A particle detection system based on laser self-mixing interference

Abstract

The application relates to the technical field of particulate matter detection, and discloses a particulate matter detection system based on laser self-mixing interference, which comprises a laser emitting device for emitting a laser beam to irradiate particulate matter, a focusing lens for focusing the laser beam on the particulate matter, a signal collecting device arranged in the laser emitting device and used for receiving a reflected light beam of the particulate matter, and a signal processing device connected to the signal collecting device and used for determining a self-mixing interference signal according to the reflected light beam of the particulate matter, performing data processing on the self-mixing interference signal, and determining particulate matter information according to the self-mixing interference signal after data processing. The self-mixing interference signal of the particulate matter is detected, the data detection efficiency is effectively improved, and the accuracy of particulate matter signal detection is enhanced.

Description

Technical Field

[0001] This application relates to the field of particulate matter detection technology, and more specifically, to a particulate matter detection system based on laser self-mixing interferometry. Background Technology

[0002] A laser emits a beam of light that strikes a particle. After being reflected by the particle, a portion of this light returns to the laser's resonant cavity, where it mixes with the existing laser beam, thus modulating the laser's output power. This phenomenon is called optical feedback self-mixing interference. Compared to traditional interference systems, systems constructed using optical feedback self-mixing interference technology have advantages such as simple and compact structure and easy collimation.

[0003] Existing technologies often struggle to identify anomalous data in mixed interference signals using conventional algorithms. Therefore, it is necessary to improve the accuracy of anomalous particulate data detection to facilitate faster and more accurate identification and correction of particulate anomalies. Summary of the Invention

[0004] This invention provides a particulate matter detection system based on laser self-mixing interferometry to solve the problem of difficulty in identifying abnormal data in particulate matter self-mixing interferometry signals in existing technologies, including:

[0005] A laser emitting device is used to emit a laser beam to irradiate particulate matter; a focusing lens is used to focus the laser beam onto the particulate matter; a signal acquisition device is located inside the laser emitting device to receive the reflected beam from the particulate matter; and a signal processing device is connected to the signal acquisition device to determine a self-mixing interference signal based on the reflected beam from the particulate matter, perform data processing on the self-mixing interference signal, and determine particulate matter information based on the processed self-mixing interference signal.

[0006] Further, the signal processing device is used to: preprocess the self-mixing interference signal of particulate matter, determine particulate matter feature data based on the preprocessed self-mixing interference signal; establish a particulate matter sample set based on particulate matter feature data within a historical preset time period, and perform particulate matter clustering based on the particulate matter feature data in the particulate matter sample set; filter the self-mixing interference signal of each particulate matter based on the clustering result, and establish a loss function based on the filtered self-mixing interference signal; establish an anomaly detection model, train the anomaly detection model based on the loss function, and detect anomalies in the self-mixing interference signal of particulate matter within the current preset time period based on the trained anomaly detection model; correct the anomalies in the self-mixing interference signal of particulate matter within the current preset time period to obtain a corrected self-mixing interference signal, and determine particulate matter information based on the corrected self-mixing interference signal.

[0007] Furthermore, the preprocessing of the self-mixing interference signal of particulate matter and the determination of particulate matter characteristic data based on the preprocessed self-mixing interference signal include: performing wavelet transform on the self-mixing interference signal of particulate matter and determining the particulate matter characteristic data based on the wavelet transform result.

[0008] Further, the particulate matter clustering based on particulate matter feature data in the particulate matter sample set includes: Step 1, randomly selecting k initial cluster centers from the particulate matter sample set; Step 2, calculating the Euclidean distance from the particulate matter feature data in the particulate matter sample set to the cluster centers, and dividing each particulate matter into its corresponding cluster based on the Euclidean distance from the particulate matter feature data in the particulate matter sample set to the cluster centers; Step 3, calculating the mean of the feature data of the particulate matter in each cluster, and reselecting cluster centers based on the mean of the feature data of the particulate matter in each cluster; repeating steps 2-3 until the cluster centers no longer change or the number of iterations reaches the preset maximum number of iterations, to obtain the clustering result of the particulate matter.

[0009] Further, the step of filtering the self-mixing interference signals of each particle based on the clustering results and establishing a loss function based on the filtered self-mixing interference signals includes: setting filter windows of different lengths; performing moving average filtering on the self-mixing interference signals of particles within each cluster based on each filter window; statistically analyzing the feature data of the filtered self-mixing interference signals of particles at different filter window lengths; calculating the difference between the feature data of the filtered self-mixing interference signals of particles at different filter window lengths and the feature data of the unfiltered self-mixing interference signals at the same filter window length, thus obtaining a feature data difference sequence of particles; acquiring the feature data of the self-mixing interference signals of particles at different filter window lengths; constructing an outlier parameter sequence based on the feature data of the self-mixing interference signals of particles at different filter window lengths; calculating the correlation coefficient between the feature data difference sequence and the outlier parameter sequence; determining the anomaly weight of the corresponding particle based on the correlation coefficient between the feature data difference sequence and the outlier parameter sequence; and establishing a loss function based on the anomaly weight of the particle.

[0010] Furthermore, the step of constructing an outlier parameter sequence based on the feature data of the self-mixing interference signal of particulate matter at different filter window lengths includes: normalizing each feature data of the self-mixing interference signal of particulate matter at different filter window lengths to obtain normalized feature data; determining the LOF value of the feature data at each filter window length based on the normalized feature data using the LOF algorithm; and constructing an outlier parameter sequence based on the LOF value of the feature data of particulate matter.

[0011] Furthermore, the step of establishing a loss function based on the abnormal weights of particulate matter includes: obtaining the abnormal weights of each particulate matter, sorting the particulate matter in descending order of abnormal weights; extracting the abnormal weights of the first N particulate matter and the next N particulate matter based on the sorting results, and establishing a loss function based on the abnormal weights of the first N particulate matter and the next N particulate matter.

[0012] Further, the step of establishing a loss function based on the anomaly weights of the first N particles and the anomaly weights of the last N particles includes:

[0013]

[0014] in, For loss function, The anomaly weight of the i-th particle among the first N particles is given by [the anomaly weight]. Let be the average LOF value of the i-th particle among the first N particles. The anomaly weight of the i-th particle among the last N particles. It represents the difference between the feature data of the i-th particle and the feature data of the remaining particles among the last N particles.

[0015] Furthermore, the step of training the abnormal data detection model based on the loss function includes: acquiring the self-mixing interference signal of particulate matter within a historical preset time period; forming a training sample set by combining the self-mixing interference signals before and after sliding filtering according to different filter window lengths; inputting the training sample set into the abnormal data detection model and outputting feature data and LOF values; after obtaining the optimal solution of the loss function, acquiring the particulate matter containing the largest average LOF value to obtain abnormal particulate matter data.

[0016] Furthermore, the step of correcting the outliers of the self-mixing interference signal of particulate matter within the current preset time period to obtain the corrected self-mixing interference signal, and determining particulate matter information based on the corrected self-mixing interference signal, includes: correcting the self-mixing interference signal of the abnormal particulate matter data based on the linear interpolation method to obtain the particulate matter information corresponding to the corrected self-mixing interference signal.

[0017] The beneficial effects of this invention are as follows:

[0018] By applying the above technical solutions, this invention detects particulate matter self-mixing interference signals, clusters different types of particulate matter, and detects abnormal data for each cluster, effectively improving detection efficiency. By establishing an abnormal data detection model to detect and correct abnormal particulate matter data, accurate particulate matter information is obtained, thereby enhancing the accuracy of particulate matter signal detection. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic diagram of a particulate matter detection system based on laser self-mixing interferometry proposed in an embodiment of the present invention is shown. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] This application provides a particulate matter detection system based on laser self-mixing interferometry, such as... Figure 1 As shown, it includes:

[0023] A laser emitting device is used to emit a laser beam to irradiate particulate matter; a focusing lens is used to focus the laser beam onto the particulate matter; a signal acquisition device is located inside the laser emitting device to receive the reflected beam from the particulate matter; and a signal processing device is connected to the signal acquisition device to determine a self-mixing interference signal based on the reflected beam from the particulate matter, perform data processing on the self-mixing interference signal, and determine particulate matter information based on the processed self-mixing interference signal.

[0024] In this embodiment, the laser emitting device is a VCSEL, and the signal acquisition device is a resonant cavity. A photodiode is integrated inside the VCSEL resonant cavity. The laser intensity inside the resonant cavity is detected by the photodiode. Infrared laser is emitted through the VCSEL. A focusing lens is placed above the VCSEL. The laser beam is focused onto the particle above the focusing lens through the focusing lens. The reflected light from the particle is focused onto the VCSEL through the focusing lens. Self-mixing interference is generated between the light inside the resonant cavity of the VCSEL and the light inside. The self-mixing interference signal is collected by the signal processing device to obtain accurate particle information.

[0025] In some embodiments of this application, the signal processing device is used to: preprocess the self-mixing interference signal of particulate matter; determine particulate matter feature data based on the preprocessed self-mixing interference signal; establish a particulate matter sample set based on particulate matter feature data within a historical preset time period; perform particulate matter clustering based on the particulate matter feature data in the particulate matter sample set; filter the self-mixing interference signal of each particulate matter based on the clustering result; establish a loss function based on the filtered self-mixing interference signal; establish an anomaly detection model; train the anomaly detection model based on the loss function; detect anomalies in the self-mixing interference signal of particulate matter within the current preset time period based on the trained anomaly detection model; correct the anomalies in the self-mixing interference signal of particulate matter within the current preset time period to obtain a corrected self-mixing interference signal; and determine particulate matter information based on the corrected self-mixing interference signal.

[0026] In this embodiment, feature data of each particulate matter within multiple historical preset time periods are first extracted using self-mixing interference signals. The particulate matter is then clustered based on the feature data. The self-mixing interference signals of the particulate matter within each cluster are filtered using the clustering results to construct a loss function. An anomaly detection model is trained based on the loss function. Anomaly detection models are then used to detect anomalies in the self-mixing interference signals of particulate matter within the current preset time period. After correcting the anomalies, accurate particulate matter information is extracted.

[0027] In some embodiments of this application, the step of preprocessing the self-mixing interference signal of particulate matter and determining particulate matter feature data based on the preprocessed self-mixing interference signal includes: performing wavelet transform on the self-mixing interference signal of particulate matter and determining particulate matter feature data based on the wavelet transform result.

[0028] In some embodiments of this application, the step of clustering particulate matter based on particulate matter feature data in a particulate matter sample set includes: Step 1, randomly selecting k initial cluster centers from the particulate matter sample set; Step 2, calculating the Euclidean distance from the particulate matter feature data in the particulate matter sample set to the cluster centers, and dividing each particulate matter into its corresponding cluster based on the Euclidean distance from the particulate matter feature data in the particulate matter sample set to the cluster centers; Step 3, calculating the mean of the feature data of the particulate matter in each cluster, and reselecting cluster centers based on the mean of the feature data of the particulate matter in each cluster; repeating steps 2-3 until the cluster centers no longer change or the number of iterations reaches the preset maximum number of iterations, to obtain the clustering result of the particulate matter.

[0029] In some embodiments of this application, the step of filtering the self-mixing interference signals of each particle based on the clustering results and establishing a loss function based on the filtered self-mixing interference signals includes: setting filter windows of different lengths; performing moving average filtering on the self-mixing interference signals of particles within each cluster based on each filter window; statistically analyzing the feature data of the filtered self-mixing interference signals of particles at different filter window lengths; calculating the difference between the feature data of the filtered self-mixing interference signals of particles at different filter window lengths and the feature data of the unfiltered self-mixing interference signals at the same filter window length to obtain a feature data difference sequence of particles; acquiring the feature data of the self-mixing interference signals of particles at different filter window lengths; constructing an outlier parameter sequence based on the feature data of the self-mixing interference signals of particles at different filter window lengths; calculating the correlation coefficient between the feature data difference sequence and the outlier parameter sequence; determining the anomaly weight of the corresponding particle based on the correlation coefficient between the feature data difference sequence and the outlier parameter sequence; and establishing a loss function based on the anomaly weight of the particle.

[0030] In this embodiment, the outlier weights of each particle are obtained by observing the changes in the characteristic data of the self-mixing interference signals of particles before and after filtering under different filter window lengths. The greater the correlation between the changes and the outlier parameters, the greater the response of the particles to the outlier data.

[0031] In some embodiments of this application, the step of constructing an outlier parameter sequence based on the feature data of the self-mixing interference signal of particulate matter at different filter window lengths includes: normalizing each feature data of the self-mixing interference signal of particulate matter at different filter window lengths to obtain normalized feature data; determining the LOF value of the feature data at each filter window length based on the normalized feature data using the LOF algorithm; and constructing an outlier parameter sequence based on the LOF value of the feature data of particulate matter.

[0032] In this embodiment, the LOF value of the feature data of each filter window length is detected based on the LOF algorithm, thereby obtaining the outlier parameter sequence.

[0033] In some embodiments of this application, the step of establishing a loss function based on the abnormal weights of particulate matter includes: obtaining the abnormal weights of each particulate matter, sorting the particulate matter in descending order of abnormal weights; extracting the abnormal weights of the first N particulate matter and the next N particulate matter based on the sorting results, and establishing a loss function based on the abnormal weights of the first N particulate matter and the next N particulate matter.

[0034] In some embodiments of this application, establishing a loss function based on the anomaly weights of the first N particles and the anomaly weights of the last N particles includes:

[0035]

[0036] in, For loss function, The anomaly weight of the i-th particle among the first N particles is given by [the anomaly weight]. Let be the average LOF value of the i-th particle among the first N particles. The anomaly weight of the i-th particle among the last N particles. It represents the difference between the feature data of the i-th particle and the feature data of the remaining particles among the last N particles.

[0037] In this embodiment, since a larger anomaly weight indicates a greater amount of constant information contained in the particles, the trained anomalous particles should be as related as possible to particles with larger anomaly weights. Therefore, a system is constructed... Meanwhile, the smaller the anomaly weight, the less abnormal information the particle contains. Therefore, the distance between the trained anomalous particles and the actual particles should be greater. Thus, a more robust and efficient training system is constructed. The final loss function is obtained by combining these methods. .

[0038] In some embodiments of this application, training the abnormal data detection model based on the loss function includes: acquiring the self-mixing interference signal of particulate matter within a historical preset time period; forming a training sample set by combining the self-mixing interference signals before and after sliding filtering with different filter window lengths; inputting the training sample set into the abnormal data detection model and outputting feature data and LOF values; obtaining the optimal solution of the loss function and acquiring the particulate matter containing the largest average LOF value to obtain abnormal particulate matter data.

[0039] In some embodiments of this application, the step of correcting the outlier values ​​of the self-mixing interference signal of particulate matter within the current preset time period to obtain the corrected self-mixing interference signal, and determining particulate matter information based on the corrected self-mixing interference signal, includes: correcting the self-mixing interference signal of the abnormal particulate matter data based on linear interpolation to obtain the particulate matter information corresponding to the corrected self-mixing interference signal.

[0040] In this embodiment, after accurately detecting abnormal particulate matter data, the self-mixing interference signal of the abnormal particulate matter data is corrected by local linear interpolation to obtain an accurate self-mixing interference signal.

[0041] By applying the above technical solutions, this invention utilizes a laser emitting device to emit a laser beam that irradiates particulate matter; a focusing lens to focus the laser beam onto the particulate matter; a signal acquisition device, located inside the laser emitting device, to receive the reflected beam from the particulate matter; and a signal processing device, connected to the signal acquisition device, to determine a self-mixing interference signal based on the reflected beam from the particulate matter, process the self-mixing interference signal, and determine particulate matter information based on the processed self-mixing interference signal. By detecting the self-mixing interference signal of the particulate matter, the data detection efficiency is effectively improved, and the accuracy of particulate matter signal detection is enhanced.

[0042] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A particulate matter detection system based on laser self-mixing interferometry, characterized in that, include: A laser emitting device is used to emit a laser beam to irradiate particulate matter. A focusing lens is used to focus a laser beam onto particulate matter; The signal acquisition device, located inside the laser emitting device, is used to receive the reflected beam of particulate matter; A signal processing device, connected to the signal acquisition device, is used to determine the self-mixing interference signal based on the reflected beam of the particulate matter, perform data processing on the self-mixing interference signal, and determine particulate matter information based on the processed self-mixing interference signal. The signal processing device is used for: The self-mixing interference signal of particulate matter is preprocessed, and the particulate matter characteristic data are determined based on the preprocessed self-mixing interference signal. A particulate matter sample set is established based on particulate matter characteristic data within a preset historical time period, and particulate matter clustering is performed based on particulate matter characteristic data in the particulate matter sample set. The self-mixing interference signal of each particle is filtered based on the clustering results, and a loss function is established based on the filtered self-mixing interference signal. An abnormal data detection model is established, and the abnormal data detection model is trained according to the loss function. The abnormal values ​​of the self-mixing interference signal of particulate matter within the current preset time period are detected based on the trained abnormal data detection model. The outliers of the self-mixing interference signal of particulate matter within the current preset time period are corrected to obtain the corrected self-mixing interference signal. The particulate matter information is then determined based on the corrected self-mixing interference signal. The step of clustering particulate matter based on particulate matter feature data in the particulate matter sample set includes: Step 1: Randomly select k initial cluster centers from the particulate matter sample set; Step 2: Calculate the Euclidean distance from the particulate feature data in the particulate sample set to the cluster center, and classify each particulate matter into the corresponding cluster based on the Euclidean distance from the particulate feature data in the particulate sample set to the cluster center. Step 3: Calculate the mean characteristic data of particles within each cluster, and reselect cluster centers based on the mean characteristic data of particles within each cluster; Repeat steps 2-3 until the cluster centers no longer change or the number of iterations reaches the preset maximum number of iterations, to obtain the clustering results of the particles; The step of filtering the self-mixing interference signal of each particle based on the clustering results, and establishing a loss function based on the filtered self-mixing interference signal, includes: Set up filter windows of different lengths, and perform moving average filtering on the self-mixing interference signal of particles in each cluster according to each filter window; The characteristic data of the self-mixing interference signal of particulate matter after statistical filtering are obtained at different filter window lengths. The difference between the characteristic data of the self-mixing interference signal of particulate matter after filtering at different filter window lengths and the characteristic data of the self-mixing interference signal before filtering at the filter window length is calculated to obtain the characteristic data difference sequence of particulate matter. The characteristic data of the self-mixing interference signal of particulate matter at different filter window lengths are obtained, and the outlier parameter sequence is constructed based on the characteristic data of the self-mixing interference signal of particulate matter at different filter window lengths. Calculate the correlation coefficient between the feature data difference sequence and the outlier parameter sequence, determine the anomaly weight of the corresponding particulate matter based on the correlation coefficient between the feature data difference sequence and the outlier parameter sequence, and establish a loss function based on the anomaly weight of the particulate matter.

2. The particulate matter detection system based on laser self-mixing interferometry according to claim 1, characterized in that, The preprocessing of the self-mixing interference signal of particulate matter, and the determination of particulate matter characteristic data based on the preprocessed self-mixing interference signal, includes: Wavelet transform is performed on the self-mixing interference signal of particulate matter, and the characteristic data of particulate matter are determined based on the wavelet transform results.

3. The particulate matter detection system based on laser self-mixing interferometry according to claim 1, characterized in that, The construction of the outlier parameter sequence based on the characteristic data of the self-mixing interference signal of particulate matter at different filter window lengths includes: The feature data of the self-mixing interference signal of particulate matter at different filter window lengths are normalized to obtain the normalized feature data. Based on the LOF algorithm, the LOF value of the feature data for each filter window length is determined according to the normalized feature data, and the outlier parameter sequence is constructed based on the LOF value of the feature data of particulate matter.

4. The particulate matter detection system based on laser self-mixing interferometry according to claim 1, characterized in that, The step of establishing a loss function based on the anomaly weights of particulate matter includes: Obtain the anomaly weight of each particulate matter and sort the particulate matter in descending order of anomaly weight; Extract the outlier weights of the first N particles and the next N particles based on the sorting results, and establish a loss function based on the outlier weights of the first N particles and the next N particles.

5. The particulate matter detection system based on laser self-mixing interferometry according to claim 4, characterized in that, The step of establishing a loss function based on the anomaly weights of the first N particles and the next N particles includes: in, For loss function, The anomaly weight of the i-th particle among the first N particles is given by [the anomaly weight]. Let be the average LOF value of the i-th particle among the first N particles. The anomaly weight of the i-th particle among the last N particles. It represents the difference between the feature data of the i-th particle and the feature data of the remaining particles among the last N particles.

6. The particulate matter detection system based on laser self-mixing interferometry according to claim 1, characterized in that, The training of the anomaly detection model based on the loss function includes: The self-mixing interference signal of particulate matter within a preset historical time period is obtained, and the self-mixing interference signal before and after sliding filtering according to different filtering window lengths is used to form a training sample set. Input the training sample set into the anomaly detection model, and the output is feature data and LOF value; After obtaining the optimal solution of the loss function, the particles containing the largest average LOF value are obtained to obtain the abnormal particle data.

7. The particulate matter detection system based on laser self-mixing interferometry according to claim 1, characterized in that, The step of correcting outliers in the self-mixing interference signal of particulate matter within the current preset time period to obtain a corrected self-mixing interference signal, and determining particulate matter information based on the corrected self-mixing interference signal, includes: The self-mixing interference signal of abnormal particulate matter data is corrected based on the linear interpolation method, and the particulate matter information corresponding to the corrected self-mixing interference signal is obtained.

Images