Equipment predictive maintenance signal detection method
By calculating the entropy and median of the signal data of the equipment operation, the limit is determined cycled and abnormal signal data is found, which solves the problem of distinguishing between normal and fault data in predictive maintenance of equipment, reduces detection costs and improves detection efficiency.
Patent Information
- Application Number
- CN202410894881.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-04
- Publication Date
- 2024-11-12
AI Technical Summary
During the predictive maintenance of equipment, it is difficult for the existing technology to effectively distinguish normal operation data from fault data, resulting in strict detection algorithms that judge normal data as faults, resulting in increased costs.
By collecting the device's running signal data, calculate the probability distribution and entropy value of the data amplitude, judge the device status based on different entropy ranges, calculate the median cyclically to determine the limit value, and find out the signal data exceeding the limit value to determine the possible equipment. The source of the problem.
It realizes more reasonable and effective detection of normal data and fault data, reduces detection costs, and improves the detection efficiency of predictive maintenance of equipment.
Smart Images

Figure CN118936885A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to equipment predictive maintenance technology, and in particular to a method for signal detection in an equipment predictive maintenance process. Background Art
[0002] In the process of predictive maintenance of equipment, normal operation often accounts for the vast majority, while failure is only a minority. Therefore, for data processing and analysis, it is crucial to effectively distinguish between normal data and failure data. Common models and algorithms often focus on failure data and try to find potential problems. This may lead to the focus deviating from the actual situation, and tolerate strict detection algorithms to judge normal operation data as failure, which will lead to increased costs for troubleshooting.
[0003] Invention patent CN116432004A discloses an automatic peak-finding method for detecting the operating status of equipment. By collecting signals under the operating status of the tested equipment, calculating the threshold of the sampled data, and performing an upper envelope on the sampled data, a first upper envelope array is formed, from which the upper envelope peak points exceeding the threshold are found, including the coordinate value and amplitude of the peak point, and the coordinate values of the two adjacent first zero-crossing points are used as the boundary. The peak point with the largest amplitude within the boundary is retained, which is the expected peak point. However, a large amount of test data is normal operating data. Detecting extreme values and checking them every time will inevitably lead to a significant increase in costs. A more reasonable detection method is needed to meet the needs of actual scenarios of equipment predictive maintenance. Summary of the invention
[0004] In order to solve the above technical problems, the present invention provides a device predictive maintenance signal detection method, comprising the following steps:
[0005] A device predictive maintenance signal detection method comprises the following steps:
[0006] Step 1: Collect equipment operation signals during equipment predictive maintenance. The signal data form an array S. Calculate the probability distribution p of the data amplitude in the array S. i (i=1, 2, ...n), where n is the number of data amplitudes in array S. The entropy value h of array S is calculated according to the following formula:
[0007]
[0008] Step 2: Calculate the median m of the data amplitude in S, and calculate the entropy values h of the amplitudes below and above m in S according to formula (1) A 、h B , and sum, compare h A 、h B The sum and the size of h:
[0009] If hA +h B ≤h, the test is terminated and the equipment is deemed to be operating normally;
[0010] If h A +h B >h, then continue with step 3;
[0011] Step 3: Calculate the median m of the data in S whose amplitude is greater than or equal to m u1 , according to formula (1), calculate the amplitude of S below m u1 and above m u1 The entropy value h u1A 、h u1B , and sum, compare h A 、h B The sum of u1A 、h u1B The size of the sum:
[0012] If h u1A +h u1B ≤h A +h B , then take the limit value t h is m u1 , and continue with step 5;
[0013] If h u1A +h u1B >h A +h B , then continue with step 4;
[0014] Step 4: Circularly calculate the amplitude in S that is greater than or equal to m uj The median of the data m uj+1 (j is an integer greater than or equal to 1), calculate the amplitude of S below m according to formula (1) uj+1 and above m uj+1 The entropy value h uj+1A 、h uj+1B , and sum, compare h ujA 、h ujB The sum of uj+1A 、h uj+1B The sum of the values is the same as the value of the sum until the condition h is met. uj+1A +h uj+1B ≤h ujA +h ujB , then take the limit value t h is m uj , and continue with step 5;
[0015] Step 5: Find the value t where the amplitude in S exceeds the limit h The signal data S x , using the signal data Sx The characteristics of the equipment determine the possible source of the problem.
[0016] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:
[0017] As a preferred technical solution of the present invention: in step 1, before calculating the probability distribution, the amplitudes are first sorted in order from large to small or from small to large to improve the calculation efficiency.
[0018] As a preferred technical solution of the present invention: in step 1, when the number of different data amplitudes is an odd number, the median is the middle position value in the order from large to small or from small to large;
[0019] When the number of different data amplitudes is even, the median is the average of the two middle position values when they are sorted from large to small or from small to large.
[0020] As a preferred technical solution of the present invention: in step 4, the number of cycles is preferably 3, that is, the maximum value of j is 2, so as to improve the detection efficiency.
[0021] As a preferred technical solution of the present invention: in step 4, there is no upper limit on the number of cycles.
[0022] As a preferred technical solution of the present invention: in step 5, the feature is a time point.
[0023] As a preferred technical solution of the present invention: in step 5, the feature is a frequency value.
[0024] As a preferred technical solution of the present invention: the data in S is multiplied by negative one to change the numerical sign, and the signal data exceeding the limit value th is obtained. x , using -S x The characteristics of the device may determine other possible sources of problems.
[0025] The present invention provides a device predictive maintenance signal detection method, which can more reasonably and effectively detect normal data and fault data and better adapt to the device predictive maintenance scenario. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of the equipment predictive maintenance signal detection method of the present invention;
[0027] Figure 2 This is a graph of the time domain detection results of the sampling data of Example 2;
[0028] Figure 3 This is a graph of the sampling data frequency domain detection result of Example 3;
[0029] Figure 4This is a graph of the time domain detection results of the sampling data of Example 4;
[0030] Figure 5 This is a graph of the time domain detection results of the sampling data of Example 5. DETAILED DESCRIPTION
[0031] The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
[0032] Example 1
[0033] like Figure 1 As shown, a device predictive maintenance signal detection method includes the following steps:
[0034] Step 1: Collect equipment operation signals during equipment predictive maintenance. The signal data form an array S. Calculate the probability distribution p of the data amplitude in the array S. i (i=1, 2, ...n), where n is the number of data amplitudes in array S. The entropy value h of array S is calculated according to the following formula:
[0035]
[0036] Step 2: Calculate the median m of the data amplitude in S, and calculate the entropy values h of the amplitudes below and above m in S according to formula (1) A 、h B , and sum, compare h A 、h B The sum and the size of h:
[0037] If h A +h B ≤h, the test is terminated and the equipment is deemed to be operating normally;
[0038] If h A +h B >h, then continue with step 3;
[0039] Step 3: Calculate the median m of the data in S whose amplitude is greater than or equal to m u1 , according to formula (1), calculate the amplitude of S below m u1 and above m u1 The entropy value h u1A 、h u1B , and sum, compare h A 、h B The sum of u1A 、h u1B The size of the sum:
[0040] If h u1A +h u1B ≤h A +h B, then take the limit value t h is m u1 , and continue with step 5;
[0041] If h u1A +h u1B >h A +h B , then continue with step 4;
[0042] Step 4: Circularly calculate the amplitude in S that is greater than or equal to m uj The median of the data m uj+1 (j is an integer greater than or equal to 1), calculate the amplitude of S below m according to formula (1) uj+1 and above m uj+1 The entropy value h uj+1A 、h uj+1B , and sum, compare h ujA 、h ujB The sum of uj+1A 、h uj+1B The sum of the values is the same as the value of the sum until the condition h is met. uj+1A +h uj+1B ≤h ujA +h ujB , then take the limit value t h is m uj , and continue with step 5;
[0043] Step 5: Find the value t where the amplitude in S exceeds the limit h The signal data S x , using the signal data S x The characteristics of the equipment determine the possible source of the problem.
[0044] Example 2
[0045] Based on the device predictive maintenance signal detection method of Example 1, the collected device operation signal is as follows: Figure 2 shown.
[0046] In step 1, the entropy value of array S is 1.28.
[0047] In step 2, h A +h B The value is 1.09. At this point, the test ends and it is determined that the equipment is operating normally.
[0048] Example 3
[0049] Equipment predictive maintenance signal detection method based on Example 1:
[0050] In step 1, the vibration acceleration signal of the motor equipment is collected, and the signal sampling frequency is 500 Hz. In step 1, the entropy value of array S is 1.11.
[0051] In step 2, h A +h B The value is 1.29, so continue with step 3.
[0052] In step 3, h u1A +h u1B The value is 1.22, then the limit value t h is m u1 Equal to 0.031.
[0053] Continue with step 5 to find the value t at which the amplitude in S exceeds the limit h The signal data S x , using the signal data S x The frequency value of the equipment determines that the possible problem is a bearing failure. The frequency domain detection results of the equipment operation data are as follows: Figure 3 shown.
[0054] Example 4
[0055] Equipment predictive maintenance signal detection method based on Example 1:
[0056] In step 1, the vibration acceleration signal of the gate closing device is collected, and the signal sampling frequency is 800 Hz. In step 1, the entropy value of array S is 1.17.
[0057] In step 2, h A +h B The value is 1.30, so continue with step 3.
[0058] In step 3, h u1A +h u1B A value of 1.33 is greater than 1.30.
[0059] In step 4, loop to calculate the amplitude in S greater than or equal to m uj The median of the data m uj+1 , when j = 4, h u5A +h u5B ≤h u4A +h u4B , then take the limit value t h is m u4 Equal to 0.22.
[0060] Continue with step 5 to find the value t at which the amplitude in S exceeds the limit h The signal data S x , using the signal data S x The frequency value of the device determines that the possible problem is that the gate closing device has an operating impact failure. The time domain detection results of the original data of the device operation are as follows: Figure 4 shown.
[0061] Example 5
[0062] Based on the equipment predictive maintenance signal data of Example 4, the signal is multiplied by negative one and then the equipment predictive maintenance signal detection method of Example 1 is used.
[0063] In step 1, the entropy value of array S in step 1 is 1.17.
[0064] In step 2, h A +h B The value is 1.30, so continue with step 3.
[0065] In step 3, h u1A +h u1B The value of 1.39 is greater than 1.30.
[0066] In step 4, loop to calculate the amplitude in S greater than or equal to m uj The median of the data m uj+1 , when j = 3, h u4A +h u4B ≤h u3A +h u3B , then take the limit value t h is m u3 Equal to 0.18.
[0067] Continue with step 5 to find the value t at which the amplitude in S exceeds the limit h The signal data S x , using the signal data S x The frequency value of the device determines that the possible problem is that the gate closing device has an operating impact failure. The time domain detection results of the original data of the device operation are as follows: Figure 5 shown.
[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for detecting equipment predictive maintenance signals, It is characterized in that The following steps are involved: Step 1: Collect equipment operation signals during equipment predictive maintenance. The signal data form an array S. Calculate the probability distribution p of the data amplitude in the array S. i , i = 1, 2, ... n, where n is the number of data amplitudes in array S, and the entropy value h of array S is calculated according to the following formula: Step 2: Calculate the median m of the data amplitude in S, and calculate the entropy values h of the amplitudes below and above m in S according to formula (1) A 、h B , and sum, compare h A 、h B The sum and the size of h: If h A +h B ≤h, the test is terminated and the equipment is deemed to be operating normally; If h A +h B >h, then continue with step 3; Step 3: Calculate the median m of the data in S whose amplitude is greater than or equal to m u1 , according to formula (1), calculate the amplitude of S below m u1 and above m u1 The entropy value h u1A 、h u1B , and sum, compare h A 、h B The sum of u1A 、h u1B The size of the sum: If h u1A +h u1B ≤h A +h B , then take the limit value t h is m u1 , and continue with step 5; If h u1A +h u1B >h A +h B , then continue with step 4; Step 4: Circularly calculate the amplitude in S that is greater than or equal to m uj The median of the data m uj+1 , j is an integer greater than or equal to 1, and the amplitudes in S below m are calculated according to formula (1). uj+1 and above m uj+1 The entropy value h uj+1A 、h uj+1B , and sum, compare h ujA 、h ujB The sum of uj+1A 、h uj+1B The sum of the values is the same as h until the condition h is met. uj+1A +h uj+1B ≤h ujA +h ujB , then take the limit value t h is m uj , and continue with step 5; Step 5: Find the value t where the amplitude in S exceeds the limit h The signal data S x , using the signal data S x The characteristics of the equipment determine the possible source of the problem.
2. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 1, before calculating the probability distribution, the amplitudes are sorted in order from large to small or from small to large.
3. The equipment predictive maintenance signal detection method according to claim 1 or 2, It is characterized in that In step 1, when the number of different data amplitudes is an odd number, the median is the middle value in the order from large to small or from small to large; When the number of different data amplitudes is even, the median is the average of the two middle position values when they are sorted from large to small or from small to large.
4. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 4, the number of cycles is preferably 3, that is, the maximum value of j is 2, in order to improve the detection efficiency.
5. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 4, there is no upper limit on the number of cycles.
6. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 5, the feature is the time point.
7. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 5, the feature is a frequency value.
8. The equipment predictive maintenance signal detection method according to claim 1, It is characterized in that In step 5, the data in S is multiplied by minus one to change the numerical sign and obtain the value exceeding the limit t h Signal data-S x , using -S x The characteristics of the device may determine other possible sources of problems.
