A data-driven device start-stop machine threshold determination method

Abstract

The application discloses a data-driven device start-stop machine threshold determination method in the technical field of signal processing, comprising the following steps: obtaining historical data within a device setting time, iteratively increasing from the minimum value of the historical data by a certain step size, the value of each iteration being a dynamic threshold, calculating the number of crossings of each dynamic threshold and the historical data, finding the iteration number corresponding to the demarcation line by analyzing the distribution change rule of the number of crossings, and accurately obtaining the start-stop machine threshold according to the iteration number meeting the determination condition. The method is completely driven by the data of the device itself, and the distribution characteristics of the historical data can be analyzed to accurately obtain the start-stop machine threshold of the device, without the need for artificial setting participation and wide applicability. The method can assist device operators in daily operation and maintenance work, improve the accuracy and reliability of the operation and maintenance process, and ensure the safe and continuous operation of the device.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of signal processing, and particularly relates to a data-driven device start-stop threshold determination method. BACKGROUND

[0002] In an automated factory, monitoring the start-stop state of a device is an important link in the full life cycle management application of the device. Accurately determining the start-stop state of the device and uploading the start-stop state of the device to a device management system can provide data support for the calculation of key indicators such as the running time of the device, the service life of the device, and the trouble-free interval time, strengthen the fine management of the enterprise on the device, improve the level of device management, and thus reduce the loss caused by unplanned downtime of the device. Taking a rotating device as an example, when the device needs to run at a low speed for a long time due to low load or other reasons, if the monitoring system always incorrectly determines that the device is in a stopped state, it will cause false judgments for the device management personnel and also cause errors in data collection and storage of the monitoring system.

[0003] In recent years, there have been many studies on the start-stop state of a device. For example, the Chinese patent with publication number CN114067297A proposes a device start-stop state recognition method based on an improved Faster RCNN. This method constructs an image data set by taking a picture of the device, constructs a recognition model, and trains a model for recognizing the start-stop state, thereby recognizing the start-stop state of the device. The Chinese patent with publication number CN112577724B proposes a method for determining the start-stop threshold of a dynamic device, a start-stop monitoring method and device. The method aggregates multiple historical acceleration effective values of the dynamic device into two categories, and finally takes the average value of the class centers of the two categories as the start-stop threshold of the dynamic device. The Chinese patent with publication number CN112446618A proposes a switch device state evaluation method and device based on joint research of multiple component indicators. The method inputs multiple component measurement indicators of the internal insulation gas of the switch device to construct a switch device fault feature value vector, thereby outputting the state of the switch device.

[0004] The existing methods or researches have one or more of the following defects and deficiencies:

[0005] 1. The image recognition method needs to additionally increase the number of cameras for shooting, which increases the workload, and the image is easily affected by external factors such as light, resulting in a large error.

[0006] 2. The model training method needs a large amount of data as a basis, and the model recognition is prone to errors, resulting in poor judgment effect.

[0007] 3. The historical data of the device may have different stages, and the clustering analysis method may not effectively separate the stop data and start data, thereby failing to obtain an accurate start-stop threshold.

[0008] 4. Manually set thresholds require experience-based judgment, and different devices have different thresholds. Manual calibration is labor-intensive and highly subjective.

[0009] 5. The operating status of the equipment may be changing. Using a fixed threshold to judge the start-up and shutdown status of the equipment is prone to misjudgment. Summary of the Invention

[0010] The purpose of this invention is to provide a data-driven method for determining device start-up and shutdown thresholds to solve the problems mentioned in the background art.

[0011] The present invention achieves the above objectives through the following technical solutions:

[0012] A data-driven method for determining device start-up and shutdown thresholds includes the following steps:

[0013] S1: Collect historical indicator data within a set time period and construct a historical indicator array. Iterate upwards with a fixed step size based on the minimum value of the indicator data in the historical indicator array, and record the value obtained in each iteration as a dynamic threshold.

[0014] S2: Construct a difference array based on the dynamic threshold and the historical index array, and construct a crossing count array based on the number of elements in the difference array that satisfy the condition that the product of the preceding and following elements is negative;

[0015] S3: Based on the preset start-stop threshold determination conditions, filter the number of crossings that meet the determination conditions in the crossing number array, and calculate the start-stop threshold based on the number of crossings.

[0016] As a further optimization of the present invention, the historical index data of the device are vibration signal data, sound signal data, temperature signal data, voltage signal data, current signal data, pressure signal data, or rotational speed signal data of the device.

[0017] As a further optimization of the present invention, the iterative process in step S1 includes:

[0018] (1) Calculate the value of the previous set proportion in the historical indicator array and use it as the upper limit value; calculate the minimum value in the historical indicator array and use it as the lower limit value.

[0019] (2) Starting from the lower limit value, perform iterative calculations upwards with a fixed step size. After several iterations, reach the upper limit value, and then record the dynamic threshold obtained in each iteration.

[0020] As a further optimization scheme of the present application, the construction of the difference array in step S2 comprises: subtracting the dynamic threshold value obtained at each iteration from the historical index array to obtain a difference value, and constructing a crossing number array based on the number of elements in the difference array that satisfy the product of the front and rear elements being negative.

[0021] As a further optimization scheme of the present application, the start-stop machine threshold value determination condition in step S3 at least comprises:

[0022] (1) The number of crossings near the start-stop machine threshold value should satisfy that a certain number of consecutive elements remain unchanged in value.

[0023] (2) Among all the elements satisfying the determination condition (1), the iteration number corresponding to the element with the smallest value should be selected, so as to ensure that the threshold value is as close to the stop value as possible.

[0024] As a further optimization scheme of the present application, the calculation of the start-stop machine threshold value in step S3 according to the crossing number comprises:

[0025] (1) According to the distribution law of the crossing number satisfying the determination condition, the corresponding iteration number is found out;

[0026] (2) Based on the lower limit value, the iteration calculation is performed with a fixed step size and the corresponding iteration number to obtain the start-stop machine threshold value.

[0027] The present application has the following advantages:

[0028] 1) The present application proposes a data-driven device start-stop machine threshold value determination method, which can trace back the historical data sequence of the device, completely based on the distribution law of the data itself, to calculate the start-stop machine threshold value; can assist the device operator in daily operation and maintenance work, improve the accuracy and reliability of the operation and maintenance process, and ensure the safe and continuous operation of the device.

[0029] 2) The present application first proposes to calculate the dynamic threshold value by iteration, and analyzes the number of crossings of each dynamic threshold value and historical data, which can ensure that the method considers all historical data and improves the accuracy of calculation;

[0030] 3) The threshold value proposed by the present application is self-adaptive to changes in data, which can adapt to devices at different stages, and ensure that the device can dynamically and adaptively adjust the threshold value under the condition of working condition fluctuation, and accurately judge the start-stop state of the device;

[0031] 4) The method proposed by the present application does not require prior knowledge of the device, does not require manual setting of the threshold value or other parameters, and completely calculates and analyzes based on the distribution characteristics of the data itself, which can effectively calculate the start-stop machine threshold value for device data with unknown parameters. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0033] Figure 2 This is a schematic diagram of the historical data sequence of vibration indicators in a specific embodiment of the present invention;

[0034] Figure 3 This is a distribution diagram showing the change in the number of traversals with the number of iterations in a specific embodiment of the present invention. Detailed Implementation

[0035] The present invention will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present invention and should not be construed as limiting the scope of protection of the present invention. Those skilled in the art can make some non-essential improvements and adjustments to the present invention based on the above-described contents.

[0036] Example 1

[0037] like Figure 1 As shown, the present invention provides a data-driven method for determining device start-up and shutdown thresholds, comprising the following steps:

[0038] S1: Collect historical indicator data within a set time period and construct a historical indicator array. Iterate upwards with a fixed step size based on the minimum value of the indicator data in the historical indicator array, and record the value obtained in each iteration as a dynamic threshold.

[0039] S2: Construct a difference array based on dynamic threshold and historical indicator array, and construct a crossing count array based on the number of elements in the difference array whose product of preceding and following elements is negative.

[0040] S3: Based on the preset start-stop threshold determination conditions, filter the number of crossings that meet the determination conditions in the crossing count array, and calculate the start-stop threshold based on the number of crossings.

[0041] In this invention, the historical index data of the equipment includes vibration signal data, sound signal data, temperature signal data, voltage signal data, current signal data, pressure signal data, or rotational speed signal data.

[0042] In this embodiment, the device server collects historical indicator data of the device over a certain period of time and constructs an array as follows: ;

[0043] The iterative process in step S1 includes:

[0044] (1) Calculate the historical indicator array L The top 20% of the values ​​are used as the upper limit, denoted as... l-up The minimum value in the historical indicator array is calculated and used as the lower limit, denoted as...l-down ;

[0045] (2) From the lower limit value l-down Start with a fixed step size step The calculation is performed iteratively upwards, reaching the upper limit value after several iterations. l-up Then, record the dynamic threshold obtained in each iteration. T i .

[0046] The total number of iterations is denoted as . During each iteration, a corresponding dynamic threshold can be obtained. ; i is the number of iterations, .

[0047] The process of constructing the traversal count array in step S2 includes:

[0048] Historical indicator array L Subtract the dynamic threshold obtained in each iteration T i Obtain the difference, and construct a difference array based on the difference. D Record the dynamic threshold at each iteration. T i and historical indicators Group L The number of times the crossing is recorded as cro i Number of crossings cro i The calculation method is as follows:

[0049] (1) Construct a dynamic threshold at each iteration. T i and historical indicator array L difference array D， in ;

[0050] (2) For the array in each iteration process D i Determine if any element in the array satisfies The number of elements in the condition is the number of traversals. cro i .

[0051] (3) When all iterations are completed, record the number of times each traversal is performed. cro i ， Construct an array of crossing times C， .

[0052] In this invention, the start / stop threshold determination conditions in step S3 include at least the following:

[0053] (1) The number of crossings near the start-stop threshold should meet the requirement that a certain number of elements remain unchanged, and the number should be at least greater than cro i the value of the number of iterations, and the number should be at least greater than i / 5, where i is the total number of iterations.

[0054] (2) Among all the elements that meet the determination condition (1), the iteration number corresponding to the element with the smallest value should be selected cro i , so as to ensure that the threshold is as close to the stop value as possible. i_b

[0055] The start-stop threshold in step S3 is calculated according to the number of crossings, including:

[0056] (1) Calculate the corresponding iteration number according to the preset start-stop threshold determination condition;

[0057] (2) Based on the lower limit value, perform iteration calculation with a fixed step size and the corresponding iteration number to obtain the start-stop threshold.

[0058] After finding the iteration number that best meets the above determination condition i_b , the corresponding start-stop threshold is calculated . The specific implementation of the method will be described below in conjunction with an actual processing example.

[0059] 1. The device server collects historical vibration acceleration index data of the device in the previous period of time, constructs an array L , and the historical data distribution is as shown in Figure 2 :

[0060] 2. According to the above step S1, the l_up of the data set is 0.14, l_down the fixed step size step is 0.00029, and the corresponding iteration number is 300 according to the iteration number calculation formula: n .

[0061] 3. According to the above step S2, the corresponding number of crossings is calculated, and an array C is constructed. The variation of the number of crossings with the iteration number is as shown in Figure 3 :

[0062] 4. According to the determination condition required by step S3, the iteration number that meets the requirements of relatively stable number of crossings and as small threshold as possible is found, and the final analysis of the iteration number that meets the requirements is i_b 75.

[0063] 5. According to the start-stop threshold​Tag The calculation formula is: , the corresponding start-stop machine threshold is calculated Tag 0.074.

[0064] From the processing results of the above cases, it can be seen that the method can accurately obtain the start-stop machine threshold of the equipment through the distribution characteristics of the historical data. The calculated 0.074 can accurately separate the stop machine data and the start machine data. The process does not require manual participation. The method has good adaptability for different types of equipment.

[0065] It should be noted that the method mentioned in the present application can be implemented in the host computer software in the form of signal processing, or can be implemented in other forms such as digital chips and hardware circuits to realize the operation process.

[0066] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the present application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application.

Claims

1. A data-driven device start-stop machine threshold determination method, characterized by: The method comprises the following steps: S1: collecting historical index data in a set time of the equipment and constructing a historical index array, performing fixed-step upward iteration based on the minimum value of the index data in the historical index array, and recording the value obtained in each iteration as a dynamic threshold value; S2: constructing a difference value array based on the dynamic threshold value and the historical index array, and constructing a crossing times array based on the number of elements in the difference value array that satisfy the product of the front and rear elements being negative; The construction process of the crossing times array in step S2 comprises: The historical index array is obtained L Subtract the dynamic threshold value obtained at each iteration T i The difference value is obtained, and a difference array is constructed based on the difference value D At each iteration, the dynamic threshold value is recorded T i The number of crossings of the historical index Group L is recorded as cro i The number of crossings cro i is calculated as follows: (1) At each iteration, construct a dynamic threshold T i and an array of historical indicators L D， where ;​ (2) For each iteration process of the array D i , the number of elements in the array elements that meet the condition is the number of traversals cro i ; (3) When all iterations are complete, record the number of crossings for each cro i ， Build the array of crossings C， ; S3: determining a start-stop machine threshold value based on a preset condition, screening the crossing times in the crossing times array that satisfy the determination condition, and calculating the start-stop machine threshold value according to the crossing times; The start-stop machine threshold value determination condition in step S3 at least comprises: (1) the crossing times near the start-stop machine threshold value should satisfy that a certain number of consecutive elements remain unchanged in value; (2) among all elements satisfying the determination condition (1), the iteration number corresponding to the element with the minimum value is selected, so as to ensure that the threshold value is as close to the stop value as possible; The calculation of the start-stop machine threshold value according to the crossing times in step S3 comprises: (1) calculating the corresponding iteration number according to the preset start-stop machine threshold value determination condition, and finding out the corresponding iteration number; (2) performing iteration calculation based on the lower limit value with a fixed step and the corresponding iteration number to obtain the start-stop machine threshold value.

2. The method of claim 1, wherein: The historical index data of the equipment is vibration signal data, sound signal data, temperature signal data, voltage signal data, current signal data, pressure signal data or rotation speed signal data of the equipment.

3. The method of claim 1, wherein: The iteration process in step S1 comprises: S1.1: calculating the values of a preset proportion in the historical index array as an upper limit value, and calculating the minimum value in the historical index array as a lower limit value; S1.2: starting from the lower limit value, performing iteration calculation upward with a fixed step, and after several iterations, reaching the upper limit value, and then recording the dynamic threshold value obtained in each iteration.

4. The method of claim 1, wherein: The construction of the difference value array in step S2 comprises: subtracting the dynamic threshold value obtained in each iteration from the historical index array to obtain a difference value, and constructing a difference value array based on the difference value.

Images