A motor rotor magnetic sheet defect detection method based on an intelligent sensing system

Abstract

The application relates to the technical field of intelligent sensing, and particularly discloses a motor rotor magnetic sheet defect detection method based on an intelligent sensing system. The method is characterized in that a plurality of ultrasonic array sensors are arranged around a motor rotor magnetic sheet, and an adaptive signal processing technology is used to filter ultrasonic echo signals, extract features, and realize beam imaging in combination with the application of a neural network to achieve motor rotor magnetic sheet crack detection. Compared with the traditional visual detection method and the X-ray detection method in the prior art, especially in complex working conditions and high-load environments, the traditional technology is limited by environmental interference and equipment, and it is difficult to realize high-precision crack detection. Through the intelligent sensing system and the adaptive signal processing method of the application, high-precision detection of motor rotor magnetic sheet cracks is realized, the misjudgment and omission problems caused by the traditional method are avoided, and the accuracy and reliability of motor rotor magnetic sheet defect detection are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent sensing, and particularly relates to a motor rotor magnetic sheet defect detection method based on an intelligent sensing system. BACKGROUND

[0002] At present, the defect detection method of the motor rotor magnetic sheet mainly relies on traditional visual detection and X-ray detection technology. Although the visual detection method can detect obvious surface cracks, it has poor detection ability for micro cracks and hidden cracks, and the crack detection effect is limited under complex surface or high load operation environment. Although the X-ray detection can provide relatively accurate internal defect images, it has high cost and complex operation, and cannot be monitored in real time, which limits its application in actual production. The existing technology cannot fully meet the needs of crack detection of the motor rotor magnetic sheet in dynamic and high load working environment, especially in real-time monitoring and high-precision detection. The problems faced by traditional technology include low crack detection accuracy, serious environmental interference, and poor adaptability to complex working conditions. Therefore, there is an urgent need for a method that can realize high-precision real-time detection of micro cracks under complex working environment, especially under high load and dynamic conditions. By using intelligent sensing technology and advanced signal processing algorithm, the application provides an efficient and accurate motor rotor magnetic sheet defect detection scheme, thereby improving the reliability of crack detection and the operation safety of motor equipment. SUMMARY

[0003] In view of the above technical deficiencies, the purpose of the present application is to provide a motor rotor magnetic sheet defect detection method based on an intelligent sensing system, which aims to solve the technical problems of the traditional visual detection method and the X-ray detection method in the prior art, especially under complex working conditions and high load environment, the traditional technology is limited by environmental interference and equipment, and it is difficult to realize high-precision crack detection.

[0004] To solve the above technical problems, the application adopts the following technical scheme: the application provides a motor rotor magnetic sheet defect detection method based on an intelligent sensing system,

[0005] The motor rotor magnetic sheet defect detection method based on the intelligent sensing system comprises:

[0006] Step S10: arranging N ultrasonic array sensors around the motor rotor magnetic sheet, synchronously emitting ultrasonic pulse waves at time t, and receiving first ultrasonic echo signals; first performing adaptive Wiener filtering on the first ultrasonic echo signals to obtain second ultrasonic echo signals, and then performing Hilbert transform on the second ultrasonic echo signals to obtain a filtered signal envelope curve;

[0007] Step S20: According to the filter signal envelope curve, the propagation characteristic difference between different ultrasonic array sensors is extracted, including time difference, waveform amplitude attenuation difference and phase difference; the propagation characteristic difference is subjected to fast Fourier transform to obtain crack sensitive frequency domain features, and a multi-directional ultrasonic feature matrix is constructed according to the propagation characteristic difference and the crack sensitive frequency domain features;

[0008] Step S30: The beam imaging method based on the delay and superposition algorithm is used for crack defect positioning, and a crack imaging feature matrix is output;

[0009] Step S40: A fusion matrix is constructed according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, a preset crack detection model is input, and a first crack detection result is obtained;

[0010] Step S50: Error analysis is performed on the first crack detection result, and a second crack detection result is output.

[0011] Preferably, in step S20, the time difference is calculated by using a weighted cross-correlation function, and the expression is:

[0012]

[0013] Wherein, ΔT ij is the time difference between the ultrasonic array sensor i and the ultrasonic array sensor j, w(t) is a weighted window function, τ is a time difference displacement variable, y i (t) and y j (t+τ) are respectively the signal of the ultrasonic array sensor i at time t and the signal of the ultrasonic array sensor j at time t+τ, and argmax is a weighted cross-correlation function; the waveform amplitude attenuation difference ΔA ij =|A i -A j |, wherein A i and A j are respectively the waveform amplitudes of the ultrasonic array sensors i and j; the phase difference Wherein are respectively the waveform phases of the ultrasonic array sensors i and j.

[0014] Preferably, in step S30, the beam imaging method based on the delay and superposition algorithm is used for crack defect positioning, and the step of outputting the crack imaging feature matrix, specifically comprising:

[0015] The beam imaging method based on the delay and superposition algorithm is used to calculate the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet to be measured:

[0016]

[0017] wherein I(x, y) is the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet to be measured; y n is the second ultrasonic echo signal received by the nth ultrasonic array sensor; r n (x, y) is the geometric distance from the position (x, y) to be measured to the nth ultrasonic array sensor; v is the propagation speed of the ultrasonic wave in the rotor magnet; A n is the weighting coefficient of the nth ultrasonic array sensor, which is determined according to the position of the ultrasonic array sensor;

[0018] A crack imaging feature matrix is constructed according to the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet to be measured.

[0019] Preferably, in step S40, the step of constructing a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, inputting a preset crack detection model, and obtaining a first crack detection result comprises:

[0020] Step S401: constructing a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, and performing two-dimensional convolution operation on the fusion matrix to obtain primary crack features;

[0021] Step S402: introducing an attention mechanism network, inputting the primary crack features into the attention mechanism network, calculating the attention weights between the primary crack features to enhance the recognition ability of crack sensitive features, and obtaining optimized crack features;

[0022] Step S403: performing maximum pooling operation on the optimized crack features to obtain a crack sensitive feature vector;

[0023] Step S404: inputting the crack sensitive feature vector into the full connection layer of the crack detection model and processing the crack sensitive feature vector through a Softmax activation function to obtain a first crack detection result, wherein the first crack detection result includes a crack category, a crack position, and a crack severity.

[0024] Preferably, in step S40, the attention mechanism network Attention adopts the formula:

[0025]

[0026] In the formula, Q, K, and V are the query matrix, key matrix, and value matrix of the attention mechanism network, which are constructed according to the primary crack features; d is the dimension of the feature matrix, which is used to normalize the scale of the feature vector; T is the matrix transpose symbol; and softmax is an activation function.

[0027] Preferably, in step S50, the step of performing error analysis on the first crack detection result and outputting a second crack detection result comprises:

[0028] Step S501: Obtain the position coordinate information of the crack from the first crack detection result, and calculate the feedback propagation characteristic difference of the ultrasonic signal between different ultrasonic array sensors based on the crack position theoretical model according to the position coordinate information of the crack, including: feedback time difference, feedback waveform amplitude value attenuation difference and feedback phase difference;

[0029] Step S502: Backtrack to obtain the propagation characteristic difference in step S20, and calculate the feedback error value according to the propagation characteristic difference and the feedback propagation characteristic difference;

[0030] Step S503: Set a first error threshold and a second error threshold, and divide the crack position confidence level according to the feedback error value:

[0031] If the feedback error value is less than or equal to the first error threshold, the crack position is defined as high confidence;

[0032] If the first error threshold is less than the feedback error value and the feedback error value is less than or equal to the second error threshold, it is defined as medium confidence;

[0033] If the feedback error value is greater than the second error threshold, it is defined as low confidence;

[0034] Step S504: Output the second crack detection result, including: crack position coordinates, crack type and size, crack detection result confidence level and feedback error value.

[0035] Preferably, in step S50, the feedback error value is calculated according to the propagation characteristic difference and the feedback propagation characteristic difference, and the formula is used:

[0036]

[0037] Wherein, And The feedback time difference, feedback amplitude value attenuation and feedback phase difference of the nth ultrasonic array sensor are respectively; And The time difference, waveform amplitude value attenuation difference and phase difference obtained in step S20 are respectively.

[0038] The application also provides a motor rotor magnetic sheet defect detection system based on an intelligent sensing system, comprising:

[0039] An envelope curve generation module is arranged around the motor rotor magnetic sheet, and is used for synchronously emitting ultrasonic pulse waves at time t and receiving first ultrasonic echo signals; the first ultrasonic echo signals are first subjected to adaptive wiener filtering to obtain second ultrasonic echo signals, and then the second ultrasonic echo signals are subjected to Hilbert transform to obtain a filtered signal envelope curve;

[0040] The multi-directional feature matrix construction module is used for extracting the propagation feature differences of ultrasonic signals between different ultrasonic array sensors from the filtered signal envelope curve, including time difference, waveform amplitude value attenuation difference and phase difference; the propagation feature differences are subjected to fast Fourier transform to obtain crack sensitive frequency domain features, and a multi-directional ultrasonic feature matrix is constructed according to the propagation feature differences and the crack sensitive frequency domain features;

[0041] The crack imaging feature matrix construction module is used for locating crack defects by using a beam imaging method based on a delay-and-sum algorithm, and outputting a crack imaging feature matrix;

[0042] The first crack detection result output module is used for constructing a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, inputting a preset crack detection model, and obtaining a first crack detection result;

[0043] The second crack detection result output module is used for error analysis on the first crack detection result and outputting a second crack detection result.

[0044] The application further provides a computer program product, comprising a motor rotor magnetic sheet defect detection program based on an intelligent sensing system, the motor rotor magnetic sheet defect detection program based on the intelligent sensing system realizes the motor rotor magnetic sheet defect detection method based on the intelligent sensing system when executed by a processor.

[0045] The application has the advantages that the intelligent sensing system is introduced, the ultrasonic array sensor and the adaptive signal processing technology are combined, and the crack defects can be detected in real time and with high precision during the operation of the motor rotor magnetic sheet.

[0046] The propagation characteristics (time difference, amplitude value attenuation difference and phase difference) of the ultrasonic signals are accurately monitored, the micro cracks can be effectively identified and the specific positions of the cracks can be located, the problems of missed detection and false detection in the traditional detection methods are avoided, and the precision of the crack detection is significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0048] Figure 1 It is a flowchart of the first embodiment of the motor rotor magnetic sheet defect detection method based on the intelligent sensing system.

[0049] Figure 2 This is a schematic diagram of equipment for a method for detecting defects in motor rotor magnetic sheets based on an intelligent sensing system according to the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] Example 1: Figure 1 2 is a flow chart of a first embodiment of a method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to the present invention, and provides a first embodiment of a method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to the present invention.

[0052] In a first embodiment, the motor rotor magnetic sheet defect detection method based on the intelligent sensing system includes:

[0053] Step S10: Arrange N ultrasonic array sensors around the motor rotor magnetic sheet, synchronously transmit ultrasonic pulse waves at time t, and receive a first ultrasonic echo signal; first perform adaptive Wiener filtering on the first ultrasonic echo signal to obtain a second ultrasonic echo signal, and then perform Hilbert transform on the second ultrasonic echo signal to obtain a filtered signal envelope curve;

[0054] It can be understood that the Wiener filtering process achieves adaptive denoising of ultrasonic echo signals by dynamically adjusting filter parameters to adapt to different noise environments, significantly improving signal quality. The Hilbert transform then further extracts the envelope information of the echo signal, allowing the signal's changes in the time domain to more clearly reflect the presence and characteristics of cracks.

[0055] It should be understood that by combining Wiener filtering and Hilbert transform processing, effective crack features can be extracted from complex and noisy backgrounds, thus avoiding the misjudgments and omissions caused by noise interference and signal distortion in traditional methods. This method improves the signal-to-noise ratio, making it possible to detect tiny cracks.

[0056] For example, in actual experiments using this method to inspect motor rotors, the signal-to-noise ratio of the ultrasonic echo signal obtained after Wiener filtering increased by approximately 30%. The envelope signal extracted through the Hilbert transform showed that crack features were more clearly separated and presented, and the detection accuracy was improved by approximately 20%. This improvement can significantly reduce detection errors caused by noise and interference, significantly improving the accuracy of microcrack detection.

[0057] Step S20: Extracting the propagation characteristic differences of ultrasonic signals between different ultrasonic array sensors based on the filtered signal envelope curve, including time difference, waveform exponential attenuation difference, and phase difference; performing a fast Fourier transform on the propagation characteristic differences to obtain crack-sensitive frequency domain features, and constructing a multi-directional ultrasonic feature matrix based on the propagation characteristic differences and the crack-sensitive frequency domain features;

[0058] It should be noted that, in step S20, the time difference is calculated using a weighted cross-correlation function, expressed as:

[0059]

[0060] Where, ΔT ij is the time difference between ultrasonic array sensor i and ultrasonic array sensor j, w(t) is the weighted window function, τ is the time difference displacement variable, y i (t), y j (t+τ) are the signals of ultrasonic array sensor i at time t and ultrasonic array sensor j at time t+τ, argmax is the weighted cross-correlation function; the waveform attenuation difference ΔA ij =|A i -A j |, where A i and A j are the wave form values ​​of ultrasonic array sensors i and j respectively; the phase difference in are the waveform phases of ultrasonic array sensors i and j respectively.

[0061] It will be appreciated that in step S20, the time difference is used to calculate the propagation characteristic differences between different ultrasonic array sensors. Specifically, the presence of cracks during ultrasonic signal propagation causes changes in propagation time, amplitude attenuation, and phase difference. By calculating the time difference between the signals received by different ultrasonic array sensors, the location of the crack can be effectively identified. This time difference is determined by the propagation speed of the ultrasonic signal and the distance between the ultrasonic array sensors, making it meaningful for crack location.

[0062] It should be understood that the time difference calculation can be understood as using a weighted cross-correlation function method, taking the time displacement of the signal as a variable, determining the best matching position between signals by maximizing the cross-correlation function, and then obtaining the time difference. This method can extract the propagation characteristics of the crack from noise and interference, enhance the distinguishability of the signal, and effectively reduce the influence of environmental noise on the detection result. The weighted window function in the time difference calculation formula makes the matching between signals more accurate, especially when the signal length is long or there are multiple reflections. By maximizing the correlation of the signals, high-precision positioning of the crack position and characteristics can be achieved. This process can greatly improve the accuracy of crack detection through cross-computation of multiple ultrasonic array sensors, especially in a dynamically changing motor rotor environment.

[0063] For example, three ultrasonic array sensors are arranged at different positions of the motor rotor magnetic sheet, and ultrasonic echo signals from the same crack are recorded. By calculating the time difference between the three ultrasonic array sensors, the propagation time difference obtained has a strong correlation with the position of the crack. Experimental data show that through the cooperation of time difference calculation and maximum cross-correlation function, the crack position can be accurately located, and the positioning accuracy is improved by about 20% compared with the traditional visual detection method.

[0064] Step S30: using a beam imaging method based on a delay-and-sum algorithm to locate the crack defect and output a crack imaging feature matrix;

[0065] It should be noted that in step S30, the step of using a beam imaging method based on a delay-and-sum algorithm to locate the crack defect and output a crack imaging feature matrix specifically includes:

[0066] Using a beam imaging method based on a delay-and-sum algorithm, the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet under test is calculated:

[0067]

[0068] where I(x, y) is the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet under test; y n is the second ultrasonic echo signal received by the nth ultrasonic array sensor; r n (x, y) is the geometric distance from the position (x, y) under test to the nth ultrasonic array sensor; v is the propagation speed of ultrasonic waves in the rotor magnet; A n is the weighting coefficient of the nth ultrasonic array sensor, determined according to the position of the ultrasonic array sensor;

[0069] A crack imaging feature matrix is constructed according to the ultrasonic imaging intensity at the two-dimensional spatial position (x, y) of the motor rotor magnetic sheet under test.

[0070] It can be understood that step S30 adopts a beam imaging method based on a time delay superposition algorithm for crack defect positioning, which is a time domain time delay superposition method by synchronously receiving multiple ultrasonic array sensor echo signals. By compensating and superimposing the propagation time of each ultrasonic array sensor signal, the echo signal of the crack can be effectively enhanced, and then a two-dimensional spatial position image of the crack on the rotor magnetic sheet is constructed through beam imaging. This method can greatly improve the accuracy and clarity of crack positioning in dynamic and complex environments.

[0071] It should be understood that the beam imaging method compensates for the time difference of each ultrasonic array sensor signal to ensure that the echo signals are aligned in space. This is a process based on the physical characteristics of ultrasonic signal propagation, which calculates the propagation time of each ultrasonic array sensor to the crack position and makes appropriate delay corrections so that each signal can be correctly superimposed. The superimposed signal can effectively reflect the accurate position and shape of the crack, not just its surface features. This technique avoids errors and biases that are prone to occur with traditional single sensor methods, especially in complex motor rotor environments.

[0072] For example, in actual experiments, when using the beam imaging method to detect cracks in a motor rotor, the positioning accuracy of the crack position image obtained by time delay compensation and weighted superposition of multiple ultrasonic array sensor signals is improved by about 30% compared to traditional methods. Five ultrasonic array sensors are arranged at different positions of the motor rotor, and the signals after time delay superposition clearly show the position of the crack, with an error reduction of more than 20% compared to traditional detection methods. This experimental result shows that the beam imaging method based on the time delay superposition algorithm can significantly improve the detection accuracy of crack defects.

[0073] Step S40: Construct a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, input a preset crack detection model, and obtain a first crack detection result;

[0074] It should be noted that in step S40, the step of constructing a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, inputting a preset crack detection model, and obtaining a first crack detection result specifically includes:

[0075] Step S401: Construct a fusion matrix according to the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, and perform a two-dimensional convolution operation on the fusion matrix to obtain a primary crack feature;

[0076] Step S402: Introduce the attention mechanism network, input the primary crack feature into the attention mechanism network, calculate the attention weight between the primary crack features to enhance the recognition ability of the crack sensitive feature, and obtain the optimized crack feature;

[0077] Step S403: Perform maximum pooling operation on the optimized crack feature to obtain a crack sensitive feature vector;

[0078] Step S404: Input the crack sensitive feature vector into the fully connected layer of the crack detection model, and process the first crack detection result through the Softmax activation function, the first crack detection result including the crack category, crack position and crack severity.

[0079] It can be understood that the key operation in step S40 is to fuse the crack imaging feature matrix with the multi-directional ultrasonic feature matrix, combine the crack information provided by both, and form a more comprehensive feature matrix as input into the crack detection model. The goal of this step is to improve the accuracy and robustness of crack detection through efficient information fusion. The fusion of the crack imaging feature matrix and the multi-directional ultrasonic feature matrix can integrate information from different ultrasonic array sensors and different perspectives. Through two-dimensional convolution operation, more rich and accurate crack features can be extracted, reducing the information loss or error that may be caused by a single feature matrix. At the same time, the purpose of adding the attention mechanism network is to dynamically adjust the weight of the feature, so as to automatically focus on the part that is crucial to crack positioning, further improving the accuracy of crack detection.

[0080] It should be understood that with the introduction of the attention mechanism network, the crack feature can strengthen those more significant features by calculating the correlation between the features, thereby optimizing the crack recognition process. Through the maximum pooling operation, the system can filter out the most representative crack features from numerous features, thereby avoiding the influence of redundant information on the subsequent classification process. Finally, through the processing of the fully connected layer and the Softmax activation function, the category, position and severity information of the crack are extracted from the original feature to obtain the first crack detection result. This series of operations ensures that the crack detection model can detect with higher accuracy and stronger adaptability.

[0081] For example, in actual experimental tests, when the crack imaging feature matrix is fused with the multi-directional ultrasonic feature matrix, and the primary features are extracted through two-dimensional convolution operation, the detection accuracy of the system is improved by about 25% after introducing the attention mechanism network compared with the scheme without using the mechanism. In addition, when processing crack features, the stability and discrimination of the features are further improved through the maximum pooling operation, and finally the error rate is reduced by 30% in the judgment of crack position, category and severity. These experimental data show that the fusion of different feature matrices and the introduction of the attention mechanism not only improve the detection accuracy, but also enhance the adaptability of the system to complex environments.

[0082] Step S50: error analysis is performed on the first crack detection result, and a second crack detection result is output.

[0083] It should be noted that in step S50, the step of performing error analysis on the first crack detection result and outputting the second crack detection result specifically includes:

[0084] Step S501: obtaining the position coordinate information of the crack from the first crack detection result, and calculating the feedback propagation feature difference between the ultrasonic signals of different ultrasonic array sensors based on the crack position theoretical model according to the position coordinate information of the crack, including: feedback time difference, feedback waveform amplitude difference and feedback phase difference;

[0085] Step S502: backtracking the propagation feature difference in step S20, and calculating the feedback error value according to the propagation feature difference and the feedback propagation feature difference;

[0086] Step S503: setting a first error threshold and a second error threshold, and dividing the crack position confidence level according to the feedback error value:

[0087] If the feedback error value is less than or equal to the first error threshold, the crack position is defined as high confidence;

[0088] If the first error threshold is less than the feedback error value and the second error threshold is less than or equal to the feedback error value, it is defined as medium confidence;

[0089] If the feedback error value is greater than the second error threshold, it is defined as low confidence;

[0090] Step S504: outputting the second crack detection result, including: crack position coordinates, crack type and size, crack detection result confidence level and feedback error value.

[0091] It can be understood that the calculation of the feedback propagation feature difference is based on the crack position coordinates obtained from the first crack detection result, and the propagation characteristics of the signal are inversely deduced through a theoretical model. By calculating the time difference, amplitude attenuation difference and phase difference of the ultrasonic signal, the potential error of the crack position can be identified. According to the feedback difference of these features, we can more accurately evaluate the actual position of the crack and classify its credibility. Specifically, if the feedback error value is small, it means that the signal propagation characteristics are highly consistent with the theoretical model, and the credibility of the crack position is high.

[0092] It should be understood that the calculation of the feedback error value is verified by combining theory and actual data through the way of backtracking to obtain the propagation feature difference. By setting the first error threshold and the second error threshold, the credibility level of the crack position can be classified according to the size of the feedback error value. If the error is small, the crack position has high credibility; otherwise, if the error is large, the credibility of the crack position is low. Ultimately, this process not only improves the accuracy of the crack detection result, but also provides valuable basis for subsequent maintenance decisions.

[0093] For example, in the experiment, we found that by setting the feedback error threshold, we can effectively distinguish the credibility of the crack by calculating the feedback error between the crack position in the first crack detection result and the actual crack position. For example, when the feedback error value is less than the first error threshold, the positioning error of the crack is less than 1mm, and the crack type and size are correctly identified; when the feedback error value is greater than the second error threshold, the error of the crack position is as high as 5mm, at which time the system automatically marks the result as low credibility and prompts further confirmation. Through this method, the crack positioning accuracy in the experiment is improved by 20%, and false judgments can be effectively avoided.

[0094] Embodiment two: In addition, the application provides a motor rotor magnetic sheet defect detection system based on an intelligent sensing system, which adopts the motor rotor magnetic sheet defect detection method based on an intelligent sensing system in the above embodiment, and can solve the technical problem of the motor rotor magnetic sheet defect detection based on an intelligent sensing system. Compared with the prior art, the beneficial effects of the motor rotor magnetic sheet defect detection system based on an intelligent sensing system provided by the application are the same as those of the motor rotor magnetic sheet defect detection method based on an intelligent sensing system provided by the above embodiment, and other technical features of the motor rotor magnetic sheet defect detection system based on an intelligent sensing system are the same as those disclosed in the above embodiment method, which will not be repeated here.

[0095] Embodiment three: The application provides a motor rotor magnetic sheet defect detection device based on an intelligent sensing system, please refer to Figure 2A motor rotor magnetic sheet defect detection device based on an intelligent sensing system includes at least one processor, and a memory connected to the at least one processor in communication. The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a motor rotor magnetic sheet defect detection method based on an intelligent sensing system according to an embodiment. The motor rotor magnetic sheet defect detection device based on an intelligent sensing system according to an embodiment can include, but is not limited to, a mobile terminal such as a mobile phone, a notebook, a digital broadcasting receiver, a PDA (Personal Digital Assistant), a PAD (Portable Application Description), a PMP (Portable Media Player), a car terminal (e.g., a car navigation terminal), and the like, and a stationary terminal such as a digital TV, a desktop computer, and the like. The motor rotor magnetic sheet defect detection device based on an intelligent sensing system is only an example, and should not impose any limitation on the function and use range of the motor rotor magnetic sheet defect detection device based on an intelligent sensing system according to an embodiment. The motor rotor magnetic sheet defect detection device based on an intelligent sensing system can include a processing device 1001 (e.g., a central processing unit, a graphic processing unit, or the like) that can perform various appropriate actions and processes according to a program stored in a ROM (Read Only Memory) 1002 or a program loaded from a storage device 1003 to a RAM (Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the motor rotor magnetic sheet defect detection device based on an intelligent sensing system are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An I / O (Input / Output) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, and the like; an output device 1008 including, for example, an LCD (Liquid Crystal Display), a speaker, a vibrator, and the like; the storage device 1003 including, for example, a magnetic tape, a hard disk, and the like; and a communication device 1009. The communication device 1009 can allow the motor rotor magnetic sheet defect detection device based on an intelligent sensing system to communicate with other devices wirelessly or wiredly to exchange data. Although the motor rotor magnetic sheet defect detection device based on an intelligent sensing system having various systems is illustrated in the drawing, it is understood that all of the illustrated systems are not required to be implemented or provided. More or less systems can be alternatively implemented or provided.

[0096] Embodiment four: the application also provides a computer program product, comprising a computer program which, when executed by a processor, implements the steps of a motor rotor magnetic sheet defect detection method based on an intelligent sensing system as described above. The computer program product provided by the application can solve the technical problem of a motor rotor magnetic sheet defect detection method based on an intelligent sensing system. Compared with the prior art, the beneficial effects of the computer program product provided by the application are the same as those of the motor rotor magnetic sheet defect detection method based on an intelligent sensing system provided by the above-described embodiments, and are not described here in detail.

[0097] In particular, according to the embodiments disclosed by the application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed by the application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by a processing device 1001, the above-mentioned functions defined in the method of the embodiments disclosed by the application are executed.

[0098] It should be understood that various parts of the application disclosed can be realized in hardware, software, firmware or a combination thereof. In the description of the above-described embodiments, specific features, structures, materials or characteristics can be combined in any appropriate manner in any one or more embodiments or examples.

[0099] Obviously, those skilled in the art can make various modifications and variations to the application without departing from the spirit and scope of the application. Thus, if these modifications and variations of the application fall within the scope of the claims of the application and their equivalent technologies, the application also intends to include these modifications and variations.

Claims

1. A method for detecting defects in motor rotor magnetic sheets based on an intelligent sensing system, characterized in that: Methods include: Step S10: Arrange N ultrasonic array sensors around the motor rotor magnetic sheet, synchronously transmit ultrasonic pulse waves at time t, and receive a first ultrasonic echo signal; first perform adaptive Wiener filtering on the first ultrasonic echo signal to obtain a second ultrasonic echo signal, and then perform Hilbert transform on the second ultrasonic echo signal to obtain a filtered signal envelope curve; Step S20: Extracting the propagation characteristic differences of ultrasonic signals between different ultrasonic array sensors based on the filtered signal envelope curve, including time difference, waveform exponential attenuation difference, and phase difference; performing a fast Fourier transform on the propagation characteristic differences to obtain crack-sensitive frequency domain features, and constructing a multi-directional ultrasonic feature matrix based on the propagation characteristic differences and the crack-sensitive frequency domain features; Step S30: using a beamforming method based on a time-delayed superposition algorithm to locate crack defects and output a crack imaging feature matrix; Step S40: constructing a fusion matrix based on the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, inputting a preset crack detection model, and obtaining a first crack detection result; Step S50: performing error analysis on the first crack detection result and outputting a second crack detection result. The step of performing error analysis on the first crack detection result and outputting the second crack detection result specifically includes: Step S501: obtaining crack location coordinate information from the first crack detection result, and calculating feedback propagation characteristic differences of ultrasonic signals between different ultrasonic array sensors based on the crack location coordinate information and a crack location theoretical model, including feedback time difference, feedback waveform attenuation difference, and feedback phase difference. Step S502: Retrospectively obtain the propagation characteristic difference in step S20, and calculate the feedback error value based on the propagation characteristic difference and the feedback propagation characteristic difference; Step S503: Set a first error threshold and a second error threshold, and divide the crack position credibility level according to the feedback error value: If the feedback error value is less than or equal to the first error threshold, the crack position is defined as having high confidence; If the first error threshold < feedback error value ≤ second error threshold, it is defined as medium credibility; If the feedback error value is greater than the second error threshold, it is defined as low confidence; Step S504: outputting the second crack detection result, including: crack position coordinates, crack type and size, reliability level of the crack detection result and feedback error value.

2. The method for detecting defects in motor rotor magnetic sheets based on an intelligent sensing system according to claim 1, wherein: In step S20, the time difference is calculated using a weighted cross-correlation function, expressed as: in, is the time difference between ultrasonic array sensor i and ultrasonic array sensor j, w(t) is the weighted window function, is the time difference displacement variable, 、 They are respectively the ultrasonic array sensor i at time The signal and ultrasonic array sensor j at time signal, argmax is the weighted cross-correlation function; the waveform modulus attenuation difference ,in, and are the wave form values ​​of ultrasonic array sensors i and j respectively; the phase difference ,in 、 are the waveform phases of ultrasonic array sensors i and j respectively.

3. The method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to claim 1, wherein: In step S30, the steps of locating crack defects using a beamforming method based on a time-delayed superposition algorithm and outputting a crack imaging feature matrix specifically include: The beam imaging method based on the time-delay superposition algorithm is used to calculate the two-dimensional spatial position of the rotor magnetic piece of the motor to be tested. Ultrasonic imaging intensity at: in, is the two-dimensional spatial position of the motor rotor magnetic piece to be tested Ultrasonic imaging intensity at For the a second ultrasonic echo signal received by an ultrasonic array sensor; The position to be measured To The geometric distance between the ultrasonic array sensors; is the propagation speed of ultrasonic waves in the rotor magnet; For the The weighting coefficient of each ultrasonic array sensor is determined according to the position of the ultrasonic array sensor; According to the two-dimensional spatial position of the rotor magnetic sheet of the motor to be tested The crack imaging feature matrix is ​​constructed based on the ultrasonic imaging intensity at .

4. The method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to claim 1, wherein: In step S40, a fusion matrix is ​​constructed based on the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, and a preset crack detection model is input to obtain a first crack detection result, which specifically includes: Step S401: constructing a fusion matrix based on the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, and performing a two-dimensional convolution operation on the fusion matrix to obtain primary crack features; Step S402: introducing an attention mechanism network, inputting primary crack features into the attention mechanism network, and calculating attention weights between primary crack features to enhance the recognition capability of crack-sensitive features, thereby obtaining optimized crack features; Step S403: performing a maximum pooling operation on the optimized crack features to obtain a crack sensitive feature vector; Step S404: Input the crack sensitive feature vector into the fully connected layer of the crack detection model, and process it through the Softmax activation function to obtain a first crack detection result, which includes the crack type, crack location and crack severity.

5. The method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to claim 4, wherein: In step S40, the attention mechanism network Using the formula: Where, The query matrix, key matrix, and value matrix of the attention mechanism network are constructed based on the primary crack features; is the dimension of the feature matrix, which is used to normalize the scale of the feature vector; is the matrix transpose symbol; is the activation function.

6. The method for detecting defects in a motor rotor magnetic sheet based on an intelligent sensing system according to claim 1, wherein: In step S50, the feedback error value is calculated based on the propagation characteristic difference and the feedback propagation characteristic difference, using the formula: in, 、 and are the nth ultrasonic array sensor Time difference, Amplitude attenuation, Phase difference; 、 and The propagation characteristic difference of the nth ultrasonic array sensor includes the time difference, the wave form attenuation difference and the phase difference.

7. A motor rotor magnetic sheet defect detection system based on an intelligent sensing system, applied to a motor rotor magnetic sheet defect detection method based on an intelligent sensing system according to any one of claims 1 to 6, characterized in that: The motor rotor magnetic sheet defect detection system based on the intelligent sensing system includes: An envelope curve generation module is configured to arrange N ultrasonic array sensors around the motor rotor magnetic sheet, synchronously transmit ultrasonic pulse waves at time t, and receive a first ultrasonic echo signal; firstly perform an adaptive Wiener filter on the first ultrasonic echo signal to obtain a second ultrasonic echo signal; then perform a Hilbert transform on the second ultrasonic echo signal to obtain an envelope curve of the filtered signal; A multi-directional feature matrix construction module is used to extract the propagation characteristic differences of ultrasonic signals between different ultrasonic array sensors based on the filtered signal envelope curve, including time difference, waveform exponential attenuation difference, and phase difference. The propagation characteristic differences are subjected to fast Fourier transform to obtain crack-sensitive frequency domain characteristics. A multi-directional ultrasonic feature matrix is ​​constructed based on the propagation characteristic differences and the crack-sensitive frequency domain characteristics. A crack imaging feature matrix construction module is used to locate crack defects using a beam imaging method based on a time-delay superposition algorithm and output a crack imaging feature matrix; A first crack detection result output module is used to construct a fusion matrix based on the crack imaging feature matrix and the multi-directional ultrasonic feature matrix, input a preset crack detection model, and obtain a first crack detection result; The second crack detection result output module is used to perform error analysis on the first crack detection result and output the second crack detection result. The steps of performing error analysis on the first crack detection result and outputting the second crack detection result specifically include: Obtaining crack location coordinate information from the first crack detection result, and calculating feedback propagation characteristic differences of ultrasonic signals between different ultrasonic array sensors based on the crack location coordinate information and a crack location theoretical model feedback, including feedback time difference, feedback waveform attenuation difference, and feedback phase difference; Retrospectively obtain the propagation characteristic difference in step S20, and calculate the feedback error value based on the propagation characteristic difference and the feedback propagation characteristic difference; Set the first error threshold and the second error threshold, and divide the crack position credibility level according to the feedback error value: If the feedback error value is less than or equal to the first error threshold, the crack position is defined as having high confidence; If the first error threshold < feedback error value ≤ second error threshold, it is defined as medium credibility; If the feedback error value is greater than the second error threshold, it is defined as low confidence; Output the second crack detection result, including: crack location coordinates, crack type and size, crack detection result credibility level and feedback error value.

8. A motor rotor magnetic sheet defect detection device based on an intelligent sensing system, characterized in that: The motor rotor magnetic sheet defect detection device based on the intelligent sensing system includes: a memory, a processor, and a motor rotor magnetic sheet defect detection program based on the intelligent sensing system stored in the memory and executable on the processor. When the motor rotor magnetic sheet defect detection program based on the intelligent sensing system is executed by the processor, a motor rotor magnetic sheet defect detection method based on the intelligent sensing system according to any one of claims 1 to 6 is implemented.

9. A computer program product, characterized in that The computer program product includes a motor rotor magnetic sheet defect detection program based on an intelligent sensing system. When the motor rotor magnetic sheet defect detection program based on an intelligent sensing system is executed by a processor, a motor rotor magnetic sheet defect detection method based on an intelligent sensing system according to any one of claims 1 to 6 is implemented.

Images