Infrared-spectroscopy-based method for monitoring temperature-induced deformation of component of intelligent moxibustion robot

Abstract

An infrared-spectroscopy-based method for monitoring temperature-induced deformation of a component of an intelligent moxibustion robot. The method comprises: collecting an infrared spectrum and a frequency spectrum of each detection position on a target component of a moxibustion robot; on the basis of frequency differences between peaks and troughs in the frequency spectrum, constructing a motion influence confidence factor of each detection position; using a box-counting method to acquire a scale relationship graph of the infrared spectrum of each detection position; on the basis of scale relationship graphs of all the detection positions, determining an overall absorbance difference index of the target component; on the basis of the overall absorbance difference index and the motion influence confidence factor, using a local outlier factor (LOF) outlier detection algorithm to calculate an LOF of each detection position in a thermal data sequence; and on the basis of a thermal alarm threshold value, determining a temperature-induced deformation risk of the target component. The method can eliminate spectral difference variations caused by shadows and reflectance properties, thereby improving the precision of determining high-temperature deformation of components.

Description

Intelligent moxibustion robot component temperature deformation monitoring method based on infrared spectrum TECHNICAL FIELD

[0001] The present application relates to the field of infrared spectrum processing technology, specifically to an intelligent moxibustion robot component temperature deformation monitoring method based on infrared spectrum. BACKGROUND

[0002] Intelligent moxibustion robots combine traditional moxibustion therapy with modern intelligent technology, representing the trend of innovation in the medical field by integrating tradition and modernity. This device not only individualizes the user's body conditioning and improves treatment effectiveness, but also monitors the user's physiological parameters in real time through intelligent sensing technology, providing scientific health advice to users. Its automation not only improves medical efficiency and reduces the burden on medical personnel, but also provides patients with the possibility of self-treatment at home. In addition, the development of intelligent moxibustion robots promotes the application of traditional Chinese medicine in modern medicine, making traditional therapy more in line with modern people's needs, while also opening up market space for the health industry. Therefore, intelligent moxibustion robots not only have important clinical significance, but also have broad prospects in business, bringing a new direction for the development of the medical field.

[0003] However, due to improper design, excessive workload, failure damage, or external environment, etc., the moxibustion robot parts may overheat and deform. Unreasonable design and material selection, long-term high-load operation, or failure damage can all cause temperature abnormalities. Infrared spectrum plays a key role in detecting high-temperature deformation of moxibustion robot components. By measuring the infrared radiation of the target component, its surface temperature distribution can be obtained. However, due to the different infrared radiation characteristics of different materials, if the infrared radiation characteristics of the target component material itself do not match the expected, it will affect the accuracy of the detection results. Moreover, since the moxibustion robot is in a continuous working state during detection, shadows will be produced on the detected area due to the robot's movements, which will affect the measurement of the infrared spectrum graph, making the detection results inaccurate.

[0004] SUMMARY

[0005] To solve the above technical problems, the present application provides an intelligent moxibustion robot component temperature deformation monitoring method based on infrared spectrum to solve the existing problems.

[0006] The intelligent moxibustion robot component temperature deformation monitoring method based on infrared spectrum of the present application adopts the following technical scheme:

[0007] One embodiment of the present application provides an intelligent moxibustion robot component temperature deformation monitoring method based on infrared spectrum, which includes the following steps:

[0008] Collecting infrared spectrum graphs of each detection position of the moxibustion robot target component;

[0009] obtain a frequency spectrum of the infrared spectrum graph by using a fast Fourier transform; output all the peaks and troughs in the frequency spectrum by using a second-order difference discrimination method; determine a peak width judgment value of each peak based on a frequency difference between adjacent troughs on left and right sides of each peak in the frequency spectrum; determine an interference peak discrimination factor of each detection position by combining the peak width judgment value and an amplitude value of each peak; and construct a motion influence confidence factor of each detection position according to the interference peak discrimination factor of each detection position and the frequency values of the peaks and the troughs.

[0010] obtain a scale relationship graph of the infrared spectrum graph of each detection position by using a box counting method based on the distribution of the spectral curve in the box under different box side lengths; obtain the number of outliers of the scale relationship graph of each detection position; determine an overall light absorption difference index of the target component according to the distance between the scale relationship graphs of all detection positions and the difference in the number of outliers; construct a heat data sequence of the target component under each wavelength; and calculate a local anomaly factor of each detection position of the heat data sequence by using a LOF anomaly detection algorithm in combination with the overall light absorption difference index and the motion influence confidence factor of each detection position.

[0011] determine the temperature deformation risk of the target component based on the local anomaly factor and a heat alarm threshold.

[0012] The determination of the peak width judgment value of each peak based on the frequency difference between adjacent troughs on left and right sides of each peak in the frequency spectrum comprises:

[0013] the frequency difference between the previous trough and the next trough of each peak is recorded as a first difference value;

[0014] the upward rounding value of the hyperbolic tangent function of the difference between the first difference value and a preset frequency range threshold is taken as the peak width judgment value of each peak.

[0015] Preferably, the determination of the interference peak discrimination factor of each detection position by combining the peak width judgment value and the amplitude value of each peak is expressed as:

[0016] wherein, Af k is the interference peak discrimination factor of the kth detection position, R1 and R2 are the number of peaks and the number of troughs in the frequency spectrum respectively, min() is a minimum function, is an upward rounding function, tanh() is a hyperbolic tangent function, T z is an amplitude threshold, is the amplitude value of the uth peak, Df u is the peak width judgment value of the uth peak at the kth detection position.

[0017] Preferably, the interference peak discrimination factor of each detection position and the frequency values of the peaks and troughs are used to construct a motion influence confidence factor of each detection position, expressed as:

[0018] wherein, Ef k is the motion influence confidence factor of the kth detection position, Af k is the interference peak discrimination factor of the kth detection position, R1 and R2 are the number of peaks and troughs in the spectrum, respectively, min() is the minimum function, F u and F u-1 are the frequency values of the uth and u-1th peaks, respectively, G u-1 and G u are the frequency values of the u-1th and uth troughs, respectively.

[0019] Preferably, the box counting method is used to obtain a scale relationship graph of the infrared spectrum of each detection position based on the distribution of the spectrum curve in the box under different box side lengths, comprising:

[0020] For the infrared spectrum of each detection position, a preset initial box side length is the side length of a square, and the center of the infrared spectrum is taken as the center of the square to construct an initial square box of the infrared spectrum;

[0021] Starting from the preset initial box side length, the unit wavelength of the horizontal coordinate is gradually decreased to obtain the box side length under different decreasing times;

[0022] For the box side length under different decreasing times, the square box formed by the box side length is tiled in the infrared spectrum, the area after tiling is greater than or equal to the area of the initial square box, the tiling position completely covers the initial square box, and there is an intersection with the initial square box; the number of square boxes after tiling that intersect with the spectrum curve in the infrared spectrum is counted;

[0023] The number of square boxes of the box side length under different decreasing times is used to form a scale relationship graph of the infrared spectrum of each detection position according to the box side length from large to small.

[0024] Preferably, the number of outliers in the scale relationship graph of each detection position is obtained by using the CURE density clustering algorithm to obtain the outliers in the scale relationship graph of each detection position; and the number of outliers in the scale relationship graph of each detection position is counted.

[0025] Preferably, the overall light absorption difference index of the target component is determined according to the distance between the scale relationship graphs of all detection positions and the difference in the number of outliers, comprising:

[0026] For the scale relationship diagram of any two detection positions, obtain the DTW distance between two scale relationship diagrams; calculate the absolute value of the difference of the number of outliers between two scale relationship diagrams;

[0027] The mean of the product of the DTW distance of all arbitrary two detection positions and the absolute value of the difference is taken as the overall light absorption difference index of the target component.

[0028] Preferably, the construction of the heat data sequence of the target component at each wavelength comprises: the absorbance of all detection positions of the target component at each wavelength is combined to form the heat data sequence of the target component at each wavelength.

[0029] Preferably, the combination of the overall light absorption difference index and the motion influence confidence factor of each detection position adopts the LOF anomaly detection algorithm to calculate the local anomaly factor of each detection position of the heat data sequence, comprising:

[0030] The reciprocal of the product of the overall light absorption difference index and the motion influence confidence factor of each detection position of the heat data sequence is taken as the index of the exponential function with the natural constant e as the base number;

[0031] The LOF anomaly detection algorithm is adopted to obtain the original local anomaly factor of each detection position of the heat data sequence; the product of the calculation result of the exponential function and the original local anomaly factor of each detection position of the heat data sequence is taken as the local anomaly factor of each detection position of the heat data sequence.

[0032] Preferably, the judgment of the temperature deformation risk of the target component based on the local anomaly factor and the heat alarm threshold value comprises:

[0033] For the local anomaly factor of each detection position of the heat data sequence, the number of local anomaly factors greater than the preset anomaly threshold value is counted;

[0034] The ratio of the number of local anomaly factors to the number of detection positions of the target component is calculated, and when the ratio is greater than or equal to the preset heat alarm threshold value, the target component has a temperature deformation risk.

[0035] The present application has at least the following beneficial effects:

[0036] The present application aims at the influence of shadow generated due to moxibustion robot movement on infrared spectrum detection, adopts fast Fourier transform to construct a frequency spectrum graph of an infrared spectrum graph, analyzes the distribution of interference peaks in the infrared spectrum graph from two levels of horizontal and vertical coordinates based on the peak and valley positions and frequency distribution in the frequency spectrum graph, constructs a motion influence confidence factor, mines the subtle and chaotic fluctuations of infrared spectrum entropy caused by the continuous change of shadow at the joint, so as to reflect the shadow influence appearing in the spectrum; secondly, aiming at the reflective properties of the surface coating of the moxibustion robot structure, an overall light absorption difference index is constructed based on the box counting method, so as to reflect the light absorption difference caused by the reflective properties on the same structure; the local outlier factor in the LOF algorithm is improved by combining the motion influence confidence factor and the overall light absorption difference index, so that the algorithm can exclude the spectrum difference changes caused by shadow and reflective properties, more accurately find out the abnormal value, realize the timely alarm of the high temperature of the part, and enhance the robustness and accuracy of the algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, below, a brief introduction will be given to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0038] Fig. 1 is a flow chart of the intelligent moxibustion robot part temperature deformation monitoring method based on infrared spectrum provided by the present application;

[0039] Fig. 2 is a setting schematic diagram of a box;

[0040] Fig. 3 is a flow chart of index construction of local outlier factor. DETAILED DESCRIPTION

[0041] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the intelligent moxibustion robot part temperature deformation monitoring method based on infrared spectrum proposed according to the present application, its specific implementation, structure, features and effects are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0043] The application provides a method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy.

[0044] The application provides a method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy.

[0045] Specifically, the method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy comprises the following steps, please refer to FIG. 1:

[0046] In step S001, infrared spectrograms of target components of the moxibustion robot are collected.

[0047] According to the structure of the moxibustion robot, the joints and connecting points of the moving components of the robot play a key role in the robot system, and the design thereof not only determines the motion flexibility, accurate positioning and load capacity of the robot, but also directly affects the controllability and working stability of the mechanical structure. A strong and reliable joint system enables the robot to perform complex tasks while ensuring torque transmission and the reliability of the mechanical structure. Therefore, a good joint design is a key factor to ensure that the robot can efficiently and stably perform tasks in various working environments, and has an important and non-negligible importance to the development and application of robot technology. If the joints and connecting points of the robot are deformed due to high temperature, the mechanical failure of the joint components will occur, the mechanical arm cannot move normally or loses the original accuracy, the accuracy and control performance of the robot are affected, and the service life of the robot is shortened. In this embodiment, a joint of the moxibustion robot is selected as a sample for temperature deformation monitoring.

[0048] In this embodiment, a scannable infrared spectrometer is used to collect real-time infrared spectrograms of the target components. The infrared spectrometer is aimed at the target components and the correct measurement distance and angle are ensured, so that the infrared spectrometer can capture the infrared radiation data of the normal operation of the moxibustion robot in real time. The scanning mode of the infrared spectrometer refers to the fact that the infrared spectrometer can scan on the surface of the sample to obtain infrared spectrogram data at different positions. In this mode, the infrared spectrometer moves along the horizontal or vertical direction to scan the sample surface point by point or line by line to collect infrared radiation data at each position. Under the accuracy of the infrared scanner, a total of K detection positions in the sample are detected, and infrared spectrograms of all positions are obtained by the infrared spectrometer. The horizontal axis of the infrared spectrogram is the wavelength, and the vertical axis is the absorbance at different wavelengths.

[0049] At this point, the infrared spectrograms of the target components of the moxibustion robot at different detection positions can be obtained.

[0050] In step S002, the infrared spectrograms at different detection positions are analyzed to construct the overall absorbance difference index of the target components and the motion influence confidence factor of each detection position.

[0051] Since the infrared scanner is used to detect the infrared spectrum of the moxibustion robot in real time, the joint sample being monitored is also in continuous motion due to the continuous movement of the moxibustion robot, and the shadow at the joint is in a state of continuous change. The infrared scanner obtains information by detecting the infrared spectrum reflected or emitted by an object. If the shadow at the joint is constantly changing, it will cause instability or distortion of the measurement signal, and interfere with the accurate acquisition of the infrared spectrum. During the continuous movement of the joint, the ideal heat condition at the joint should be continuously rising with the continuous movement of the joint. Compared with the ideal situation, the change of the shadow at the joint will cause the actual situation detected by the infrared spectrometer to be in continuous fluctuation. The present embodiment selects the kth detection position as an example for analysis

[0052] The infrared spectrum of the kth detection position is converted into a frequency spectrum using fast Fourier transform (FFT). Fast Fourier transform is a known technology and will not be described again. Since the continuous change of the shadow at the joint analyzed above is manifested as subtle and chaotic fluctuations on the infrared spectrum, these fluctuations will usually be presented in the form of interference peaks on the frequency spectrum when converted into a frequency spectrum by fast Fourier transform (FFT). The present embodiment sets an amplitude threshold T z , T z is valued as , where A k，max and A k，min are the maximum amplitude and minimum amplitude on the frequency spectrum, respectively. Then, using the second-order difference discrimination method, the frequency spectrum is inputted to output all the peaks and valleys of the frequency spectrum. The second-order difference discrimination method is a known technology and will not be described again. Let R1 be the total number of peaks and R2 be the total number of valleys in the frequency spectrum. Set a frequency range threshold T p , T p is valued as , where F u and G u are the frequency values of the uth peak and the uth valley, respectively, i.e. the value of T p is the average of the absolute values of the frequency difference between adjacent peaks and valleys. Then, the motion influence confidence factor Ef k of the kth detection position is constructed:

[0053] where Ef k is the motion influence confidence factor of the kth detection position, representing the influence and degree of influence of the change of the shadow at the joint on the infrared spectrum during the movement of the moxibustion robot, Af k is the interference peak discrimination factor of the kth detection position, R1 and R2 are the number of peaks and the number of valleys in the frequency spectrum, respectively, min() is the minimum function, and Fu , F u-1 are the frequency values of the u-th, u-1-th peak, respectively, G u-1 , G u and G u+1 are the frequency values of the u-1-th, u-th and u+1-th valley, respectively;

[0054] tanh() is the hyperbolic tangent function, T z is the amplitude threshold value, is the amplitude value of the u-th peak, Df u is the peak width judgment value of the u-th peak at the k-th detection position;

[0055] T p is the frequency range threshold value, is the rounding up function, and u is the summation function index.

[0056] It should be noted that, due to the uncertainty of the number of peaks and valleys in the spectrum, a judgment function is introduced to distinguish the two valleys adjacent to the u-th peak. In the calculation of the motion influence confidence factor Ef k , the interference peak discrimination factor Af k is used to count the number of all interference peaks. When , it means that the amplitude of the u-th peak is less than the threshold value, and at this time it is considered that the amplitude of the u-th peak is less than the average level of the peak value, which means that the amplitude of the peak is within the amplitude range of the interference peak, and at this time the value of is 1, and vice versa: 0; and the peak width judgment value Df u of the u-th peak at the k-th detection position is used to determine whether the peak belongs to the interference peak by the width of the u-th peak. The frequency difference between the two peaks adjacent to the u-th peak is calculated, and the difference with the frequency range threshold value is used to judge whether the peak width is greater than the threshold value. If it is greater than the threshold value, the value obtained after the hyperbolic tangent function and the rounding up function is 1, and vice versa: 0, that is, the interference peak discrimination factor Af k By concatenating the height and width of the peak and summing them up, the number of peaks belonging to the interference peak is determined, and the proportion of the interference peak is obtained by dividing the total number of peaks. The greater the value represents the greater the proportion of the subtle and chaotic fluctuations caused by the shadow change, and the greater the Ef k ; In the spectrum, the main peak usually presents a clear symmetrical shape, and the interference peak may present an irregular shape, |(G u -F u )-(F u -G u-1) The irregularity of the interference peak is determined by judging the frequency value difference of the u-th wave peak and its adjacent two wave troughs. When the u-th wave peak shape is symmetrical, the value is 0. When the u-th wave peak shape is more asymmetrical, the value is larger, representing that the irregularity of the interference peak is larger, and the influence degree of the fine and chaotic fluctuation caused by the shadow change on the infrared spectrum is larger. Ef k is larger.

[0057] Since the moxibustion robot is in a motion state, the coating at its joints has different reflections at different motion positions. The robot at different motion positions may cause changes in the coating at the joints, affecting the signal of the infrared spectrum. Different positions produce different peak or valley values in the spectrum. During the motion of the robot joints, the temperature changes of the entire joint part may be influenced by multiple factors.

[0058] First, normal motion is usually accompanied by uniform energy distribution, so that the energy change of the entire joint area is relatively uniform; second, the robot joints often use metals or other materials with good thermal conductivity, which promotes the rapid conduction of heat at the joint part, thereby making the temperature changes of the entire joint area consistent; in addition, a reasonably designed mechanical structure and stable working environment also help to disperse and uniformly transmit the heat generated by motion, further maintaining the similarity of the temperature of the joint part. These factors work together to make the robot joints in motion exhibit similar temperature change characteristics as a whole.

[0059] However, the reflection of the joint coating will cause differences in the infrared spectrum detected by the infrared spectrometer at different detection positions. Taking the infrared spectrum graph at the k-th detection position as input, when analyzing the infrared spectrum graph using the box counting method, first place the infrared spectrum graph at the k-th detection position in an initial square box. The side length of this initial box must be large enough to completely contain the value range of the horizontal and vertical axis data of the spectrum graph. In order to construct a gradually shrinking box sequence, start with the initial box side length , and decrease by unit wavelength of the horizontal coordinate until the side length is reduced to unit wavelength. Take the initial box side length as the side length of the square, and take the center of the infrared spectrum graph as the center of the square to construct the initial square box of the infrared spectrum graph; with each reduction of the box side length, a plurality of square boxes composed of the box side length obtained by each reduction are tiled in the initial square box. The area after each tiling needs to be greater than or equal to the area of the initial square box, and needs to completely cover the initial square box. Among them, Take the maximum value of the horizontal and vertical axis data of the spectrum graph.

[0060] In each process of reducing the side length of the box, it is checked whether there is a point on the spectrum curve in each box to count the intersection part of the box and the spectrum curve. For the size of the box and the number of boxes containing the curve in each process of reducing the size of the box, a coordinate system is established, in which the horizontal axis represents the size of the box and the vertical axis represents the number of corresponding boxes. Finally, by drawing the scale relationship diagram of the infrared spectrum of the kth detection position, the change trend of the box counting method in analyzing the spectrum diagram can be more intuitively understood. The setting of the box is shown in FIG. 2.

[0061] With the scale relationship diagram of the kth detection position and the first detection position as input, the DTW distance DTW k，l between the two scale relationship diagrams is calculated.

[0062] wherein Ps is the overall absorbance difference index of the joint part, representing the difference degree of the overall infrared light absorption and reflection of the joint due to the special reasons such as coating or reflection, DTW k，l is the DTW distance in the two scale relationship diagrams of the kth detection position and the first detection position, V k and V l represent the number of outliers in the scale relationship diagram of the kth detection position and the first detection position respectively, K is the number of detection positions of the target component, and k and l are both summation function indexes.

[0063] In the calculation of the overall absorbance difference index, the size of the DTW distance reflects the degree of time stretching or compression of one of the sequences in the process of finding the optimal matching path. The greater the distance, the more difficult the matching between the sequences, the smaller the similarity, and the greater the DTW k，l distance, which represents the greater difference between the detection position k and the detection position l, i.e., the two detection points at the same joint have more different infrared spectrum graphs, representing that the infrared spectrum difference between the two detection points due to factors such as coating reflection is greater, and Ps increases accordingly. The number of boxes containing the curve in the scale relationship diagram gradually increases with the decrease of the size of the box. If there are outliers in the scale relationship diagram, it represents that there are irregular and dramatic changes on the infrared spectrum represented by the scale relationship diagram. The more outliers represent the more dramatic changes on the infrared spectrum, and the greater the value of |V k -V l represents the greater difference between the scale relationship diagrams of the two detection positions, i.e., the greater difference between the infrared spectra of the two detection positions, and Ps increases accordingly.

[0064] Step S003, the absorbance of different detection positions at each wavelength is detected for abnormality, the local outlier factor in the LOF algorithm is improved, and the high temperature deformation warning of the moxibustion robot is realized.

[0065] When monitoring the high temperature deformation of the joint area, in order to avoid the deformation caused by local friction and local heating, the absorbance of all detection positions at each wavelength at the same time is recorded in real time to form a heat data sequence at the corresponding wavelength. And using the heat data sequence as input, the original local outlier factor of the detection position k on the sequence is redefined using the LOF anomaly detection algorithm:

[0066] Wherein, LoF k，new is the local outlier factor of the kth detection position, EF k is the motion influence confidence factor of the kth detection position, Ps is the overall absorbance difference index of the joint part, LOF k is the original local outlier factor of the kth detection position calculated by the traditional LOF anomaly detection algorithm on the heat data sequence. The index construction flow chart of the local outlier factor is shown in Figure 3. The LOF anomaly detection algorithm is a known technology and will not be described again

[0067] When the motion influence confidence factor of the kth detection position is larger, it represents that the shadow caused by the motion of the moxibustion robot has greater interference on the absorbance detected by the infrared spectrometer. At this time, in order to exclude the influence of the shadow, the position is given smaller abnormal confidence, and LOF k，new is reduced accordingly; and Ps represents the difference degree of the overall infrared light absorption and reflection of the joint caused by the coating or reflection and other special reasons. The greater the value, the greater the influence of the detected position joint absorbance. At this time, the calculation of the overall abnormal factor is reduced to exclude the influence of the infrared light absorption of the infrared spectrometer on the joint caused by the coating or reflection and other special reasons, and LOF k，new is reduced accordingly.

[0068] Set the abnormal threshold, the value of this embodiment is the standard deviation of the local outlier factor of all detection positions in the heat data sequence, and finally the number of local outlier factors greater than the abnormal threshold in the heat data sequence is counted. Suppose that the heat data sequence detects H local outlier factors, set the heat alarm threshold T, T takes the value of 0.15, when , it represents that the detected component joint appears high temperature deformation risk, at this time, the alarm reminds the staff to pay extra attention to the joint of the detected position. Wherein, K is the number of detection positions of the target component.

[0069] In summary, the embodiment of the present application aims at the influence of the shadow generated by the movement of the moxibustion robot on the infrared spectrum detection, adopts fast Fourier transform to construct the frequency spectrum of the infrared spectrum graph, analyzes the distribution of the interference peak in the infrared spectrum graph from the horizontal and vertical coordinates, constructs the motion influence confidence factor, and excavates the subtle and chaotic fluctuations of the infrared spectrum entropy caused by the continuous change of the shadow at the joint, so as to reflect the influence of the shadow in the spectrum; secondly, aiming at the reflection of the surface coating of the moxibustion robot structure, the overall light absorption difference index is constructed based on the box counting method, so as to reflect the light absorption difference caused by the reflection on the same structure; combined with the motion influence confidence factor and the overall light absorption difference index, the local outlier factor in the LOF algorithm is improved, so that the algorithm can exclude the spectrum difference changes caused by the shadow and the reflection, more accurately find the abnormal value, realize the timely alarm of the high temperature of the part, and enhance the robustness and accuracy of the algorithm.

[0070] It should be noted that the above-mentioned embodiments of the present application are only for description, not representing the advantages and disadvantages of the embodiments. And the above-mentioned description of the specific embodiments of the present application is described, and 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, multi-task processing and parallel processing are also possible or may be advantageous.

[0071] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0072] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; the technical solutions recorded in the above-described embodiments are modified, or some technical features are replaced, without changing the essence of the corresponding technical solutions out of the scope of the technical solutions of the embodiments of the present application, which should be included in the protection scope of the present application.

Claims

1. A method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy, characterized in that: The method comprises the following steps: Collect infrared spectra of each detection position of the target components of the moxibustion robot; A fast Fourier transform is used to obtain a spectrum of the infrared spectrum; a second-order difference discrimination method is used to output all peaks and troughs in the spectrum; a peak width judgment value of each peak is determined based on the frequency difference between the adjacent troughs on the left and right sides of each peak in the spectrum; an interference peak discrimination factor of each detection position is determined based on the peak width judgment value and the amplitude value of each peak; and a motion influence confidence factor of each detection position is constructed based on the interference peak discrimination factor of each detection position and the frequency values ​​of the peaks and troughs. The box counting method is used to obtain the scale relationship diagram of the infrared spectrum at each detection position based on the distribution of the spectral curve in the box with different box side lengths. The number of outliers in the scale relationship diagram at each detection position is obtained. The overall absorbance difference index of the target component is determined based on the distance between the scale relationship diagrams of all detection positions and the difference in the number of outliers. The thermal data series of the target component at each wavelength is constructed. The local anomaly factor (LOF) anomaly detection algorithm is used to calculate the local anomaly factor of each detection position in the thermal data series, combining the overall absorbance difference index and the motion influence confidence factor of each detection position. Determine the temperature deformation risk of the target component based on local abnormal factors and thermal alarm thresholds; The determining of the peak width judgment value of each peak based on the frequency difference of the troughs adjacent to the left and right sides of each peak in the spectrum graph includes: The frequency difference between the next trough and the previous trough of each peak is recorded as the first difference; The rounded-up value of the hyperbolic tangent function of the difference between the first difference and the preset frequency range threshold is used as the peak width judgment value of each peak.

2. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The interference peak discrimination factor of each detection position is determined by combining the peak width judgment value and the amplitude value of each peak, and the expression is: Among them, Af k is the interference peak discrimination factor of the kth detection position, R1 and R2 are the number of peaks and troughs in the spectrum respectively, min() is the minimum value function, is the upward rounding function, tanh() is the hyperbolic tangent function, T z is the amplitude threshold, is the amplitude value of the u-th peak, Df u is the peak width judgment value of the u-th peak at the k-th detection position.

3. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The motion influence confidence factor of each detection position is constructed based on the interference peak discrimination factor of each detection position and the frequency values ​​of the peak and trough, and the expression is: Among them, Ef k is the confidence factor of the motion impact of the kth detection position, Af k is the interference peak discrimination factor of the kth detection position, R1 and R2 are the number of peaks and troughs in the spectrum respectively, min() is the minimum function, F u 、F u-1 are the frequency values ​​of the u-th and u-1-th peaks respectively, G u-1 , G u are the frequency values ​​of the u-1th and uth troughs respectively.

4. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The box counting method is used to obtain a scale relationship diagram of the infrared spectrum at each detection position based on the distribution of the spectrum curve in the box under different box side lengths, including: For the infrared spectrum of each detection position, the initial box side length is set as the side length of the square, and the center of the infrared spectrum is used as the center of the square to construct the initial square box of the infrared spectrum; Starting from the preset initial box side length, gradually decrease according to the unit wavelength of the horizontal axis to obtain the box side length at different decreasing times; For each box side length at different decreasing times, the square box formed by the box side length is tiled on the infrared spectrum. The area after tiling is greater than or equal to the area of ​​the initial positive box. The tiled position completely covers the initial square box and intersects with the initial square box. The number of square boxes that intersect with the spectrum curve in the infrared spectrum after tiling is counted. The number of square boxes with different decreasing times of box side lengths is used to form a scale relationship diagram of the infrared spectrum of each detection position according to the box side length from large to small.

5. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The obtaining of the number of outliers in the scale relationship graph of each detection position includes: using a CURE density clustering algorithm to obtain the outliers in the scale relationship graph of each detection position; and counting the number of outliers in the scale relationship graph of each detection position.

6. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The determining of the overall absorbance difference index of the target component according to the distances between the scale relationship graphs of all detection positions and the difference in the number of outliers includes: For any two scale relationship graphs of detection positions, obtain the DTW distance between the two scale relationship graphs; calculate the absolute value of the difference in the number of outliers between the two scale relationship graphs; The average of the products of the DTW distance and the absolute value of the difference between any two detection positions is used as the overall absorbance difference index of the target component.

7. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The constructing of the heat data sequence of the target component at each wavelength includes: forming the heat data sequence of the target component at each wavelength by the absorbance of all detection positions of the target component at each wavelength.

8. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The LOF anomaly detection algorithm is used to calculate the local anomaly factor of each detection position in the thermal data sequence by combining the overall absorbance difference index and the motion impact confidence factor of each detection position, including: Calculate the inverse of the product of the overall absorbance difference index and the confidence factor of the motion effect at each detection position of the thermal data sequence as the exponent of an exponential function with the natural constant e as the base; The LOF anomaly detection algorithm is used to obtain the original local anomaly factor of each detection position in the heat data sequence; the product of the calculation result of the exponential function and the original local anomaly factor of each detection position in the heat data sequence is used as the local anomaly factor of each detection position in the heat data sequence.

9. The method for monitoring temperature deformation of components of an intelligent moxibustion robot based on infrared spectroscopy according to claim 1, wherein: The determining of the temperature deformation risk of the target component based on the local abnormality factor and the thermal alarm threshold includes: For the local anomaly factors of each detection position in the heat data sequence, the number of local anomaly factors greater than the preset anomaly threshold is counted; The ratio of the number of the local abnormal factors to the number of detection positions of the target component is calculated. When the ratio is greater than or equal to a preset thermal alarm threshold, the target component has a risk of temperature deformation.

Images