Intelligent crankshaft machining online detection system and method

Abstract

The invention relates to the technical field of crankshaft online detection, in particular to an intelligent crankshaft machining online detection system and method. The method comprises the steps that physical parameter data in the crankshaft machining process are obtained in real time to serve as original signals, and the original signals are converted into electronic signals through a multi-level self-adaptive signal conversion and enhancement algorithm; the electronic signals are preprocessed, and finally collected physical parameter data are obtained; analyzing the finally acquired physical parameter data by using a multi-level self-adaptive hybrid neural network analysis algorithm, and comparing standard specifications to obtain an evaluation result of the crankshaft processing quality; and an adjustment suggestion is generated based on the evaluation result, a machining parameter adjustment instruction of the machining equipment is generated based on the adjustment suggestion, and the machining parameter adjustment instruction is fed back to the machining equipment. The problems that in the crankshaft machining online detection process, crankshaft physical parameters are not accurately obtained, and the detection and analysis accuracy is not high are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of crankshaft online detection, and in particular to an intelligent crankshaft processing online detection system and method. Background Art

[0002] In modern manufacturing, crankshafts are one of the important components of transportation engines and other mechanical equipment. Their processing quality directly affects the performance and life of the equipment. The traditional crankshaft processing includes multiple process links such as turning, grinding, and drilling, involving multiple physical parameters and operation steps. These processes require high-precision control, but in actual production, due to the influence of various factors during the processing, such as material properties, equipment status, operator experience differences, changes in ambient temperature, etc., it often leads to instability in processing accuracy, thus affecting the quality of the final product.

[0003] As the manufacturing industry's requirements for precision machining increase, traditional manual control and experience-based judgment methods can no longer meet the needs of modern crankshaft machining. Traditional machining methods usually rely on manual measurement and adjustment, which has a large possibility of manual errors and improper operation. In addition, when faced with complex machining processes and a large amount of real-time data, the processing capacity and response speed of traditional methods are also limited. Therefore, an intelligent crankshaft machining detection method is urgently needed. Summary of the invention

[0004] The present invention provides an intelligent crankshaft machining online detection system and method to solve the problems of inaccurate acquisition of crankshaft physical parameters and low detection and analysis accuracy during crankshaft machining online detection.

[0005] The present invention provides an intelligent crankshaft processing online detection system and method, which specifically includes the following technical solutions: An intelligent crankshaft processing online detection method comprises the following steps: S1. Real-time acquisition of physical parameter data during crankshaft machining as raw signals, and conversion of raw signals into electronic signals using a multi-level adaptive signal conversion and enhancement algorithm; pre-processing of electronic signals to obtain the final collected physical parameter data; S2. Analyze the physical parameter data finally collected using a multi-level adaptive hybrid neural network analysis algorithm, and obtain the evaluation results of the crankshaft processing quality by comparing with standard specifications; generate adjustment suggestions based on the evaluation results, generate processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and feed back the processing parameter adjustment instructions to the processing equipment.

[0006] Preferably, the S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the original signal is standardized to obtain a standardized signal, and the standardized signal is processed by a multi-dimensional convolution filter to obtain an output signal processed by the multi-dimensional convolution filter.

[0007] Preferably, the S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the output signal processed by the multi-dimensional convolution filter is subjected to nonlinear enhancement processing through a composite function. The specific formula is: , in, is the signal after nonlinear enhancement processing; is the dynamic gain factor; is the dynamic activation rate factor; It's in time Output signal after processing by multi-dimensional convolution filter; is a nonlinear power exponent; is the dynamic logarithmic adjustment factor.

[0008] Preferably, the S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the amplitude and characteristics of the signal after nonlinear enhancement processing are adjusted through adaptive reconstruction to obtain the adaptively reconstructed signal; the adaptive reconstruction formula is as follows: , in, is the adaptively reconstructed signal; is the adjustment factor for the rate of change of the control signal; is the instantaneous rate of change of the signal after nonlinear enhancement processing; It is a sensitivity adjustment factor, which is used to adjust the nonlinear effect of the instantaneous change rate of the signal after nonlinear enhancement processing on the final signal amplitude.

[0009] Preferably, the S1 specifically includes: The amplitude correction mechanism is introduced to adjust the amplitude of the adaptively reconstructed signal to obtain a corrected signal, and the corrected signal is converted into an electronic signal through digital processing; the formula of the amplitude correction mechanism is: , in, is the corrected signal; is the second-order derivative of the adaptively reconstructed signal, which represents the acceleration of the adaptively reconstructed signal; is the correction rate adjustment factor; is the corrected acceleration adjustment factor.

[0010] Preferably, the S2 specifically includes: In the process of implementing the multi-level adaptive hybrid neural network analysis algorithm, the physical parameter data finally collected is normalized to obtain normalized data; the normalized data is subjected to dimensionality reduction to obtain reduced dimensionality data, and the reduced dimensionality data is analyzed by the hybrid neural network.

[0011] Preferably, the S2 specifically includes: The hybrid neural network combines the convolutional neural network and the long short-term memory network. The convolutional neural network is used to extract the local features of the data after dimensionality reduction, and the final output of the convolutional neural network is input into the long short-term memory network.

[0012] Preferably, the S2 specifically includes: Based on the outputs of the convolutional neural network and the long short-term memory network, an adaptive weighting mechanism is introduced to fuse the outputs of the two networks to obtain the final fusion result; based on the final fusion result and the standard specification data, the error of the processing quality is calculated; based on the error and the set threshold, the quality level is divided to obtain the evaluation result.

[0013] An intelligent crankshaft processing online detection system includes the following parts: Intelligent sensor module, data acquisition module, data depth analysis module, real-time feedback module, intelligent control module; The intelligent sensor module senses the physical parameter data during the crankshaft processing in real time, and then converts the physical parameter data into electronic signals through a multi-level adaptive signal conversion and enhancement algorithm; and sends the electronic signals to the data acquisition module; The data acquisition module pre-processes the electronic signal to obtain the final collected physical parameter data, and sends the final collected physical parameter data to the data depth analysis module; The data in-depth analysis module uses a multi-level adaptive hybrid neural network analysis algorithm to analyze the physical parameter data finally collected, and outputs the evaluation results of the processing quality by comparing with the standard specifications, and sends the evaluation results to the real-time feedback module; A real-time feedback module determines whether the processing technology needs to be adjusted based on the evaluation results, generates adjustment suggestions based on the evaluation results, generates processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and sends the processing parameter adjustment instructions to the intelligent control module; The intelligent control module adjusts the processing equipment based on the processing parameter adjustment instruction and in combination with the processing parameter limitation during the adjustment process of the processing equipment.

[0014] The beneficial effects of the technical solution of the present invention are: 1. By integrating multiple sensor devices such as laser sensors, vibration sensors, temperature sensors and roughness sensors, comprehensive real-time perception of various physical parameter data in the crankshaft processing process, such as size, form and position tolerances, surface roughness, vibration, temperature, etc., is achieved. This multi-dimensional perception can ensure the accurate capture of all key variables during the processing process and provide high-quality original signal input for subsequent data processing. By applying multi-level adaptive signal conversion and enhancement algorithms, the original signal can be pre-processed to reduce noise interference and improve signal quality. In addition, multi-dimensional convolution filtering can effectively extract key information from the signal and optimize the signal quality, laying the foundation for the next step of data analysis.

[0015] 2. By introducing principal component analysis for dimensionality reduction, redundant information is reduced and the most representative features are retained, making subsequent analysis more efficient and accurate. By introducing a multi-level adaptive hybrid neural network analysis algorithm, the processing quality can be accurately evaluated and reliable adjustment suggestions can be generated, thereby improving the stability and controllability of the processing quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a structural diagram of an intelligent crankshaft processing online detection system according to the present invention; Figure 2 This is a flow chart of an intelligent crankshaft processing online detection method described in the present invention. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The specific scheme of an intelligent crankshaft processing online detection system and method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] See attached Figure 1 , which shows a structure diagram of an intelligent crankshaft processing online detection system provided by an embodiment of the present invention, the system includes the following parts: Intelligent sensor module, data acquisition module, data depth analysis module, real-time feedback module, intelligent control module; The intelligent sensor module senses various physical parameter data in the crankshaft processing process in real time, and then converts the physical parameter data into electronic signals through multi-level adaptive signal conversion and enhancement algorithms; and sends the electronic signals to the data acquisition module; Physical parameter data include size, geometric tolerance, surface roughness, vibration, temperature, etc.; The data acquisition module performs pre-processing such as noise reduction and cleaning on the electronic signal to obtain the final collected physical parameter data, and sends the final collected physical parameter data to the data depth analysis module; The data in-depth analysis module uses a multi-level adaptive hybrid neural network analysis algorithm to conduct in-depth analysis on the physical parameter data finally collected, and outputs the evaluation results of the processing quality by comparing with the standard specifications, and sends the evaluation results to the real-time feedback module; A real-time feedback module determines whether the processing technology needs to be adjusted based on the evaluation results, generates adjustment suggestions based on the evaluation results, generates processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and sends the processing parameter adjustment instructions to the intelligent control module; The intelligent control module automatically adjusts the processing equipment based on the processing parameter adjustment instructions and in combination with the processing parameter limitations during the adjustment process of the processing equipment.

[0021] Refer to the attached Figure 2 , which shows a flow chart of an intelligent crankshaft processing online detection method provided by an embodiment of the present invention, the method comprising the following steps: S1. Real-time acquisition of physical parameter data during crankshaft machining as raw signals, and conversion of raw signals into electronic signals using a multi-level adaptive signal conversion and enhancement algorithm; preprocessing of electronic signals to obtain the final collected physical parameter data; During the crankshaft machining process, in order to sense various physical parameter data, including size, form and position tolerance, surface roughness, vibration, temperature, etc., sensors such as laser sensors, vibration sensors, temperature sensors, roughness sensors, etc. are used to sense and obtain the physical parameter data during the crankshaft machining process as the original signal.

[0022] Furthermore, the original signal is converted into an electronic signal using a multi-level adaptive signal conversion and enhancement algorithm; the multi-level adaptive signal conversion and enhancement algorithm enhances and converts the original signal through multi-level convolution filtering, nonlinear enhancement, dynamic correction and other processing, and makes the original signal adapt to changes in the processing process. The specific implementation process is as follows: First, the physical parameter data as the original signal Perform standardization to obtain a standardized signal Through standardization, the deviation and noise of the original signal are reduced, and the original signal is converted to a unified scale for subsequent processing.

[0023] Furthermore, in order to better extract key information from the standardized signal, a multi-dimensional convolution filter is used; the multi-dimensional convolution filter not only works in the time domain, but also considers the frequency and space dimensions. The goal of the multi-dimensional convolution filter is to process the standardized signal through multiple convolution kernels to obtain its data information in different dimensions. It can be expressed as: , in, It's in time The output signal after multi-dimensional convolution filter processing represents the output signal of different convolution kernels. For input signal The results are weighted and integrated; Indicates at time Normalized signal; is the number of convolution kernels; It's in time Upper delay After convolution kernel, convolution kernel Represents a specific mode of signal filtering, such as the spectral response or time-localized characteristics of the filter, determined by expert experience; It is the spatial / temporal coherence coefficient, which is used to weight different parts of the input signal and reflects the coherence degree of the input signal at different time points. It is related to the noise characteristics of the input signal or specific system characteristics and is used to strengthen or weaken the impact of the input signal at certain time points. is the time-integrated variable.

[0024] In order to further improve the quality and availability of the signal, the output signal processed by the multi-dimensional convolution filter is enhanced through nonlinear enhancement and adaptive reconstruction to enhance the details and adapt to the changes that may occur during the processing. Nonlinear enhancement is achieved by introducing a composite function. The nonlinear enhancement process combines exponential functions, power functions, and logarithmic functions in order to improve the signal's detail changes and adjust the signal's response through an adaptive mechanism. The specific formula is: , in, is the signal after nonlinear enhancement processing; It is the dynamic gain factor, which represents the dynamic adjustment factor of the signal gain during the enhancement process. It is dynamically adjusted according to the time-varying characteristics of the signal, environmental changes, etc., and is used to control the intensity of signal enhancement; , is the initial gain factor; is the gain adjustment factor, which indicates the effect of the signal change rate on the gain factor adjustment. is the rate of change of the output signal after processing by the multidimensional convolution filter over time; It is a dynamic activation rate factor, which is used to control the response rate of the signal when it passes through a nonlinear activation function (Sigmoid function), and reflects the degree of change of the signal; ,in, is the initial activation rate, is the rate adjustment factor, is a small constant that prevents the denominator from being zero; It is a nonlinear power exponent, which is used to adjust the power to make the signal enhancement more accurate. It is a nonlinear factor in the enhancement process. By adjusting it, the nonlinear degree of enhancement can be controlled. is the dynamic logarithmic adjustment factor, which is used to control the sensitivity of the signal amplitude adjustment in the logarithmic transformation; ,in, and Respectively represent the variance and mean of the output signal after processing by the multi-dimensional convolution filter, is the logarithmic adjustment factor, is the initial logarithmic adjustment factor.

[0025] Furthermore, in order to make the signal more in line with the actual needs of the processing equipment, the amplitude and detail characteristics of the signal after nonlinear enhancement processing are adjusted through adaptive reconstruction technology. This process automatically adjusts the gain of the signal according to the changes of the signal after nonlinear enhancement processing at different time points, so that the signal can respond to small changes in the processing process in real time.

[0026] The adaptive reconstruction formula is as follows: , in, is the adaptively reconstructed signal; It is the adjustment factor that controls the rate of change of the signal, which indicates the sensitivity of the signal after nonlinear enhancement processing to time changes, that is, the degree of influence of the instantaneous rate of change of the signal after nonlinear enhancement processing on the final signal amplitude; It is the instantaneous rate of change of the signal after nonlinear enhancement processing, indicating the speed of change of the signal after nonlinear enhancement processing; It is a sensitivity adjustment factor, which is used to adjust the nonlinear effect of the instantaneous change rate of the signal after nonlinear enhancement processing on the final signal amplitude, and controls the degree of amplification or compression of the instantaneous change rate of the signal after nonlinear enhancement processing. Through this process, the amplitude of the adaptively reconstructed signal can be automatically adjusted in time to ensure its stability and sensitivity during the processing. The core purpose of adaptive reconstruction is to enable the adaptively reconstructed signal to adapt to different processing conditions and provide high-quality feedback.

[0027] Furthermore, in order to make the adaptively reconstructed signal more suitable for subsequent analysis and acquisition processes, a variation amplitude correction mechanism is introduced to adjust the amplitude of the adaptively reconstructed signal according to the dynamic change characteristics of the signal. The core purpose of the variation amplitude correction mechanism is to reduce the interference of high-frequency noise on the adaptively reconstructed signal and dynamically adjust its amplitude when the adaptively reconstructed signal changes rapidly, so as to maintain the stability and availability of the adaptively reconstructed signal. The formula of the variation amplitude correction mechanism is: , in, is the corrected signal; is the second-order derivative of the adaptively reconstructed signal, which represents the acceleration of the adaptively reconstructed signal; is the correction rate adjustment factor, which is used to control the rate of correction of the signal after adaptive reconstruction; is the correction acceleration adjustment factor, which is used to control the acceleration rate of amplitude correction. Through the above formula, the adaptively reconstructed signal is corrected in the area with large acceleration, thereby avoiding the drastic fluctuation of the adaptively reconstructed signal and improving the accuracy of the signal.

[0028] Finally, in order to convert the corrected signal into an electronic signal format for subsequent analysis and feedback, the following digital processing is used for the final conversion: , in, It is the digital signal value, that is, the electronic signal, which represents the time The signal that has been sampled and quantized at all times is the signal format used for subsequent processing and control systems; The corrected signal The mean of The corrected signal The standard deviation of It is the resolution of the ADC (Analog-to-Digital Converter) used in analog-to-digital processing, which represents the number of bits of the digital signal, that is, the accuracy of the smallest unit that can be expressed when converting the analog signal to a digital signal. Through the above formula, the amplitude of the corrected signal is scaled to the corresponding ADC input range, so that it can be collected and processed in digital format.

[0029] Furthermore, the electronic signal is subjected to a preprocessing process such as noise reduction and cleaning to obtain the final collected physical parameter data. The preprocessing process adopts existing technical means and will not be described in detail here.

[0030] S2. Analyze the physical parameter data finally collected by using a multi-level adaptive hybrid neural network analysis algorithm, and obtain the evaluation results of the crankshaft processing quality by comparing with standard specifications; generate adjustment suggestions based on the evaluation results, generate processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and feed back the processing parameter adjustment instructions to the processing equipment.

[0031] The multi-level adaptive hybrid neural network analysis algorithm is used to deeply analyze the physical parameter data finally collected during the crankshaft processing process, and the evaluation results are obtained by comparing with the standard specifications. Based on the evaluation results, accurate adjustment suggestions are generated. The specific implementation process is as follows: First, the physical parameter data finally collected is normalized to eliminate the differences caused by different physical quantity dimensions, so that all data can be compared under the same dimension to obtain normalized data. , and further apply the existing principal component analysis (PCA) technology to reduce the dimension of the normalized data to obtain the reduced-dimensional data ; The reduced-dimensional data is further analyzed by a hybrid neural network, which combines two neural networks: a convolutional neural network (CNN) and a long short-term memory network (LSTM). CNN is used to extract local features of the reduced-dimensional data, and LSTM is used to capture the long-term dependencies of the output data of the convolutional neural network.

[0032] The main function of convolutional neural network (CNN) is to extract local features of data, which is especially suitable for analyzing instantaneous changes in crankshaft machining. The characteristics of are calculated as follows: , in, It is The output of the convolutional neural network layer represents the data processed by this layer and passed to the next layer; It is The weight matrix of a layer represents the connection strength or weight coefficient between the neurons in this layer and the neurons in the previous layer; It is The output of the layer; in addition, ; Represents the convolution operation; It is The bias term of the layer; is an activation function, such as ReLU. The function of a convolutional neural network is to gradually extract local features by sliding the convolution kernel over the input data.

[0033] Convolutional neural network after Output after layer processing is passed to LSTM, which is used to capture long-term dependencies. Recorded as ; The update formula of LSTM is as follows: , , , , , , in, It is the current moment The activation value of the forget gate; is the hidden state output at the previous moment; is the weight matrix corresponding to the forget gate, connecting the input and the hidden state of the previous moment to the forget gate; is the bias term of the forget gate; is the activation value of the input gate; is the weight matrix of the input gate, connecting the input and the hidden state of the previous moment to the input gate; is the bias term of the input gate; is a candidate memory unit; is the weight matrix of the candidate memory unit, connecting the input and the hidden state of the previous moment to the candidate memory unit; is the bias term of the candidate memory unit; is the state of the memory unit at the current moment; It was the last moment The state of the memory unit; is the activation value of the output gate; is the weight matrix of the output gate, connecting the input and the hidden state of the previous moment to the output gate; is the bias term of the output gate; It is the hidden state at the current moment, representing the network’s output of input and memory at the current moment. LSTM controls the information flow through a gating mechanism to ensure that long-term dependencies are effectively captured.

[0034] Based on the output of CNN and LSTM, an adaptive weighting mechanism is introduced to fuse the feature outputs of the two networks to obtain the final fusion result. , the calculation formula is: , in, is the adaptive weight factor.

[0035] Next, the error in processing quality is calculated by comparing with the standard specifications. .error The calculation formula is: , in, It is The final fusion result of samples, wherein the samples are crankshaft samples detected online; It is the standard normative data set according to the expert experience method; is the number of samples; error The mean squared error (MSE) is used to measure the difference between the predicted results and the actual standard.

[0036] Based on error , the machining quality of the crankshaft is evaluated. The expert experience method is used to set multiple thresholds to divide different quality levels and obtain the evaluation results, which are as follows: when , it means that the processing quality is very good, close to the standard specification, and belongs to the excellent grade; when When , it means that the processing quality is good, but there is still a certain deviation, which belongs to the good grade; when When , it means that the processing quality is average, with obvious deviation, and belongs to the general grade; when When , it means that the processing quality is poor and needs to be adjusted and improved, which is a poor grade.

[0037] Adjustment suggestions are automatically generated based on the evaluation results, including: equipment adjustment, such as adjusting the machine tool's rotation speed, feed speed, cutting depth, etc.; process adjustment, such as optimizing the selection of cutting tools, adjusting the flow of cutting fluid, using more efficient tool materials, etc.; process monitoring, increasing the frequency of real-time monitoring data collection to ensure that abnormal changes in the processing process can be captured in a timely manner.

[0038] Finally, based on the generated adjustment suggestions, the online detection system will automatically generate the processing parameter adjustment instructions for the processing equipment and feed back the processing parameter adjustment instructions to the processing equipment. The format of the adjustment instructions includes the specific operation instructions of the equipment, the adjustment range, the target parameters, etc. For example: Equipment adjustment instructions: "Adjust the machine tool speed from 3000 RPM to 3300 RPM, and the feed speed from 0.5 mm / s to 0.475 mm / s." Process monitoring instructions: "Increase the tool monitoring frequency to collect data every 5 minutes." Processing parameter adjustment instructions will be transmitted directly to the processing equipment through the existing industrial control system to ensure that the equipment can make adaptive adjustments based on the evaluation results.

[0039] In summary, an intelligent crankshaft machining online detection system and method have been completed.

[0040] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0041] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0042] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An intelligent crankshaft processing online detection method, characterized in that: The following steps are involved: S1. Real-time acquisition of physical parameter data during crankshaft machining as raw signals, and conversion of raw signals into electronic signals using a multi-level adaptive signal conversion and enhancement algorithm; pre-processing of electronic signals to obtain the final collected physical parameter data; S2. Analyze the physical parameter data finally collected using a multi-level adaptive hybrid neural network analysis algorithm, and obtain the evaluation results of the crankshaft processing quality by comparing with standard specifications; generate adjustment suggestions based on the evaluation results, generate processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and feed back the processing parameter adjustment instructions to the processing equipment.

2. The intelligent crankshaft processing online detection method according to claim 1 is characterized in that: The S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the original signal is standardized to obtain a standardized signal, and the standardized signal is processed by a multi-dimensional convolution filter to obtain an output signal processed by the multi-dimensional convolution filter.

3. The intelligent crankshaft processing online detection method according to claim 2 is characterized in that: The S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the output signal processed by the multi-dimensional convolution filter is subjected to nonlinear enhancement processing through a composite function. The specific formula is: ， in, is the signal after nonlinear enhancement processing; is the dynamic gain factor; is the dynamic activation rate factor; It's in time Output signal after processing by multi-dimensional convolution filter; is a nonlinear power exponent; is the dynamic logarithmic adjustment factor.

4. The intelligent crankshaft processing online detection method according to claim 3 is characterized in that: The S1 specifically includes: In the process of implementing the multi-level adaptive signal conversion and enhancement algorithm, the amplitude and characteristics of the signal after nonlinear enhancement processing are adjusted through adaptive reconstruction to obtain the adaptively reconstructed signal; the adaptive reconstruction formula is as follows: ， in, is the adaptively reconstructed signal; is the adjustment factor for the rate of change of the control signal; is the instantaneous rate of change of the signal after nonlinear enhancement processing; It is a sensitivity adjustment factor, which is used to adjust the nonlinear effect of the instantaneous change rate of the signal after nonlinear enhancement processing on the final signal amplitude.

5. The intelligent crankshaft processing online detection method according to claim 4 is characterized in that: The S1 specifically includes: The amplitude correction mechanism is introduced to adjust the amplitude of the adaptively reconstructed signal to obtain a corrected signal, and the corrected signal is converted into an electronic signal through digital processing; the formula of the amplitude correction mechanism is: ， in, is the corrected signal; is the second-order derivative of the adaptively reconstructed signal, which represents the acceleration of the adaptively reconstructed signal; is the correction rate adjustment factor; is the corrected acceleration adjustment factor.

6. The intelligent crankshaft processing online detection method according to claim 1 is characterized in that: The S2 specifically includes: In the process of implementing the multi-level adaptive hybrid neural network analysis algorithm, the physical parameter data finally collected is normalized to obtain normalized data; the normalized data is subjected to dimensionality reduction to obtain reduced dimensionality data, and the reduced dimensionality data is analyzed by the hybrid neural network.

7. The intelligent crankshaft processing online detection method according to claim 6 is characterized in that: The S2 specifically includes: The hybrid neural network combines the convolutional neural network and the long short-term memory network. The convolutional neural network is used to extract the local features of the data after dimensionality reduction, and the final output of the convolutional neural network is input into the long short-term memory network.

8. The intelligent crankshaft processing online detection method according to claim 7 is characterized in that: The S2 specifically includes: Based on the outputs of the convolutional neural network and the long short-term memory network, an adaptive weighting mechanism is introduced to fuse the outputs of the two networks to obtain the final fusion result; based on the final fusion result and the standard specification data, the error of the processing quality is calculated; based on the error and the set threshold, the quality level is divided to obtain the evaluation result.

9. An intelligent crankshaft processing online detection system, applied to the intelligent crankshaft processing online detection method as claimed in claim 1, characterized in that: Includes the following parts: Intelligent sensor module, data acquisition module, data depth analysis module, real-time feedback module, intelligent control module; The intelligent sensor module senses the physical parameter data during the crankshaft processing in real time, and then converts the physical parameter data into electronic signals through a multi-level adaptive signal conversion and enhancement algorithm; the electronic signals are sent to the data acquisition module; The data acquisition module pre-processes the electronic signal to obtain the final collected physical parameter data, and sends the final collected physical parameter data to the data depth analysis module; The data in-depth analysis module uses a multi-level adaptive hybrid neural network analysis algorithm to analyze the physical parameter data finally collected, and outputs the evaluation results of the processing quality by comparing with the standard specifications, and sends the evaluation results to the real-time feedback module; A real-time feedback module determines whether the processing technology needs to be adjusted based on the evaluation results, generates adjustment suggestions based on the evaluation results, generates processing parameter adjustment instructions for the processing equipment based on the adjustment suggestions, and sends the processing parameter adjustment instructions to the intelligent control module; The intelligent control module adjusts the processing equipment based on the processing parameter adjustment instruction and in combination with the processing parameter limitation during the adjustment process of the processing equipment.

Images