A method for industrial valve failure prediction based on multiple monitoring parameters

Abstract

The application discloses an industrial valve fault prediction method based on multiple monitoring parameters, and relates to the technical field of industrial equipment fault prediction.The application realizes real-time collection of multi-dimensional operation data of a target valve, combines a fault prediction model to periodically output prediction data, accurately determines maintenance requirements based on comparison of a fault occurrence probability and a maintenance threshold, generates maintenance information and prompt information containing historical data traceability, realizes early prediction of fault risks, effectively avoids resource waste during a low fault occurrence period and intervention lag problems during a high fault occurrence period of traditional fixed cycle prediction, determines a plurality of intervention combinations and a prediction intervention coefficient through data analysis, dynamically determines whether intervention in the prediction process is needed based on a fault probability floating value of a continuous prediction cycle, determines an interval length of next fault prediction, realizes adaptive optimization of the prediction cycle, matches the fault prediction frequency with dynamic changes in risks, and improves the prediction response speed during a high risk stage.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment fault prediction technology, specifically to an industrial valve fault prediction method based on multiple monitoring parameters. Background Technology

[0002] In industrial production systems, industrial valves are core components for fluid transport, pressure regulation, and media shut-off. Their operating status directly affects the stability, safety, and economy of the production process. With the increasing scale and automation of industrial production, the complexity of industrial valve operating conditions has increased significantly. They face various failure risks such as media corrosion, pressure fluctuations, mechanical wear, and seal aging. Once a failure occurs, it may lead to production interruption, energy waste, or even safety accidents. To achieve early prediction of industrial valve failures, existing technologies generally employ prediction schemes based on multiple monitoring parameters. This involves deploying various sensors such as pressure, temperature, vibration, and flow to collect multi-dimensional monitoring data during the operation of industrial valves. Based on preset rules, failure risk is assessed. To ensure prediction accuracy, some schemes attempt to use multi-frequency measurement strategies. However, traditional multi-frequency strategies are mostly fixed-cycle patterns and do not consider the dynamic changes in failure risk. When the probability of failure is low, fixed high-frequency prediction can lead to data redundancy and wasted computing resources. When the probability of failure is high or the risk is rapidly increasing, fixed low-frequency prediction may miss the best intervention opportunity, leading to the escalation of the failure. To address the above problems, this invention proposes a solution. Summary of the Invention

[0003] The purpose of this invention is to provide an industrial valve fault prediction method based on multiple monitoring parameters, in order to solve the problems mentioned in the background art.

[0004] This invention provides a method for predicting industrial valve failures based on multiple monitoring parameters, comprising the following steps: S1: Real-time acquisition of the target valve's operating data; S2: Every prediction cycle, a fault prediction is performed on the target valve based on all the operating data of the target valve collected within the prediction cycle, and the prediction data of the target valve in the prediction cycle is obtained. The interval of the prediction cycle is preset by the management personnel. S3: For each prediction data of the target valve in a prediction cycle, determine whether to perform fault maintenance on the target valve based on the prediction data. If it is determined that fault maintenance is required for the target valve, generate several prompt messages and one maintenance message for the target valve; otherwise, do not perform any processing. S4: After the maintenance information of the target valve reaches a preset fixed amount, perform data analysis on all the maintenance information of the target valve to determine several intervention combinations of the target valve and their predicted intervention coefficients. S5: For each prediction data of the target valve in a prediction cycle, based on the prediction data and the previous prediction data, determine whether to intervene in the failure prediction process of the target valve. If it is determined that intervention is needed in the failure prediction process of the target valve, then intervene according to the preset intervention steps.

[0005] Further, step S1 specifically includes: The operation data of the target valve is collected in real time by deploying several sensors at the corresponding monitoring points of the target valve. The types of sensors deployed include pressure sensors, temperature sensors, vibration sensors and flow sensors. The collected operation data contains several types of monitoring data, and the types of monitoring data correspond one-to-one with the types of sensors. Each type of monitoring data contains the monitoring values ​​of several monitoring parameters.

[0006] Further, step S2 specifically includes: Every prediction cycle, all operating data of the target valve collected within the prediction cycle is input into the pre-trained fault prediction model. The fault prediction model performs a fault prediction on the target valve based on all the operating data, and obtains the prediction result of the fault prediction model for all the operating data. The prediction result is used as the prediction data of the target valve in the prediction cycle. The prediction data contains several fault types, and each fault type has a probability of occurrence within a preset fixed time period.

[0007] Furthermore, in step S3, the specific steps for determining whether to perform fault maintenance on the target valve are as follows: Extract the occurrence probability of any fault type from the predicted data. Based on the risk level predefined for the fault type, extract the maintenance probability threshold corresponding to the risk level. Compare the occurrence probability with the maintenance probability threshold. If the occurrence probability is greater than or equal to the maintenance probability threshold, it is determined that fault maintenance is required for the target valve. Generate a prompt message for the target valve based on the fault type and its occurrence probability. Simultaneously, starting from the prediction period corresponding to the predicted data, backtrack for P1 prediction periods, obtain all operating data of the target valve collected within the backtracked P1 prediction periods, and generate a maintenance message for the target valve based on the predicted data of the backtracked P1 prediction periods. P1 is a preset standard recording quantity.

[0008] Furthermore, in step S3, for any generated prompt message, the prompt message is displayed to the management personnel for review, and the management personnel can set a maintenance plan for the target valve themselves.

[0009] Furthermore, in step S4, the specific steps are as follows: All predicted data are extracted sequentially from all generated maintenance information. For each maintenance information, all predicted data are extracted to construct several parameter combinations and calculate their impact coefficients on the fluctuation of failure probability. Then, all constructed parameter combinations and their impact coefficients on the fluctuation of failure probability are merged to obtain several intervention combinations for the target valve and their predicted intervention coefficients.

[0010] Furthermore, step S4 specifically includes the following steps: S41: Label all maintenance information as A1, A2, ..., Aa, where a≥1, in the order they were generated; label the monitoring parameters corresponding to all monitoring values ​​contained in the running data as B1, B2, ..., Bb, where b≥1; S42: Extract all prediction data from maintenance information A1, and label all prediction data as C1, C2, ..., Cc in chronological order of the corresponding prediction periods, where c≥1; S43: Based on the prediction period corresponding to the prediction data C2, within the operating data of the target valve in the prediction period, according to the monitoring parameters corresponding to all abnormal monitoring values, a parameter combination D1 is formed, and the influence coefficient E1 of the parameter combination D1 on the failure probability fluctuation is calculated and obtained. S45: Construct parameter combinations D2, D3, ..., Dc-1 based on the prediction periods corresponding to the prediction data C3, C4, ..., Cc, according to S42 to S43, and calculate the influence coefficients of parameter combinations D2, D3, ..., Dc-1 on the fluctuation of fault probability. S46: Extract all prediction data from maintenance information A2, A3, ..., Aa in sequence. For each maintenance information, extract all prediction data and construct several parameter combinations according to S42 to S45 and calculate their influence coefficient on the failure probability fluctuation. Then, merge all constructed parameter combinations and their influence coefficients on the failure probability fluctuation to obtain several intervention combinations for the target valve and their prediction intervention coefficients.

[0011] Furthermore, in step S5, the specific content to be determined is as follows: Extract the occurrence probability of any fault type from the predicted data and the previously obtained predicted data. Perform a difference operation on the two extracted occurrence probabilities and take the absolute value of the operation result to obtain the probability fluctuation value of the fault type in two consecutive prediction periods. According to the risk level predefined for the fault type, extract the probability fluctuation threshold corresponding to the risk level. Compare the probability fluctuation value with the probability fluctuation threshold. If the probability fluctuation value is greater than or equal to the probability fluctuation threshold, it indicates that the risk fluctuation of the fault type has exceeded the allowable range of the corresponding risk level, and the fault risk may enter a rapid evolution stage. At this time, it is determined that the fault prediction process of the target valve needs to be intervened. The fault type is marked as the intervention benchmark type. Calculate and obtain the interval Tnew of the prediction period corresponding to the next fault prediction of the target valve. The next fault prediction begins after Tnew time has elapsed since the end of the current fault prediction. Conversely, if the risk fluctuation of this type of failure is in a stable range and the failure evolution trend is gentle, it is determined that there is no need to intervene in the failure prediction process of the target valve, and the next failure prediction will continue at the preset interval.

[0012] Furthermore, in step S5, the specific intervention content is as follows: S51: Obtain all intervention benchmark types calibrated after obtaining the prediction data; based on each obtained intervention benchmark type, sequentially obtain the occurrence probability J1 and J2 of the intervention benchmark type in the prediction period and the previous prediction period, and calculate the probability fluctuation value ΔJ1=|J1-J2| of the intervention benchmark type in the prediction period. S52: Using the formula The calculation yields the following formula: r represents the total number of all intervention benchmark types obtained; ΔJq represents the probability fluctuation value of each obtained intervention benchmark type in the prediction period; JKq corresponds to the preset probability fluctuation threshold for each intervention benchmark type; Kq corresponds to the preset maximum allowable probability fluctuation threshold for each intervention benchmark type; βq represents the preset weight for each intervention benchmark type; and L1 is the predicted intervention coefficient of the intervention combination that best matches all the obtained intervention benchmark types, where best match refers to the intervention combination that contains the largest number of all the intervention benchmark types.

[0013] Compared with existing technologies, it has the following advantages: This invention collects multi-dimensional operational data of target valves in real time and combines it with a pre-trained fault prediction model to periodically output prediction data containing fault types and occurrence probabilities. Based on the comparison between the fault occurrence probability and the maintenance threshold, it accurately determines maintenance needs and generates maintenance information and prompts containing historical data traceability. This not only achieves early prediction of fault risks but also provides comprehensive data support for maintenance decisions, effectively avoiding the resource waste in low-fault situations and the intervention lag in high-fault situations caused by traditional fixed-period prediction. This invention uses data analysis to mine and form several intervention combinations and their predictive intervention coefficients. At the same time, it dynamically determines whether intervention is needed in the prediction process based on the fluctuating value of the failure probability in the continuous prediction cycle. By combining the weight of the intervention benchmark type, the probability fluctuation characteristics, and the predictive intervention coefficient of the most suitable intervention combination, the interval of the next failure prediction is accurately determined, realizing adaptive optimization of the prediction cycle. This makes the failure prediction frequency match the dynamic changes in risk, which not only improves the prediction response speed in high-risk stages, but also reduces the calculation and data redundancy in low-risk stages, ensuring the stability, safety and economy of industrial valve operation. Attached Figure Description

[0014] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 This application provides a method for predicting industrial valve failures based on multiple monitoring parameters, comprising the following steps: S1: Real-time acquisition of the target valve's operating data; Step S1 is as follows: The system collects real-time operating data of the target valve by deploying several sensors at the corresponding monitoring points. The types of sensors deployed include pressure sensors, temperature sensors, vibration sensors, and flow sensors. The collected operating data contains several types of monitoring data, and the types of monitoring data correspond one-to-one with the types of sensors. Each type of monitoring data contains the monitoring values ​​of several monitoring parameters. These monitoring parameters are pre-selected by the management personnel based on the monitoring characteristics of the corresponding type of sensor and the actual operating monitoring needs of the industrial valve. In this application, the monitoring data corresponding to the pressure sensor includes the monitoring values ​​of parameters such as the inlet pressure, outlet pressure, and internal pressure of the valve body cavity of the target valve; the monitoring data corresponding to the temperature sensor includes the monitoring values ​​of parameters such as the valve body wall temperature, the medium inlet and outlet temperature, and the valve stem seal temperature; the monitoring data corresponding to the vibration sensor includes the monitoring values ​​of parameters such as the radial vibration amplitude of the valve body, the axial vibration frequency of the valve stem, and the vibration acceleration of the valve actuator; and the monitoring data corresponding to the flow sensor includes the monitoring values ​​of parameters such as the valve inlet medium flow rate, the outlet medium flow rate, the instantaneous medium flow rate, and the cumulative medium flow rate. S2: Every prediction cycle, a fault prediction is performed on the target valve based on all the operating data of the target valve collected within the prediction cycle, and the prediction data of the target valve in the prediction cycle is obtained. The interval of the prediction cycle is preset by the management personnel. Step S2, specifically: Every prediction cycle, all operating data of the target valve collected within the prediction cycle is input into the pre-trained fault prediction model. The fault prediction model performs a fault prediction on the target valve based on all the operating data, and obtains the prediction result of the fault prediction model for all the operating data. The prediction result is used as the prediction data of the target valve in the prediction cycle. The prediction data contains several fault types, and each fault type corresponds to a probability of occurrence within a preset fixed time. Any fault type involved in the fault prediction result is a fault type that the fault prediction model has learned during the training phase and has the ability to make effective predictions. It should be noted that the occurrence probability of any fault type refers to the probability that the target valve may have the fault type within a fixed time period, as predicted by the fault prediction model, and the preset fixed time period corresponding to the occurrence probability of each fault type is consistent. S3: For each prediction data of the target valve in a prediction cycle, determine whether to perform fault maintenance on the target valve based on the prediction data. If it is determined that fault maintenance is required for the target valve, generate several prompt messages and one maintenance message for the target valve; otherwise, do not perform any processing. In step S3, the specific steps for determining whether to perform fault maintenance on the target valve are as follows: The probability of occurrence of any fault type is extracted from the predicted data. Based on the risk level predefined for the fault type, the maintenance probability threshold corresponding to the risk level is extracted. The probability of occurrence is compared with the maintenance probability threshold. If the probability of occurrence is greater than or equal to the maintenance probability threshold, it is determined that the target valve needs to be maintained. A prompt message for the target valve is generated based on the fault type and its probability of occurrence. At the same time, starting from the prediction period corresponding to the predicted data, P1 prediction periods are traced back. All operating data of the target valve collected within the traced P1 prediction periods and the prediction data of the traced P1 prediction periods are obtained to generate a maintenance message for the target valve. P1 is a preset standard record quantity. In this application, the value of P1 is set by the management personnel based on the actual data requirements of the traceability analysis and the maintenance statistical rules of similar equipment in the industry. In step S3, for any generated prompt message, the prompt message is displayed to the management personnel for viewing. The management personnel can set a maintenance plan for the target valve, which includes maintenance time, maintenance personnel, and maintenance process. S4: After the maintenance information of the target valve reaches a preset fixed amount, perform data analysis on all the maintenance information of the target valve to determine several intervention combinations of the target valve and their predicted intervention coefficients. The fixed amount is preset by the management personnel to ensure the sufficiency of the sample size and the completeness of the data dimensions, and to ensure the validity of the data analysis results. In step S4, specifically: All predicted data are extracted sequentially from all generated maintenance information. For each maintenance information, all predicted data are extracted to construct several parameter combinations and calculate their impact coefficient on the fluctuation of failure probability. Then, all constructed parameter combinations and their impact coefficients on the fluctuation of failure probability are merged to obtain several intervention combinations of the target valve and their predicted intervention coefficients. Step S4 specifically includes the following steps: S41: Label all maintenance information as A1, A2, ..., Aa, where a≥1, in the order they were generated; label the monitoring parameters corresponding to all monitoring values ​​contained in the running data as B1, B2, ..., Bb, where b≥1; S42: Extract all prediction data from maintenance information A1, and label all prediction data as C1, C2, ..., Cc in chronological order of the corresponding prediction periods, where c≥1; S43: Based on the prediction period corresponding to the prediction data C2, within the operating data of the target valve in the prediction period, according to the monitoring parameters corresponding to all abnormal monitoring values, a parameter combination D1 is formed, and the influence coefficient E1 of the parameter combination D1 on the failure probability fluctuation is calculated and obtained. The identification criteria for abnormal changes in the monitoring value of any monitoring parameter are as follows: Set the normal operating threshold range [Emin, Emax] for the monitoring parameters. This range is set by the management personnel in combination with the industrial valve design parameters, industry standards and historical operating data. Calculate and obtain the average change of the monitoring parameter during the prediction period. In the formula, n is the total number of monitored values ​​of the monitored parameters in the running data; Gi is the monitoring value of the monitoring parameter collected at the i-th moment within the monitoring period, where i is a positive integer and 2≤i≤n, and Gi-1 is the monitoring value of the monitoring parameter collected at the moment preceding the i-th moment in the monitoring period. Set a threshold value k for the rate of change, with a range of 0.1-0.3. Adjust the value according to the type of monitoring parameter, such as k=0.2 for pressure parameter and k=0.15 for temperature parameter. If any monitoring value of the monitoring parameter in the prediction period exceeds the normal operation threshold range, or the average change ΔF1 ≥ k, then the monitoring value of the monitoring parameter is determined to be abnormal; otherwise, the monitoring value of the monitoring parameter is determined to be normal. In this application, the formula for calculating E1 is as follows: In the formula Let ωh be the total change of the prediction period parameter combination D1, e be the total number of monitoring parameters within parameter combination D1, and ωh be the weight of the h-th monitoring parameter within parameter combination D1, which is set by management personnel based on the importance of the monitoring parameters to the fault anomaly. ; ΔIh is the average change of the h-th monitoring parameter during the monitoring period. ΔIh is calculated by summing the differences between the monitoring values ​​of the monitoring parameter at pairwise adjacent times during the monitoring period and then averaging the results. E1 is the historical average operating value of the h-th monitoring parameter, which is preset by the management personnel based on the historical data of the monitoring parameter; μ is a preset dimensionless factor to ensure that E1 is a dimensionless value. PC1 and PC2 are the occurrence probabilities of the target valve in the prediction period corresponding to the predicted data C1 and C2, respectively. α is a correction coefficient with a value range of 0.9-1.1. In this application, the value of α is preferably 1.05. S45: Construct parameter combinations D2, D3, ..., Dc-1 based on the prediction periods corresponding to the prediction data C3, C4, ..., Cc, according to S42 to S43, and calculate the influence coefficients of parameter combinations D2, D3, ..., Dc-1 on the fluctuation of fault probability. S46: Extract all prediction data from maintenance information A2, A3, ..., Aa in sequence. For each maintenance information, extract all prediction data and construct several parameter combinations according to S42 to S45 and calculate their influence coefficient on the failure probability fluctuation. Then, merge all constructed parameter combinations and their influence coefficients on the failure probability fluctuation to obtain several intervention combinations of the target valve and their prediction intervention coefficients. The merged content is as follows: For a unique combination of parameters, it is labeled as an intervention combination, and its influence coefficient on the fluctuation of the failure probability is used as the predictive intervention coefficient of the intervention combination. For non-unique parameter combinations, duplicates are removed, and the unique parameter combinations after deduplication are used as intervention combinations. The average of the influence coefficients of all non-unique parameter combinations on the failure probability fluctuation before deduplication is calculated, and the calculation result is used as the prediction intervention coefficient of the intervention combination. Step S5: For each prediction data of the target valve in a prediction cycle, based on the prediction data and the previous prediction data, determine whether to intervene in the next failure prediction process of the target valve. If it is determined that intervention is needed in the next failure prediction process of the target valve, then intervene according to the preset intervention steps. The previous prediction data is the prediction data obtained in the previous adjacent prediction cycle of the prediction cycle. In step S5, the specific content to be determined is as follows: Extract the occurrence probability of any fault type from the predicted data and the previously obtained predicted data. Perform a difference operation on the two extracted occurrence probabilities and take the absolute value of the result to obtain the probability fluctuation value of the fault type in two consecutive prediction periods. Based on the risk level predefined for the fault type, extract the probability fluctuation threshold corresponding to the risk level. Compare the probability fluctuation value with the probability fluctuation threshold. If the probability fluctuation value is greater than or equal to the probability fluctuation threshold, it indicates that the risk fluctuation of the fault type has exceeded the allowable range of the corresponding risk level, and the fault risk may enter a rapid evolution stage. At this time, it is determined that the fault prediction process of the target valve needs to be intervened. The fault type is marked as the intervention benchmark type. Calculate and obtain the interval Tnew of the prediction period corresponding to the next fault prediction of the target valve. The next fault prediction begins after Tnew time from the time when the current fault prediction ends. Each intervention only intervenes in the interval of the next fault prediction, and does not intervene in the interval of other fault predictions. Conversely, it indicates that the risk fluctuation of this type of failure is in a stable range, the failure evolution trend is gentle, and it is determined that there is no need to intervene in the next failure prediction process of the target valve, and the next failure prediction will continue at the preset interval. In this application, after the failure prediction model predicts the occurrence of each type of failure, the management personnel classify all predictable failure types into three risk levels: low, medium, and high, based on the degree of impact and harmful consequences that may be caused to the operation of the target valve, equipment system, and even production operation. They also set matching probability fluctuation thresholds for different risk levels to ensure that the intervention judgment in the failure prediction process has clear and unified quantitative standards, guarantee the objectivity and rationality of the intervention judgment results, and realize differentiated early warning and handling of failures with different risk levels. In step S5, the specific intervention content is as follows: S51: Obtain all intervention benchmark types calibrated after obtaining the prediction data; based on each obtained intervention benchmark type, sequentially obtain the occurrence probability J1 and J2 of the intervention benchmark type in the prediction period and the previous prediction period, and calculate the probability fluctuation value ΔJ1=|J1-J2| of the intervention benchmark type in the prediction period. S52: Using the formula The calculation is as follows: r is the total number of all intervention benchmark types after acquisition; ΔJq refers to the probability fluctuation value of each acquired intervention benchmark type in the prediction period; JKq corresponds to the preset probability fluctuation threshold of each intervention benchmark type; Kq corresponds to the preset maximum allowable probability fluctuation threshold of each intervention benchmark type; βq refers to the preset weight of each intervention benchmark type; and L1 is the predicted intervention coefficient of the intervention combination that best matches all the acquired intervention benchmark types, where best match refers to the intervention combination that contains the largest number of all the intervention benchmark types. It should be noted that if there are multiple intervention combinations that match the number of all the intervention benchmark types obtained, the intervention combination with the largest sum of risk weights is selected as the most suitable intervention combination based on the preset risk weights of each intervention benchmark type.

[0017] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0018] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for predicting industrial valve failures based on multiple monitoring parameters, characterized in that, Includes the following steps: S1: Real-time acquisition of the target valve's operating data; S2: Every prediction cycle, a fault prediction is performed on the target valve based on all the operating data of the target valve collected within the prediction cycle, and the prediction data of the target valve in the prediction cycle is obtained. The interval of the prediction cycle is preset by the management personnel. S3: For each prediction data of the target valve in a prediction cycle, determine whether to perform fault maintenance on the target valve based on the prediction data. If it is determined that fault maintenance is required for the target valve, generate several prompt messages and one maintenance message for the target valve; otherwise, do not perform any processing. S4: After the maintenance information of the target valve reaches a preset fixed amount, perform data analysis on all the maintenance information of the target valve to determine several intervention combinations of the target valve and their predicted intervention coefficients. S5: For each prediction data of the target valve in a prediction cycle, based on the prediction data and the previous prediction data, determine whether to intervene in the failure prediction process of the target valve. If it is determined that intervention is needed in the failure prediction process of the target valve, then intervene according to the preset intervention steps.

2. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, Step S1 is as follows: The operation data of the target valve is collected in real time by deploying several sensors at the corresponding monitoring points of the target valve. The types of sensors deployed include pressure sensors, temperature sensors, vibration sensors and flow sensors. The collected operation data contains several types of monitoring data, and the types of monitoring data correspond one-to-one with the types of sensors. Each type of monitoring data contains the monitoring values ​​of several monitoring parameters.

3. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, Step S2, specifically: Every prediction cycle, all operating data of the target valve collected within the prediction cycle is input into the pre-trained fault prediction model. The fault prediction model performs a fault prediction on the target valve based on all the operating data, and obtains the prediction result of the fault prediction model for all the operating data. The prediction result is used as the prediction data of the target valve in the prediction cycle. The prediction data contains several fault types, and each fault type has a probability of occurrence within a preset fixed time period.

4. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, In step S3, the specific steps for determining whether to perform fault maintenance on the target valve are as follows: Extract the occurrence probability of any fault type from the predicted data. Based on the risk level predefined for the fault type, extract the maintenance probability threshold corresponding to the risk level. Compare the occurrence probability with the maintenance probability threshold. If the occurrence probability is greater than or equal to the maintenance probability threshold, it is determined that fault maintenance is required for the target valve. Generate a prompt message for the target valve based on the fault type and its occurrence probability. Simultaneously, starting from the prediction period corresponding to the predicted data, backtrack for P1 prediction periods, obtain all operating data of the target valve collected within the backtracked P1 prediction periods, and generate a maintenance message for the target valve based on the predicted data of the backtracked P1 prediction periods. P1 is a preset standard recording quantity.

5. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, In step S3, for any generated prompt message, the prompt message is displayed to the management personnel for review, and the management personnel can set a maintenance plan for the target valve themselves.

6. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, In step S4, specifically: All predicted data are extracted sequentially from all generated maintenance information. For each maintenance information, all predicted data are extracted to construct several parameter combinations and calculate their impact coefficients on the fluctuation of failure probability. Then, all constructed parameter combinations and their impact coefficients on the fluctuation of failure probability are merged to obtain several intervention combinations for the target valve and their predicted intervention coefficients.

7. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 6, characterized in that, Step S4 specifically includes the following steps: S41: Label all maintenance information as A1, A2, ..., Aa, where a≥1, in the order they were generated; label the monitoring parameters corresponding to all monitoring values ​​contained in the running data as B1, B2, ..., Bb, where b≥1; S42: Extract all prediction data from maintenance information A1, and label all prediction data as C1, C2, ..., Cc in chronological order of the corresponding prediction periods, where c≥1; S43: Based on the prediction period corresponding to the prediction data C2, within the operating data of the target valve in the prediction period, according to the monitoring parameters corresponding to all abnormal monitoring values, a parameter combination D1 is formed, and the influence coefficient E1 of the parameter combination D1 on the failure probability fluctuation is calculated and obtained. S45: Construct parameter combinations D2, D3, ..., Dc-1 based on the prediction periods corresponding to the prediction data C3, C4, ..., Cc, according to S42 to S43, and calculate the influence coefficients of parameter combinations D2, D3, ..., Dc-1 on the fluctuation of fault probability. S46: Extract all prediction data from maintenance information A2, A3, ..., Aa in sequence. For each maintenance information, extract all prediction data and construct several parameter combinations according to S42 to S45 and calculate their influence coefficient on the failure probability fluctuation. Then, merge all constructed parameter combinations and their influence coefficients on the failure probability fluctuation to obtain several intervention combinations for the target valve and their prediction intervention coefficients.

8. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 1, characterized in that, In step S5, the specific content to be determined is as follows: Extract the occurrence probability of any fault type from the predicted data and the previously obtained predicted data. Perform a difference operation on the two extracted occurrence probabilities and take the absolute value of the operation result to obtain the probability fluctuation value of the fault type in two consecutive prediction periods. According to the risk level predefined for the fault type, extract the probability fluctuation threshold corresponding to the risk level. Compare the probability fluctuation value with the probability fluctuation threshold. If the probability fluctuation value is greater than or equal to the probability fluctuation threshold, it indicates that the risk fluctuation of the fault type has exceeded the allowable range of the corresponding risk level, and the fault risk may enter a rapid evolution stage. At this time, it is determined that the fault prediction process of the target valve needs to be intervened. The fault type is marked as the intervention benchmark type. Calculate and obtain the interval Tnew of the prediction period corresponding to the next fault prediction of the target valve. The next fault prediction begins after Tnew time has elapsed since the end of the current fault prediction. Conversely, if the risk fluctuation of this type of failure is in a stable range and the failure evolution trend is gentle, it is determined that there is no need to intervene in the failure prediction process of the target valve, and the next failure prediction will continue at the preset interval.

9. The industrial valve fault prediction method based on multiple monitoring parameters according to claim 8, characterized in that, In step S5, the specific intervention content is as follows: S51: Obtain all intervention benchmark types calibrated after obtaining the prediction data; based on each obtained intervention benchmark type, sequentially obtain the occurrence probability J1 and J2 of the intervention benchmark type in the prediction period and the previous prediction period, and calculate the probability fluctuation value ΔJ1=|J1-J2| of the intervention benchmark type in the prediction period. S52: Using the formula The calculation yields the following formula: r represents the total number of all intervention benchmark types obtained; ΔJq represents the probability fluctuation value of each obtained intervention benchmark type in the prediction period; JKq corresponds to the preset probability fluctuation threshold for each intervention benchmark type; Kq corresponds to the preset maximum allowable probability fluctuation threshold for each intervention benchmark type; βq represents the preset weight for each intervention benchmark type; and L1 is the predicted intervention coefficient of the intervention combination that best matches all the obtained intervention benchmark types, where best match refers to the intervention combination that contains the largest number of all the intervention benchmark types.

Images