A real-time detection method for insufficient pressure of high-pressure mixed gas
By collecting and filtering laser cutting current and air pressure data in real time, combined with the double-layer traversal judgment of density clustering algorithm, the accuracy of detection of air pressure in high-pressure mixed gas is solved, real-time detection and alarm in high-pressure environments are realized, and the improvement of real-time detection and alarm in high-pressure environments is achieved. Detection accuracy and cost-effectiveness.
Patent Information
- Application Number
- CN202411364288.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-29
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2044-09-29
AI Technical Summary
In high-pressure environments, it is difficult for the prior art to accurately detect the insufficient air pressure of the high-pressure mixed gas, resulting in a degradation of equipment performance and a decrease in product yield, and may even cause safety accidents.
By collecting laser cutting current and air pressure data in real time, data filtering and fitting are performed, and double-layer traversal judgment is performed based on density clustering algorithm to achieve real-time detection and alarm of air pressure deficiency.
It improves detection accuracy, reduces environmental interference, reduces detection costs, and realizes accurate detection and alarm for insufficient air pressure of high-pressure mixed gas.
Smart Images
Figure CN118885837B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent welding and cutting, and in particular relates to a real-time detection method for insufficient gas pressure of high-pressure mixed gas. Background Art
[0002] In the existing laser cutting process, the precise ratio of high-pressure mixed gas and the stable air pressure are the key factors to ensure the operation of the cutting process and the quality and stability of the cut products. Especially in a high-pressure environment, insufficient mixed gas pressure may lead to decreased equipment performance, reduced product yield, and even cause safety accidents. Therefore, it is particularly important to detect insufficient mixed gas pressure during the cutting process.
[0003] At present, the market mainly uses detection methods such as direct detection by pressure sensor and gas flow reverse estimation to detect the pressure of high-pressure mixed gas. The above methods are easily interfered in complex high-pressure environments, affecting the detection accuracy. Summary of the invention
[0004] Purpose of the invention: In view of the problems existing in the above-mentioned background technology, the present invention provides a real-time detection method for insufficient pressure of high-pressure mixed gas, which collects current and pressure data in real time and performs data filtering. Density clustering is performed based on the pressure peak set, and double-layer traversal is performed based on the clustering results to determine whether to issue an insufficient pressure alarm.
[0005] Technical solution: A real-time detection method for insufficient pressure of high-pressure mixed gas, comprising the following steps:
[0006] Step S1, collecting and filtering the laser cutting current and air pressure time series data at the inlet in real time, extracting the peak value sequence based on the filtered air pressure time series data; judging whether to perform function relationship fitting based on the peak value sequence;
[0007] Step S2, fitting a linear function relationship between the peak value and time, and judging whether to perform insufficient air pressure detection based on the slope of the linear function;
[0008] Step S3, start the air pressure shortage detection, cluster the peak value sequence based on the density clustering algorithm and analyze the clustering results, and judge whether there is air pressure shortage based on the double-layer traversal. When there is air pressure shortage, start the alarm.
[0009] Furthermore, the specific method of extracting the peak value sequence in step S1 includes:
[0010] Set a collection current_list for real-time storage of current data and a collection inpressure_list for real-time storage of air pressure data; set a current sliding window with a window length of ws; when ws current points are received in current_list, filter all current points that are less than a preset reference current m, and filter the air pressure points at the corresponding time sequence position in inpressure_list; based on the filtered inpressure_list, extract all air pressure peak values in the collection to generate a peak value set inpressure_peak_list.
[0011] Furthermore, in step S2, when the number of peak values in inpressure_peak_list is greater than ws / 10, and the absolute values of the first-order difference values of adjacent peak values are all less than the preset threshold th1, a linear function relationship between the peak value sequence and time is fitted, and the slope k of the linear function relationship is obtained; when the slope k is less than the preset threshold k_th, further air pressure deficiency detection is performed.
[0012] Furthermore, the specific method for detecting insufficient air pressure in step S3 includes:
[0013] Step S3.1, when the inpressure_list filtered in step S1 satisfies that the difference between the maximum value of the first th2 air pressure values and the maximum value of the last th2 air pressure values is less than th3, density clustering is performed on the peak values in the corresponding inpressure_peak_list, otherwise, the insufficient air pressure detection is no longer performed;
[0014] Step S3.2: After completing the density clustering, obtain n labels, where each label includes several peak values, and each peak value corresponds to a position index; perform the following processing on the clustering results:
[0015] First, extract all labels with the maximum continuous length greater than 2 in the density clustering results, and store the labels in the label_filter set in ascending order; when the number of labels in label_filter is less than 2, directly turn on the low pressure alarm, otherwise execute step S3.3;
[0016] Step S3.3, set the counter count, the initial value is 0, and count is performed based on a double-layer traversal; the first layer traverses the label value in label_filter, and the second layer traverses the corresponding position index and label value in inpressure_peak_list; when the labels of the first and second layers are the same, count is increased by 1, otherwise count is set to 0; when count is greater than 10, record the position index A and the corresponding label B in the second layer at this time; traverse the subsequent position indexes of the second layer based on the recorded position index A, and when there is a label greater than B in the label value corresponding to the subsequent position index, no insufficient air pressure detection is performed, otherwise the insufficient air pressure alarm is turned on.
[0017] Compared with the prior art solutions, the technical solution adopted by the present invention has the following beneficial effects:
[0018] The present invention collects high-pressure mixed gas output current and inlet pressure data in real time, filters the data, and determines whether to perform insufficient air pressure detection based on the fitted function relationship. The present invention also designs an insufficient air pressure detection method based on density clustering, performs a double-layer traversal based on the clustering results, and performs insufficient air pressure detection based on the traversal results. The method provided by the present invention will not be affected by the on-site environment, and relies on an accurate data analysis model to achieve the effect of real-time detection and alarm. Compared with existing detection methods, it is simpler and has lower use costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 This is a schematic diagram of the principle of the real-time detection method for insufficient pressure of high-pressure mixed gas provided by the present invention. DETAILED DESCRIPTION
[0020] The present invention provides a real-time detection method for insufficient air pressure of high-pressure mixed gas, which performs data preprocessing based on real-time collected current and inlet pressure data, filters pressure data in a non-cutting state and extracts the peak value of the inlet pressure data, fits the relationship between the peak value and time and makes a judgment whether to enter the insufficient air pressure detection. The present invention provides a method for insufficient air pressure detection based on density clustering, which performs density clustering and analysis and judgment on the peak value set, and issues a corresponding alarm when the insufficient air pressure condition is met.
[0021] The present invention will be further explained below in conjunction with the accompanying drawings.
[0022] like Figure 1 As shown, the present invention provides a real-time detection method for insufficient pressure of high-pressure mixed gas, comprising the following steps:
[0023] Step S1, collecting and filtering the time series data set of the laser cutting current and the air pressure at the inlet in real time, and extracting the peak value sequence based on the filtered air pressure data.
[0024] In this embodiment, the current data and the air pressure data at the inlet that best reflect the cutting state and air pressure state during the laser cutting process are first collected, and the two correspond to each other in time series. Preprocessing is performed based on the above time series data, specifically including:
[0025] Set a set current_list for real-time storage of current data and a set inpressure_list for real-time storage of inlet air pressure data; set a current sliding window with a window length of ws. When ws current points are received in current_list, filter all current points that are less than the preset reference current m, and filter the air pressure points at the corresponding time sequence position of inpressure_list. Based on the filtered inpressure_list, extract all the air pressure peak values in the set to generate a peak value set inpressure_peak_list.
[0026] Step S2: When there are multiple cutting events of the same process in inpressure_peak_list, fit the linear function relationship between the peak value and time, and obtain the slope k; and determine whether to perform insufficient pressure detection in combination with the slope judgment standard. Specifically,
[0027] When the number of peak values in inpressure_peak_list is greater than ws / 10, and the absolute values of the first-order difference values of adjacent peak values are all less than the preset threshold value th1, it means that there are multiple cutting events of the same process in the sliding window. If the air pressure drops at this time, other factors can be excluded. At this time, the linear function relationship between the peak value and time is fitted. Based on the slope k, it can be seen that the change law of air pressure with the passage of events. When the slope k is less than the preset threshold value k_th, it means that as time goes by, the air pressure drops more and more, and insufficient air pressure may occur. At this time, further judgment is made. In this embodiment, k_th<0.
[0028] Step S3: After the slope judgment standard in step S2 is met, the insufficient air pressure detection is performed. Specifically,
[0029] Step S3.1, when the inpressure_list filtered in step S1 satisfies that the difference between the maximum value of the first th2 pressure values and the maximum value of the last th2 pressure values is less than th3, it means that the pressure value in the set does not fluctuate much and there is no process switching. At this time, the corresponding peak values in the inpressure_peak_list are density clustered, otherwise the insufficient pressure detection is no longer performed. In this embodiment, the peak values are clustered by the density clustering method to obtain the process identification result. Different clustering results represent different processes. Based on the pressure difference between different processes, a refined judgment is made on whether to perform insufficient pressure detection.
[0030] In this embodiment, the core parameters of the density clustering algorithm are set to eps=0.5 and min_samples=10, and the obtained clustering results include n labels, each label includes several peak values, and each peak value has a corresponding position index.
[0031] Step S3.2: After completing the density clustering, perform clustering result processing. First, extract all labels with the maximum continuous length greater than 2 in the density clustering result, and store the labels in the label_filter set in ascending order. In this embodiment, the labels are arranged from small to large, which means that the peak pressure values are arranged from small to large.
[0032] When the number of labels in label_filter is less than 2, it means that there is only one process at this time, and the air pressure has been decreasing. At this time, the insufficient air pressure alarm is directly turned on, otherwise further traversal judgment is performed.
[0033] Step S3.3, when the number of labels in label_filter is greater than or equal to 2, set the counter count with an initial value of 0, and perform count based on a double-layer traversal. The first layer traverses the label value in label_filter, and the second layer traverses the corresponding position index and label value in inpressure_peak_list. When the labels of the first and second layers are the same, count is increased by 1, otherwise count is set to 0. When count is greater than 10, record the position index A and the corresponding label B in the second layer at this time. Traverse the subsequent position indexes in the second layer based on the recorded position index A. When there is a label greater than B in the label value corresponding to the subsequent position index, it means that the process has been changed and there is no need to perform insufficient air pressure detection. Otherwise, the insufficient air pressure alarm is turned on.
[0034] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A real-time detection method for insufficient pressure of high-pressure mixed gas, characterized in that: The following steps are involved: Step S1, collecting and filtering the laser cutting current and air pressure time series data at the inlet in real time, and extracting the peak value sequence based on the filtered air pressure time series data; Determine whether to perform functional relationship fitting based on the peak value sequence; Step S2, fitting a linear function relationship between the peak value and time, and judging whether to perform insufficient air pressure detection based on the slope of the linear function; Step S3, start the air pressure shortage detection, cluster the peak value sequence based on the density clustering algorithm and analyze the clustering results, judge whether there is air pressure shortage based on the double-layer traversal; turn on the alarm when there is air pressure shortage; specifically, Step S3.1, when the inpressure_list filtered in step S1 satisfies that the difference between the maximum value of the first th2 air pressure values and the maximum value of the last th2 air pressure values is less than th3, density clustering is performed on the peak values in the corresponding inpressure_peak_list, otherwise, the insufficient air pressure detection is no longer performed; Step S3.2: After completing the density clustering, obtain n labels, where each label includes several peak values, and each peak value corresponds to a position index; perform the following processing on the clustering results: First, extract all labels with the maximum continuous length greater than 2 in the density clustering results, and store the labels in the label_filter set in ascending order; when the number of labels in label_filter is less than 2, directly turn on the low pressure alarm, otherwise execute step S3.3; Step S3.3, set the counter count, the initial value is 0, and count is performed based on a double-layer traversal; the first layer traverses the label value in label_filter, and the second layer traverses the corresponding position index and label value in inpressure_peak_list; when the labels of the first and second layers are the same, count is increased by 1, otherwise count is set to 0; when count is greater than 10, record the position index A and the corresponding label B in the second layer at this time; traverse the subsequent position indexes of the second layer based on the recorded position index A, and when there is a label greater than B in the label value corresponding to the subsequent position index, no insufficient air pressure detection is performed, otherwise the insufficient air pressure alarm is turned on.
2. A real-time detection method for insufficient pressure of high-pressure mixed gas according to claim 1, characterized in that: The specific method of extracting the peak value sequence in step S1 includes: Set a collection current_list for real-time storage of current data and a collection inpressure_list for real-time storage of air pressure data; set a current sliding window with a window length of ws; when ws current points are received in current_list, filter all current points that are less than a preset reference current m, and filter the air pressure points at the corresponding time sequence position in inpressure_list; based on the filtered inpressure_list, extract all air pressure peak values in the collection to generate a peak value set inpressure_peak_list.
3. A real-time detection method for insufficient pressure of high-pressure mixed gas according to claim 2, characterized in that: In step S2, when the number of peak values in inpressure_peak_list is greater than ws / 10, and the absolute values of the first-order difference values of adjacent peak values are all less than the preset threshold value th1, a linear function relationship between the peak value sequence and time is fitted, and the slope k of the linear function relationship is obtained; when the slope k is less than the preset threshold value k_th, further insufficient air pressure detection is performed.