Signal Time-Domain Feature Analysis Method, Device, Electronic Device and Storage Medium

Abstract

The present invention provides a method, apparatus, electronic device and storage medium for signal time-domain feature analysis, relating to the field of signal analysis and processing. In the present invention, first, a one-dimensional time-domain discrete signal to be analyzed is obtained, the signal to be analyzed is divided to obtain a plurality of signal segments, then, according to the signal values of each signal segment, a convolution kernel of each signal segment is determined, and then, according to the convolution kernel of each signal segment, a convolution operation is performed on each signal segment to obtain a feature analysis value of each signal segment, and finally, according to the feature analysis values of all signal segments of the signal to be analyzed, the signal to be analyzed is analyzed to determine an abnormal signal in the signal to be analyzed. Through the above method, the signal change features of the signal to be analyzed can be extracted more comprehensively and accurately, and according to the signal change features, the abnormal signal in the signal to be analyzed can be determined more accurately.

Description

Technical Field

[0001] The present invention relates to the field of signal analysis and processing, and in particular, to a method, apparatus, electronic device, and storage medium for analyzing time-domain characteristics of signals. Background Art

[0002] In the industrial field, the working conditions and states of equipment can be obtained by analyzing signal data such as the real-time working voltage and current of electric drive equipment. Among them, when the equipment is working normally, the signals will show regular periodic fluctuations. When abnormal conditions occur during the operation of the equipment or the equipment itself, such as motor bearing failures, coil failures, interruption of shielding gas during the welding process, welding abnormalities caused by wire feeding failures, etc., abnormal signals will be generated. Theoretically, compared with normal signals, abnormal signals are manifested as changes in signal periods or signal values. However, in actual industrial production, abnormal signals and normal signals are superimposed together. Using the existing signal feature analysis method of extracting variance, it is impossible to accurately determine abnormal signals based on the extracted signal features. Summary of the Invention

[0003] Embodiments of the present invention provide a method, apparatus, terminal device, and storage medium for analyzing time-domain characteristics of signals, which can overcome the deficiencies of the prior art.

[0004] To achieve the above object, the technical solutions adopted in the embodiments of the present invention are as follows:

[0005] In a first aspect, embodiments of the present invention provide a method for analyzing time-domain characteristics of signals, the method for analyzing time-domain characteristics of signals comprising:

[0006] Obtain a signal to be analyzed that is discrete in one-dimensional time domain;

[0007] Divide the signal to be analyzed to obtain a plurality of signal segments;

[0008] Determine a convolution kernel for each signal segment according to the signal values of each signal segment;

[0009] Perform a convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain a feature analysis value for each signal segment;

[0010] Analyze the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed to determine an abnormal signal in the signal to be analyzed.

[0011] As a possible implementation, each signal segment includes a plurality of moments, and each moment corresponds to a signal value. The step of determining a convolution kernel for each signal segment according to the signal values of each signal segment includes:

[0012] Calculate the mean value of the signal values at all times in each of the signal segments to obtain the signal mean of each of the signal segments;

[0013] Based on the signal mean of each of the signal segments and the signal value at each of the times, obtain the signal deviation degree at each of the times;

[0014] Based on the signal deviation degrees at all times of each of the signal segments, construct the convolution kernel of each of the signal segments.

[0015] As a possible implementation manner, the expression of the convolution kernel of any target signal segment is:

[0016]

[0017] wherein, the starting time of the target signal segment is τ, the ending time is τ + ω, x t is the signal value of the target signal segment at time t, μ is the signal mean of the target signal segment, (x t - μ) 2 is the signal deviation degree of the target signal segment at time t, a is a preset constant, and f is the convolution kernel of the target signal segment.

[0018] As a possible implementation manner, the steps of dividing the signal to be analyzed to obtain multiple signal segments include:

[0019] Starting from the starting time of the signal to be analyzed, slide a preset analysis window at a preset step length, and sequentially intercept signal segments from the signal to be analyzed to obtain multiple signal segments, wherein each time the preset analysis window is slid, one of the signal segments is obtained.

[0020] As a possible implementation manner, the steps of performing a convolution operation on each of the signal segments according to the convolution kernel of each of the signal segments to obtain the feature analysis value of each of the signal segments include:

[0021] For any target signal segment, according to the convolution kernel of the target signal segment, use the formula to perform a convolution operation to obtain the feature analysis value of the target signal segment, wherein f is the convolution kernel of the target signal segment, g is the target signal segment, the starting time of the target signal segment is τ, the ending time is τ + ω, g(t) = x t is the signal value of the target signal segment at time t, and λ is the feature analysis value of the target signal segment.

[0022] As a possible implementation manner, the step of analyzing the signal to be analyzed and determining the abnormal signal in the signal to be analyzed according to the characteristic analysis values of all signal segments of the signal to be analyzed includes:

[0023] Generating a characteristic curve of the signal to be analyzed according to the characteristic analysis values of all signal segments of the signal to be analyzed;

[0024] Analyzing the characteristic curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed.

[0025] As a possible implementation manner, the abnormal signal includes an abnormal time and an abnormal signal value. The step of analyzing the characteristic curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed includes:

[0026] Taking the signal segment corresponding to the characteristic analysis value greater than the preset threshold in the characteristic curve of the signal to be analyzed as an abnormal signal segment;

[0027] Taking the starting time of the abnormal signal segment as the abnormal time;

[0028] Taking the signal value at the abnormal time as the abnormal signal value.

[0029] In a second aspect, an embodiment of the present invention provides a signal time-domain feature analysis device, which includes:

[0030] An acquisition module, configured to acquire a signal to be analyzed that is discrete in a one-dimensional time domain;

[0031] A segmentation module, configured to divide the signal to be analyzed to obtain a plurality of signal segments;

[0032] A determination module, configured to determine a convolution kernel for each signal segment according to the signal value of each signal segment;

[0033] An operation module, configured to perform a convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain a characteristic analysis value of each signal segment;

[0034] An analysis module, configured to analyze the signal to be analyzed according to the characteristic analysis values of all signal segments of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed.

[0035] In a third aspect, an embodiment of the present invention provides an electronic device, which includes a memory and a processor. The memory is used to store a computer program; the processor is configured to execute the signal time-domain feature analysis method provided by the embodiment of the present invention when calling the computer program.

[0036] Fourthly, an embodiment of the present invention provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the signal time-domain feature analysis method provided by the embodiment of the present invention is implemented.

[0037] Compared with the prior art, a signal time-domain feature analysis method, device, electronic device and storage medium provided by an embodiment of the present invention first obtains a to-be-analyzed signal that is discrete in the one-dimensional time domain, divides the to-be-analyzed signal to obtain a plurality of signal segments, then determines a convolution kernel for each signal segment according to the signal values of each signal segment, and then performs a convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain a feature analysis value of each signal segment. Finally, according to the feature analysis values of all signal segments of the to-be-analyzed signal, the to-be-analyzed signal is analyzed to determine an abnormal signal in the to-be-analyzed signal. Since the embodiment of the present invention constructs a convolution kernel adapted to the signal change characteristics for each signal segment, and performs a convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain a feature analysis value of each signal segment, the signal change characteristics of the to-be-analyzed signal extracted are more comprehensive and accurate, and the abnormal signal in the to-be-analyzed signal is determined more accurately according to the signal change characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained according to these drawings without creative efforts.

[0039] Figure 1 It is a flowchart of a signal time-domain feature analysis method provided in this embodiment;

[0040] Figure 2 It is a schematic diagram of dividing a to-be-analyzed signal into a plurality of signal segments provided by an embodiment of the present invention;

[0041] Figure 3 It is a flowchart of determining a convolution kernel based on the signal values of signal segments provided by an embodiment of the present invention;

[0042] Figure 4 It is a schematic diagram of a convolution operation provided by an embodiment of the present invention;

[0043] Figure 5 It is a flowchart of determining an abnormal signal in a to-be-analyzed signal based on the feature analysis values of signal segments provided by an embodiment of the present invention;

[0044] Figure 6AIt is a diagram showing the normal welding result provided by the embodiment of the present invention;

[0045] Figure 6B It is a schematic diagram of an application effect provided by the embodiment of the present invention;

[0046] Figure 6C It is a diagram showing the abnormal welding result provided by the embodiment of the present invention;

[0047] Figure 6D It is another schematic diagram of an application effect provided by the embodiment of the present invention;

[0048] Figure 6E It is a comparison diagram of the application effects of two methods provided by the embodiment of the present invention;

[0049] Figure 7 It is a block diagram of the signal time-domain feature analysis device provided by the embodiment of the present invention;

[0050] Figure 8 It is a schematic block diagram of the structure of an electronic device provided by the embodiment of the present invention.

[0051] Icons: 100 - Signal time-domain feature analysis device; 101 - Acquisition module; 102 - Segmentation module; 103 - Determination module; 104 - Operation module; 105 - Analysis module; 200 - Electronic device; 210 - Memory; 220 - Processor. Detailed implementation manners

[0052] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0054] It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0055] In the description of the present invention, it should be noted that if terms such as "upper", "lower", "inner", "outer", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the invention is customarily placed during use, it is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation on the present invention.

[0056] In addition, if terms such as "first", "second", etc. are only used for distinguishing descriptions, they cannot be understood as indicating or implying relative importance.

[0057] It should be noted that, without conflict, the features in the embodiments of the present invention can be combined with each other.

[0058] In the industrial field, the working voltage or current data of real-time collected devices can reflect the working conditions and states of the devices. Taking a welding machine device as an example, by obtaining the working voltage or current data of the welding machine during the welding process, the working state of the welding machine during the welding process can be judged, and welding abnormalities caused by motor bearing failures, coil failures, shielding gas blockages, wire feeding failures, and burn-throughs can be detected in a timely manner. The working voltage or current data of the welding machine during the welding process collected in real time can be regarded as a one-dimensional time-domain discrete signal. When the working state of the welding machine is normal, the signal usually shows regular periodic fluctuations, and such a signal can be regarded as a normal signal. When an abnormality occurs during the welding process, the period or signal value of the signal will change, and such a signal can be regarded as an abnormal signal. Since the abnormal signal and the normal signal are superimposed together, in the method of extracting signal change features by variance in the prior art, the difference between the normal signal and the abnormal signal in the extracted signal change features is not obvious, and the moment when the abnormal signal appears cannot be accurately located and the abnormal signal cannot be separated from the signal, which is not conducive to subsequent analysis of the factors causing the abnormal signal.

[0059] In order to overcome the deficiencies of the prior art, signal change features can also be extracted by performing a convolution operation on the signal. The process of the convolution operation is to use a convolution kernel to perform a sliding weighted sum with the signal, and a series of values obtained from multiple slidings are used to characterize the change features of the signal. Although the signal change features extracted using different convolution kernels are different, since the convolution kernel is fixed during the convolution operation, no matter which convolution kernel is used, the change features of the signal cannot be comprehensively and accurately extracted.

[0060] The embodiments of the present invention provide a signal time-domain feature analysis method, device, electronic device, and storage medium, which can comprehensively and accurately extract the change features of the signal and accurately determine the abnormal signal, and the following will describe it in detail.

[0061] Please refer to Figure 1 ,Figure 1 The flowchart of a signal time-domain feature analysis method provided in this embodiment includes steps S101 to S105.

[0062] Step S101: Obtain the signal to be analyzed that is discrete in the one-dimensional time domain.

[0063] In the embodiment of the present invention, the signal to be analyzed is a discrete signal in the one-dimensional time domain, including multiple discrete moments, and the time interval between every two adjacent moments is equal, and each moment corresponds to a signal value.

[0064] Step S102: Divide the signal to be analyzed to obtain multiple signal segments.

[0065] In the embodiment of the present invention, to determine whether the signal value at a certain moment is a normal signal or an abnormal signal, it is necessary to combine the signal values corresponding to several other moments adjacent to this moment, that is, the signal values corresponding to multiple pairwise adjacent moments form a signal segment, and a signal to be analyzed can be divided into at least one signal segment.

[0066] Step S103: Determine the convolution kernel of each signal segment according to the signal values of each signal segment.

[0067] In the embodiment of the present invention, since each moment in the signal to be analyzed corresponds to a signal value, and the moments included in each signal segment are different, the change characteristics of the signals in each signal segment are different. Based on the signal values corresponding to each moment in each signal segment, the convolution kernel of each signal segment is constructed.

[0068] Step S104: Perform convolution operations on each signal segment according to the convolution kernel of each signal segment to obtain the feature analysis value of each signal segment;

[0069] In the embodiment of the present invention, each signal segment corresponds to a convolution kernel. The convolution kernels are sequentially convolved with the signal segments corresponding to them in the order of signal segment division, and the multiple convolution operation results obtained are the feature analysis values of the respective signal segments.

[0070] Step S105: Analyze the signal to be analyzed according to the feature analysis values of all the signal segments of the signal to be analyzed to determine the abnormal signals in the signal to be analyzed.

[0071] In the embodiment of the present invention, the feature analysis value of each signal segment characterizes the change characteristics of the signal within the signal segment. According to the change characteristics of the signals of all the signal segments, the change characteristics of the signal to be analyzed are obtained, and the abnormal signals are determined according to the change characteristics of the signal to be analyzed.

[0072] The above method provided by the embodiments of the present invention has the beneficial effect that a convolution kernel adapted to the signal change characteristics is constructed for each signal segment, and each signal segment is convolved using the convolution kernel of each signal segment to obtain the feature analysis value of each signal segment, thereby making the signal change characteristics of the signal to be analyzed extracted more comprehensive and accurate, and determining the abnormal signal in the signal to be analyzed more accurately according to the signal change characteristics.

[0073] Based on Figure 1 , the embodiments of the present invention also provide a specific implementation manner for dividing the signal to be analyzed. Step S102 can be implemented in the following manner:

[0074] Starting from the starting moment of the signal to be analyzed, a preset analysis window is slid according to a preset step size, and signal segments are sequentially intercepted from the signal to be analyzed to obtain a plurality of signal segments, where each time the preset analysis window is slid, a signal segment is obtained.

[0075] In the embodiments of the present invention, a signal segment is intercepted using a preset analysis window. The size of the preset analysis window represents the number of discrete moments included in the intercepted signal segment, and its size can be determined by the period of the signal to be analyzed. If the period of the signal to be analyzed is not easy to determine, it can be set according to experience. The preset step size is used to represent the number of discrete moments separated by sliding the preset analysis window once.

[0076] To more intuitively show the process of dividing the signal to be analyzed into multiple signal segments, in the embodiments of the present invention, taking the signal x(t)=x t , t = 0, 1, 2, 3,..., n as the signal to be analyzed, the size of the preset analysis window is 4, that is, the number of discrete moments included in the signal segment intercepted using this preset analysis window is 4, and the preset step size is 1, that is, the number of discrete moments separated by sliding the preset analysis window once is 1, as an example for illustration, please refer to Figure 2 , Figure 2 is a schematic diagram of dividing the signal to be analyzed into multiple signal segments provided by the embodiments of the present invention.

[0077] First, align the starting position of the preset analysis window with the t = 0 moment of the signal to be analyzed x(t), and align the ending position of the preset analysis window with the t = 3 moment of the signal to be analyzed x(t);

[0078] Then, intercept the signal values at the t = 0, t = 1, t = 2, and t = 3 moments of the signal to be analyzed x(t) within the preset analysis window as a signal segment g0(t)=x t , t = 0, 1, 2, 3;

[0079] Next, slide the preset analysis window to the right once according to the preset step size. At this time, the starting position of the preset analysis window aligns with the moment of t = 1 of the signal x(t) to be analyzed, and the ending position of the preset analysis window aligns with the moment of t = 4 of the signal x(t) to be analyzed. Then, intercept the signal values at the moments of t = 1, t = 2, t = 3, and t = 4 of the signal x(t) within the preset analysis window as a signal segment g1(t) = x t , t = 1, 2, 3, 4;

[0080] Then, slide the preset analysis window to the right once again according to the preset step size. At this time, the starting position of the preset analysis window aligns with the moment of t = 2 of the signal x(t) to be analyzed, and the ending position of the preset analysis window aligns with the moment of t = 5 of the signal x(t) to be analyzed. Then, intercept the signal values at the moments of t = 2, t = 3, t = 4, and t = 5 of the signal x(t) within the preset analysis window as a signal segment g2(t) = t ， t = 2, 3, 4, 5;

[0081] Each time the preset analysis window is slid to the right, a signal segment is intercepted. When the ending position of the preset analysis window aligns with the moment of t = n of the signal x(t) to be analyzed, the sliding of the preset analysis window is ended, and the division of the signal x(t) to be analyzed is completed.

[0082] The above division method is a specific implementation method for dividing the pre-acquired signal. In fact, in the embodiments of the present invention, the signal to be analyzed obtained in real time can also be divided by a method of dividing while acquiring.

[0083] Based on Figure 1 , the embodiments of the present invention also provide a specific implementation method of step S103. Please refer to Figure 3 , Figure 3 FIG. is a flowchart of determining a convolution kernel based on the signal values of signal segments provided by the embodiments of the present invention. Step S103 further includes sub-steps S103-1 to sub-step S103-3.

[0084] Sub-step S103-1: Calculate the mean value of the signal values at all moments in each signal segment to obtain the signal mean value of each signal segment.

[0085] In the embodiments of the present invention, the signal mean value of each signal segment refers to the sum of the signal values at all moments of the signal segment divided by the number of moments included in the signal segment, which reflects the central tendency of the signal values of the signal segment.

[0086] Sub-step S103-2: Obtain the signal deviation degree at each moment according to the signal mean value of each signal segment and the signal value at each moment.

[0087] In an embodiment of the present invention, for any signal segment, the signal deviation degree at each moment refers to the square value of the difference between the signal value at each moment and the mean value of the signal segment, which reflects the deviation degree between the signal value at each moment and the mean value of the signal segment.

[0088] Sub-step S103-3: Construct a convolution kernel for each signal segment according to the signal deviation degrees at all moments of each signal segment.

[0089] In an embodiment of the present invention, the change characteristics of the signals in each signal segment are different. Based on the signal deviation degrees corresponding to each moment in each signal segment, a convolution kernel for each signal segment is constructed.

[0090] As a specific implementation manner, the convolution kernel expression of any target signal segment is:

[0091]

[0092] In the above expression, the starting moment of the target signal segment is τ, the ending moment is τ + ω, x t is the signal value of the target signal segment at time t, μ is the signal mean value of the target signal segment, (x t -μ) 2 is the signal deviation degree of the target signal segment at time t, a is a preset constant, and f is the convolution kernel of the target signal segment.

[0093] The convolution kernel of the target signal segment is the ratio of the exponential function of the signal deviation degree at each moment to the linear function of the signal deviation degree. When the signal value at each moment in the target signal segment is a normal signal, the signal deviation degrees (x t -μ) 2 at each moment in the target signal segment are close to a constant. When there is an abnormal signal in the target signal segment, the signal deviation degree (x t -μ) 2 at the abnormal moment is larger than that at the normal moment. By calculating the ratio of the exponential function of the signal deviation degree at each moment to the linear function of the signal deviation degree, the difference between the normal signal and the abnormal signal is amplified.

[0094] Taking Figure 2 the signal segment g1(t) = x t , t = 1, 2, 3, 4 in as an example of the target signal segment, it can be understood that the signal mean value of this signal segment The expression of the convolution kernel corresponding to this signal segment is:

[0095]

[0096] Based on Figure 1, The implementation of the present invention also provides a specific implementation manner of step S104.

[0097] Specifically, for any target signal segment, according to the convolution kernel of the target signal segment, the following formula is used for convolution operation.

[0098]

[0099] In the above formula, f is the convolution kernel of the target signal segment, g is the target signal segment, the starting time of the target signal segment is τ, the ending time is τ + ω, g(t) = x t is the signal value of the target signal segment at time t, and λ is the characteristic analysis value of the target signal segment.

[0100] To more intuitively show the process of performing convolution operation on the target signal segment according to the convolution kernel of the target signal segment, the embodiment of the present invention uses Figure 2 the signal segment signal g1(t) = x t , t = 1, 2, 3, 4 in Figure 4 , Figure 4 as an example of the target signal segment for illustration. Please refer to

[0101] which is a schematic diagram of a convolution operation provided by the embodiment of the present invention.

[0102] Based on Figure 1 , the implementation of the present invention also provides a specific implementation manner of step S105. Please refer to Figure 5 , Figure 5 which is a flowchart for determining abnormal signals in the signal to be analyzed based on the characteristic analysis values of signal segments provided by the embodiment of the present invention. Step S105 includes sub-step S105-1 and sub-step S105-2.

[0103] Sub-step S105-1, generate a characteristic curve of the signal to be analyzed according to the characteristic analysis values of all signal segments of the signal to be analyzed.

[0104] In an embodiment of the present invention, the characteristic analysis values of all signal segments are arranged in the order of extraction of the signal segments, and a characteristic curve of the signal to be analyzed is generated using statistical software such as EXCEL.

[0105] Sub-step S105-2: Analyze the characteristic curve of the signal to be analyzed to determine the abnormal signals in the signal to be analyzed.

[0106] In an embodiment of the present invention, the magnitude of the characteristic analysis value reflects the number of abnormal signals contained in the corresponding signal segment. The larger the characteristic analysis value, the more abnormal signals are contained in the signal segment, and the smaller the characteristic analysis value, the fewer abnormal signals are contained in the signal segment.

[0107] As a specific implementation manner, the steps for determining the abnormal signals in the signal to be analyzed are as follows:

[0108] First, use the signal segments corresponding to the characteristic analysis values greater than the preset threshold in the characteristic curve of the signal to be analyzed as abnormal signal segments;

[0109] Second, use the starting moment of the abnormal signal segment as the abnormal moment;

[0110] Finally, use the signal value at the abnormal moment as the abnormal signal value.

[0111] In an embodiment of the present invention, the preset threshold refers to the upper limit value of the characteristic analysis value. Considering that there are many factors affecting the generation of abnormal signals in industrial production and the magnitudes of abnormal signals caused by different factors also vary, any signal segment in the actually collected signals contains more or less abnormal signals. When the characteristic analysis value is less than the preset value, the abnormal signals contained in the corresponding signal segment are relatively weak and within the allowable range. When the characteristic analysis value is greater than the preset value, the abnormal signals contained in the corresponding signal segment have exceeded the allowable range, then the starting moment of the signal segment is used as the abnormal moment, and the signal value at this moment is used as the abnormal signal value.

[0112] To better demonstrate the application effect of the above method, the embodiment of the present invention applies the above method to the welding current data in actual industrial production. Among them, the current data when the welding process is normal is normal welding current data, and the current data when the welding process is abnormal is abnormal welding current data. The embodiment of the present invention selects 89 welding current data, including 66 welding current data corresponding to normal welding results and 23 welding current data corresponding to abnormal welding results. The preset analysis window size is set to 60, the preset step size is set to 1, and the preset threshold is set to 60.

[0113] Figure 6A For the normal welding result display diagram provided by the embodiment of the present invention, from Figure 6AIt can be seen that there is no area of ​​the welded material that is melted through. Figure 6A The welding current data corresponding to the normal welding result shown in FIG. 1 is subjected to the signal time domain characteristic analysis method provided by the embodiment of the present invention. The results are shown in FIG. Figure 6B ,from Figure 6B It can be seen that the welding current data corresponding to the normal welding result shows regular periodic fluctuations. The characteristic curve extracted by the signal time domain characteristic analysis method provided in the embodiment of the present invention can be approximately regarded as a horizontal straight line, and is far less than the preset threshold.

[0114] Figure 6C The abnormal welding result display diagram provided by the embodiment of the present invention is as follows: Figure 6C It can be seen that there are multiple melted-through areas of varying degrees on the welded material. Figure 6C The welding current data corresponding to the abnormal welding result shown in FIG. 1 is subjected to the signal time domain characteristic analysis method provided by the embodiment of the present invention. The results are shown in FIG. Figure 6D ,from Figure 6D It can be seen that the characteristic curve extracted by the signal time domain feature analysis method provided by the embodiment of the present invention includes multiple smoothly changing areas and multiple peaks of different sizes, and the values ​​of some peaks have exceeded the preset threshold, wherein the data segments corresponding to the characteristic analysis values ​​of the smoothly changing areas are normal data segments, and the data segments corresponding to the characteristic analysis values ​​of the peak areas are abnormal data segments, and the characteristic analysis values ​​corresponding to the normal data segments and the abnormal data segments have clear boundaries on the characteristic curve.

[0115] Respectively Figure 6C The welding current data corresponding to the abnormal welding result shown in FIG. 1 is applied to the signal time domain feature analysis method provided by the embodiment of the present invention and the existing method for extracting signal change features through variance. The results are shown in FIG. Figure 6E ,from Figure 6E It can be seen that the variance curve extracted by the existing method only amplifies and slows down the change process of the current value of the welding current data. The variance values ​​corresponding to the normal data segment and the abnormal data segment have no clear boundary on the variance curve, and the abnormal data segment in the welding current data cannot be accurately located.

[0116] The signal time domain characteristic analysis method provided by the embodiment of the present invention is applied to the selected 89 welding current data to extract characteristic curves, detect and distinguish welding results, wherein the distinguishing principle is that for any characteristic curve of welding current data, if all characteristic analysis values ​​on the characteristic curve are less than the preset threshold, then the welding current data corresponds to a normal welding result, and if there is a characteristic analysis value on the characteristic curve greater than the preset threshold, then the welding current data corresponds to an abnormal welding result. The comparison conclusion between the detection situation and the actual situation is shown in the following table:

[0117]

[0118] Among the 66 welding current data corresponding to normal welding results, the detection of 64 cases conforms to the actual situation, with an accuracy rate of 96.97%. The detection of all 23 welding current data corresponding to abnormal welding results conforms to the actual situation, with an accuracy rate of 100%. That is, among the 89 welding current data, the detection of 87 cases is consistent with the actual situation, and the accuracy rate is 97.75%.

[0119] To execute the corresponding steps in the above embodiments and each possible implementation manner, the following provides an implementation manner of a signal time-domain feature analysis device 100. Please refer to Figure 7 , Figure 7 FIG. shows a block diagram of a signal time-domain feature analysis device 100 provided by an embodiment of the present invention. It should be noted that the basic principle and the technical effects generated by the signal time-domain feature analysis device 100 provided by the embodiment of the present invention are the same as those of the above embodiments. For the sake of brief description, the embodiments of the present invention do not mention them.

[0120] The signal time-domain feature analysis device 100 includes an acquisition module 101, a segmentation module 102, a determination module 103, an operation module 104, and an analysis module 105.

[0121] The acquisition module 101 is configured to acquire a one-dimensional time-domain discrete signal to be analyzed.

[0122] The segmentation module 102 is configured to divide the signal to be analyzed to obtain a plurality of signal segments.

[0123] As a specific implementation manner, the segmentation module 102 is specifically configured to start from the starting moment of the signal to be analyzed, slide a preset analysis window according to a preset step size, and sequentially intercept signal segments from the signal to be analyzed to obtain a plurality of signal segments, where each time the preset analysis window is slid, one of the signal segments is obtained.

[0124] The determination module 103 is configured to determine a convolution kernel for each signal segment according to the signal values of each signal segment.

[0125] As a specific implementation manner, the determination module 103 is specifically configured to calculate the mean value of the signal values at all moments in each signal segment to obtain the signal mean value of each signal segment; obtain the signal deviation degree at each moment according to the signal mean value of each signal segment and the signal value at each moment; and construct a convolution kernel for each signal segment according to the signal deviation degrees at all moments of each signal segment.

[0126] The operation module 104 is configured to perform a convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain a feature analysis value of each signal segment.

[0127] As a specific implementation, the operation module 104 is specifically configured to, for any target signal segment, according to the convolution kernel of the target signal segment, use the formula to perform convolution operation to obtain the feature analysis value of the target signal segment, where f is the convolution kernel of the target signal segment, g is the target signal segment, the starting time of the target signal segment is τ, the ending time is τ + ω, g(t) = x t is the signal value of the target signal segment at time t, and λ is the feature analysis value of the target signal segment.

[0128] The analysis module 105 is configured to analyze the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed, and determine the abnormal signal in the signal to be analyzed.

[0129] As a specific implementation, the analysis module 105 is specifically configured to generate a feature curve of the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed; analyze the feature curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed.

[0130] As a specific implementation, when the analysis module 105 is used to analyze the feature curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed, it is specifically configured to: use the signal segment corresponding to the feature analysis value greater than the preset value in the feature curve of the signal to be analyzed as the abnormal signal segment; use the starting time of the abnormal signal segment as the abnormal time; use the signal value at the abnormal time as the abnormal signal value.

[0131] Further, please refer to Figure 8 , Figure 8 which is a schematic structural diagram of an electronic device 200 provided by an embodiment of the present invention. The electronic device 200 may include a memory 210 and a processor 220.

[0132] Wherein, the processor 220 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the signal time-domain feature analysis method provided in the following method embodiments.

[0133] The memory 210 can be a ROM or other types of static storage devices that can store static information and instructions, a RAM or other types of dynamic storage devices that can store information and instructions, or can also be an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 210 can exist independently and be connected to the processor 220 through a communication bus. The memory 210 can also be integrated with the processor 220. Among them, the memory 210 is used to store machine-executable instructions for implementing the solution of this application. The processor 220 is used to execute the machine-executable instructions stored in the memory 210 to implement the foregoing method embodiments.

[0134] Since the electronic device 200 provided by the embodiments of the present invention is another implementation form of the signal time-domain feature analysis method provided by the foregoing method embodiments, the technical effects that can be obtained therefrom can refer to the above method embodiments and will not be elaborated herein.

[0135] The embodiments of the present invention also provide a readable storage medium containing computer-executable instructions, and the computer-executable instructions can be used to perform related operations in the signal time-domain feature analysis method provided by the foregoing method embodiments when executed.

[0136] In summary, for the signal time-domain feature analysis method, device, electronic device, and storage medium provided by the embodiments of the present invention, first, a one-dimensional time-domain discrete signal to be analyzed is obtained, the signal to be analyzed is divided to obtain multiple signal segments, then, according to the signal values of each signal segment, the convolution kernel of each signal segment is determined, and then, according to the convolution kernel of each signal segment, a convolution operation is performed on each signal segment to obtain the feature analysis value of each signal segment, and finally, according to the feature analysis values of all the signal segments of the signal to be analyzed, the signal to be analyzed is analyzed to determine the abnormal signal in the signal to be analyzed. Compared with the prior art, in the embodiments of the present invention, a convolution kernel adapted to the signal change feature of each signal segment is constructed for each signal segment, and a convolution operation is performed on each signal segment using the convolution kernel of each signal segment to obtain the feature analysis value of each signal segment, so that the signal change features of the signal to be analyzed extracted are more comprehensive and accurate, and the abnormal signal in the signal to be analyzed is determined more accurately according to the signal change features.

[0137] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims described above.

Claims

1. A method for signal time-domain feature analysis, characterized in that, The method includes the following steps: Obtain a one-dimensional discrete signal to be analyzed in the time domain, where the signal to be analyzed is the current data of the welding machine during real-time welding process acquisition; Divide the signal to be analyzed to obtain multiple signal segments; Determine the convolution kernel of each signal segment according to the signal value of each signal segment; Perform convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain the feature analysis value of each signal segment; Analyze the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed, and determine the abnormal signal in the signal to be analyzed; Each signal segment includes multiple moments, and each moment corresponds to a signal value. The step of determining the convolution kernel of each signal segment according to the signal value of each signal segment includes: Calculate the mean value of the signal values at all moments in each signal segment to obtain the signal mean value of each signal segment; Obtain the signal deviation degree of each moment according to the signal mean value of each signal segment and the signal value of each moment; Construct the convolution kernel of each signal segment according to the signal deviation degrees of all moments of each signal segment; The expression of the convolution kernel of any target signal segment is: Among them, the start time of the target signal segment is , and the end time is , is the signal value of the target signal segment at time, is the signal mean value of the target signal segment, is the signal deviation degree of the target signal segment at time, is a preset constant, is the convolution kernel of the target signal segment; The step of analyzing the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed and determining the abnormal signal in the signal to be analyzed includes: Generate a feature curve of the signal to be analyzed according to the feature analysis values of all signal segments of the signal to be analyzed, and the feature curve of the signal to be analyzed is extracted by analyzing the current data; Analyze the feature curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed, and the abnormal signal characterizes the abnormal welding result corresponding to the current data.

2. The method according to claim 1, wherein The step of dividing the signal to be analyzed to obtain multiple signal segments includes: Starting from the starting moment of the signal to be analyzed, slide a preset analysis window at a preset step length, and sequentially intercept signal segments from the signal to be analyzed to obtain multiple signal segments, where each time the preset analysis window is slid, one signal segment is obtained.

3. The method according to claim 1, wherein The step of performing convolution operation on each signal segment according to the convolution kernel of each signal segment to obtain the feature analysis value of each signal segment includes: For any target signal segment, according to the convolution kernel of the target signal segment, the formula is used to perform a convolution operation to obtain the feature analysis value of the target signal segment, where is the convolution kernel of the target signal segment, is the target signal segment, and the starting time of the target signal segment is , the ending time is , is the signal value of the target signal segment at time, and is the feature analysis value of the target signal segment.

4. The method according to claim 1, wherein The abnormal signal includes an abnormal moment and an abnormal signal value. The step of analyzing the feature curve of the signal to be analyzed to determine the abnormal signal in the signal to be analyzed includes: Take the signal segment corresponding to the feature analysis value greater than the preset threshold in the feature curve of the signal to be analyzed as an abnormal signal segment; Take the starting moment of the abnormal signal segment as the abnormal moment; Take the signal value of the abnormal moment as the abnormal signal value.

5. A signal time-domain feature analysis device, characterized in that, The device includes: An acquisition module for acquiring a one-dimensional discrete signal to be analyzed in the time domain, where the signal to be analyzed is the current data of the welding machine during real-time welding process acquisition; A segmentation module for dividing the signal to be analyzed to obtain multiple signal segments; A determination module, configured to determine a convolution kernel for each of the signal segments according to the signal value of each of the signal segments; An operation module, configured to perform a convolution operation on each of the signal segments according to the convolution kernel of each of the signal segments to obtain a feature analysis value for each of the signal segments; An analysis module, configured to analyze the signal to be analyzed according to the feature analysis values of all the signal segments of the signal to be analyzed to determine an abnormal signal in the signal to be analyzed; Each of the signal segments includes a plurality of moments, and each moment corresponds to a signal value. The determining module is specifically configured to calculate the mean value of the signal values of all moments in each signal segment to obtain the signal mean value of each signal segment; obtain the signal deviation degree of each moment according to the signal mean value of each signal segment and the signal value of each moment; construct the convolution kernel of each signal segment according to the signal deviation degrees of all moments of each signal segment. The expression of the convolution kernel of any target signal segment is: ; where the starting moment of the target signal segment is , and the ending moment is , is the signal value of the target signal segment at moment, is the signal mean value of the target signal segment, is the signal deviation degree of the target signal segment at moment, is a preset constant, is the convolution kernel of the target signal segment; Specifically, the analysis module is configured to: generate a feature curve of the signal to be analyzed according to the feature analysis values of all the signal segments of the signal to be analyzed, where the feature curve of the signal to be analyzed is obtained by analyzing the current data; analyze the feature curve of the signal to be analyzed to determine an abnormal signal in the signal to be analyzed, and the abnormal signal characterizes an abnormal welding result corresponding to the current data.

6. An electronic device, characterized in that, Including: A memory and a processor, where the memory is configured to store a computer program; the processor is configured to execute the method according to any one of claims 1-4 when calling the computer program.

7. A readable storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by the processor, the method according to any one of claims 1-4 is implemented.

Images