Device fault detection method and apparatus, and storage medium
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI QIYAO SCREW MACHINERY
- Filing Date
- 2022-04-14
- Publication Date
- 2026-07-07
AI Technical Summary
Current fault diagnosis methods for process screw compressors rely on experience-based judgment, resulting in a high rate of misdiagnosis. They are not suitable for remote fault diagnosis and lack effective intelligent diagnostic algorithms.
Based on the historical statistical data of the target equipment, the prior probability and likelihood probability of the fault cause are calculated, the posterior probability is calculated using Bayes' theorem, the detection order of the fault cause is determined, and the fault cause is judged by combining the time series data of the feature parameters.
It improves the reliability and accuracy of fault diagnosis, supports remote fault detection, and reduces the misdiagnosis rate.
Smart Images
Figure CN114841382B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of process screw compressor technology, and in particular to a method, apparatus and storage medium for equipment fault detection. Background Technology
[0002] Current condition monitoring and fault diagnosis technologies for large-scale generating units are widely used in industries such as petroleum, chemical, and power generation. The equipment involved includes steam generator sets and centrifugal compressors. After practical application, both accident rates and maintenance costs have been significantly reduced. Process screw compressors occupy a critical position in petrochemical production processes; their stability is crucial to the safety of the entire process. A malfunction in a process screw compressor can easily lead to safety accidents, causing significant property damage and even loss of life. Due to the complex structure and harsh operating conditions of process screw compressors, current research on their fault diagnosis algorithms is limited, and mature industrial applications have not yet been developed.
[0003] Currently, fault diagnosis methods for process screw compressors still rely on experience-intensive and unreliable methods such as observing temperature, listening to sound, and observing current fluctuations. This results in a high rate of misdiagnosis and hinders the realization of remote fault diagnosis for process screw compressors. Summary of the Invention
[0004] This invention provides a method, apparatus, and storage medium for equipment fault detection, which can effectively solve the problem that current fault diagnosis methods for process screw compressors still rely on experience-intensive and unreliable methods such as observing temperature, listening to sound, and observing current fluctuations, resulting in a high misdiagnosis rate and hindering remote fault diagnosis of process screw compressors.
[0005] According to one aspect of the present invention, a device fault detection method is provided, the method comprising: acquiring all fault causes corresponding to a target fault associated with a target device; determining a prior probability of each fault cause occurring and a likelihood probability of the target fault occurring under the condition that each fault cause occurs based on historical statistical data of the target device; calculating a posterior probability corresponding to the occurrence of each fault cause when the target fault occurs based on the prior probability and the likelihood probability; and determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause.
[0006] Furthermore, determining the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs based on the historical statistical data of the target device includes: determining the total operating time of the target device based on the historical statistical data; obtaining the fault duration corresponding to each fault cause; and calculating the proportion of the fault duration corresponding to each fault cause in the total operating time, and using the proportion in the total operating time as the prior probability of each fault cause.
[0007] Furthermore, determining the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs based on the historical statistical data of the target device further includes: determining the number of times each fault cause occurs based on the historical statistical data; obtaining the number of times the target fault occurs due to each fault cause; calculating the proportion of the number of times the target fault occurs due to each fault cause to the number of times the fault cause occurs, and using the proportion as the likelihood probability of the target fault occurring under the condition that each fault cause occurs.
[0008] Further, the step of calculating the posterior probability corresponding to each fault cause when the target fault occurs based on the prior probability and the likelihood probability includes: for each fault cause B i The cause of the fault, B, is calculated according to the following formula. i The corresponding posterior probability:
[0009] Where n is the total number of causes of failure, P(B i A) is the cause of the malfunction. B) i For the posterior probability of target fault A, P(A|B) i The cause of the malfunction is B. i For the likelihood probability of target fault A, P(B) i The cause of the malfunction is B. i The corresponding prior probability.
[0010] Further, determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause includes: obtaining the posterior probability corresponding to the occurrence of each fault cause; arranging the posterior probabilities from largest to smallest, and sequentially sorting the fault causes according to the order of their corresponding posterior probabilities to form the detection order of all fault causes.
[0011] Furthermore, the following steps are taken: First, a fault-free operating condition is obtained; second, the trend components of multiple characteristic parameters of the target device corresponding to the fault-free operating condition are obtained within a preset range; third, at least one characteristic parameter associated with each fault cause is determined; fourth, the trend components of the time-series data of the monitored multiple characteristic parameters are obtained during the period in which the target fault occurs; fifth, during the period in which the target fault occurs, the single fault-free operating condition that the target device should be in is determined.
[0012] During the period in which the target fault occurs, determine whether the trend component of the time-series data of at least one characteristic parameter associated with the target fault falls within a preset range of the trend component corresponding to the fault-free operating condition.
[0013] Further, if the value of the trend component of the time series data of the at least one feature parameter does not fall within the preset range corresponding to the fault-free operating condition during the period in which the target fault occurs, the cause of the target fault is confirmed to be a fault cause associated with the at least one feature parameter; wherein the value not falling within the preset range of the trend component corresponding to the fault-free operating condition includes: the trend component of the time series data of the at least one feature parameter is continuously greater than the preset range of the trend component within a first preset time period; or
[0014] The trend component of the time series data of at least one characteristic parameter exceeds the preset range of the trend component more than a preset number of times within a second preset time period.
[0015] Furthermore, the method also includes: periodically updating the prior probability of each fault cause occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the latest statistical data.
[0016] According to another aspect of the present invention, a device fault detection apparatus is provided, comprising: a fault cause acquisition unit, configured to acquire all fault causes corresponding to a target fault associated with a target device; a first calculation unit, configured to determine the prior probability of each fault cause occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on historical statistical data of the target device; a second calculation unit, configured to calculate the posterior probability corresponding to the occurrence of each fault cause when the target fault occurs, based on the prior probability and the likelihood probability; and a fault detection unit, configured to determine the detection order of all fault causes when the target fault occurs, based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause.
[0017] According to another aspect of the present invention, a storage medium is provided, wherein a plurality of instructions are stored therein, the instructions being adapted to be loaded by a processor to execute the device fault detection method according to any embodiment of the present invention.
[0018] The advantages of this invention are twofold: First, it determines the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs based on historical statistical data of the target device. Then, it calculates the posterior probability corresponding to each fault cause when the target fault occurs based on the prior probability and the likelihood probability. Finally, it determines the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to each fault cause. Second, it monitors whether the time-series data of the monitoring feature parameters conforms to a preset range under the current operating conditions, thereby improving the reliability of subsequent fault causes. Attached Figure Description
[0019] The technical solution and other beneficial effects of the present invention will become apparent from the following detailed description of specific embodiments of the invention, in conjunction with the accompanying drawings.
[0020] Figure 1 The flowchart illustrates the steps of a device fault detection method provided in Embodiment 1 of the present invention.
[0021] Figure 2 The flowchart of the sub-steps of step S120 provided in the embodiment of the present invention.
[0022] Figure 3 A flowchart of another sub-step of step S120 provided in an embodiment of the present invention.
[0023] Figure 4 This is a flowchart of the steps of a device fault detection method provided in Embodiment 2 of the present invention.
[0024] Figure 5 This is a schematic diagram of a device for detecting equipment faults according to Embodiment 3 of the present invention.
[0025] Figure 6 This is a schematic diagram of target fault detection provided in an embodiment of the present invention. Detailed Implementation
[0026] 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 a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0027] Now refer to Figure 1 , Figure 1 A device fault detection method provided in Embodiment 1 of the present invention. The method includes:
[0028] Step S110: Obtain all fault causes corresponding to the target fault associated with the target device.
[0029] For example, the target device can be a screw compressor. The principle analysis and fault mechanism analysis of the screw compressor are performed, whereby the target fault is a high discharge temperature fault in the screw compressor, including all fault causes and the formation mechanism of the high discharge temperature fault. (See also...) Figure 6 All the causes of the failure can be understood as follows: Let event A be "high discharge temperature of the screw compressor", and event B be... i (i = 1, 2, 3…N) represents all possible causes, B1, …, B N To complete the event group.
[0030] Step S120: Determine the prior probability of each fault cause occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the historical statistical data of the target device.
[0031] For example, the operating data of a screw compressor generally includes operating data and fault data of thousands of screw compressors over many years. Of course, in some embodiments, the operating data of a screw compressor may also be the current operating data and fault data of a screw compressor.
[0032] In some embodiments, in conjunction with reference Figure 2 Step S120 specifically includes the following steps:
[0033] Step S210: Determine the total operating time of the target device based on the historical statistical data.
[0034] Step S220: Obtain the fault duration corresponding to each of the aforementioned fault causes.
[0035] Step S230: Calculate the proportion of the fault duration corresponding to each fault cause in the total running time, and use the proportion in the total running time as the prior probability of each fault cause occurring.
[0036] For example, the times when the characteristic parameters of a certain on-site unit are lower than the inlet steam condensation point, lower than the alarm value, and higher than the alarm value within a certain period are t1, t2, and t3, respectively. That is, the fault duration corresponding to each fault cause is also t1, t2, and t3, respectively. The total operating time of the unit is T, that is, the total operating time of the target equipment is determined based on the historical statistical data. Then, P(B2) = t2 / T, and so on, P(B1) and P(B3) can be obtained.
[0037] In some embodiments, in conjunction with reference Figure 3 Step S120 further includes the following steps:
[0038] Step S310: Determine the number of times each of the aforementioned fault causes occurs based on the historical statistical data.
[0039] Step S320: Obtain the number of times the target fault occurs due to each of the aforementioned fault causes.
[0040] Step S330: Calculate the proportion of the number of times the target fault occurs caused by each fault cause to the number of times each fault cause occurs, and use the proportion of the number of times the fault cause occurs as the likelihood probability of the target fault occurring under the condition that each fault cause occurs.
[0041] For example, the frequency of occurrence of B1, B2, and B3 is counted from the historical statistical data, and then the high discharge temperature fault records of the screw compressor caused by the occurrence of B1, B2, and B3 are counted respectively to obtain P(A|B i For example, if the number of times the internal and external pressure ratio of a certain field unit increases over a period of time is M, and the number of times the increase in external pressure ratio causes exhaust temperature failure is m, then P(A|B2)=m / M.
[0042] Step S130: Calculate the posterior probability corresponding to each cause of the target fault when the target fault occurs, based on the prior probability and the likelihood probability.
[0043] For example, according to the formula Where P(C) is the posterior probability, P(B) is the likelihood probability, and P(A) is the prior probability. P(C) = P(B) i |A),
[0044] P(B) = P(A|B) i ),P(A)=P(B i That is, for each cause B of the failure. i The cause of the fault, B, is calculated according to the following formula. i The corresponding posterior probability:
[0045]
[0046] Where n is the total number of causes of failure, P(B i A) is the cause of the malfunction. B) i For the posterior probability of target fault A, P(A|B) i The cause of the malfunction is B. i For the likelihood probability of target fault A, P(B) i The cause of the malfunction is B. i The corresponding prior probability.
[0047] Step S140: Determine the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause.
[0048] In some embodiments, determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause includes: obtaining the posterior probability corresponding to the occurrence of each fault cause, arranging the posterior probabilities from largest to smallest, and sequentially sorting the fault causes according to the order of their corresponding posterior probabilities to form the detection order of all fault causes.
[0049] Example 1 determines the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs based on the historical statistical data of the target device. It calculates the posterior probability corresponding to each fault cause when the target fault occurs based on the prior probability and the likelihood probability. It determines the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to each fault cause.
[0050] Figure 4 This invention provides a device fault detection method according to Embodiment 2 of the present invention. The method includes:
[0051] Step S410: Obtain all fault causes corresponding to the target fault associated with the target device.
[0052] For example, the target device can be a screw compressor. The principle analysis and fault mechanism analysis of the screw compressor are performed, whereby the target fault is a high discharge temperature fault in the screw compressor, including all fault causes and the formation mechanism of the high discharge temperature fault. (See also...) Figure 6 All the causes of the failure can be understood as follows: Let event A be "high discharge temperature of the screw compressor", and event B be... i (i = 1, 2, 3…N) represents all possible causes, B1, …, B N To complete the event group.
[0053] Step S420: Determine the prior probability of each fault cause occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the historical statistical data of the target device.
[0054] For example, the operating data of a screw compressor generally includes operating data and fault data of thousands of screw compressors over many years. Of course, in some embodiments, the operating data of a screw compressor may also be the current operating data and fault data of a screw compressor.
[0055] In some embodiments, in conjunction with reference Figure 2 Step S120 specifically includes the following steps:
[0056] Step S210: Determine the total operating time of the target device based on the historical statistical data.
[0057] Step S220: Obtain the fault duration corresponding to each of the aforementioned fault causes.
[0058] Step S230: Calculate the proportion of the fault duration corresponding to each fault cause in the total running time, and use the proportion in the total running time as the prior probability of each fault cause occurring.
[0059] For example, the times when the characteristic parameters of a certain on-site unit are lower than the inlet steam condensation point, lower than the alarm value, and higher than the alarm value within a certain period are t1, t2, and t3, respectively. That is, the fault duration corresponding to each fault cause is also t1, t2, and t3, respectively. The total operating time of the unit is T, that is, the total operating time of the target equipment is determined based on the historical statistical data. Then, P(B2) = t2 / T, and so on, P(B1) and P(B3) can be obtained.
[0060] In some embodiments, in conjunction with reference Figure 3 Step S420 further includes the following steps:
[0061] Step S310: Determine the number of times each of the aforementioned fault causes occurs based on the historical statistical data.
[0062] Step S320: Obtain the number of times the target fault occurs due to each of the aforementioned fault causes.
[0063] Step S330: Calculate the proportion of the number of times the target fault occurs caused by each fault cause to the number of times each fault cause occurs, and use the proportion of the number of times the fault cause occurs as the likelihood probability of the target fault occurring under the condition that each fault cause occurs.
[0064] For example, the frequency of occurrence of B1, B2, and B3 is counted from the historical statistical data, and then the high discharge temperature fault records of the screw compressor caused by the occurrence of B1, B2, and B3 are counted respectively to obtain P(A|B i For example, if the number of times the internal and external pressure ratio of a certain field unit increases over a period of time is M, and the number of times the increase in external pressure ratio causes exhaust temperature failure is m, then P(A|B2)=m / M.
[0065] Step S430: Calculate the posterior probability corresponding to each cause of the target fault when the target fault occurs, based on the prior probability and the likelihood probability.
[0066] For example, according to the formula Where P(C) is the posterior probability, P(B) is the likelihood probability, and P(A) is the prior probability. P(C) = P(B) i |A),
[0067] P(B) = P(A|B) i ),P(A)=P(B i That is, for each cause B of the failure. i The cause of the fault, B, is calculated according to the following formula. i The corresponding posterior probability:
[0068]
[0069] Where n is the total number of causes of failure, P(B i A) is the cause of the malfunction. B) i For the posterior probability of target fault A, P(A|B) i The cause of the malfunction is B. i For the likelihood probability of target fault A, P(B) i The cause of the malfunction is B. i The corresponding prior probability.
[0070] Step S440: Determine the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause.
[0071] In some embodiments, determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause includes: obtaining the posterior probability corresponding to the occurrence of each fault cause, arranging the posterior probabilities from largest to smallest, and sequentially sorting the fault causes according to the order of their corresponding posterior probabilities to form the detection order of all fault causes.
[0072] Step S450: Obtain the preset range of trend components of multiple characteristic parameters of the target device corresponding to the fault-free operating condition.
[0073] Step S460: Determine at least one characteristic parameter associated with each of the aforementioned fault causes.
[0074] Step S470: Acquire the trend components of the time series data of multiple monitored characteristic parameters during the period in which the target fault occurs.
[0075] Step S480: During the period when the target fault occurs, determine whether the trend component of the time series data of at least one characteristic parameter associated with the target fault falls within the preset range of the trend component corresponding to the fault-free operating condition.
[0076] Step S490: If the value of the trend component of the time series data of the at least one feature parameter does not fall within the preset range corresponding to the fault-free condition during the period when the target fault occurs, the cause of the target fault is confirmed to be a fault cause associated with the at least one feature parameter.
[0077] In some embodiments, the step of obtaining a fault-free operating condition may be included before step S450.
[0078] In some embodiments, during the period in which the target fault occurs, it is also necessary to determine the single fault-free operating condition of the target device to determine the operating condition of the target device during the period in which the target fault occurs.
[0079] In some embodiments, the preset range of the trend component that does not fall within the fault-free operating condition includes:
[0080] The trend component of the time series data of at least one characteristic parameter is continuously greater than a preset range of trend components within a first preset time period; or
[0081] The trend component of the time series data of at least one characteristic parameter exceeds the preset range of the trend component more than a preset number of times within a second preset time period.
[0082] Step S500: Periodically update the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the latest statistical data.
[0083] like Figure 6 As shown, taking "intake air temperature" as an example, trend analysis is performed. The time-series data of intake air temperature is decomposed using STL to obtain periodic components, trend components, and residual components. If the change in the trend component continuously exceeds the limit value, i.e., it does not fall within the preset range corresponding to the fault-free operating condition (to determine whether it falls within the preset range, which can be taken as 10% of the rated intake air temperature value, and if it repeatedly crosses the preset range, then the change in "intake air temperature" is considered a possible cause of event A; otherwise, it is considered an impossible cause). This method is tested under fault conditions to verify the fault detection rate. If the fault detection rate meets the requirements, an initial model for high exhaust temperature faults is obtained. If it does not meet the requirements, the threshold of the clustering algorithm and the STL decomposition algorithm need to be optimized.
[0084] In Example 2, on the one hand, based on historical statistical data of the target device, the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs are determined. The posterior probability corresponding to each fault cause when the target fault occurs is calculated based on the prior probability and the likelihood probability. The detection order of all fault causes when the target fault occurs is determined based on the magnitude of the posterior probability corresponding to each fault cause. On the other hand, the time-series data of monitoring feature parameters is judged to see if it conforms to a preset range under the current operating conditions, thereby improving the reliability of subsequent fault causes.
[0085] Figure 5 This is a device fault detection apparatus provided in Embodiment 3 of the present invention. The apparatus includes: a fault cause acquisition unit 100, a first calculation unit 200, a second calculation unit 300, and a fault detection unit 400.
[0086] The fault cause acquisition unit is used to acquire all fault causes corresponding to the target fault associated with the target device. For example, the target device can be a screw compressor. The unit performs principle analysis and fault mechanism analysis on the screw compressor, where the target fault is all fault causes corresponding to the high exhaust temperature fault of the screw compressor, the formation mechanism of the high exhaust temperature fault, etc. (See also...) Figure 6 All the causes of the failure can be understood as follows: Let event A be "high discharge temperature of the screw compressor", and event B be... i (i = 1, 2, 3…N) represents all possible causes, B1, …, B N To complete the event group.
[0087] The first calculation unit is used to determine the prior probability of each fault cause occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the historical statistical data of the target device. For example, the operating data of the screw compressor generally includes operating data and fault data of thousands of screw compressors over many years. Of course, in some embodiments, the operating data of the screw compressor can also be the current operating data and fault data of the screw compressor.
[0088] In some embodiments, the first calculation unit performs the following steps: determining the total operating time of the screw compressor based on the operating data of the screw compressor; obtaining the fault duration corresponding to each fault cause; calculating the proportion of the fault duration corresponding to each fault cause in the total operating time, and using the proportion in the total operating time as the prior probability of each fault cause occurring.
[0089] For example, the times when the characteristic parameters of a certain on-site unit are lower than the inlet steam condensation point, lower than the alarm value, and higher than the alarm value within a certain period are t1, t2, and t3, respectively. That is, the fault duration corresponding to each fault cause is also t1, t2, and t3, respectively. The total operating time of the unit is T, that is, the total operating time of the target equipment is determined based on the historical statistical data. Then, P(B2) = t2 / T, and so on, P(B1) and P(B3) can be obtained.
[0090] For example, the frequency of occurrence of B1, B2, and B3 is counted from the historical statistical data, and then the high discharge temperature fault records of the screw compressor caused by the occurrence of B1, B2, and B3 are counted respectively to obtain P(A|B i For example, if the number of times the internal and external pressure ratio of a certain field unit increases over a period of time is M, and the number of times the increase in external pressure ratio causes exhaust temperature failure is m, then P(A|B2)=m / M.
[0091] In some embodiments, the first computing unit further performs the following step: determining the number of times each of the fault causes occurs based on the operating data of the screw compressor.
[0092] Obtain the number of times the target fault occurs due to each of the aforementioned fault causes. Calculate the proportion of the number of times the target fault occurs due to each of the aforementioned fault causes relative to the total number of times the fault cause occurs, and use this proportion as the likelihood probability of the target fault occurring given that each fault cause occurs.
[0093] For example, the frequency of occurrence of B1, B2, and B3 is counted from the historical statistical data, and then the high discharge temperature fault records of the screw compressor caused by the occurrence of B1, B2, and B3 are counted respectively to obtain P(A|B i For example, if the number of times the internal and external pressure ratio of a certain field unit increases over a period of time is M, and the number of times the increase in external pressure ratio causes exhaust temperature failure is m, then P(A|B2)=m / M.
[0094] The second calculation unit is used to calculate the posterior probability corresponding to each cause of the target fault when the target fault occurs, based on the prior probability and the likelihood probability. For example, according to the formula... Where P(C) is the posterior probability, P(B) is the likelihood probability, and P(A) is the prior probability. P(C) = P(B) i |A), P(B) = P(A|B) i ),P(A)=P(B i That is, for each cause B of the failure. i The cause of the fault, B, is calculated according to the following formula.i The corresponding posterior probability:
[0095]
[0096] Where n is the total number of causes of failure, P(B i A) is the cause of the malfunction. B) i For the posterior probability of target fault A, P(A|B) i The cause of the malfunction is B. i For the likelihood probability of target fault A, P(B) i The cause of the malfunction is B. i The corresponding prior probability.
[0097] The fault detection unit is used to determine the detection order of all fault causes when the target fault occurs, based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause.
[0098] In some embodiments, determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause includes: obtaining the posterior probability corresponding to the occurrence of each fault cause, arranging the posterior probabilities from largest to smallest, and sequentially sorting the fault causes according to the order of their corresponding posterior probabilities to form the detection order of all fault causes.
[0099] Example 3: Based on the historical statistical data of the target device, the prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs are determined. The posterior probability corresponding to each fault cause when the target fault occurs is calculated according to the prior probability and the likelihood probability. The detection order of all fault causes when the target fault occurs is determined according to the magnitude of the posterior probability corresponding to each fault cause.
[0100] The present invention also provides a storage medium storing a plurality of instructions adapted for loading by a processor to execute the device fault detection method according to any embodiment of the present invention.
[0101] In summary, although the present invention has been disclosed above with reference to preferred embodiments, the above preferred embodiments are not intended to limit the present invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the scope defined in the claims.
Claims
1. A method for detecting equipment faults, characterized in that, include: Obtain all fault causes associated with the target fault of the target device; Based on the historical statistical data of the target device, determine the prior probability of each of the fault causes occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs; Determining the prior probability includes: determining the total operating time of the target device based on the historical statistical data, and using the proportion of the fault cause in the total operating time as the prior probability of each fault cause occurring; determining the likelihood probability includes: determining the number of times each fault cause occurs based on the historical statistical data, and using the proportion of the number of times the fault cause occurs as the likelihood probability of the target fault occurring under the condition that each fault cause occurs. Calculate the posterior probability corresponding to each cause of the target fault when the target fault occurs based on the prior probability and the likelihood probability; and The detection order of all fault causes when the target fault occurs is determined based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause. The process involves: acquiring a preset range of trend components for multiple characteristic parameters of the target device corresponding to a fault-free operating condition; determining at least one characteristic parameter associated with each fault cause; acquiring the trend components of time-series data of the monitored multiple characteristic parameters during the period in which the target fault occurs; and determining whether the trend component of the time-series data of at least one characteristic parameter associated with the target fault falls within the preset range of the trend components corresponding to the fault-free operating condition during the period in which the target fault occurs. The trend components are obtained by decomposing the time-series data of the characteristic parameters using STL. If the trend component of the time series data of at least one of the characteristic parameters does not fall within the preset range corresponding to the fault-free condition during the period in which the target fault occurs, the cause of the target fault is confirmed to be a fault cause associated with the at least one of the characteristic parameters. The trend component preset range that does not fall within the fault-free operating condition includes: the trend component of the time series data of at least one of the feature parameters is continuously greater than the trend component preset range within a first preset time period; or the trend component of the time series data of at least one of the feature parameters is greater than the trend component preset range more than a preset number of times within a second preset time period.
2. The equipment fault detection method according to claim 1, characterized in that, Determining the percentage of the total operating time caused by the fault includes: Obtain the fault duration corresponding to each of the aforementioned fault causes; and Calculate the proportion of the fault duration corresponding to each fault cause in the total operating time, and use the proportion in the total operating time as the prior probability of each fault cause occurring.
3. The equipment fault detection method according to claim 1, characterized in that, Determining the percentage of occurrences of the aforementioned cause of failure includes: Obtain the number of times the target fault occurs due to each of the aforementioned fault causes; Calculate the percentage of the number of times the target fault occurs due to each fault cause out of the total number of occurrences of each fault cause.
4. The equipment fault detection method according to claim 1, characterized in that, The step of calculating the posterior probability corresponding to each cause of the target fault when the target fault occurs based on the prior probability and the likelihood probability includes: For each cause of failure The cause of the fault can be calculated using the following formula. The corresponding posterior probability: P(B i |A)= ; Where n is the total number of causes of failure, P(B i A) is the cause of the fault. For the posterior probability of target fault A, Cause of the malfunction For the likelihood probability of target fault A, Cause of the malfunction The corresponding prior probability.
5. The equipment fault detection method according to claim 1, characterized in that, The step of determining the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause includes: Obtain the posterior probability of each of the aforementioned fault causes; The posterior probabilities are arranged from largest to smallest, and the causes of failure are sequentially sorted according to the order of their corresponding posterior probabilities to form the detection order of all causes of failure.
6. The equipment fault detection method according to any one of claims 1-5, characterized in that, The method further includes: The prior probability of each fault cause and the likelihood probability of the target fault occurring under the condition that each fault cause occurs are periodically updated based on the latest statistical data.
7. A device for detecting equipment faults, characterized in that, include: The fault cause acquisition unit is used to acquire all fault causes corresponding to the target fault associated with the target device. The first calculation unit is used to determine the prior probability of each of the fault causes occurring and the likelihood probability of the target fault occurring under the condition that each fault cause occurs, based on the historical statistical data of the target device. Determining the prior probability includes: determining the total operating time of the target device based on the historical statistical data, and using the proportion of the fault cause in the total operating time as the prior probability of each fault cause occurring; determining the likelihood probability includes: determining the number of times each fault cause occurs based on the historical statistical data, and using the proportion of the number of times the fault cause occurs as the likelihood probability of the target fault occurring under the condition that each fault cause occurs. The second calculation unit is configured to calculate, based on the prior probability and the likelihood probability, the posterior probability corresponding to the occurrence of each fault cause when the target fault occurs; and A fault detection unit is configured to: determine the detection order of all fault causes when the target fault occurs based on the magnitude of the posterior probability corresponding to the occurrence of each fault cause; acquire a preset range of trend components of multiple characteristic parameters of the target device corresponding to a fault-free operating condition; determine at least one characteristic parameter associated with each fault cause; acquire the trend components of the time-series data of the monitored multiple characteristic parameters during the period when the target fault occurs; and determine whether the trend component of the time-series data of at least one characteristic parameter associated with the target fault falls within the preset range of the trend component corresponding to the fault-free operating condition during the period when the target fault occurs, wherein the trend component is obtained by decomposing the time-series data of the characteristic parameter using STL. If the trend component of the time series data of at least one of the characteristic parameters does not fall within the preset range corresponding to the fault-free condition during the period in which the target fault occurs, the cause of the target fault is confirmed to be a fault cause associated with the at least one of the characteristic parameters. The trend component preset range that does not fall within the fault-free operating condition includes: the trend component of the time series data of at least one of the feature parameters is continuously greater than the trend component preset range within a first preset time period; or the trend component of the time series data of at least one of the feature parameters is greater than the trend component preset range more than a preset number of times within a second preset time period.
8. A storage medium, characterized in that, The storage medium stores a plurality of instructions, which are adapted to be loaded by a processor to execute the device fault detection method according to any one of claims 1 to 6.