Photoelectric circumference lens detection method and system based on pollution degree index system model

Abstract

The invention provides a photoelectric circumference lens detection method and system based on a pollution degree index system model, and the method comprises the following steps: 100, obtaining the photosensitive voltage data of a dual-channel CCD sensor and the circumference data of a cigarette, and carrying out the data cleaning and preprocessing; 200, based on the preprocessed data obtained in the step 1, a pollution degree index system model is established, and the model comprises calculation of a stable trend of a photosensitive voltage curve, quantification offset and comprehensive scoring to generate a pollution degree index; step 300, pollution positioning is carried out according to the pollution degree index and the orientation mapping relation, and a visual alarm signal containing the pollution orientation and the pollution level is generated; and step 400, constructing a time sequence model based on the historical pollution data, predicting the pollution development trend of the photoelectric circumference lens and generating maintenance suggestions. Through a big data technology, the method focuses on research and mining of distribution and trend change performance of cigarette making machine sensor data, and early warning analysis is carried out on the state of a sensor part.

Description

Technical Field

[0001] The present invention relates to the technical field of photoelectric circular lens detection, and in particular to a photoelectric circular lens detection method and system based on a contamination index system model. Background Art

[0002] Currently, tobacco factories lack corresponding intelligent data supervision for the self-inspection and self-test maintenance of cigarette machine sensors, especially photoelectric circular lenses. Existing maintenance technology cannot provide operators and related technicians with historical data related to photoelectric circular lens detection in a timely manner, and there is a lack of overall display and research on long-term or short-term significant abnormalities.

[0003] At the same time, as a component in the cigarette making machine sensor that is more susceptible to contamination and requires operators to carry out long-term and continuous planning for cleaning, predicting its future usage is also a key issue in the efficient production of cigarette making machines. Summary of the Invention

[0004] This invention proposes a photoelectric circular lens detection method and system based on a contamination index system model, aiming to partially resolve the existing technical problems. The specific technical solution is as follows:

[0005] On the one hand, a photoelectric circumferential lens detection method based on a contamination index system model is proposed, comprising the following steps:

[0006] Step 100: Acquire the photosensitive voltage data and cigarette circumference data of the dual-channel CCD sensor, and perform data cleaning and preprocessing;

[0007] Step 200: Based on the pre-processed data obtained in step 1, a pollution index system model is established, wherein the model includes calculating the steady trend of the photosensitive voltage curve, quantifying the offset, and generating a pollution index by comprehensive scoring;

[0008] Step 300: Locate the pollution based on the pollution index and the position mapping relationship, and generate a visual alarm signal including the pollution position and pollution level;

[0009] Step 400: Build a time series model based on historical pollution data to predict the pollution development trend of the optoelectronic circumferential lens and generate maintenance recommendations.

[0010] As a preferred technical solution, step 100 specifically includes:

[0011] Step 110: synchronously collecting the rated circumference value, the actual circumference value of the produced cigarettes, and the original voltage signal of the dual-channel CCD sensor;

[0012] Step 120: Using a sliding window method to remove outliers from the raw data, including: setting a ±3σ range to remove voltage spike noise; using cubic spline interpolation to fill in missing data; and performing a moving average filter on the cigarette circumference data.

[0013] Step 130: Create a data set with timestamp alignment to form a multi-dimensional monitoring data table with second-level granularity.

[0014] As a preferred technical solution, step 200 specifically includes the following steps:

[0015] Step 210: aligning the reference curves of the dual-channel CCD photosensitive voltage values ​​using a dynamic time warping algorithm;

[0016] Step 220: extracting the trend component of the photosensitive voltage curve through wavelet transform;

[0017] Step 230: Calculate the Hausdorff distance between the real-time curve and the reference curve as an offset;

[0018] Step 240: Construct a pollution index based on the fuzzy comprehensive evaluation method: pollution index = α × offset + β × mutation frequency + γ × duration; where α + β + γ = 1, α = 0.6, β = 0.3, γ = 0.1.

[0019] As a preferred technical solution, the offset calculation in step 230 specifically includes:

[0020] Step 231: performing piecewise linearization processing on the reference curve to divide it into n characteristic intervals;

[0021] Step 232: extracting the integral area S_real of the corresponding interval from the real-time curve;

[0022] Step 233: Calculate the area deviation rate δ = (S_real - S_base) / S_base × 100%;

[0023] Step 234: When δ exceeds the threshold value for 5 consecutive seconds, a trend abnormality flag is triggered, wherein the threshold value range is 3% to 4%.

[0024] As a preferred technical solution, the establishment of the orientation mapping relationship in step 300 includes:

[0025] Step 310: Establish a lens orientation coordinate system and divide it into four quadrants with the detection point as the origin;

[0026] Step 311: The left channel CCD corresponds to the X-axis direction, and the right channel CCD corresponds to the X+ axis direction;

[0027] Step 312: The up and down directions correspond to the Y+ and Y- axis directions respectively;

[0028] Step 320: Calculate the pollution azimuth angle θ=arctan(ΔV_right / ΔV_left) based on the curve mutation position, where ΔV_right is the photosensitive voltage offset of the right channel CCD sensor, and ΔV_left is the photosensitive voltage offset of the left channel CCD sensor;

[0029] Step 330: Determine the specific pollution direction based on the 45° fan-shaped area where the θ value falls.

[0030] As a preferred technical solution, a pollution diffusion prediction model is also included, including:

[0031] Step 340: Construct the pollution diffusion equation:

[0032] Where C is the pollution concentration, D is the diffusion coefficient, t is time, which describes the dynamic process of pollution diffusion in seconds, x and y are spatial coordinates, representing the two-dimensional position of the lens surface in millimeters, and v is the flow velocity vector, which represents the flow velocity of the clean air flow or other flow that drives the migration of pollutants;

[0033] Step 350: Optimize model parameters based on historical cleaning records to predict pollutant diffusion paths.

[0034] As a preferred technical solution, step 400 includes:

[0035] Step 410: Construct an LSTM neural network prediction model, with input features including: pollution index time series, ambient temperature and humidity data, equipment operating time, and historical cleaning cycles;

[0036] Step 420: Output prediction results including: optimal cleaning time window, estimated cleaning time, and recommended cleaning agent type.

[0037] On the other hand, a photoelectric circumferential lens contamination detection system based on the above method is also proposed, comprising:

[0038] The data acquisition module is equipped with a dual-channel CCD sensor and a cigarette circumference measurement device; the signal processing module includes an AD converter and a digital filter to perform data preprocessing; and the analysis and decision-making module is equipped with:

[0039] Trend analysis unit, which calculates the dynamic characteristics of the photosensitive voltage curve in real time; pollution degree calculation unit, which executes the fuzzy comprehensive evaluation algorithm; orientation positioning unit, which calculates the pollution coordinates based on the geometric mapping relationship; human-computer interaction module, which includes a touch screen and an audible and visual alarm device.

[0040] As a preferred technical solution, the analysis and decision-making module also includes: a self-learning unit (34), which uses a reinforcement learning algorithm to optimize model parameters, specifically including: defining a state space S = {pollution degree, environmental parameters, equipment status}, and defining an action space A = {immediate cleaning, delayed cleaning, adjustment of detection parameters}.

[0041] On the third aspect, a computer-readable storage medium is also proposed, which stores a computer program. When the program is executed by a processor, it implements the steps of the above method, including: a real-time data acquisition interface program; a multi-threaded data processing engine; a curve analysis module based on OpenCV; a visual alarm generator; and a predictive maintenance decision tree.

[0042] Compared with the prior art, the present invention has achieved the following technical effects: Through big data technology, the present invention focuses on the research and mining of the distribution and trend change performance of cigarette machine sensor data (mainly the photosensitive voltage data of the photoelectric circumferential lens), and conducts early warning analysis on the status of sensor components (mainly the pollution, occlusion and other abnormalities of the photoelectric circumferential lens). Through the study of basic data, it is possible to indirectly discover abnormal phenomena of the equipment sensor components, and issue clear reminders to the operators in a timely manner, which acts as an alarm; at the same time, after the alarm is resolved, the corresponding data in the early warning phenomenon is stored, and later abnormal data modeling and learning are carried out, and predictive reminders are issued in time before the next abnormal phenomenon occurs, so that the operators can repair the equipment in time before the abnormality occurs, reducing the losses that should have occurred. Through the cycle of the two action feedback modes, a long-term big data analysis is comprehensively conducted on the equipment components targeted by the research, the overall equipment efficiency is improved, the information data is effectively and reasonably used to maintain the lens components, unnecessary losses are reduced, and the equipment is kept low in consumption and operates efficiently.

[0043] Other features and advantages of the present application will be set forth in the following description, and in part will be apparent from the description, or may be learned by practicing the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a photoelectric circular lens detection method based on a contamination index system model proposed in Example 1 of the present invention;

[0045] Figure 2 Schematic diagram of the data cleaning and preprocessing process in step 100 in Example 1 of the present invention;

[0046] Figure 3 This is a schematic diagram of the process of establishing a pollution index system model in step 200 of Example 1 of the present invention;

[0047] Figure 4This is a schematic diagram of an algorithm flow for establishing an orientation mapping relationship in step 300 in embodiment 1 of the present invention;

[0048] Figure 5 A schematic diagram of a process for establishing a pollution diffusion prediction model in Example 1 of the present invention;

[0049] Figure 6 Schematic diagram of the layout structure of the dual-channel CCD, cigarette and photoelectric circumferential lens in Example 1 of the present invention;

[0050] Figure 7 The curve trend of the photosensitive voltage value of the CCD1 channel of the photoelectric circular lens of a cigarette making machine in Example 2 of the present invention per unit time is shown;

[0051] Figure 8 The change in the azimuth contamination index of the photoelectric circumferential lens of a cigarette making machine in Example 2 of the present invention over a period of time;

[0052] Figure 9 The pollution position of the photoelectric circumferential lens of a cigarette making machine in Example 2 of the present invention over a period of time;

[0053] Figure 10 The trend of the photosensitive voltage value curve of the CCD2 channel of the photoelectric circular lens of a cigarette making machine in Example 3 of the present invention per unit time;

[0054] Figure 11 The change in the azimuth contamination index of the photoelectric circumferential lens of a cigarette making machine in Example 3 of the present invention over a period of time;

[0055] Figure 12 The pollution position of the photoelectric circumferential lens of a cigarette making machine in Example 3 of the present invention over a period of time;

[0056] Figure 13 This is a schematic diagram of a photoelectric circumferential lens contamination detection system based on the above method proposed in Example 4 of the present invention.

[0057] Description of reference numerals:

[0058] Data acquisition module 10; signal processing module 20; analysis and decision module 30; human-computer interaction module 40. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention and corresponding drawings. In the description of the present invention, it should be noted that the term "or" is generally used in the sense of including "and / or" unless the content clearly indicates otherwise.

[0060] It should be understood that the various steps described in the method embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0061] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Definitions of other terms are provided below.

[0062] It should be noted that the modifications of "one" and "plurality" mentioned in this application are illustrative rather than restrictive. Those skilled in the art will understand that unless the context clearly indicates otherwise, they should be understood as "one or more." "Plurality" should be understood as two or more.

[0063] In the present invention, the technical solution mainly includes: obtaining single cigarette data and photosensitive voltage data, for the two CCD photoelectric detection channels (distributed on both sides of the cigarette, distributed in the form of plane X-axis and Y-axis with the direction of cigarette transmission), taking the circular mapping of the detected cigarette as the center for photosensitive data analysis and research, and establishing a pollution index system model based on big data. Real-time photoelectric circumferential lens status detection is carried out, and when the pollution level reaches the threshold, an early warning reminder is issued, and the occurrence and location of lens pollution are displayed based on the photosensitive feedback information. Long-term supervision is implemented, and lens status health reports are provided on schedule according to the needs of the cigarette factory. Based on information such as pollution data and the time and location of occurrence, a corresponding comprehensive lens analysis report is given, so that operators can better rationalize and predict lens pollution and deal with it.

[0064] The aforementioned light-sensing data analysis analyzes the specific light-sensing data reflected by the photosensitive voltage on the circumference of the detection photoelectric circle, and determines whether the lens is contaminated or obstructed by the data changes. Based on electrical principles, the specific value of the photosensitive voltage depends on the light sensitivity of the detected object. The stronger the light sensitivity (the less obstructed the object), the corresponding voltage value is smaller; conversely, the darker the area, the corresponding photosensitive voltage value is larger.

[0065] The above-mentioned pollution index system model uses the aforementioned photosensitive data analysis as the basis for big data modeling. In actual production, the photosensitive voltage data shows a trend of being lower on both sides and higher in the middle, which is consistent with the fact that cigarettes are basically blocked at the center of the lens, resulting in a peak in the photosensitive value. When the photosensitive voltage trend obtained during the normal, lens-clean production stage tends to be stable, the sudden appearance of long-term high values ​​of data points on both sides can be considered to be caused by occlusion contamination of the lens. Based on this principle, the sensitivity of the lens is used as the pollution analysis standard for pollution analysis. The establishment of the pollution index is based on the stable trend when the lens performs well. After the changing trend curve is numerically quantified by a specific algorithm, the pollution index is given with a scoring system as the core.

[0066] The lens status detection system, based on the contamination index system model, eliminates invalid values ​​from the photosensitive voltage data of the two channels and divides the lens orientation. Using the upper and lower limits of the photosensitive voltage as a benchmark, it determines the current photoelectric data, including whether a detection is currently occurring, whether the cigarette cross section is aligned or offset, and whether the specific positioning of the lens on both sides of the channel is correct. The system monitors the photoelectric circumference of the lens in real time, displays the presence and location of lens contamination based on the photosensitive feedback, and issues an early warning when the contamination level reaches a threshold.

[0067] This long-term monitoring, including monitoring of photoelectric circumferential lenses, utilizes real-time analysis to achieve coordinated management and early warning as lenses age. This allows operators to promptly clean the circumferential lenses and replace components when necessary. This effectively and rationally utilizes information data to maintain lens components, reduce unnecessary wear and tear, and maintain efficient operation with minimal consumption.

[0068] For detailed technical solutions, please refer to the following examples.

[0069] Example 1

[0070] like Figure 1 The photoelectric circumferential lens detection method based on the contamination index system model proposed in this embodiment includes the following steps:

[0071] Step 100: Acquire the photosensitive voltage data and cigarette circumference data from the dual-channel CCD sensor and perform data cleaning and preprocessing. Specifically, this includes statistical classification and visualization of the raw photosensitive voltage data, acquiring single cigarette data and photosensitive voltage data, performing data cleaning and preprocessing, and then statistically displaying their changes.

[0072] Step 200: Based on the preprocessed data obtained in Step 1, a pollution index system model is established. This model calculates the steady trend of the photosensitive voltage curve, quantifies the offset, and generates a pollution index through comprehensive scoring. Specifically, the model analyzes and monitors the trend of the photosensitive voltage data in the two CCD photoelectric detection channels, and uses a specific algorithm to derive the pollution index model.

[0073] Step 300: Based on the pollution index and the orientation mapping relationship, pollution is located, generating a visual alarm signal that includes the pollution location and pollution level. Specifically, with the mapping point of the detected cigarette as the center, the orientation of the corresponding lens of the two channels is analyzed to determine the specific pollution obstruction situation.

[0074] Step 400: Build a time series model based on historical pollution data to predict the pollution development trend of the optoelectronic circumferential lens and generate maintenance recommendations.

[0075] Preferably, step 100 specifically includes:

[0076] Step 110: Synchronously collect the rated circumference value, the actual circumference value of produced cigarettes, and the original voltage signal of the dual-channel CCD sensor.

[0077] Step 120: Use the sliding window method to remove outliers from the original data, including: setting the ±3σ range to remove voltage spike noise; using cubic spline interpolation to fill in missing data; and performing moving average filtering on the cigarette circumference data. Among them, the sliding window is a data structure and algorithm problem-solving technique that is suitable for problems involving arrays or lists. 3σ criterion: The 68-95-99.7 principle is the percentage of data in a normal distribution that is less than one standard deviation, two standard deviations, and three standard deviations from the mean. More precise numbers are 68.27%, 95.45%, and 99.73%. This is the 3σ criterion:

[0078] The probability that the value is distributed in (uo,u+o) is 0.6827;

[0079] The probability that the value is distributed in (u-20,u+20) is 0.9545;

[0080] The probability that the value is distributed in (u-3σ,u+3σ) is 0.9973.

[0081] Step 130: Create a data set with timestamp alignment to form a multi-dimensional monitoring data table with second-level granularity.

[0082] Preferably, step 200 specifically includes the following steps:

[0083] Step 210: Using a dynamic time warping algorithm to align the reference curves of the dual-channel CCD photosensitive voltage values.

[0084] Step 220: extracting the trend component of the photosensitive voltage curve through wavelet transform.

[0085] Step 230: Calculate the Hausdorff distance between the real-time curve and the reference curve as an offset.

[0086] Step 240: Construct a pollution index based on the fuzzy comprehensive evaluation method: pollution index = α × offset + β × mutation frequency + γ × duration; where α + β + γ = 1, α = 0.6, β = 0.3, γ = 0.1.

[0087] Preferably, the offset calculation in step 230 specifically includes:

[0088] Step 231: performing piecewise linearization processing on the reference curve to divide it into n characteristic intervals;

[0089] Step 232: extracting the integral area S_real of the corresponding interval from the real-time curve;

[0090] Step 233: Calculate the area deviation rate δ = (S_real - S_base) / S_base × 100%;

[0091] Step 234: When δ exceeds the threshold value for 5 consecutive seconds, a trend abnormality flag is triggered, wherein the threshold range is set to 3% to 4%.

[0092] Preferably, the establishment of the orientation mapping relationship in step 300 specifically includes:

[0093] Step 310: Establish a lens orientation coordinate system and divide it into four quadrants with the detection point as the origin.

[0094] Step 311: The left channel CCD corresponds to the X-axis direction, and the right channel CCD corresponds to the X+ axis direction;

[0095] Step 312: The up and down directions correspond to the Y+ and Y- axis directions respectively;

[0096] Step 320: Calculate the pollution azimuth angle θ=arctan(ΔV_right / ΔV_left) based on the curve mutation position, where ΔV_right is the photosensitive voltage offset of the right channel CCD sensor, and ΔV_left is the photosensitive voltage offset of the left channel CCD sensor;

[0097] Step 330: Determine the specific pollution direction based on the 45° fan-shaped area where the θ value falls.

[0098] Specifically, the coordinate system is defined as follows: X-axis (left-right direction): the left CCD (L) and the right CCD (R) are symmetrically distributed on both sides of the cigarette. Y-axis (transmission direction): the cigarette lies horizontally and moves along the Y-axis (direction of the arrow).

[0099] Z-axis (direction of the optical path, the light source is integrated inside the photoelectric circular lens, emitting parallel light of a specific wavelength, such as near-infrared 850nm): the light source (photoelectric lens) emits light along the Z-axis, and the axis of the cigarette is orthogonal to the optical path. Position of key components: Center of the cigarette: located at the origin of the coordinate system (O), lying horizontally in the Y-axis direction. Dual-channel CCD: installed in the negative direction of the X-axis (left CCD) and the positive direction of the X-axis (right CCD) respectively. Photoelectric lens: located in the negative direction of the Z-axis, emitting the detection light path (positive direction of the Z-axis) to the cigarette. Detection process: The cigarette is transmitted horizontally along the Y-axis, cross-cutting the Z-axis optical path. The left and right CCDs monitor the light intensity distribution after penetrating the cigarette in real time, and determine the direction of contamination by differential signals.

[0100] Preferably, a pollution diffusion prediction model is also included, including:

[0101] Step 340: Construct the pollution diffusion equation:

[0102] Where C is the pollution concentration, D is the diffusion coefficient, t is time, which describes the dynamic process of pollution diffusion in seconds, x and y are spatial coordinates, representing the two-dimensional position of the lens surface in millimeters, and v is the flow velocity vector, which represents the flow velocity of the clean air flow or other flow that drives the migration of pollutants;

[0103] Step 350: Optimize model parameters based on historical cleaning records to predict pollutant diffusion paths.

[0104] As a preferred technical solution, step 400 includes:

[0105] Step 410: Construct an LSTM neural network prediction model, with input features including: pollution index time series, ambient temperature and humidity data, equipment operating time, and historical cleaning cycles;

[0106] Step 420: Output prediction results including: optimal cleaning time window, estimated cleaning time, and recommended cleaning agent type.

[0107] Example 2

[0108] This embodiment provides an example of applying the above method to the lens detection analysis of channel CCD1. Figure 7 As shown, the photosensitive voltage of the CCD1 channel shows a normal cigarette detection performance of high in the middle and low on both sides.

[0109] like Figure 8As shown in the figure, a small peak interval with a significant sudden increase appears on the right side of the curve of the CCD1 channel, which is the manifestation of the sudden increase of pollution. By monitoring its pollution index over a period of time, it can be concluded that the contamination of the lens has not been cleaned in time, and the operator needs to check and clean it as soon as possible.

[0110] like Figure 9 As shown, the location of the contamination is relatively stable, at a distance of 12.2% from the rightmost side of the lens. There is no random occlusion, and manual cleaning is required by the operator.

[0111] Example 3

[0112] This embodiment provides an example of applying the above method to the lens detection analysis of channel CCD2. Figure 10 As shown, the photosensitive voltage of the CCD2 channel shows a normal cigarette detection performance of high in the middle and low on both sides.

[0113] like Figure 11 As shown in the figure, there is a small peak interval with a significant sudden increase on the left side of the curve of the CCD2 channel, which is the manifestation of the sudden increase wave of pollution. By monitoring its pollution index over a period of time, it can be concluded that the contamination of the lens has not been cleaned in time, and the operator needs to check and clean it as soon as possible.

[0114] like Figure 12 As shown, the location of the contamination is relatively stable, at a ratio of 13.8% from the leftmost side of the lens. There is no random occlusion, and manual cleaning is required by the operator.

[0115] Example 4

[0116] like Figure 13 As shown, this embodiment proposes a photoelectric circumferential lens contamination detection system based on the above method, including:

[0117] The data acquisition module 10 is equipped with a dual-channel CCD sensor and a cigarette circumference measuring device;

[0118] The signal processing module 20 includes an AD converter and a digital filter to perform data preprocessing;

[0119] The analysis and decision module 30 is configured with: a trend analysis unit that calculates the dynamic characteristics of the photosensitive voltage curve in real time; a pollution degree calculation unit that executes a fuzzy comprehensive evaluation algorithm; and an orientation positioning unit that calculates pollution coordinates based on geometric mapping relationships.

[0120] The human-computer interaction module 40 includes a touch screen display and an audible and visual alarm device.

[0121] Preferably, the analysis and decision module 30 also includes: a self-learning unit (34), which uses a reinforcement learning algorithm to optimize model parameters, specifically including: defining a state space S = {pollution degree, environmental parameters, equipment status}, and defining an action space A = {immediate cleaning, delayed cleaning, adjustment of detection parameters}.

[0122] On the other hand, this embodiment also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in Example 1, including: a real-time data acquisition interface program; a multi-threaded data processing engine; an OpenCV-based curve analysis module; a visual alarm generator; and a predictive maintenance decision tree.

[0123] Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

Claims

1. A photoelectric circumferential lens detection method based on a pollution index system model, characterized in that: The following steps are involved: Step 100: Acquire photosensitive voltage data and cigarette circumference data from a dual-channel CCD sensor, and perform data cleaning and preprocessing. Step 200: Establish a pollution index system model based on the preprocessed data obtained in step 1. The model includes calculating the stable trend of the photosensitive voltage curve, quantifying the offset, and generating a pollution index through comprehensive scoring. Step 300: Locate the pollution based on the pollution index and the position mapping relationship, and generate a visual alarm signal including the pollution position and pollution level; Step 400: Build a time series model based on historical pollution data to predict the pollution development trend of the optoelectronic circumferential lens and generate maintenance recommendations.

2. The photoelectric circumferential lens detection method based on the pollution index system model according to claim 1 is characterized in that: The step 100 specifically includes: Step 110: Synchronously collect the rated circumference value, the actual circumference value of produced cigarettes, and the raw voltage signal of the dual-channel CCD sensor; Step 120: Use a sliding window method to remove outliers from the raw data, including: setting a ±3σ range to remove voltage spike noise; using cubic spline interpolation to fill in missing data; and performing moving average filtering on the cigarette circumference data; Step 130: Create a data set with timestamp alignment to form a multi-dimensional monitoring data table with second-level granularity.

3. The method according to claim 2, characterized in that The step 200 specifically includes the following steps: Step 210: aligning the reference curves of the dual-channel CCD photosensitive voltage values ​​using a dynamic time warping algorithm; Step 220: extracting the trend component of the photosensitive voltage curve through wavelet transform; Step 230: Calculate the Hausdorff distance between the real-time curve and the reference curve as an offset; Step 240: Construct a pollution index based on the fuzzy comprehensive evaluation method: pollution index = α × offset + β × mutation frequency + γ × duration; where α + β + γ = 1, α = 0.6, β = 0.3, γ = 0.

1.

4. The method according to claim 3, characterized in that The offset calculation in step 230 specifically includes: Step 231: performing piecewise linearization processing on the reference curve to divide it into n characteristic intervals; Step 232: extracting the integral area S_real of the corresponding interval from the real-time curve; Step 233: Calculate the area deviation rate δ = (S_real - S_base) / S_base × 100%; Step 234: When δ exceeds the threshold value for 5 consecutive seconds, a trend abnormality flag is triggered, wherein the threshold value range is 3% to 4%.

5. The method according to claim 1, wherein The establishment of the orientation mapping relationship in step 300 includes: Step 310: Establish a lens orientation coordinate system and divide it into four quadrants with the detection point as the origin; Step 311: The left channel CCD corresponds to the X-axis direction, and the right channel CCD corresponds to the X+ axis direction; Step 312: The up and down directions correspond to the Y+ and Y- axis directions respectively; Step 320: Calculate the pollution azimuth angle θ=arctan(ΔV_right / ΔV_left) based on the curve mutation position, where ΔV_right is the photosensitive voltage offset of the right channel CCD sensor, and ΔV_left is the photosensitive voltage offset of the left channel CCD sensor; Step 330: Determine the specific pollution direction according to the 45° fan-shaped area in which the azimuth angle θ value falls.

6. The method according to claim 5, characterized in that Also included are pollution spread prediction models, including: Step 340: Construct the pollution diffusion equation: Where C is the pollution concentration, D is the diffusion coefficient, t is time, which describes the dynamic process of pollution diffusion in seconds, x and y are spatial coordinates, representing the two-dimensional position of the lens surface in millimeters, and v is the flow velocity vector, which represents the flow velocity of the clean air flow or other flow that drives the migration of pollutants; Step 350: Optimize model parameters based on historical cleaning records to predict pollutant diffusion paths.

7. The method according to claim 1, characterized in that The step 400 includes: Step 410: Construct an LSTM neural network prediction model, with input features including: pollution index time series, ambient temperature and humidity data, equipment operating time, and historical cleaning cycles; Step 420: Output prediction results including: optimal cleaning time window, estimated cleaning time, and recommended cleaning agent type.

8. A system based on the photoelectric circumferential lens contamination detection method according to any one of claims 1 to 7, characterized in that: include: Data acquisition module, equipped with a dual-channel CCD sensor and cigarette circumference measurement device; The signal processing module includes an AD converter and a digital filter, and performs the data preprocessing as described in any one of claims 1 to 7; the analysis and decision-making module is configured with: a trend analysis unit, which calculates the dynamic characteristics of the photosensitive voltage curve in real time, a pollution degree calculation unit, which executes a fuzzy comprehensive evaluation algorithm, and an orientation positioning unit, which calculates the pollution coordinates based on the geometric mapping relationship; and a human-computer interaction module, which includes a touch display screen and an audio-visual alarm device.

9. The system according to claim 8, characterized in that The analysis and decision module also includes: a self-learning unit (34), which uses a reinforcement learning algorithm to optimize model parameters, specifically including: defining a state space S = {pollution degree, environmental parameters, equipment status}, and defining an action space A = {immediate cleaning, delayed cleaning, adjustment of detection parameters}.

10. A computer-readable storage medium, characterized in that A computer program is stored, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 7, including: a real-time data acquisition interface program; a multi-threaded data processing engine; an OpenCV-based curve analysis module; a visual alarm generator; and a predictive maintenance decision tree.

Images