Rotating machinery health evaluation method and monitoring system based on multi-parameter composite empowerment

Abstract

This invention discloses a health assessment method and monitoring system for rotating machinery based on multi-parameter composite weighting, primarily relating to the field of rotating machinery condition monitoring technology. The method includes: collecting vibration parameters and current harmonic parameters, preprocessing them through anomaly removal, sliding filtering, and degradation degree normalization; calculating subjective and objective weights using the analytic hierarchy process (AHP) and grey relational analysis, respectively, and fusing them through an adaptive weighted product method guided by confidence factors to obtain combined weights; inputting the degradation degree into a Gaussian membership function to calculate the comprehensive membership degree of four health levels, outputting the health status and health index; the monitoring system adopts a hybrid transmission architecture of BLE and 4G, combined with dynamic sampling and sleep strategies to achieve low-power operation. The beneficial effects of this invention are: achieving coordinated coverage of mechanical and electrical faults, improving assessment accuracy and robustness, reducing false alarm and false negative rates, extending sensor battery life, and providing intuitive and quantitative operation and maintenance guidance for rotating machinery.

Description

Technical Field

[0001] This invention relates to the field of technology, specifically to a method and monitoring system for assessing the health of rotating machinery based on multi-parameter composite weighting. Background Technology

[0002] Rotating machinery is a core piece of equipment in industrial production, and its operating status directly affects production efficiency and safety. Accurately assessing the health status of rotating machinery and achieving early warning of faults has become an important issue in the field of intelligent equipment operation and maintenance.

[0003] Currently, existing health assessment technologies for rotating machinery have the following main shortcomings: 1. Single monitoring parameter: Most solutions rely solely on vibration signals to assess equipment status, which is insufficient to cover electrical faults such as aging of motor winding insulation and uneven air gap. Furthermore, under complex working conditions such as forging and impact, fluctuations in a single parameter can easily lead to inaccurate assessment results. 2. Lack of rationality in weighting methods: Traditional subjective weighting methods rely on expert experience and are easily influenced by human preferences; while objective weighting methods can reflect the characteristics of data distribution, they ignore the physical importance of parameters in industrial scenarios. Both are difficult to scientifically allocate the weight of each monitoring parameter, resulting in insufficient reliability of health assessment. 3. Vague health level classification: Most existing technologies output binary results of "normal" or "abnormal", lacking a quantitative health level system. This fails to reflect the continuous evolution of equipment from minor deterioration to serious failure, which is not conducive to maintenance personnel to formulate differentiated maintenance strategies. 4. Significant issues with on-site deployment and energy consumption: Wired sensors have complex wiring and are difficult to retrofit, while continuous wireless transmission consumes a lot of power and cannot meet the battery life requirements for long-term online monitoring.

[0004] Therefore, there is an urgent need for a rotating machinery health assessment method and monitoring system that can integrate multi-parameter collaborative monitoring, subjective and objective combined weighting, quantitative level assessment, and low-power wireless deployment. Summary of the Invention

[0005] The purpose of this invention is to provide a rotating machinery health assessment method and monitoring system based on multi-parameter composite weighting. It can achieve coordinated coverage of mechanical and electrical faults, improve assessment accuracy and robustness, reduce false alarm rate and false alarm rate, extend sensor battery life, and provide intuitive and quantitative operation and maintenance guidance for rotating machinery.

[0006] To achieve the above objectives, the present invention employs the following technical solution: On the one hand, the present invention provides a method for health assessment of rotating machinery based on multi-parameter composite weighting, comprising the following steps: Step S1: Collect vibration parameters and current harmonic parameters of the rotating machinery. Perform outlier removal, sliding filtering, time alignment, and degradation normalization preprocessing on the collected parameters. Map each parameter to... interval; Step S2: The subjective weights of each parameter are calculated using the analytic hierarchy process (AHP), and the objective weights of each parameter are calculated using the grey relational analysis method. The subjective and objective weights are then combined using the weighted product method and normalized to obtain the combined weights. Step S3: Input the degradation degree of each parameter obtained in step S1 into the Gaussian membership function, calculate the membership degree of each parameter to the four health levels of good, average, attention, and severe, calculate the comprehensive membership degree of each health level using the weighted average method, take the level corresponding to the maximum value as the current health status output, calculate the health index, execute multi-level early warning based on the health index and output operation and maintenance suggestions. Step S4: Execute multi-level early warnings and generate operation and maintenance suggestions based on health indices.

[0007] Preferably, in step S1, the vibration parameters include the effective value of acceleration and kurtosis, and the current harmonic parameters include the content of the 2nd to 10th harmonics. RMS acceleration Represented as ,in For the first One acceleration sample value, This represents the number of sampling points within a single sampling window. cliff Represented as ,in The arithmetic mean of the acceleration sample values; No. Subharmonic content Represented as ,in For the first The effective value of the subharmonic current. This is the effective value of the fundamental current. The value of is an integer ranging from 2 to 10.

[0008] Preferably, in step S1, the normalized degradation degree is expressed as: ; in, These are the measured values ​​of the parameters. and These are the lower and upper limits of the parameter under normal operating conditions, respectively, obtained based on historical data from 30 days of continuous operation of rotating machinery under normal conditions; and automatically updated every 30 days based on the health status assessment results of the most recent 90 days. and The update rule is as follows: when the percentage of days assessed as good in the past 30 days is greater than or equal to 80%, the original boundary remains unchanged; when the percentage of days assessed as attention or serious is greater than or equal to 20%, the boundary value is recalculated.

[0009] Preferably, in step S2, the calculation of the subjective weight includes: A three-layer hierarchical structure consisting of a target layer, a criterion layer, and a parameter layer is constructed. Pairwise comparisons are performed on the parameters of the criterion layer to construct a judgment matrix. The comparison results are quantified using a 1-9 scaling method. The subjective weight vector is obtained by calculating the eigenvectors of the judgment matrix. And perform a consistency check on the judgment matrix, and the consistency ratio. ,in , To determine the largest eigenvalue of a matrix, For the number of parameters, As a random consistency indicator, when The matrix is ​​valid at that time. After all the judgment matrices pass the consistency test, the eigenvectors are geometrically averaged and then normalized to obtain the final result. .

[0010] Preferably, in step S2, the calculation of the objective weight includes: Ideal reference sequence corresponding to a state of perfect health As a reference sequence, measured data under normal and fault conditions were selected as the comparison sequence. Calculate the correlation coefficient between each comparison sequence and the reference sequence. : ; in, The resolution coefficient is set to 0.5. After calculating the correlation degree of each parameter, normalization is performed to obtain the objective weight vector. .

[0011] Preferably, in step S2, the weighted product fusion adopts an adaptive variable weight fusion mechanism guided by a confidence factor, specifically including: Step S21: Calculate the subjective confidence factor vector: For the The evaluation parameter is determined based on the judgment matrix. row element Weight ratio corresponding to that row The degree of deviation, constructing local consistency deviation. , represented as ,in For the total number of parameters, and The first The and the first Subjective weights of each parameter; Based on the local consistency deviation Calculate the first Subjective confidence factor for each parameter : ; in, The preset allowable deviation threshold is between 0.1 and 0.2. This is the attenuation coefficient, with a value between 0.2 and 0.5. This yields the subjective confidence factor vector corresponding to each parameter. ; Step S22: Calculate the objective confidence factor vector: In grey relational analysis, several comparison sequences under different working conditions are selected, and the correlation coefficients between each parameter and the ideal reference sequence are calculated. Furthermore, for each parameter... Obtain the standard deviation of its correlation coefficient sequence This is used to measure the volatility of the correlation of this parameter; when Greater than the preset threshold When, calculate the first Objective confidence factor for each parameter : ,otherwise ; in For parameters The largest observed value of the standard deviation of the correlation coefficient in historical assessments. This is the fluctuation tolerance threshold, with a value between 0.02 and 0.08. This is the attenuation coefficient, with a value between 0.2 and 0.5. This yields the objective confidence factor vector corresponding to each parameter. ; Step S3: Merge to obtain combined weights: The weights of each parameter combination are calculated using the weighted product method adjusted by the aforementioned vectorized confidence factor. : ; in and The first The subjective and objective weights of each parameter. This represents the total number of parameters.

[0012] Preferably, in step S3, the Gaussian membership function is expressed as: ,in The central values ​​for each health level are... The width value represents the center value of each health level, and the corresponding relationship between the center value and the degree of deterioration range is as follows: Good grade The corresponding interval is [0, 0.35]. General level The corresponding interval is [0, 0.65]. Note the level The corresponding interval is [0.35, 0.85]. Severity level The corresponding interval is [0.6, 1.0]. All values ​​are taken as 0.18; The health index is set as follows: , which is represented as ,in For the first The overall membership degree of each health level, For the first The quantitative scores for the four health levels are as follows: Good (100 points), Average (75 points), Attention (50 points), and Serious (25 points). The value range is a continuous number between 25 and 100.

[0013] Preferably, step S4 is as follows: When the health status is assessed as general, the sampling interval of vibration parameters and current harmonic parameters is shortened from 10 seconds under normal operating conditions to 1 second, and maintenance prompts containing a list of recommended inspection items are generated. The list of recommended inspection items is generated based on the fault type corresponding to the parameter with the highest value among the parameters of deterioration. When the health status is assessed as the attention level, a local audible and visual warning signal is triggered. The complete waveform data for 60 seconds before and after the warning time is stored on the local edge computing node, and the health status and at least two parameters with the highest degradation are uploaded to the cloud platform through the 4G module. When the health status is assessed as severe, the health status, degradation parameter list, and spectrum analysis results of at least three fault characteristic frequencies are simultaneously uploaded to the cloud platform, and an immediate shutdown and maintenance instruction is pushed to the operation and maintenance personnel's terminal. The immediate shutdown and maintenance instruction includes fault location information and maintenance operation instructions.

[0014] On the other hand, the present invention also provides a rotating machinery health assessment system based on multi-parameter composite weighting, for implementing the rotating machinery health assessment method based on multi-parameter composite weighting as described above, including: The sensing layer includes at least one vibration sensor and at least one current harmonic sensor. The vibration sensor is installed on the main bearing end and gearbox of the rotating machinery, and the current harmonic sensor is clamped in the three-phase input cable of the motor. The transmission layer includes a BLE gateway module and a 4G module. Each sensor in the perception layer establishes a wireless connection with the BLE gateway module through the BLE 5.0 protocol. The BLE gateway module establishes a data transmission channel with the cloud platform through the 4G module. The edge computing layer, including edge computing nodes, is used to perform preprocessing operations and composite weighted evaluation calculations; The cloud platform layer, including cloud servers, is used to store time-series data from various sensors, health level assessment results at various time points, historical early warning records, and operation logs of maintenance personnel, and provides data display and query interfaces.

[0015] Preferably, the sensing layer employs a dynamic power management strategy: When the health status assessment of the rotating machinery is good and the degradation of each parameter is no greater than 0.2, the sampling interval of each sensor is 10 seconds and the data upload cycle is 60 seconds. When the degradation degree of any parameter is greater than 0.2 and not greater than 0.5, the sampling interval is shortened to 5 seconds and the data upload cycle is shortened to 30 seconds. When the degradation of any parameter is greater than 0.5, the sampling interval is shortened to 1 second and the data upload cycle is shortened to 10 seconds; When there is no data transmission, the sensor enters a deep sleep mode. In the sleep mode, all functional modules inside the sensor except the real-time clock module are powered off, and the wake-up cycle is 1 second. The transmission layer uses the LZ77 algorithm to perform lossless compression on the uploaded data, achieving a compression ratio of 3:1. Only the original waveform data of parameters with a degradation degree greater than 0.3 and the health level assessment results are uploaded. The remaining data is discarded after preprocessing in the edge computing layer. The battery life of a single sensor in the system is no less than 11 months.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention simultaneously collects vibration parameters and current harmonic parameters, enabling coordinated monitoring of mechanical faults such as gear wear and bearing pitting, as well as electrical faults such as winding aging and uneven air gap. This overcomes the limitations of traditional solutions that rely on only a single parameter and significantly reduces the risk of misjudgment under complex working conditions.

[0017] 2. This invention employs the analytic hierarchy process (AHP) and grey relational analysis to calculate subjective and objective weights respectively, and then fuses them using an adaptive weighted product method guided by confidence factors. This approach retains the emphasis of expert experience on key parameters while also taking into account the objective laws of actual operating data. This composite weighting strategy enhances the scientific nature of weight allocation, making the health assessment results more closely reflect the actual condition of the equipment.

[0018] 3. This invention maps the degree of degradation to four health levels—Good, Average, Attention, and Critical—using a Gaussian membership function, and outputs a continuous health index. Compared to traditional binary judgment results, the four-level system clearly reflects the gradual process of equipment degradation from minor to severe failure, providing clear quantitative basis for maintenance personnel to formulate preventive maintenance plans.

[0019] 4. This invention adopts a hybrid BLE and 4G wireless transmission architecture, eliminating the need for complex wiring and enabling rapid upgrades to older equipment. Combined with dynamic sampling interval adjustment, data compression transmission, and a deep sensor sleep mechanism, it effectively reduces overall system power consumption and significantly extends the continuous operating time of on-site monitoring equipment.

[0020] 5. This invention triggers differentiated early warnings and maintenance suggestions based on different health levels, realizing closed-loop management from routine monitoring, encrypted sampling, local early warning to remote push of maintenance instructions, thereby improving the timeliness and pertinence of equipment maintenance. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0022] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined in this application.

[0023] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0024] Example: This embodiment takes a rotating machinery group consisting of multiple ring rolling machines in a forging workshop as the monitoring object, constructs a health assessment system based on multi-parameter composite weighting, and applies the method of this invention for online monitoring and intelligent operation and maintenance.

[0025] like Figure 1 As shown, this embodiment provides a method for assessing the health of rotating machinery based on multi-parameter composite weighting, including the following steps: Step S1: Collect vibration parameters and current harmonic parameters of the rotating machinery. Perform outlier removal, sliding filtering, time alignment, and degradation normalization preprocessing on the collected parameters. Map each parameter to... interval; Step S2: The subjective weights of each parameter are calculated using the analytic hierarchy process (AHP), and the objective weights of each parameter are calculated using the grey relational analysis method. The subjective and objective weights are then combined using the weighted product method and normalized to obtain the combined weights. Step S3: Input the degradation degree of each parameter obtained in step S1 into the Gaussian membership function, calculate the membership degree of each parameter to the four health levels of good, average, attention, and severe, calculate the comprehensive membership degree of each health level using the weighted average method, take the level corresponding to the maximum value as the current health status output, calculate the health index, execute multi-level early warning based on the health index and output operation and maintenance suggestions. Step S4: Execute multi-level early warnings and generate operation and maintenance suggestions based on health indices.

[0026] like Figure 2 As shown, this embodiment also provides a rotating machinery health assessment system based on multi-parameter composite weighting, including: The sensing layer includes at least one vibration sensor and at least one current harmonic sensor. The vibration sensor is installed on the main bearing end and gearbox of the rotating machinery, and the current harmonic sensor is clamped in the three-phase input cable of the motor. The transmission layer includes a BLE gateway module and a 4G module. Each sensor in the perception layer establishes a wireless connection with the BLE gateway module through the BLE 5.0 protocol. The BLE gateway module establishes a data transmission channel with the cloud platform through the 4G module. The edge computing layer, including edge computing nodes, is used to perform preprocessing operations and composite weighted evaluation calculations; The cloud platform layer, including cloud servers, is used to store time-series data from various sensors, health level assessment results at various time points, historical early warning records, and operation logs of maintenance personnel, and provides data display and query interfaces.

[0027] The perception layer employs a dynamic power management strategy: When the health status assessment of the rotating machinery is good and the degradation of each parameter is no greater than 0.2, the sampling interval of each sensor is 10 seconds and the data upload cycle is 60 seconds. When the degradation degree of any parameter is greater than 0.2 and not greater than 0.5, the sampling interval is shortened to 5 seconds and the data upload cycle is shortened to 30 seconds. When the degradation of any parameter is greater than 0.5, the sampling interval is shortened to 1 second and the data upload cycle is shortened to 10 seconds; When there is no data transmission, the sensor enters a deep sleep mode. In the sleep mode, all functional modules inside the sensor except the real-time clock module are powered off, and the wake-up cycle is 1 second. The transport layer uses the LZ77 algorithm to perform lossless compression on the uploaded data, achieving a compression ratio of 3:1. Only the original waveform data of parameters with a degradation degree greater than 0.3 and the health level assessment results are uploaded. The remaining data is discarded after preprocessing at the edge computing layer. The battery life of a single sensor in the system is no less than 11 months.

[0028] Specifically: First, perform step S1: install vibration sensors on the main bearing end of the main roller gearbox bearing and on the gearbox housing of each ring rolling mill; clamp current harmonic sensors at the three-phase input cable of the main roller motor; and connect the sensors via BLE. The 5.0 protocol establishes a wireless connection with the on-site BLE gateway, forming a sensing layer. Vibration sensors continuously collect acceleration signals, and edge computing nodes calculate the effective value of acceleration and kurtosis as vibration parameters. Current harmonic sensors collect current waveforms and extract the content of harmonics from the 2nd to the 10th order as current harmonic parameters. The edge computing nodes perform preprocessing on the two types of parameters: first, based on the 3σ principle, abnormal sudden values ​​caused by instantaneous impacts are eliminated; then, random noise is smoothed by sliding window filtering while retaining fault characteristics. The window length and overlap rate are preset according to the machine's operating speed. Then, the vibration parameter sequence and the current harmonic parameter sequence are time-aligned according to the data timestamp to eliminate the fusion deviation caused by transmission delay. Finally, the degradation degree normalization formula is used to map each parameter to the [0,1] interval. The upper and lower limits of normal operation are obtained based on the historical data statistics of the ring rolling mill under continuous normal operating conditions for one month, and the boundary values ​​are dynamically updated based on the health assessment results over a period of time. When the proportion of days with good recent conditions is high, the original boundary is maintained; when the proportion of days with warning or severe conditions increases, the boundary values ​​are recalculated to ensure that the degradation degree benchmark remains reasonable as the equipment status changes.

[0029] The process then proceeds to step S2, composite weighting calculation. First, multiple rotating machinery diagnostic experts construct a three-layer structure—target layer, criterion layer, and parameter layer—using the analytic hierarchy process (AHP). Criterion layer parameters, such as effective acceleration values, kurtosis, and dominant harmonic content, are compared pairwise to construct a judgment matrix. The relative importance is quantified using a 1-9 scale. The eigenvector of the judgment matrix is ​​calculated to obtain the subjective weight vector, and the validity of the judgment matrix is ​​verified using the consistency ratio (CR). Next, the objective weight is calculated using the grey relational analysis method. An ideal healthy sequence with zero degradation of all parameters is used as the reference sequence. Multiple sets of measured data under normal operating conditions and known fault conditions are used to construct a comparison sequence. The correlation coefficient and correlation degree between each comparison sequence and the reference sequence are calculated. After normalization, the objective weight vector is obtained. To address the impact of fluctuations in the consistency of subjective weights and changes in the dispersion of objective weight data on the reliability of fusion, this embodiment introduces an adaptive variable weight fusion mechanism guided by parameter-by-parameter confidence factors. For subjective weights, a differentiated subjective confidence factor is assigned to each evaluation parameter based on the degree of local consistency deviation when comparing each parameter with other parameters in the judgment matrix. For objective weights, a differentiated objective confidence factor is assigned to each parameter based on the coefficient of variation of the grey relational degree sequence of each parameter within the sliding window. Then, a weighted product method is used to multiply the subjective weight and objective weight of each parameter by their respective corresponding confidence factors and calculate the product. After normalization, the final combined weight is obtained. This achieves a dynamic balance between expert experience and data patterns, enhancing the robustness of weighting.

[0030] Step S3 uses a Gaussian membership function to map the degradation degree of each parameter to four health levels: good, average, attention, and severe. Each level has a pre-set center value and width value. The center value corresponds to the typical degree of slight degradation, initial degradation, significant degradation, and severe degradation, respectively. The membership degree of each parameter to the four levels is calculated to form a membership matrix. A weighted average method is used to synthesize the combined weights obtained in step S2 with the membership matrix to obtain the comprehensive membership degree of each health level. The level corresponding to the highest comprehensive membership degree is taken as the current health status output. Simultaneously, a continuous health index is calculated based on the comprehensive membership degree and the preset quantitative scores for each level. A higher health index indicates that the equipment is closer to complete health, while a lower index indicates a deeper degree of degradation. Step S4 executes multi-level early warnings based on the health status and health index. The system recommends that when the health status is normal, the system automatically shortens the sampling interval of each sensor from the longer period under normal operating conditions to a high-frequency mode, and generates a list of suggested inspection items pointing to the fault type corresponding to the parameter with the highest current degradation. This list is then pushed to the on-site display screen to remind maintenance personnel to pay attention. When the health status is critical, the local edge computing node triggers an audible and visual warning, and simultaneously records complete waveform data for a period of time before and after the warning and saves it locally. The system also uploads the health status and parameters with prominent degradation to the cloud platform via the 4G module. When the health status is severe, in addition to uploading the health status and all degradation parameters, the system also simultaneously uploads the spectrum analysis results of the fault characteristic frequency and pushes an immediate shutdown and maintenance command containing fault location information and maintenance operation instructions to the bound mobile terminal.

[0031] The cloud platform layer provides storage and query interfaces for time-series data and health assessment results through cloud servers. Operation and maintenance personnel can view the health index change curves, heat maps of the deterioration degree of each parameter, and health level labels of each ring rolling machine through the web or mobile terminal. They can also trace historical warning records and operation logs to achieve closed-loop management from perception, assessment, warning to maintenance decision-making.

[0032] In this embodiment, the sensing layer adopts a dynamic power management strategy. When the sensor is in good health and the degradation of each parameter is low, it samples and uploads data periodically at longer intervals. The sensor also controls the internal functional modules, except for the real-time clock, to be in deep sleep, requiring only milliampere-level standby current. When the degradation of any parameter enters the medium range, the sampling and uploading cycle is gradually shortened. When the degradation exceeds a higher threshold, it switches to the highest frequency sampling and the shortest uploading cycle to capture the transient characteristics of fault development. The transmission layer uses a lossless compression algorithm for the uploaded data, uploading only the original waveform data of parameters whose degradation exceeds the set threshold and the health assessment results. Other process data is discarded after being processed by the edge computing layer. Through the above low-power design, a single sensor battery in the system can work continuously for more than a year without replacement, meeting the long-term maintenance-free monitoring needs of industrial sites.

[0033] In actual operation, when early wear of the gears in the cone roller gearbox of a certain ring rolling mill occurred, the degradation degree of the effective vibration value and the corresponding harmonic content degradation degree increased simultaneously. The system assessed it as a warning level and issued an audible and visual warning. After shutdown and inspection, it was confirmed that there was slight pitting on the gear tooth surface, which was consistent with the assessment conclusion. On another ring rolling mill, the main roller motor caused the high-order harmonic content degradation degree to enter the severe range due to slow aging of the winding insulation. The system immediately pushed a shutdown command containing fault location and maintenance guidance, avoiding a sudden insulation breakdown accident, and verifying the engineering effectiveness of the method and system of the present invention.

[0034] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A method for health assessment of rotating machinery based on multi-parameter composite weighting, characterized in that, Includes the following steps: Step S1: Collect vibration parameters and current harmonic parameters of the rotating machinery. Perform outlier removal, sliding filtering, time alignment, and degradation normalization preprocessing on the collected parameters. Map each parameter to... interval; Step S2: The subjective weights of each parameter are calculated using the analytic hierarchy process (AHP), and the objective weights of each parameter are calculated using the grey relational analysis method. The subjective and objective weights are then combined using the weighted product method and normalized to obtain the combined weights. Step S3: Input the degradation degree of each parameter obtained in step S1 into the Gaussian membership function, calculate the membership degree of each parameter to the four health levels of good, fair, attention, and serious, calculate the comprehensive membership degree of each health level using the weighted average method, take the level corresponding to the maximum value as the current health status output, and calculate the health index. Step S4: Execute multi-level early warnings and generate operation and maintenance suggestions based on health indices.

2. The method for assessing the health of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S1, the vibration parameters include the effective value of acceleration and kurtosis, and the current harmonic parameters include the content of the 2nd to 10th harmonics. RMS acceleration Represented as ,in For the first One acceleration sample value, This represents the number of sampling points within a single sampling window. cliff Represented as ,in The arithmetic mean of the acceleration sample values; No. Subharmonic content Represented as ,in For the first The effective value of the subharmonic current. This is the effective value of the fundamental current. The value of is an integer ranging from 2 to 10.

3. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S1, the normalized degradation degree is expressed as: ； in, These are the measured values ​​of the parameters. and These are the lower and upper limits of the parameter under normal operating conditions, respectively, obtained based on historical data from 30 days of continuous operation of rotating machinery under normal conditions; and automatically updated every 30 days based on the health status assessment results of the most recent 90 days. and The update rule is as follows: when the percentage of days assessed as good in the past 30 days is greater than or equal to 80%, the original boundary remains unchanged; when the percentage of days assessed as attention or serious is greater than or equal to 20%, the boundary value is recalculated.

4. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S2, the calculation of the subjective weight includes: A three-layer hierarchical structure consisting of a target layer, a criterion layer, and a parameter layer is constructed. Pairwise comparisons are performed on the parameters of the criterion layer to construct a judgment matrix. The comparison results are quantified using a 1-9 scaling method. The subjective weight vector is obtained by calculating the eigenvectors of the judgment matrix. And perform a consistency check on the judgment matrix, and the consistency ratio. ,in , To determine the largest eigenvalue of a matrix, For the number of parameters, As a random consistency indicator, when The matrix is ​​valid at that time. After all the judgment matrices pass the consistency test, the eigenvectors are geometrically averaged and then normalized to obtain the final result. .

5. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S2, the calculation of the objective weight includes: Ideal reference sequence corresponding to a state of perfect health As a reference sequence, measured data under normal and fault conditions were selected as the comparison sequence. Calculate the correlation coefficient between each comparison sequence and the reference sequence. : ； in, The resolution coefficient is set to 0.

5. After calculating the correlation degree of each parameter, normalization is performed to obtain the objective weight vector. .

6. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S2, the weighted product fusion adopts an adaptive variable weight fusion mechanism guided by the confidence factor, specifically including: Step S21: Calculate the subjective confidence factor vector: For the The evaluation parameter is determined based on the judgment matrix. row element Weight ratio corresponding to that row The degree of deviation, constructing local consistency deviation. , represented as ,in For the total number of parameters, and The first The and the first Subjective weights of each parameter; Based on the local consistency deviation Calculate the first Subjective confidence factor for each parameter : ； in, The preset allowable deviation threshold is between 0.1 and 0.

2. This is the attenuation coefficient, with a value between 0.2 and 0.

5. This yields the subjective confidence factor vector corresponding to each parameter. ; Step S22: Calculate the objective confidence factor vector: In grey relational analysis, several comparison sequences under different working conditions are selected, and the correlation coefficients between each parameter and the ideal reference sequence are calculated. Furthermore, for each parameter... Obtain the standard deviation of its correlation coefficient sequence This is used to measure the volatility of the correlation of this parameter; when Greater than the preset threshold When, calculate the first Objective confidence factor for each parameter : ,otherwise ; in For parameters The largest observed value of the standard deviation of the correlation coefficient in historical assessments. This is the fluctuation tolerance threshold, with a value between 0.02 and 0.

08. This is the attenuation coefficient, with a value between 0.2 and 0.

5. This yields the objective confidence factor vector corresponding to each parameter. ; Step S3: Merge to obtain combined weights: The weights of each parameter combination are calculated using the weighted product method adjusted by the aforementioned vectorized confidence factor. : ； in and The first The subjective and objective weights of each parameter. This represents the total number of parameters.

7. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, In step S3, the Gaussian membership function is expressed as: ,in The central values ​​for each health level are... The width value represents the center value of each health level, and the corresponding relationship between the center value and the degree of deterioration range is as follows: Good grade The corresponding interval is [0, 0.35]. General level The corresponding interval is [0, 0.65]. Note the level The corresponding interval is [0.35, 0.85]. Severity level The corresponding interval is [0.6, 1.0]. All values ​​are taken as 0.18; The health index is set as follows: , which is represented as ,in For the first The overall membership degree of each health level, For the first The quantitative scores for the four health levels are as follows: Good (100 points), Average (75 points), Attention (50 points), and Serious (25 points). The value range is a continuous number between 25 and 100.

8. The method for health assessment of rotating machinery based on multi-parameter composite weighting according to claim 1, characterized in that, Step S4 is as follows: When the health status is assessed as general, the sampling interval of vibration parameters and current harmonic parameters is shortened from 10 seconds under normal operating conditions to 1 second, and maintenance prompts containing a list of recommended inspection items are generated. The list of recommended inspection items is generated based on the fault type corresponding to the parameter with the highest value among the parameters of deterioration. When the health status is assessed as the attention level, a local audible and visual warning signal is triggered. The complete waveform data for 60 seconds before and after the warning time is stored on the local edge computing node, and the health status and at least two parameters with the highest degradation are uploaded to the cloud platform through the 4G module. When the health status is assessed as severe, the health status, degradation parameter list, and spectrum analysis results of at least three fault characteristic frequencies are simultaneously uploaded to the cloud platform, and an immediate shutdown and maintenance instruction is pushed to the operation and maintenance personnel's terminal. The immediate shutdown and maintenance instruction includes fault location information and maintenance operation instructions.

9. A rotating machinery health assessment system based on multi-parameter composite weighting, used to implement the rotating machinery health assessment method based on multi-parameter composite weighting as described in any one of claims 1-8, characterized in that, include: The sensing layer includes at least one vibration sensor and at least one current harmonic sensor. The vibration sensor is installed on the main bearing end and gearbox of the rotating machinery, and the current harmonic sensor is clamped in the three-phase input cable of the motor. The transmission layer includes a BLE gateway module and a 4G module. Each sensor in the perception layer establishes a wireless connection with the BLE gateway module through the BLE 5.0 protocol. The BLE gateway module establishes a data transmission channel with the cloud platform through the 4G module. The edge computing layer, including edge computing nodes, is used to perform preprocessing operations and composite weighted evaluation calculations; The cloud platform layer, including cloud servers, is used to store time-series data from various sensors, health level assessment results at various time points, historical early warning records, and operation logs of maintenance personnel, and provides data display and query interfaces.

10. The rotating machinery health assessment system based on multi-parameter composite weighting according to claim 9, characterized in that, The perception layer employs a dynamic power management strategy: When the health status assessment of the rotating machinery is good and the degradation of each parameter is no greater than 0.2, the sampling interval of each sensor is 10 seconds and the data upload cycle is 60 seconds. When the degradation degree of any parameter is greater than 0.2 and not greater than 0.5, the sampling interval is shortened to 5 seconds and the data upload cycle is shortened to 30 seconds. When the degradation of any parameter is greater than 0.5, the sampling interval is shortened to 1 second and the data upload cycle is shortened to 10 seconds; When there is no data transmission, the sensor enters a deep sleep mode. In the sleep mode, all functional modules inside the sensor except the real-time clock module are powered off, and the wake-up cycle is 1 second. The transmission layer uses the LZ77 algorithm to perform lossless compression on the uploaded data, achieving a compression ratio of 3:

1. Only the original waveform data of parameters with a degradation degree greater than 0.3 and the health level assessment results are uploaded. The remaining data is discarded after preprocessing in the edge computing layer. The battery life of a single sensor in the system is no less than 11 months.

Images