Air compressor control method and control device capable of adaptively adjusting load changes

Abstract

The present invention discloses an air compressor control method and control device with adaptive load changes, which relates to the technical field of air compressors. The method includes: inputting the production schedule into a predetermined twin bottle blowing production line to simulate demand and determine the simulated demand; predicting interference based on environmental factors and generating a predicted interference coefficient; compensating for the simulated demand and optimizing air compressor parameters according to the optimized predicted load configuration; analyzing production deviations according to the bottle body data set and determining production deviation values; outputting real-time air compressor control deviations; and adaptively compensating for the air compressor parameters of the previous node. The method solves the technical problems of the existing air compressor control that it is difficult to accurately predict and cope with the dynamic changes of the air compressor load during the production process, which leads to poor production efficiency and product quality, and achieves accurate prediction of the air compressor load and adaptive adjustment, achieving the technical effect of improving overall production efficiency and improving product quality stability.

Description

Technical Field

[0001] The present application relates to the technical field related to air compressors, and in particular to an air compressor control method and a control device that are adaptive to load changes. Background Art

[0002] In modern industrial production, air compressors are one of the key equipment. The stability of their performance and the level of their efficiency directly affect the operating efficiency and product quality of the entire production line. With the rapid development of the intelligent manufacturing industry, traditional air compressor control methods have gradually been unable to meet the complex and changeable production needs. Especially in the bottle blowing production line that has extremely high requirements for air pressure stability, the adaptive control of air compressors is particularly important. As an important part of the plastic packaging industry, the production efficiency and product quality of the bottle blowing production line are directly affected by the stability of the air supply of the air compressor. However, in the actual production process, due to changes in production schedules, fluctuations in production environment factors, and wear and aging of the equipment itself, the load of the air compressor often presents a large degree of uncertainty. Traditional air compressor control methods are often difficult to accurately predict and adapt to the dynamic changes of the load during the production process, resulting in energy waste and reduced production efficiency. It may also lead to production deviations, such as bottle size deviations and uneven wall thickness, which affect the final product quality.

[0003] Therefore, in the current air compressor control related technologies, there are technical problems that make it difficult to accurately predict and respond to the dynamic changes in the air compressor load during the production process, which in turn leads to poor production efficiency and product quality. Summary of the invention

[0004] The present application provides an air compressor control method and control device with adaptive load changes, and adopts technical means such as integration of digital twins, environmental factor prediction, and real-time data analysis to solve the technical problems of existing air compressor control, such as difficulty in accurately predicting and coping with the dynamic changes of air compressor load during the production process, which leads to poor production efficiency and product quality. The application realizes accurate prediction of air compressor load and adaptive adjustment, thereby achieving the technical effect of improving overall production efficiency and enhancing product quality stability.

[0005] The present application provides an air compressor control method that is adaptive to load changes, the method comprising: calling a production schedule within a predetermined period, inputting the production schedule into a predetermined bottle blowing twin production line to simulate demand, and determining a simulated demand, wherein the predetermined bottle blowing twin production line simulates and constructs a bottle blowing machine production line through a digital twin; real-time collection of environmental factors of a production workshop, interference prediction based on the environmental factors, and generating a predicted interference coefficient; compensating the simulated demand based on a predicted leakage coefficient and the predicted interference coefficient, determining an optimized predicted load, and optimizing air compressor parameters according to the optimized predicted load; and The optimized air compression parameters execute air compressor control within a predetermined period, and collect bottle body information at a fixed point during the control process, perform production deviation analysis based on the bottle body data set, and determine the production deviation value, wherein the production deviation value includes bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient; the bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient are input into a control deviation predictor, and a real-time air compression control deviation is output; a control compensation parameter is determined based on the real-time air compression control deviation, and the air compression parameter of the previous node is adaptively compensated according to the control compensation parameter.

[0006] In a possible implementation, the predicted leakage coefficient is obtained and the following processing is also performed: query the production log of the bottle blowing machine to obtain a control monitoring data sequence within a predetermined time window, wherein the control monitoring data includes historical air pressure parameters and historical output parameters; perform leakage proportion analysis based on the control monitoring data sequence to generate a leakage proportion sequence, wherein the leakage proportion is the ratio of the difference between the historical air pressure parameters and the historical output parameters to the historical air pressure parameters; take the monitoring node as the X-axis and the leakage proportion as the Y-axis, construct a leakage trend prediction curve based on the leakage proportion sequence, determine the predicted leakage proportion within a predetermined period according to the leakage trend prediction curve, and set the predicted leakage proportion as the predicted leakage coefficient.

[0007] In a possible implementation, the production deviation value is determined by performing the following processing: at a predetermined monitoring node, using a dual-element monitoring device to collect N bottle bodies to obtain N bottle body image sets and N bottle body point cloud sets, wherein N is an integer greater than 10; performing image stitching of the N bottle body image sets to obtain N complete bottle body images, and using a convolutional neural network to extract contour features of the N complete bottle body images to obtain N bottle body contours; performing point cloud fitting on the N bottle body point cloud sets based on the N bottle body contours, and generating N three-dimensional bottle body models according to the point cloud fitting results; performing production deviation analysis based on the N complete bottle body images and the N three-dimensional bottle body models to determine the production deviation value.

[0008] In a possible implementation, point cloud fitting is performed on the N bottle point cloud sets based on the N bottle body contours, and the following processing is performed: a first bottle body contour is randomly selected from the N bottle body contours, and a first bottle body point cloud set of the first bottle body contour is obtained; with the first bottle body contour as a constraint, the first bottle body point cloud set is randomly fitted to obtain a first fitting result, and a first fitting degree is calculated, wherein the first fitting degree is a ratio of the number of point clouds falling within the first bottle body contour to the total number of point clouds in the first bottle body point cloud set; with the first bottle body contour as a constraint again, the first bottle body point cloud set is randomly fitted to obtain a second fitting result and a second fitting degree; iterative fitting is performed until a predetermined number of iterations is reached, and a fitting result with the highest fitting degree is output as a first optimal fitting result.

[0009] In a possible implementation, production deviation analysis is performed based on the N complete bottle body images and the N three-dimensional bottle body models, and the following processing is further performed: grayscale processing and grid division are performed on the N complete bottle body images to obtain N grayscale grid images of the bottle body; a standard bottle body grayscale grid image is obtained, and based on the standard bottle body grayscale image, bottle body wall thickness deviation and bottle body wall thickness uniformity analysis is performed on the N bottle body grayscale grid images to obtain N wall thickness deviation ratios and N wall thickness deviation variances, and the mean of the N wall thickness deviation ratios and the N wall thickness deviation variances are respectively calculated to obtain the bottle body wall thickness deviation and the bottle body wall thickness uniformity coefficient; a standard three-dimensional bottle body model is obtained, and based on the standard three-dimensional bottle body model, bottle body size deviation and bottle body size uniformity analysis is performed on the N three-dimensional bottle body models to obtain N size deviation ratios and N size deviation variances, and the mean of the N size deviation ratios and the N size deviation variances are respectively calculated to obtain the bottle body size deviation and the bottle body size uniformity coefficient.

[0010] In a possible implementation, a control deviation predictor is constructed, and the following processing is also performed: taking the bottle blowing machine production line and the bottle blowing air compressor as constraints, homologous positive sample retrieval is performed based on big data to obtain a sample data set, wherein the sample data includes the sample bottle body size deviation, the sample bottle body size uniformity coefficient, the sample bottle body wall thickness deviation, the sample bottle body wall thickness uniformity coefficient and the sample pressure deviation, and the sample flow deviation; the sample data set is subjected to differential dimensionality reduction to obtain a standard sample data set, the standard sample data set is divided into K equal parts, and K parts are selected with replacement to construct a first training set, and K training sets are generated by iterative selection K times; the BP neural network is supervised and trained using the K training sets to obtain K deviation prediction units that meet the expected convergence constraints, and the control deviation predictor is constructed based on the K deviation prediction units, wherein the output of the control deviation predictor is the mean of the output results of the K deviation prediction units.

[0011] In a possible implementation, the sample data set is subjected to differential dimensionality reduction to obtain a standard sample data set, and the following processing is also performed: a first sample data set is randomly selected from the sample data set, wherein the first sample data set includes a first sample bottle body size deviation, a first sample bottle body size uniformity coefficient, a first sample bottle body wall thickness deviation, and a first sample bottle body wall thickness uniformity coefficient; a difference operation is performed between the first sample bottle body size deviation and the bottle body size deviation of other sample data in the sample data set, and the frequency of the difference being greater than a predetermined size difference is counted and set as a first difference degree; if the first difference degree is greater than a predetermined threshold, the first sample bottle body size uniformity coefficient is differenced with the bottle body size uniformity coefficient of other sample data in the sample data set. The method further comprises the following steps: performing a value operation, and performing a statistical operation on the frequency of a difference value greater than a predetermined size uniformity coefficient difference, which is set as a second difference; if the second difference is greater than a predetermined threshold, performing a difference operation on the wall thickness deviation of the first sample bottle body and the wall thickness deviation of other sample data in the sample data set, and performing a statistical operation on the frequency of a difference value greater than a predetermined wall thickness difference, which is set as a third difference; if the third difference is greater than a predetermined threshold, performing a difference operation on the wall thickness uniformity coefficient of the first sample bottle body and the wall thickness uniformity coefficient of other sample data in the sample data set, and performing a statistical operation on the frequency of a difference value greater than a predetermined wall thickness uniformity coefficient, which is set as a fourth difference; if the fourth difference is greater than a predetermined threshold, adding the first sample data to the standard sample data set to obtain the standard sample data set.

[0012] The present application also provides an air compressor control device that adapts to load changes, including: a simulation demand determination module, the simulation demand determination module is used to call the production schedule within a predetermined period, input the production schedule into a predetermined bottle blowing twin production line to simulate the demand, and determine the simulated demand, wherein the predetermined bottle blowing twin production line simulates and constructs the bottle blowing machine production line through digital twin; a prediction interference coefficient generation module, the prediction interference coefficient generation module is used to collect environmental factors of the production workshop in real time, perform interference prediction based on the environmental factors, and generate a prediction interference coefficient; an optimization air compressor parameter configuration module, the optimization air compressor parameter configuration module is used to compensate for the simulated demand based on the predicted leakage coefficient and the predicted interference coefficient, determine the optimized predicted load, and configure the optimized air compressor parameters according to the optimized predicted load; production deviation value confirmation A determination module, wherein the production deviation value determination module is used to execute air compressor control within a predetermined period according to the optimized air pressure parameters, and to collect bottle body information at a fixed point during the control process, and to perform production deviation analysis according to the bottle body data set to determine the production deviation value, wherein the production deviation value includes bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient; a real-time air pressure control deviation output module, wherein the real-time air pressure control deviation output module is used to input the bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient into a control deviation predictor, and output a real-time air pressure control deviation; a control compensation parameter determination module, wherein the control compensation parameter determination module is used to determine the control compensation parameter based on the real-time air pressure control deviation, and to perform adaptive compensation on the air pressure parameter of the previous node according to the control compensation parameter.

[0013] The proposed method and device for air compressor control with adaptive load changes in this application will input the production schedule into the scheduled twin bottle blowing production line to simulate the demand and determine the simulated demand; predict interference based on environmental factors to generate a predicted interference coefficient; compensate for the simulated demand and optimize the air compressor parameters according to the optimized predicted load configuration; analyze the production deviation based on the bottle data set to determine the production deviation value; output the real-time air compressor control deviation; and adaptively compensate the air compressor parameters of the previous node. This solves the technical problem of the existing air compressor control that it is difficult to accurately predict and cope with the dynamic changes in the air compressor load during the production process, which leads to poor production efficiency and product quality. It achieves accurate prediction of the air compressor load and adaptive adjustment, achieving the technical effect of improving overall production efficiency and improving product quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the device according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A schematic flow chart of an air compressor control method for adaptive load changes provided in an embodiment of the present application;

[0016] Figure 2 A schematic structural diagram of an air compressor control device with adaptive load changes provided in an embodiment of the present application.

[0017] Explanation of the reference numerals: simulation demand determination module 10, prediction interference coefficient generation module 20, air compression parameter optimization configuration module 30, production deviation value determination module 40, real-time air compression control deviation output module 50, control compensation parameter determination module 60. DETAILED DESCRIPTION

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, device, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0021] The present application embodiment provides an air compressor control method that is adaptive to load changes, such as Figure 1 As shown, the method includes:

[0022] Step S100, call the production schedule within a predetermined period, input the production schedule into the predetermined blowing twin production line to simulate the demand, and determine the simulated demand, wherein the predetermined blowing twin production line simulates the blowing machine production line through digital twin. The production schedule refers to a detailed production plan formulated based on production orders, production capacity, material supply and other factors, which stipulates the production tasks, production quantities, production time, etc. within a predetermined period in the future (such as one week, one month, etc.), such as the number of different blowing orders, bottle types, bottle body sizes, etc.; the blowing machine production line refers to the physical production line actually used to produce blow-molded bottles (such as beverage bottles, cosmetic bottles, etc.), including blowing machines, molds, conveyor belts and other equipment and facilities; digital twin is a simulation process that uses physical models, sensor updates, operation history and other data to integrate multi-disciplinary, multi-physical quantities, multi-scales, and multi-probabilities, and completes mapping in virtual space, thereby reflecting the corresponding full life cycle of the physical equipment. In the manufacturing industry, digital twin technology is widely used in the simulation and optimization of production lines; the predetermined blowing twin production line refers to a virtual space constructed using digital twin technology that is similar to the actual blowing machine The virtual production line corresponding to the production line can simulate the operating status of the actual production line, including the working conditions of the equipment, the flow of materials, the production process, etc. Specifically, the production plan data within a specific time period is extracted from the production planning system, and the production schedule is input into the scheduled twin blowing production line for demand simulation. A blowing production line contains multiple blowing machines. Each blowing machine produces different types of bottles at different speeds in different time periods, which will cause fluctuations in the load of the air compressor. Changes in production speed and type will cause changes in the demand for compressed air. Therefore, the air demand in the actual production process is simulated through simulation operation. The simulation process will take into account multiple factors such as the production capacity, production speed, and equipment failure rate of the production line. After the simulation run, the amount of air required to produce blow-molded bottles in the future scheduled period is calculated based on the simulation results, that is, the simulated demand.

[0023] Step S200, collect environmental factors of the production workshop in real time, perform interference prediction based on the environmental factors, and generate a predicted interference coefficient. By arranging various sensors and monitoring equipment in the production workshop, environmental factors are collected in real time, wherein the sensors may include temperature sensors, humidity sensors, barometers, light sensors, air quality monitors, etc. Environmental factors refer to various factors in the environment that affect a specific process or system, including but not limited to temperature, humidity, air pressure, light, air quality (such as dust particle count, harmful gas concentration), water quality (if production water is involved), noise level, vibration, etc., which have an important impact on the normal operation of the production workshop and product quality. The collected environmental factor data will be input into the data analysis system for preprocessing, cleaning and conversion to extract useful information for analyzing the relationship between environmental factors and their possible impact on the production process. An interference prediction model is constructed based on machine learning (such as decision trees, random forests, neural networks, etc.), and the historical data of the production line (including historical environmental factors) is used to generate an interference prediction model. The interference prediction model is trained based on the environmental record data and historical production data. The model should be able to learn the complex relationship between environmental factors and production interference. The current environmental factor data is then input into the trained interference prediction model for analysis to predict possible production interference in the future, thereby generating a predicted interference coefficient. The predicted interference coefficient is a quantitative indicator used to indicate the degree of impact of the predicted interference on the production process. According to the predicted interference coefficient, the production scheduler can adjust the production plan to avoid or reduce the impact of the interference on production. For example, if it is predicted that the air quality will decline within a certain time period, the processes that are sensitive to air quality can be arranged in advance to be carried out in other time periods. According to the predicted interference coefficient, the environmental control system of the production workshop can also be adjusted, such as adjusting the temperature, humidity and air quality, to create environmental conditions that are more conducive to production, thereby improving the stability of the production process and product quality.

[0024] Step S300, compensating the simulated demand based on the predicted leakage coefficient and the predicted interference coefficient, determining the optimized predicted load, and configuring the optimized air compression parameters according to the optimized predicted load. The predicted leakage coefficient is obtained by analyzing the difference between historical air pressure parameters and historical output parameters, combined with the leakage trend prediction curve. It reflects the proportion of air loss caused by equipment leakage and other reasons to the air supply in the future. After determining the predicted leakage coefficient and predicted interference coefficient, these two coefficients need to be applied to the simulated demand for compensation. Specifically, the simulated demand is multiplied by the coefficient of (1+predicted leakage coefficient+predicted interference coefficient) to obtain an optimized predicted load after considering leakage and interference factors. This optimized predicted load is closer to the actual production demand, which helps to avoid production interruptions or waste of resources due to failure to consider leakage and interference. Then, according to the optimized predicted load, the air pressure parameters are configured and optimized, such as adjusting the compressor's pressure setting value, flow setting value, operating time and other parameters to match the needs of the optimized predicted load. When adjusting the parameters, it is necessary to comprehensively consider multiple factors such as the performance limitations, operating costs, and production efficiency of the equipment to achieve the optimal configuration effect, ensuring that while meeting production needs, energy consumption and wear are minimized as much as possible, and the stability of the production process and the overall operating efficiency of the production line are improved.

[0025] In a possible implementation, step S300 further includes step S310, querying the production log of the bottle blowing machine to obtain a control monitoring data sequence within a predetermined time window, wherein the control monitoring data includes historical air pressure parameters and historical output parameters. The time range of the query is clarified, i.e., a predetermined time window, such as the past week, month, or a specific time period, and the predetermined time window is input as a query condition through the production log system or enterprise resource planning (ERP) system of the bottle blowing machine, using the corresponding query interface, and extracting the required control monitoring data from the production log, mainly referring to the leakage trend analysis of air compression, including historical air pressure parameters and historical output parameters, and obtaining control monitoring data from the most recent maintenance node to obtain a control monitoring data sequence, wherein the historical air pressure parameters include exhaust pressure, intake flow, etc., and the historical output parameters include bottle size, bottle wall thickness, etc.

[0026] Step S320, based on the control monitoring data sequence, leakage ratio analysis is performed to generate a leakage ratio sequence, wherein the leakage ratio is the ratio of the difference between the historical air pressure parameters and the historical output parameters to the historical air pressure parameters. The difference between the historical air pressure parameters and the historical output parameters is calculated, which represents the degree of deviation between the air pressure parameters and the expected output parameters due to some reason (such as leakage, wear, improper adjustment, etc.), and then the leakage ratio analysis is performed. Specifically, the leakage ratio is calculated by dividing the above difference by the historical air pressure parameters, reflecting the proportion of invalid or lost parts in the air pressure parameters due to leakage and other reasons. The higher the leakage ratio, the more serious the leakage problem, and the greater the impact on production efficiency and product quality. The leakage ratio of each monitoring node (such as each production cycle, each time interval, etc.) is arranged in chronological order to form a leakage ratio sequence, which provides intuitive information about the change of leakage problems over time, which is helpful for trend analysis and prediction, timely discovery and solution of leakage problems in the production process of bottle blowing machines, and improve production efficiency and product quality.

[0027] Step S330, taking the monitoring node as the X-axis and the leakage proportion as the Y-axis, constructing a leakage trend prediction curve based on the leakage proportion sequence, determining the predicted leakage proportion within a predetermined period according to the leakage trend prediction curve, and setting the predicted leakage proportion as the predicted leakage coefficient. Each monitoring node (such as timestamp, production batch, etc.) is taken as a point on the X-axis, representing different time points or production stages of data collection. The leakage ratio corresponding to each monitoring node is taken as the value on the Y-axis, reflecting the severity of the leakage problem at different monitoring nodes. By representing each monitoring node and its corresponding leakage ratio with a point (X, Y) on the coordinate graph, and connecting these points to form a curve (or broken line), a visual representation of the leakage ratio sequence is obtained, namely, a leakage trend graph. Based on the leakage trend graph, a time series analysis is further performed to construct a leakage trend prediction curve. The prediction curve should be able to capture the trend of leakage ratio changing over time, including rising, falling or stable patterns. After obtaining the leakage trend prediction curve, a predetermined period (such as the next week, month, etc.) is determined according to actual needs, and the corresponding predicted leakage ratio value in the period is found on the prediction curve, which represents the expected severity of the leakage problem in the future predetermined period. Finally, the predicted leakage ratio in the predetermined period is set as the predicted leakage coefficient, which is used to evaluate the possibility and severity of leakage problems in the future.

[0028] Step S400, according to the optimized air pressure parameters, the air compressor control is executed within a predetermined period, and the bottle body information is collected at a fixed point during the control process, and the production deviation analysis is performed according to the bottle body data set to determine the production deviation value, wherein the production deviation value includes the bottle body size deviation, the bottle body size uniformity coefficient, the bottle body wall thickness deviation and the bottle body wall thickness uniformity coefficient. Within the predetermined period, the air compressor is controlled according to the optimized air pressure parameters (the control parameters of the air compressor are optimized by optimizing the predicted load determined by the predicted leakage coefficient and the predicted interference coefficient), including monitoring the operating status of the air compressor, adjusting its output to meet production needs, and ensuring that it operates stably within the predetermined parameter range. Specifically, during the air compressor control process, the bottle body information is collected at a fixed point, that is, by setting a specific detection point on the production line, using sensors or visual recognition systems and other equipment to perform real-time detection and data collection on the bottle body. The collected bottle body information may include multiple aspects such as the bottle body size, wall thickness, weight, and appearance quality. The collected bottle body information is organized into a bottle body data set, and the production deviation analysis is performed based on the bottle body data set to determine the production deviation value, wherein The production deviation values ​​include bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient. Bottle size deviation refers to the difference between the actual bottle size and the standard size, reflecting the accuracy of size control in the production process; bottle size uniformity coefficient is used to evaluate the consistency of sizes of various parts of the bottle. The higher the uniformity coefficient, the more uniform the bottle size; bottle wall thickness deviation refers to the difference between the actual bottle wall thickness and the standard wall thickness, reflecting the accuracy of wall thickness control in the production process; bottle wall thickness uniformity coefficient is used to evaluate the consistency of wall thickness of various parts of the bottle. The higher the uniformity coefficient, the more uniform the bottle wall thickness; evaluating the stability of the production process and product quality according to the production deviation value, timely discovering and solving problems in the production process, will help improve production efficiency and reduce production costs.

[0029] In a possible implementation, step S400 further includes step S410, at a predetermined monitoring node, using a dual-element monitoring device to collect N bottle bodies to obtain N bottle body image sets and N bottle body point cloud sets, where N is an integer greater than 10. A predetermined monitoring node refers to a pre-set time point or production stage. Under the predetermined monitoring node, a dual-element monitoring device is used to collect data from N bottle bodies to obtain N bottle body image sets and N bottle body point cloud sets. The dual-element monitoring device is a device with image monitoring and point cloud collection functions, which can simultaneously collect image data and point cloud data of the bottle body to obtain N bottle body image sets and N bottle body point cloud sets. This batch monitoring method can improve efficiency and reduce the time and cost required for monitoring a single bottle body. For each monitored bottle body, the dual-element monitoring device will capture multiple images (for example, taken from different angles and under different lighting conditions) to form an image set. At the same time, a certain technology (such as laser scanning, structured light, etc.) is used to obtain the three-dimensional point cloud data of the bottle body, which is composed of a large number of three-dimensional coordinate points, which together describe the three-dimensional shape and surface features of the bottle body. Each bottle body corresponds to a point cloud set.

[0030] Step S420, performing image stitching of the N bottle body image sets to obtain N complete bottle body images, and extracting contour features of the N complete bottle body images using a convolutional neural network to obtain N bottle body contours. Image stitching refers to merging these multiple images into a complete, seamless bottle body image, that is, using image processing technology to identify the overlapping areas between images, and accurately stitching them together through an algorithm, each bottle body will correspond to a complete image, showing a comprehensive view of the bottle body, and obtaining N complete bottle body images, and then using a convolutional neural network (CNN) to extract contour features from these images. The convolutional neural network (CNN) can learn and identify specific features in an image, such as edges, shapes, textures, etc. By inputting each complete bottle body image into a trained CNN model, the model will automatically extract the contour features of the bottle body and output the corresponding contour image or contour data to obtain N bottle body contours, which can be a binary image (i.e., an image containing only black and white colors, where white represents the contour and black represents the background), or a data set containing contour coordinate points.

[0031] Step S430, performing point cloud fitting on the N bottle point cloud sets based on the N bottle contours, and generating N three-dimensional bottle models according to the point cloud fitting results. Based on the extracted N bottle contours, point cloud fitting is performed on the corresponding N bottle point cloud sets. Point cloud fitting is to convert point cloud data (multiple three-dimensional coordinate points) into a more continuous and smoother surface model. According to the point cloud fitting results, N three-dimensional bottle models are generated, which are accurate representations of the bottle in three-dimensional space.

[0032] Step S440, performing production deviation analysis based on the N complete bottle images and the N three-dimensional bottle models to determine the production deviation value. The production deviation analysis is performed using the N complete bottle images and the N three-dimensional bottle models. Specifically, the size, shape, surface quality and other aspects of the bottle are compared to determine the deviation value in the production process, such as size deviation, shape deviation, etc., to identify potential problems in the production process, such as equipment failure, improper process parameter settings, etc., so as to take timely measures to adjust and improve.

[0033] In a possible implementation, step S430 further includes step S431, randomly selecting a first bottle body contour from the N bottle body contours, and obtaining a first bottle body point cloud set of the first bottle body contour. The first bottle body contour is randomly selected from the N bottle body contours, and the corresponding first bottle body point cloud set is the three-dimensional coordinate point set of the first bottle body.

[0034] Step S432: With the first bottle body contour as a constraint, the first bottle body point cloud set is randomly fitted to obtain a first fitting result, and a first fitting degree is calculated, wherein the first fitting degree is the ratio of the number of point clouds falling within the first bottle body contour to the total number of point clouds in the first bottle body point cloud set. With the first bottle body contour as a constraint, the first bottle body point cloud set is randomly fitted for the first time, for example, a random sampling consistency algorithm is used to estimate the best fitting shape of the point cloud data, while ensuring that the shape does not exceed the given bottle body contour, to obtain a first fitting result, and the first fitting degree is obtained by calculating the ratio of the number of point clouds falling within the first bottle body contour to the total number of point clouds in the first bottle body point cloud set. The larger the first fitting degree, the closer the fitting result is to the actual bottle body shape and does not exceed the contour limit.

[0035] Step S433, again using the first bottle body contour as a constraint, performing random fitting on the first bottle body point cloud set to obtain a second fitting result and a second fitting degree. Similarly, again using the first bottle body contour as a constraint, performing a second random fitting on the first bottle body point cloud set to obtain a second fitting result and calculate the corresponding second fitting degree.

[0036] Step S434, perform iterative fitting until a predetermined number of iterations is reached, and output the fitting result with the highest degree of fit as the first optimal fitting result. Repeat the above random fitting and fitting calculation process, each fitting is based on the same bottle body contour and point cloud data, but uses different random initial conditions, and each iteration will output a new fitting result and corresponding degree of fit until a predetermined number of iterations is reached, then compare the degrees of fit of all fitting results, and find the fitting result with the highest degree of fit as the first optimal fitting result, which represents the best fitting shape of the first bottle body point cloud set under the given bottle body contour constraints.

[0037] In a possible implementation, step S440 further includes step S441, grayscale processing and grid division of the N complete bottle body images to obtain N grayscale grid images of the bottle body. Grayscale processing is performed on the N complete bottle body images, that is, the color image is converted into a grayscale image, each pixel in the grayscale image has only one brightness value (usually between 0 and 255), then the grayscale image is grid divided, the image is divided into multiple small areas (i.e., grids), and N grayscale grid images of the bottle body are obtained, each grid contains a part of the image, which is convenient for subsequent local analysis of the image, such as evaluating the wall thickness or size deviation in each grid.

[0038] Step S442, obtaining a standard bottle body grayscale grid image, performing bottle body wall thickness deviation and bottle body wall thickness uniformity analysis on the N bottle body grayscale grid images based on the standard bottle body grayscale image, obtaining N wall thickness deviation ratios and N wall thickness deviation variances, performing mean calculation on the N wall thickness deviation ratios and the N wall thickness deviation variances, respectively, to obtain the bottle body wall thickness deviation and the bottle body wall thickness uniformity coefficient. The standard bottle body grayscale grid image represents the ideal bottle body wall thickness distribution and is used for comparison with N bottle body grayscale grid images. Based on the standard bottle body grayscale grid image, a wall thickness deviation analysis is performed on each bottle body grayscale grid image. Specifically, the grayscale values ​​in the corresponding grids (or the wall thickness values ​​derived therefrom) are compared to calculate the wall thickness deviation ratio of each grid. Then, the wall thickness deviation ratios of the entire bottle body are summarized to obtain N wall thickness deviation ratios. The uniformity of the wall thickness is also evaluated by calculating the variance of the wall thickness deviation of each bottle body. The smaller the variance, the more uniform the bottle body wall thickness. Then, N wall thickness deviation variances are obtained. Finally, the mean of the N wall thickness deviation ratios and the N wall thickness deviation variances are calculated respectively to obtain the wall thickness deviation and wall thickness uniformity coefficient of the entire batch of bottles, which reflects the overall quality level of the batch of bottles in terms of wall thickness.

[0039] Step S443, obtain a standard three-dimensional bottle model, perform bottle size deviation and bottle size uniformity analysis on the N three-dimensional bottle models based on the standard three-dimensional bottle model, obtain N size deviation ratios and N size deviation variances, perform mean calculation on the N size deviation ratios and N size deviation variances, and obtain bottle size deviation and bottle size uniformity coefficient. The standard three-dimensional bottle model represents the ideal bottle size and shape. Based on the standard three-dimensional bottle model, perform size deviation analysis on each three-dimensional bottle model. Specifically, compare the size differences of corresponding parts (such as height, diameter, etc.) to calculate the size deviation ratio of each bottle to obtain N size deviation ratios. Similarly, evaluate the size uniformity by calculating the variance of each bottle size deviation. The smaller the variance, the more consistent the bottle size is, and obtain N size deviation variances. Finally, perform mean calculation on the N size deviation ratios and N size deviation variances, and obtain the size deviation and size uniformity coefficient of the entire batch of bottles, which reflects the overall quality level of the batch of bottles in terms of size.

[0040] Step S500, input the bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient into the control deviation predictor to output the real-time air compressor control deviation. Input the bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient into the control deviation predictor to output the real-time air compressor control deviation, and dynamically adjust the control parameters of the air compressor using the bottle quality parameters in actual production to ensure the stability of the production process and the consistency of product quality, wherein the control deviation predictor is a prediction unit, which receives the above input parameters as input, and performs calculations and predictions according to a preset algorithm or model, that is, predicts the control deviation of the air compressor according to the current bottle quality parameters. Specifically, a large amount of historical data is collected, and these data should include the air compressor Under different operating conditions, the exhaust pressure, exhaust flow and other control parameters, as well as the corresponding bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation, bottle wall thickness uniformity coefficient and other quality parameters are collected to clean the historical data, remove noise and outliers, ensure the accuracy and reliability of the data, divide the cleaned data into training set, validation set and test set, and then build a prediction model based on machine learning models (such as neural networks, support vector machines, random forests, etc.), design the model structure, including determining the structure of the model's input layer, hidden layer and output layer, and selecting appropriate activation functions, optimization algorithms and losses. Function, etc. For the neural network model, it is also necessary to determine the number of network layers, the number of nodes in each layer, and the connection method, etc., and then initialize the model parameters, such as weights and biases, and use the training set data to iteratively train the model. In each iteration, the input data is input into the model, and the error between the output of the model and the true value (such as mean square error, absolute error, etc.) is calculated. Then, the parameters of the model are adjusted according to the error back propagation algorithm to reduce the error. During the training process, the performance of the model is regularly evaluated using the validation set data, and the parameters or structure of the model are adjusted according to the evaluation results to avoid problems such as overfitting or underfitting. The performance of the trained model is evaluated using the test set data to obtain a control deviation predictor. The real-time air compression control deviation output by the control deviation predictor includes exhaust pressure deviation and exhaust flow deviation. The exhaust pressure deviation represents the difference between the actual exhaust pressure and the expected exhaust pressure, which is used to guide the adjustment of the exhaust pressure of the air compressor; the exhaust flow deviation represents the difference between the actual exhaust flow and the expected exhaust flow, which is used to guide the adjustment of the exhaust flow of the air compressor; then, the exhaust pressure and exhaust flow of the air compressor are adjusted in real time according to these deviation information to reduce the deviation and make the production process more stable, thereby improving production efficiency and product quality.

[0041] In a possible implementation, step S500 further includes step S510, taking the bottle blowing machine production line and the bottle blowing air compressor as constraints, performing homologous positive sample retrieval based on big data, and obtaining a sample data set, wherein the sample data includes sample bottle size deviation, sample bottle size uniformity coefficient, sample bottle wall thickness deviation, sample bottle wall thickness uniformity coefficient, sample pressure deviation, and sample flow deviation. Taking the bottle blowing machine production line and the bottle blowing air compressor as constraints, performing homologous positive sample retrieval through big data technology, that is, retrieving qualified sample data from the historical records and real-time monitoring data of the two systems, the collected sample data includes multiple key indicators, such as sample bottle size deviation, sample bottle size uniformity coefficient, sample bottle wall thickness deviation, sample bottle wall thickness uniformity coefficient, as well as sample pressure deviation and sample flow deviation, which reflect the quality of the bottle manufacturing process and the working status of the air compressor.

[0042] Step S520, perform differential dimensionality reduction on the sample data set to obtain a standard sample data set, divide the standard sample data set into K equal parts, select K times with replacement, construct the first training set, iterate and select K times, and generate K training sets. Perform differential dimensionality reduction processing on the collected sample data set, including feature selection (selecting the most important features from the original features) and feature extraction (converting the original features into a new, lower-dimensional feature space through some transformation), reduce the dimension of the data, remove redundant information, and highlight the differences or differential features in the data, that is, those features that can distinguish different samples or categories, to obtain a standard sample data set, and then divide the standard sample data set into K equal parts, and select with replacement (also called bootstrap sampling), that is, randomly select one from the K data sets as the first training set, then put it back, and randomly select another one as the second training set, and so on, until we have constructed K training sets.

[0043] Step S530, using the K training sets to perform supervised training on the BP neural network to obtain K deviation prediction units that meet the expected convergence constraints, and constructing the control deviation predictor based on the K deviation prediction units, wherein the output of the control deviation predictor is the mean of the output results of the K deviation prediction units. BP neural network is a multi-layer feedforward neural network. The weights and biases in the network are trained by the back propagation algorithm. During the training process, the network adjusts its parameters according to the input training data and the expected output (i.e., label or target value) to minimize the prediction error. Specifically, K training sets are used as input data, and the corresponding deviation values ​​are used as target outputs. Then, the prediction values ​​are calculated through the forward propagation of the network, and the weights and biases of the network are adjusted through the back propagation algorithm to reduce the error between the prediction value and the target value. After supervised training, each training set will obtain a trained BP neural network model, that is, K deviation prediction units are obtained. Each deviation prediction unit can predict the corresponding deviation value according to the input data. Based on the K trained deviation prediction units, an integrated predictor is constructed, that is, a control deviation predictor. The output of the control deviation predictor is the mean of the output results of the K deviation prediction units. For example, when a new input sample needs to predict the deviation, this sample is input into all K deviation prediction units to obtain K prediction results, and then the average value of these prediction results is calculated as the final deviation prediction output.

[0044] In a possible implementation, step S520 further includes step S521, randomly selecting first sample data from the sample data set, wherein the first sample data includes the first sample bottle size deviation, the first sample bottle size uniformity coefficient, the first sample bottle wall thickness deviation, and the first sample bottle wall thickness uniformity coefficient. The first sample data is randomly selected from the sample data set, and includes the bottle size deviation, the bottle size uniformity coefficient, the bottle wall thickness deviation, and the bottle wall thickness uniformity coefficient.

[0045] Step S522, performing a difference operation on the bottle size deviation of the first sample and the bottle size deviation of other sample data in the sample data set, and counting the frequency of the difference being greater than a predetermined size difference, which is set as a first difference degree. Perform a difference operation on the bottle size deviation of the first sample and the bottle size deviation of all other samples in the data set, and counting the frequency of the difference being greater than a predetermined size difference, which is set as the first difference degree, reflecting the uniqueness of the first sample in terms of bottle size deviation, wherein the predetermined size difference is a threshold value set according to historical data, which is used to determine the degree of size difference of the deviation, and the greater the frequency of being greater than the predetermined size difference, the greater the difference between the bottle size deviation of the first sample and the overall difference.

[0046] Step S523, if the first difference is greater than a predetermined threshold, the first sample bottle size uniformity coefficient is subjected to a difference operation with the bottle size uniformity coefficients of other sample data in the sample data set, and the frequency of the difference being greater than the predetermined size uniformity coefficient difference is counted, and the difference is set as the second difference. If the first difference is greater than the predetermined threshold, the bottle size uniformity coefficient of the first sample is continuously analyzed, the bottle size uniformity coefficient value of the first sample is taken out, and the difference operation is performed one by one with the bottle size uniformity coefficient values ​​of all other samples in the data set, and for each calculated difference, it is checked whether it is greater than the predetermined size uniformity coefficient difference, and if the difference is greater than the predetermined size uniformity coefficient difference, the situation is counted as a valid difference, and the process is repeated for all other samples, and the total number of all valid differences is counted, and the total number of times counted is set as the second difference, which reflects the significant difference degree of the first sample in terms of the bottle size uniformity coefficient compared with other samples in the data set, wherein the predetermined size uniformity coefficient difference is a threshold value set according to historical data, and is used to determine the difference degree of the size uniformity coefficient of the deviation.

[0047] Step S524, if the second difference is greater than a predetermined threshold, the difference between the first sample and the other sample data in the sample data set is calculated, and the frequency of the difference greater than the predetermined wall thickness difference is counted, which is set as the third difference. If the second difference is greater than the predetermined threshold, the bottle wall thickness deviation value of the first sample is taken out, and the difference is calculated with the bottle wall thickness deviation values ​​of all other samples in the data set one by one. For each calculated difference, check whether it is greater than a predetermined wall thickness difference. If the difference is greater than the predetermined wall thickness difference, the situation is counted as a valid difference, and the process is repeated for all other samples, and the total number of all valid differences is counted. The total number of times counted is set as the third difference, which reflects the degree of significant difference in the bottle wall thickness deviation of the first sample compared with other samples in the data set, wherein the predetermined wall thickness difference is a threshold set based on historical data, which is used to determine what degree of wall thickness deviation difference is considered significant.

[0048] Step S525, if the third difference is greater than a predetermined threshold, the first sample bottle wall thickness uniformity coefficient is subjected to a difference operation with the bottle wall thickness uniformity coefficients of other sample data in the sample data set, and the frequency of the difference being greater than the predetermined wall thickness uniformity coefficient is counted, which is set as the fourth difference. If the third difference is greater than a predetermined threshold, the bottle wall thickness uniformity coefficient value of the first sample is taken out, and the difference operation is performed one by one with the bottle wall thickness uniformity coefficient values ​​of all other samples in the data set, and for each calculated difference, it is checked whether it is greater than a predetermined wall thickness uniformity coefficient difference, and if the difference is greater than the predetermined wall thickness uniformity coefficient difference, the situation is counted as a valid difference, and the process is repeated for all other samples, and the total number of all valid differences is counted, and the total number of times counted is set as the fourth difference, which reflects the degree of significant difference in bottle wall thickness uniformity between the first sample and other samples in the data set, wherein the predetermined wall thickness uniformity coefficient is a set threshold used to determine what degree of wall thickness uniformity difference is considered significant.

[0049] Step S526, if the fourth difference is greater than the predetermined threshold, the first sample data is added to the standard sample data set to obtain the standard sample data set. When the fourth difference (i.e., the degree of difference between the first sample and other samples in the data set in terms of bottle wall thickness uniformity) is greater than the predetermined threshold, it indicates that the first sample has shown significant uniqueness in these key features. The first sample data is added to the standard sample data set. If any of the four differences is less than its corresponding predetermined threshold, the first sample data is discarded. By continuously adding samples that meet specific conditions to the standard sample data set, a more complete and representative data set is gradually constructed, that is, the final standard sample data set is obtained.

[0050] Step S600, based on the real-time air pressure control deviation, the control compensation parameters are determined, and the air pressure parameters of the previous node are adaptively compensated according to the control compensation parameters. Based on the real-time air pressure control deviation, the control compensation parameters are determined, and the air pressure parameters of the previous node are adaptively compensated according to these control compensation parameters to ensure the stable operation of the system and achieve the predetermined production target. Specifically, according to the real-time air pressure control deviation, a suitable control compensation strategy is selected, which may include adjusting the air intake of the air compressor, adjusting the opening of the exhaust valve, changing the operating state of the cooling system, etc. According to the selected compensation strategy, the specific control compensation parameters are calculated, and then according to the calculated control compensation parameters, the air pressure parameters of the previous node are adaptively adjusted, that is, real-time and dynamic air pressure parameter adjustment is performed to quickly respond to system deviations and reduce their impact on the production process. During the adjustment process, the operating state and deviation of the system are continuously monitored, and the control compensation parameters are fine-tuned according to the new deviation information to ensure that the system always remains in the best operating state. By adaptively compensating the air pressure parameters, the stability and production efficiency of the system can be significantly improved, the production cost can be reduced, and the energy utilization efficiency can be improved.

[0051] In the above, refer to Figure 1 A method for controlling an air compressor according to an embodiment of the present invention is described in detail. Figure 2 An air compressor control device capable of adaptively adapting to load changes according to an embodiment of the present invention is described.

[0052] An air compressor control device with adaptive load changes according to an embodiment of the present invention is used to solve the technical problem that it is difficult to accurately predict and cope with the dynamic changes of the air compressor load during the production process in the existing air compressor control, which leads to poor production efficiency and product quality. It can accurately predict the air compressor load and make adaptive adjustments, thereby achieving the technical effect of improving overall production efficiency and improving product quality stability. An air compressor control device with adaptive load changes includes: a simulation demand determination module 10, a prediction interference coefficient generation module 20, an air compressor parameter optimization configuration module 30, a production deviation value determination module 40, a real-time air compressor control deviation output module 50, and a control compensation parameter determination module 60.

[0053] A simulation demand determination module 10, wherein the simulation demand determination module 10 is used to call a production schedule within a predetermined period, input the production schedule into a predetermined bottle blowing twin production line to simulate demand, and determine a simulated demand, wherein the predetermined bottle blowing twin production line simulates and constructs a bottle blowing machine production line through a digital twin;

[0054] A predicted interference coefficient generation module 20, which is used to collect environmental factors of the production workshop in real time, perform interference prediction based on the environmental factors, and generate a predicted interference coefficient;

[0055] An optimized air compression parameter configuration module 30, the optimized air compression parameter configuration module 30 is used to compensate the simulated demand based on the predicted leakage coefficient and the predicted interference coefficient, determine the optimized predicted load, and configure the optimized air compression parameters according to the optimized predicted load;

[0056] A production deviation value determination module 40, the production deviation value determination module 40 is used to execute air compressor control within a predetermined period according to the optimized air compression parameters, collect bottle body information at a fixed point during the control process, perform production deviation analysis according to the bottle body data set, and determine the production deviation value, wherein the production deviation value includes bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient;

[0057] A real-time air pressure control deviation output module 50, wherein the real-time air pressure control deviation output module 50 is used to input the bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient into a control deviation predictor, and output a real-time air pressure control deviation;

[0058] The control compensation parameter determination module 60 is used to determine the control compensation parameter based on the real-time air pressure control deviation, and to adaptively compensate the air pressure parameter of the previous node according to the control compensation parameter.

[0059] The specific configuration of the air pressure parameter optimization configuration module 30 will be described in detail below. The air pressure parameter optimization configuration module 30 may further include: querying the production log of the bottle blowing machine to obtain a control monitoring data sequence within a predetermined time window, wherein the control monitoring data includes historical air pressure parameters and historical output parameters; performing leakage ratio analysis based on the control monitoring data sequence to generate a leakage ratio sequence, wherein the leakage ratio is the ratio of the difference between the historical air pressure parameters and the historical output parameters to the historical air pressure parameters; taking the monitoring node as the X-axis and the leakage ratio as the Y-axis, constructing a leakage trend prediction curve based on the leakage ratio sequence, determining the predicted leakage ratio within a predetermined period according to the leakage trend prediction curve, and setting the predicted leakage ratio as the predicted leakage coefficient.

[0060] The specific configuration of the production deviation value determination module 40 will be described in detail below. The production deviation value determination module 40 further includes: at a predetermined monitoring node, using a dual-element monitoring device to collect N bottle bodies, to obtain N bottle body image sets and N bottle body point cloud sets, where N is an integer greater than 10; performing image stitching of the N bottle body image sets to obtain N complete bottle body images, using a convolutional neural network to extract contour features of the N complete bottle body images to obtain N bottle body contours; performing point cloud fitting on the N bottle body point cloud sets based on the N bottle body contours, and generating N three-dimensional bottle body models according to the point cloud fitting results; performing production deviation analysis based on the N complete bottle body images and the N three-dimensional bottle body models to determine the production deviation value.

[0061] The specific configuration of the production deviation value determination module 40 will be described in detail below. The production deviation value determination module 40 may further include: randomly selecting a first bottle body contour from the N bottle body contours, and obtaining a first bottle body point cloud set of the first bottle body contour; taking the first bottle body contour as a constraint, randomly fitting the first bottle body point cloud set to obtain a first fitting result, and calculating a first fitting degree, wherein the first fitting degree is the ratio of the number of point clouds falling within the first bottle body contour to the total number of point clouds in the first bottle body point cloud set; again taking the first bottle body contour as a constraint, randomly fitting the first bottle body point cloud set to obtain a second fitting result and a second fitting degree; performing iterative fitting until a predetermined number of iterations is reached, and outputting the fitting result with the highest fitting degree as the first optimal fitting result.

[0062] The specific configuration of the production deviation value determination module 40 will be described in detail below. The production deviation value determination module 40 may further include: grayscale processing and grid division of the N complete bottle body images to obtain N bottle body grayscale grid images; obtaining a standard bottle body grayscale grid image, performing bottle body wall thickness deviation and bottle body wall thickness uniformity analysis on the N bottle body grayscale grid images based on the standard bottle body grayscale image, obtaining N wall thickness deviation ratios and N wall thickness deviation variances, respectively performing mean calculation on the N wall thickness deviation ratios and N wall thickness deviation variances, and obtaining bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient; obtaining a standard three-dimensional bottle body model, performing bottle body size deviation and bottle body size uniformity analysis on the N three-dimensional bottle body models based on the standard three-dimensional bottle body model, obtaining N size deviation ratios and N size deviation variances, respectively performing mean calculation on the N size deviation ratios and N size deviation variances, and obtaining bottle body size deviation and bottle body size uniformity coefficient.

[0063] The specific configuration of the real-time air compressor control deviation output module 50 will be described in detail below. The real-time air compressor control deviation output module 50 may further include: taking the bottle blowing machine production line and the bottle blowing air compressor as constraints, performing homologous positive sample retrieval based on big data to obtain a sample data set, wherein the sample data includes the sample bottle body size deviation, the sample bottle body size uniformity coefficient, the sample bottle body wall thickness deviation, the sample bottle body wall thickness uniformity coefficient and the sample pressure deviation, and the sample flow deviation; performing differential dimensionality reduction on the sample data set to obtain a standard sample data set, dividing the standard sample data set into K equal parts, and selecting K times with replacement to construct a first training set, iteratively selecting K times to generate K training sets; using the K training sets to supervise the BP neural network training to obtain K deviation prediction units that meet the expected convergence constraints, and constructing the control deviation predictor based on the K deviation prediction units, wherein the output of the control deviation predictor is the mean of the output results of the K deviation prediction units.

[0064] The specific configuration of the real-time air pressure control deviation output module 50 will be described in detail below. The real-time air pressure control deviation output module 50 may further include: randomly selecting a first sample data in the sample data set, wherein the first sample data includes a first sample bottle size deviation, a first sample bottle size uniformity coefficient, a first sample bottle wall thickness deviation, and a first sample bottle wall thickness uniformity coefficient; performing a difference operation on the first sample bottle size deviation and the bottle size deviation of other sample data in the sample data set, and counting the frequency of the difference being greater than a predetermined size difference, which is set as a first difference degree; if the first difference degree is greater than a predetermined threshold, performing a difference operation on the first sample bottle size uniformity coefficient and the bottle size uniformity coefficient of other sample data in the sample data set, and counting the difference being greater than a predetermined value; The frequency of the difference in the uniformity coefficient of the fixed size is set as the second difference degree; if the second difference degree is greater than a predetermined threshold value, the difference operation is performed on the wall thickness deviation of the first sample bottle body and the wall thickness deviation of the bottle body of other sample data in the sample data set, and the frequency of the difference value being greater than the predetermined wall thickness difference is set as the third difference degree; if the third difference degree is greater than a predetermined threshold value, the difference operation is performed on the wall thickness uniformity coefficient of the first sample bottle body and the wall thickness uniformity coefficient of other sample data in the sample data set, and the frequency of the difference value being greater than the predetermined wall thickness uniformity coefficient is set as the fourth difference degree; if the fourth difference degree is greater than a predetermined threshold value, the first sample data is added to the standard sample data set to obtain the standard sample data set.

[0065] An air compressor control device with adaptive load changes provided in an embodiment of the present invention can execute an air compressor control method with adaptive load changes provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0066] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0067] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. An air compressor control method that is adaptive to load changes, characterized in that: include: Calling a production schedule within a predetermined period, inputting the production schedule into a predetermined twin blowing production line to simulate demand, and determining a simulated demand, wherein the predetermined twin blowing production line simulates and constructs a blowing machine production line through a digital twin; Collect environmental factors of the production workshop in real time, perform interference prediction based on the environmental factors, and generate a predicted interference coefficient; The simulated demand is compensated based on the predicted leakage coefficient and the predicted interference coefficient, an optimized predicted load is determined, and an optimized air compression parameter is configured according to the optimized predicted load; Execute air compressor control within a predetermined period according to the optimized air compression parameters, collect bottle body information at a fixed point during the control process, perform production deviation analysis according to the bottle body data set, and determine the production deviation value, wherein the production deviation value includes bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient; Input the bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient into a control deviation predictor, and output a real-time air pressure control deviation; A control compensation parameter is determined based on the real-time air pressure control deviation, and the air pressure parameter of the previous node is adaptively compensated according to the control compensation parameter.

2. The air compressor control method according to claim 1, characterized in that: Get predicted leakage coefficients, including: Query the production log of the bottle blowing machine to obtain the control monitoring data sequence within the predetermined time window, where the control monitoring data includes historical air pressure parameters and historical output parameters; Performing leakage ratio analysis based on the control monitoring data sequence to generate a leakage ratio sequence, wherein the leakage ratio is the ratio of the difference between the historical air pressure parameter and the historical output parameter to the historical air pressure parameter; With the monitoring node as the X-axis and the leakage proportion as the Y-axis, a leakage trend prediction curve is constructed based on the leakage proportion sequence, the predicted leakage proportion within a predetermined period is determined according to the leakage trend prediction curve, and the predicted leakage proportion is set as the predicted leakage coefficient.

3. The air compressor control method according to claim 1, characterized in that: Determine production deviation values, including: At a predetermined monitoring node, a dual-element monitoring device is used to collect N bottle bodies to obtain N bottle body image sets and N bottle body point cloud sets, where N is an integer greater than 10; Perform image stitching of the N bottle body image sets to obtain N complete bottle body images, and use a convolutional neural network to extract contour features of the N complete bottle body images to obtain N bottle body contours; Performing point cloud fitting on the N bottle point cloud sets based on the N bottle body contours, and generating N three-dimensional bottle body models according to the point cloud fitting results; A production deviation analysis is performed based on the N complete bottle body images and the N three-dimensional bottle body models to determine the production deviation value.

4. The air compressor control method according to claim 3, characterized in that: Performing point cloud fitting on the N bottle body point cloud sets based on the N bottle body contours includes: Randomly select a first bottle body contour from the N bottle body contours, and obtain a first bottle body point cloud set of the first bottle body contour; Taking the first bottle body contour as a constraint, randomly fitting the first bottle body point cloud set to obtain a first fitting result, and calculating a first fitting degree, wherein the first fitting degree is a ratio of the number of point clouds falling within the first bottle body contour to the total number of point clouds in the first bottle body point cloud set; Again, taking the first bottle body contour as a constraint, randomly fitting the first bottle body point cloud set to obtain a second fitting result and a second fitting degree; Iterative fitting is performed until a predetermined number of iterations is reached, and the fitting result with the highest fitting degree is output as the first optimal fitting result.

5. The air compressor control method according to claim 3, characterized in that: Performing production deviation analysis based on the N complete bottle body images and the N three-dimensional bottle body models, including: Performing grayscale processing and grid division on the N complete bottle body images to obtain N bottle body grayscale grid images; Obtain a standard bottle body grayscale grid image, perform bottle body wall thickness deviation and bottle body wall thickness uniformity analysis on the N bottle body grayscale grid images based on the standard bottle body grayscale grid image, obtain N wall thickness deviation ratios and N wall thickness deviation variances, perform mean calculation on the N wall thickness deviation ratios and the N wall thickness deviation variances, and obtain the bottle body wall thickness deviation and the bottle body wall thickness uniformity coefficient; A standard three-dimensional bottle body model is obtained, and bottle body size deviation and bottle body size uniformity analysis is performed on the N three-dimensional bottle body models based on the standard three-dimensional bottle body model to obtain N size deviation ratios and N size deviation variances, and the means of the N size deviation ratios and the N size deviation variances are respectively calculated to obtain the bottle body size deviation and the bottle body size uniformity coefficient.

6. The air compressor control method according to claim 1, characterized in that: Build a control deviation predictor, including: Taking the bottle blowing machine production line and bottle blowing air compressor as constraints, homologous positive sample retrieval is performed based on big data to obtain a sample data set, where the sample data includes sample bottle size deviation, sample bottle size uniformity coefficient, sample bottle wall thickness deviation, sample bottle wall thickness uniformity coefficient, sample pressure deviation, and sample flow deviation; Perform differential dimensionality reduction on the sample data set to obtain a standard sample data set, divide the standard sample data set into K equal parts, select K parts with replacement to construct a first training set, iterate and select K times to generate K training sets; The K training sets are used to perform supervised training on the BP neural network to obtain K deviation prediction units that meet the expected convergence constraints, and the control deviation predictor is constructed based on the K deviation prediction units, wherein the output of the control deviation predictor is the mean of the output results of the K deviation prediction units.

7. The air compressor control method according to claim 6, characterized in that: Performing differential dimensionality reduction on the sample data set to obtain a standard sample data set, including: Randomly selecting first sample data from the sample data set, wherein the first sample data includes a first sample bottle size deviation, a first sample bottle size uniformity coefficient, a first sample bottle wall thickness deviation, and a first sample bottle wall thickness uniformity coefficient; Performing a difference operation on the first sample bottle size deviation and the bottle size deviation of other sample data in the sample data set, and counting the frequency of the difference being greater than a predetermined size difference, and setting it as a first difference degree; If the first difference is greater than a predetermined threshold, performing a difference operation on the uniformity coefficient of the bottle size of the first sample and the uniformity coefficient of the bottle size of other sample data in the sample data set, and counting the frequency of the difference being greater than the predetermined size uniformity coefficient difference, which is set as a second difference; If the second difference is greater than a predetermined threshold, performing a difference operation on the wall thickness deviation of the first sample bottle body and the wall thickness deviation of the bottle body of other sample data in the sample data set, and counting the frequency of the difference being greater than the predetermined wall thickness difference, which is set as a third difference; If the third difference is greater than a predetermined threshold, performing a difference operation on the uniformity coefficient of the bottle wall thickness of the first sample and the uniformity coefficient of the bottle wall thickness of other sample data in the sample data set, and counting the frequency of the difference being greater than the predetermined uniformity coefficient of the wall thickness, which is set as a fourth difference; If the fourth difference is greater than a predetermined threshold, the first sample data is added to a standard sample data set to obtain the standard sample data set.

8. An air compressor control device capable of self-adapting to load changes, characterized in that: The device is used to implement an air compressor control method for adaptive load changes according to any one of claims 1 to 7, and the device comprises: A simulation demand determination module, the simulation demand determination module is used to call a production schedule within a predetermined period, input the production schedule into a predetermined bottle blowing twin production line to simulate demand, and determine a simulated demand, wherein the predetermined bottle blowing twin production line simulates and constructs a bottle blowing machine production line through a digital twin; A predicted interference coefficient generation module, which is used to collect environmental factors of the production workshop in real time, perform interference prediction based on the environmental factors, and generate a predicted interference coefficient; An optimized air compression parameter configuration module, the optimized air compression parameter configuration module is used to compensate the simulated demand based on the predicted leakage coefficient and the predicted interference coefficient, determine the optimized predicted load, and configure the optimized air compression parameters according to the optimized predicted load; A production deviation value determination module, the production deviation value determination module is used to execute air compressor control within a predetermined period according to the optimized air compression parameters, collect bottle body information at a fixed point during the control process, perform production deviation analysis according to the bottle body data set, and determine the production deviation value, wherein the production deviation value includes bottle body size deviation, bottle body size uniformity coefficient, bottle body wall thickness deviation and bottle body wall thickness uniformity coefficient; A real-time air pressure control deviation output module, wherein the real-time air pressure control deviation output module is used to input the bottle size deviation, bottle size uniformity coefficient, bottle wall thickness deviation and bottle wall thickness uniformity coefficient into a control deviation predictor, and output a real-time air pressure control deviation; A control compensation parameter determination module is used to determine a control compensation parameter based on the real-time air pressure control deviation, and to adaptively compensate the air pressure parameter of the previous node according to the control compensation parameter.

Images