model. The secondary digital model sets secondary indicators based on equipment type, production function and / or operating environment, and sets corresponding detection sensors on corresponding parts of the equipment based on the secondary digital model; Based on the real-time detection samples from the sensors, analytical data is constructed and preprocessed to remove invalid operational data, abnormal interference data from the field, and sensor data acquisition errors from the detection signals. Feature extraction is performed on the preprocessed signal to extract the physical quantities corresponding to the secondary indicators required for analyzing the health status of the equipment, forming an analysis feature matrix, and the analysis feature matrix is ​​normalized. Based on the importance of key components in production, the consequences of failure, the probability of historical failure, the difficulty of maintenance, and the safety risks, a custom weight for key components is introduced into the normalized feature values ​​of the same secondary indicators monitored by similar sensors on different components of the equipment. To account for abnormal fluctuations during equipment operation, an adaptive weight correction is introduced; the weighted average of the normalized feature values ​​is calculated as the calibration value. After normalization, the calibration value reflects the degree of danger in equipment operation; the larger the value, the worse the equipment health. For the calibration values ​​of primary indicators and the same secondary indicators of equipment, the subjective weights corresponding to the primary indicators are calculated based on the FAHP method. The objective weights corresponding to the secondary indicators are calculated based on the CRITIC method. The comprehensive risk value of the component is calculated based on the corrected calibration values ​​of the monitoring data, as well as the weights of the primary and secondary indicators. This comprehensive risk value is then converted inversely into a health value ranging from 0 to 100 points. The calibration value is a risk-type calibration value; a higher value indicates a higher equipment risk. The health value and the comprehensive risk value change inversely. The formula for calculating the health value is: In the formula Health correction factor; double summation term Calculate the comprehensive risk value of the k-th component, with a value ranging from [0,1]. A larger value indicates a higher risk for the equipment. Here, n is the number of primary indicators. Let be the number of secondary indicators under the i-th primary indicator. The final comprehensive weight of the j-th secondary indicator is... This is the normalized calibration value of the index corresponding to the k-th component; Determine the health status of the equipment based on its health values.

2. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, The primary indicators include vibration, electrical, and process indicators. The secondary indicators for vibration include one or more of the following: time domain, frequency domain, waveform, and demodulation. The secondary indicators for process indicators include one or more of the following: temperature, pressure, flow rate, thermal stress risk indicators, and dust wear risk indicators. The thermal stress risk indicators are determined by one or more of the following: temperature deviation, temperature gradient, and temperature rise rate. The dust wear risk indicators are determined by one or more of the following: dust concentration, filtration or sealing pressure difference, and high-frequency impact characteristics. The secondary indicators for electrical indicators include one or more of the following: voltage, current, and power. The sensors installed on each equipment component include one or more of the following: velocity sensors, acceleration sensors, displacement sensors, temperature sensors, voltage sensors, current sensors, pressure sensors, flow sensors, and dust concentration sensors.

3. The comprehensive health evaluation method for key coking equipment according to claim 2, characterized in that, For the key monitoring sections of the rotating equipment, two orthogonal vibration sensors are arranged according to the full vector spectrum requirements to collect signals in the X and Y directions respectively. The full vector spectrum vibration signals are also preprocessed by X / Y channel synchronization correction, X / Y channel sensitivity correction, X / Y direction polarity correction, phase delay correction, and order tracking when the rotation speed changes. The full vector spectrum is obtained by constructing complex vibration signals for each monitoring section, and the vibration characteristics corresponding to the monitoring section are extracted from the full vector spectrum. When preprocessing the full vector spectrum vibration signal, dual-channel installation angle compensation and baseline drift compensation are also included. Dual-channel installation angle compensation is used to correct two vibration signals that are not strictly orthogonal to orthogonal vibration signals, and baseline drift compensation is used to eliminate the effects of zero-point drift and temperature drift during long-term operation of the sensor.

4. The comprehensive health evaluation method for key coking equipment according to claim 3, characterized in that, The vibration characteristics include one or more of the following: total vector amplitude, principal vector, secondary vector, forward and reverse precession components, and principal direction angle.

5. A comprehensive health evaluation method for key coking equipment according to claim 3 or 4, characterized in that, For vibration-related indicators, dynamic hazard values ​​are used for normalization, and the calculation formula is as follows: ,in, Let j be the feature value of the j-th secondary index of the k-th component. These are reference values ​​for normal operation. These are industry-standard alarm values. This is the dynamic coefficient for operating conditions, determined based on equipment load, speed deviation, and temperature conditions, with a value range of 0.8 to 1.

5. Specifically, the value is 0.8 to 0.95 for light load stable conditions; 0.95 to 1.05 for rated stable conditions; 1.15 to 1.30 for heavy load and high temperature conditions; and 1.30 to 1.50 for extreme heavy load or drastic fluctuation conditions. For other constraint-related indicators, an upper limit normalization method is used, and the calculation formula is as follows: ,in This represents the risk value corresponding to the indicator.

6. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, The process of obtaining the subjective weights of primary indicators using the FAHP method includes the following steps: Construct a hierarchical structure: with equipment health status assessment as the target layer, first-level indicators as the criterion layer, and second-level indicators as the indicator layer, forming a three-level hierarchical structure; Constructing fuzzy judgment matrices: For the first-level indicators of the criterion layer, based on the overall goal of equipment health status, multiple coking experts were invited to conduct pairwise comparisons of the relative importance between indicators using a fuzzy scale of 0.1 to 0.9, and fuzzy judgment matrices were constructed accordingly. Where n is the number of primary indicators. The fuzzy scaling value represents two indicators of equal importance. A value of 0.5 indicates that the two indicators are equally important, while a larger value indicates that the former indicator is more important than the latter, and the following conditions must be met: ; When an expert's score deviates from other experts' scores by more than a set threshold, the expert's weight can be reduced or a re-score can be required. Calculating subjective weights: Normalize the fuzzy judgment matrix, sum the normalized columns, and then divide the summed vector by the matrix order n to obtain the subjective weight vector of the first-order index. The calculation formula is as follows: The resulting weight vector satisfies the normalization constraint. ; Consistency check: Calculate the consistency index of the judgment matrix, where is the largest eigenvalue of the judgment matrix; combine with the average consistency index RI to calculate the consistency ratio CR; When CR < 0.1, the judgment matrix passes the consistency test, and the subjective weights of the obtained primary indicators are valid; If the test fails, the fuzzy judgment matrix is ​​readjusted until the test is passed; through the above process, the final weights of the primary indicators are finally obtained. ,Right now .

7. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, The process of obtaining the objective weights of secondary indicators using the CRITIC method includes the following steps: Zero value determination of calibration value: When the calibration value of a certain secondary indicator is zero on all components, it is determined to be a normal indicator with all zeros. The information content of the indicator in the current period is zero and it does not participate in the objective weight competition. When only some components of a secondary indicator have a value of zero, it is classified as a locally zero indicator. The low-risk meaning represented by the zero value is retained, and a very small positive number is added to the denominator of the coefficient of variation calculation to avoid calculation instability. That is, if a column of a secondary indicator has a value of 0, a very small positive number is uniformly added to all values ​​in that column. The corrected data does not change the discrete nature of the data or the correlation between indicators; the correction formula is: ,in Let j be the standard deviation of the j-th indicator. The average value of the j-th indicator; Comparison strength of indicators: The coefficient of variation is used to characterize the comparison strength of indicators. A large coefficient of variation indicates that the indicator is more dispersed among different components, has a stronger ability to distinguish the health status of equipment, and contributes more to the health assessment. Calculating the conflict between indicators: Based on the Pearson correlation coefficient, the conflict between secondary indicators belonging to the same primary indicator is calculated. A higher conflict rate indicates lower information overlap and higher independent information content between the indicator and other indicators. The conflict quantification value of the j-th secondary indicator is: ( ),in Let be the Pearson correlation coefficient between the j-th and h-th secondary indicators, and m be the number of secondary indicators under the primary indicator; The larger the value, the greater the conflict between this indicator and other indicators at the same level; Calculate the comprehensive information content of the indicators ,in Let be the coefficient of variation of the j-th secondary indicator: taking into account the comparative strength and conflict, calculate the comprehensive information content of the secondary indicator. The larger the comprehensive information content, the higher the importance of the indicator in the health assessment. Calculating the objective weights of secondary indicators: The comprehensive information content of all secondary indicators belonging to the same primary indicator is normalized to obtain the corresponding weights of the secondary indicators. The calculation formula is as follows: ,in Let the objective weights of the j-th secondary indicators under the i-th primary indicator satisfy the normalization constraints. Through the above process, the final weights of the secondary indicators under each primary indicator are calculated. ,Right now .

8. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, When calculating the final comprehensive weight of the secondary indicators, a subjective and objective weight balance distribution coefficient is introduced. ,in The combined secondary weight of the j-th secondary indicator under the i-th primary indicator is the weighted sum of the expert subjective secondary weight and the CRITIC objective secondary weight. The final comprehensive weight is the product of the primary indicator FAHP weight and the combined secondary weight. The formula for calculating the final comprehensive weight is as follows: ,in Let be the final comprehensive weight of the j-th secondary indicator under the i-th primary indicator in the entire evaluation system; This represents the FAHP weight of the i-th primary indicator to which the secondary indicator belongs; The expert subjective weight is the j-th secondary indicator under the i-th primary indicator; Let be the objective weight of the CRITIC for the j-th secondary indicator under the i-th primary indicator.

9. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, Before determining the equipment health value according to the weakest link principle, the health values ​​of each component are first filtered using a moving average, and the validity of single abnormal measurement points is judged. When a single measurement point experiences communication interruption, constant signal, over-range, zero-point drift anomaly, or isolated sudden change, the measurement point is masked and the component calibration value is recalculated. When the number of valid measurement points is lower than the set lower limit, a sensor fault alarm is output, instead of directly judging the entire machine as a shutdown state. Finally, the lowest score of each component in the equipment is selected as the component health value, and the lowest health value of the component is the equipment health value. The formula for calculating the smoothed health value after moving average filtering is as follows: ,in This represents the health value of the k-th component in the t-th evaluation period, L is the sliding window filter length, and q represents the time offset index within the sliding window, i.e., which evaluation period to trace back.

10. The comprehensive health evaluation method for key coking equipment according to claim 1, characterized in that, The equipment health level is divided into four levels: a health value in the range of (80, 100) is good, a health value in the range of (60, 80) is usable, a health value in the range of (40, 60) requires maintenance, and a health value ≤40 requires shutdown.

Images