System and method for multi-parameter cooperative control in PVC steel wire hose production

Abstract

The invention discloses a system and a method for multi-parameter cooperative control in PVC steel wire hose production, and relates to the technical field of plastic hose production, the system comprises a parameter monitoring module, a central processing module, a parameter adjusting module and a quality detection module, the parameter monitoring module is a high-precision intelligent sensor and has a self-adaptive filtering function, and key parameters are measured and transmitted to the central processing module; the central processing module uses an improved multi-objective optimization algorithm to calculate an optimal cooperation value, a simulated annealing-genetic hybrid algorithm is used for iteration, the parameter adjustment module uses intelligent PID, adaptive variable structure control and model prediction control adjustment equipment, and the quality detection module uses multi-modal fusion deep learning and high-precision physical performance detection equipment. And if not, recalculating the adjustment parameters. The production quality and efficiency of the PVC steel wire hose are improved, the cost is reduced, parameters are precisely and cooperatively controlled, the hose is stable in physical performance and few in surface defect, the raw material and equipment maintenance and energy consumption cost is reduced, and the enterprise revenue and market competitiveness are improved.

Description

Technical Field

[0001] The invention relates to the technical field of plastic hose production, and in particular to a system and method for coordinated control of multiple parameters in the production of PVC steel wire hoses. Background Art

[0002] PVC steel wire hose is widely used in many fields due to its excellent flexibility, corrosion resistance and certain strength, such as industrial transportation for conveying liquids and gases; agricultural irrigation as water pipelines; construction field for wire threading, etc. However, there are many problems to be solved in the production process of PVC steel wire hose.

[0003] At present, the multi-parameter control technology in the production process of PVC steel wire hose is relatively backward. Traditional production methods mostly rely on manual experience to set and adjust production parameters, such as temperature, pressure, speed, etc. It is difficult to achieve precise control through manual adjustment, which can easily lead to large fluctuations in parameters. For example, when controlling the plasticizing temperature of PVC raw materials, if the temperature is too high, the raw materials will decompose and affect the physical properties and chemical stability of the hose; if the temperature is too low, the plasticization will be insufficient, resulting in insufficient hose strength and rough surface. In terms of pressure control, inaccurate pressure will cause uneven winding of the steel wire, resulting in large differences in local strength of the hose, reducing the overall quality of the product. At the same time, manual adjustment has a slow response speed and cannot respond to sudden changes in the production process in a timely manner, resulting in low production efficiency and increased scrap rate.

[0004] Although some existing automated control methods can achieve automatic adjustment of some parameters, these methods often independently control a single parameter, ignoring the interrelationship and synergy between the parameters. In actual production, there is a complex coupling relationship between parameters such as temperature, pressure, and speed. For example, changes in the pressure in the extruder will affect the plasticization effect and extrusion speed of the PVC raw material, thereby affecting the uniformity of the steel wire winding and the molding quality of the hose. In addition, the existing automated control methods have weak coping capabilities for interference factors such as raw material quality fluctuations and equipment aging, and cannot fundamentally guarantee the stability of the production process and the reliability of product quality, and it is difficult to meet the growing market demand for high-quality PVC steel wire hoses. Therefore, the development of a system and method that can achieve multi-parameter coordinated control and effectively deal with various interference factors has become a key issue that needs to be urgently solved in the field of PVC steel wire hose production. Summary of the invention

[0005] The present invention proposes a system and method for coordinated control of multiple parameters in the production of PVC steel wire hoses to solve the problems mentioned in the above-mentioned prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A multi-parameter collaborative control system in the production of PVC steel wire hoses, including the following modules:

[0008] Parameter monitoring module: Use intelligent sensors to monitor the key parameters of PVC raw material production in real time. This sensor has an adaptive filtering function, automatically adjusts the filtering parameters according to the interference of the production environment, reduces the influence of noise, and transmits the data to the central processing module;

[0009] Central processing module: Receive the data from the parameter monitoring module, calculate the optimal collaborative value of each parameter using an improved multi-objective optimization algorithm, construct the objective function F = αE + βQ - γC with production efficiency E, product quality Q, and cost C, and at the same time introduce a dynamic weight adjustment mechanism, which is adjusted in real time according to the production stage and product requirements. The weight coefficients α, β, and γ are calculated iteratively using the simulated annealing-genetic hybrid algorithm to obtain the optimal temperature T opt 、the optimal pressure P opt 、the optimal wire winding speed V wire-opt 、the optimal hose extrusion speed V hose-opt ;

[0010] Parameter adjustment module: Adjust the plasticizing temperature of the PVC raw material to T through a heating and cooling device with intelligent PID adjustment. This device has a feed-forward compensation function to compensate for temperature fluctuations; use a pressure regulating valve with adaptive variable structure control to control the pressure in the extruder to P opt ,automatically adjusts the control structure according to the pressure change rate and deviation; adjust the wire winding speed to V with a motor speed regulation device based on model predictive control opt ,and adjust the hose extrusion speed to V wire-opt ,predict the future speed change by establishing a motor dynamic model and make an advance adjustment; hose-opt ,

[0011] Quality inspection module: Use a machine vision system based on multi-modal fusion deep learning to detect the appearance defects of the hoses, and test the hose indicators through physical property testing equipment. The equipment has an automatic calibration and error correction function, and feeds back the test results to the central processing module.

[0012] Furthermore, it also includes a raw material quality analysis module. This module uses hyperspectral imaging-machine learning fusion technology to detect the composition and purity of PVC raw materials in real time. Obtain the spectral characteristic image of the raw material through hyperspectral imaging, and use the support vector machine combined with the improved kernel function algorithm to classify and calculate the purity of the spectral data to obtain the raw material quality parameter M and transmit it to the central processing module. The central processing module introduces a raw material quality correction coefficient K M ,and the objective function becomes F' = αE + βQ - γC × K M ,where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, and y j is the standard raw material quality characteristic value, are the corresponding means respectively.

[0013] Furthermore, it also includes an equipment status monitoring module, which monitors the operating status parameters of production equipment in real time, and establishes an equipment fault prediction model P using an improved long short-term memory network combined with an attention mechanism fault = f(I, U, P). When it exceeds the preset threshold, it gives an early warning and feeds back the equipment status data to the central processing module, and the central processing module adjusts the parameters in combination with the equipment remaining life prediction model.

[0014] Furthermore, the sensors in the parameter monitoring module adopt self-calibration technology, and regularly self-calibrate according to the calibration formula X calibrated = X measured ×(1 + △X), where ΔX is calculated by the Bayesian estimation method in combination with the probability distribution of the sensor historical measurement data and the standard value.

[0015] Furthermore, the central processing module uses a deep reinforcement learning-fuzzy logic hybrid algorithm to adjust the parameters, uses the deep reinforcement learning network to learn the long-term optimal strategy of the production process, combines the fuzzy logic rules to respond and adjust to short-term fluctuations, and dynamically adjusts the parameters according to the production data and quality feedback.

[0016] Furthermore, the quality inspection module uses an algorithm combining a generative adversarial network and transfer learning to evaluate the quality of the hose. It uses the generative adversarial network to generate quality inspection samples, adapts the pre-trained model in related fields to the PVC steel wire hose quality inspection task through transfer learning, processes the machine vision images, identifies appearance defects, and predicts the long-term performance in combination with the physical property detection data.

[0017] Furthermore, the method of the multi-parameter collaborative control system in the production of PVC steel wire hoses includes the following steps:

[0018] Parameter monitoring step: Use intelligent sensors to monitor the plasticizing temperature T of PVC raw materials, the pressure P in the extruder, the wire winding speed V wire , and the hose extrusion speed V hose parameters in real time. The sensor automatically adjusts the filtering parameters according to environmental interference and transmits the data to the central processing unit;

[0019] Parameter calculation steps: The central processing unit receives parameter monitoring data, applies an improved multi-objective optimization algorithm, constructs an objective function F = αE + βQ - γC using production efficiency E, product quality Q, and cost C, introduces a dynamic weight adjustment mechanism to adjust the α, β, and γ weight coefficients in real time, and uses a simulated annealing-genetic hybrid algorithm for iterative calculation. First, perform a global search through the simulated annealing algorithm, and then use the genetic algorithm for local search to obtain the optimal temperature T opt , the optimal pressure P opt , the optimal wire winding speed V wire-opt , the optimal hose extrusion speed V hose-opt ;

[0020] Parameter adjustment steps: Use a heating and cooling device with intelligent PID adjustment and feedforward compensation function to adjust the PVC raw material plasticization temperature to T opt ; Use a pressure regulating valve with adaptive variable structure control to control the pressure in the extruder to P opt ; Use a motor speed regulation device based on model predictive control to adjust the wire winding speed to V wire-opt , and adjust the hose extrusion speed to V hose-opt ;

[0021] Quality inspection steps: Use a machine vision system based on multi-modal fusion deep learning to detect hose appearance defects, and identify defects through an improved residual network combined with an attention mechanism; Use a physical property detection device with automatic calibration and error correction functions to test hose indicators, and feed the detection results back to the central processing unit;

[0022] Feedback adjustment steps: The central processing unit judges whether the product quality meets the standard according to the quality inspection feedback results. If it does not meet the standard, recalculate the optimal parameters and return to the parameter adjustment steps for adjustment until it meets the standard.

[0023] Furthermore, before the parameter monitoring step, it also includes a raw material quality analysis step. Use a hyperspectral imaging-machine learning fusion technology to detect the composition and purity of PVC raw materials in real time. Obtain the raw material spectral feature image through hyperspectral imaging, and use a support vector machine combined with an improved kernel function algorithm to classify and calculate the purity of the spectral data to obtain the raw material quality parameter M. Introduce a raw material quality correction coefficient K in the parameter calculation step M , and the objective function becomes F' = αE + βQ - γC × K M where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, y j is the standard raw material quality characteristic value, are the corresponding means respectively.

[0024] Further, while performing the parameter monitoring step, a device status monitoring step is carried out to monitor the motor current I, voltage U, and operating state parameters of the heating device power P of the production equipment in real time. An improved long short-term memory network combined with an attention mechanism is used to establish a device fault prediction model P fault = f(I, U, P), predict the fault probability, give an early warning when it exceeds the preset threshold, and combine the equipment remaining life prediction model in the parameter calculation step.

[0025] Further, in the quality inspection step, an algorithm combining a generative adversarial network and transfer learning is used to evaluate the quality of the hose. The generative adversarial network is used to generate more quality inspection samples, and through transfer learning, the model pre-trained in the related field is adapted to the quality inspection task of the PVC steel wire hose, process the machine vision image, identify appearance defects, and predict the long-term performance in combination with the physical performance detection data.

[0026] Compared with the existing technologies, the beneficial effects of the present invention are as follows:

[0027] In terms of improving product quality, through multi-parameter collaborative control, the key parameters in the production process can be accurately adjusted. The system can accurately adjust parameters such as temperature, pressure, and speed in real time according to factors such as the characteristics of the PVC raw material and the operating state of the equipment, ensuring that the PVC raw material can be fully and evenly plasticized, enabling the steel wire to be evenly wound, thereby effectively avoiding problems such as insufficient strength and surface defects of the hose. The produced PVC steel wire hose shows more stable and excellent physical properties such as compression resistance and tensile strength, greatly improving the quality and reliability of the product, and being able to better meet the requirements of different application scenarios.

[0028] In terms of improving production efficiency, the present application can quickly respond to changes in the production process. When encountering situations such as fluctuations in raw material quality and minor equipment failures, the system can automatically adjust parameters in a timely manner, avoiding production stagnation or product quality decline caused by unreasonable parameters. At the same time, the optimized parameter adjustment mechanism reduces the debugging time and scrap rate in the production process, making the production process smoother and the production rhythm more stable, greatly improving the production efficiency, shortening the production cycle of the product, helping the enterprise to quickly respond to market demands, and enhancing the market competitiveness.

[0029] In terms of cost reduction, due to the improvement of product quality, the scrap rate is significantly reduced, reducing the waste of raw materials and the additional costs brought by scrap treatment. Moreover, the efficient and stable production process reduces the frequent debugging and maintenance requirements of the equipment, reducing the equipment loss and maintenance costs. In addition, the precise control of production parameters by the system can also effectively reduce energy consumption, further saving production costs and improving the economic benefits of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic block diagram of a multi-parameter collaborative control system in the production of PVC steel wire hoses proposed by the present invention;

[0031] Figure 2 It is a schematic block diagram of a multi-parameter collaborative control method in the production of PVC steel wire hoses proposed by the present invention;

[0032] Figure 3 It is a broken line graph comparing the appearance defects of products before and after the application of this technology proposed by the present invention;

[0033] Figure 4 It is a bar graph of the production efficiency and cost changes after the application of this technology proposed by the present invention. Specific implementation manners

[0034] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0035] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation to the present invention.

[0036] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality" means two or more, unless otherwise specifically defined. In addition, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations. Next, the present invention will be further described in detail with reference to the accompanying drawings.

[0037] Reference Figures 1 to 4 : A system for multi-parameter collaborative control in the production of PVC steel wire hoses, including the following modules:

[0038] Parameter monitoring module: Using high-precision intelligent sensors, which are manufactured based on advanced micro-electro-mechanical system (MEMS) technology and have excellent sensitivity and stability. It can monitor the plasticizing temperature T of PVC raw materials in real time and accurately. Through the built-in thermocouple sensor, it can precisely capture the subtle changes in temperature with an accuracy of ±0.1°C. For the monitoring of the pressure P in the extruder, a piezoresistive pressure sensor is used, which can achieve a measurement accuracy of ±0.5% FS within the range of 0 - 100 MPa. The wire winding speed V wire and the hose extrusion speed V hose are monitored by means of high-precision optical encoders, with a resolution of up to 0.01 r / min. The sensor also integrates an adaptive filtering algorithm. Based on real-time environmental interference data, through fast Fourier transform (FFT) analysis, it automatically adjusts the filtering parameters, effectively filters out noise, and accurately transmits the data to the central processing module through a high-speed serial interface.

[0039] Central processing module: Receives the data from the parameter monitoring module and calculates the optimal collaborative values of each parameter using an improved multi-objective optimization algorithm. Constructs the objective function F = αE + βQ - γC with production efficiency E, product quality Q, and cost C, and at the same time introduces a dynamic weight adjustment mechanism to adjust the weight coefficients α, β, and γ in real time according to the production stage and product requirements. Uses a simulated annealing-genetic hybrid algorithm for iterative calculation. First, performs a global search through the simulated annealing algorithm to quickly approach the optimal solution region, and then uses the genetic algorithm for local fine search to obtain the optimal temperature T opt 、optimal pressure P opt 、optimal wire winding speed V wire-opt 、optimal hose extrusion speed V hose-opt .

[0040] Parameter adjustment module: According to the optimal parameter values calculated by the central processing module, adjusts the plasticizing temperature of PVC raw materials to T opt through a heating and cooling device with intelligent PID adjustment. This device has a feed-forward compensation function and can compensate in advance for temperature fluctuations caused by environmental temperature changes, etc.; controls the pressure in the extruder to P opt using a pressure regulating valve with adaptive variable structure control, which can automatically adjust the control structure according to the pressure change rate and deviation; adjusts the wire winding speed to V wire-opt 、the hose extrusion speed to V hose-opt using a motor speed regulation device based on model predictive control, and predicts future speed changes and adjusts in advance by establishing a motor dynamic model.

[0041] Quality inspection module: An advanced machine vision system based on multi-modal fusion deep learning is adopted. This system integrates a high-resolution visible light camera and a high-sensitivity infrared thermal imager, and can synchronously collect the visible light image and infrared light image information of the hose. After preprocessing the image data, it is input into an improved Residual Network (ResNet) architecture, which incorporates an Attention Mechanism. By assigning higher weights to the key regions of the image, it can more sensitively capture subtle defects such as tiny cracks and bubbles, and the defect recognition accuracy is greatly improved. For the physical property detection, high-precision physical property detection equipment is used. Taking the compressive strength and tensile strength tests as examples, the force sensors of the equipment adopt high-precision strain gauge technology and have an automatic calibration function. They are regularly calibrated by built-in standard weights and combined with an error correction algorithm to compensate for the errors caused by environmental factors or equipment aging in real time, ensuring the accuracy of the test data. The detection results are fed back to the central processing module in real time through a high-speed data interface. Once the detection fails to meet the standard, the central processing module quickly calls an optimization algorithm to recalculate and accurately adjust the production parameters.

[0042] In the present invention, a raw material quality analysis module is further included. This module uses hyperspectral imaging-machine learning fusion technology to detect the composition and purity of PVC raw materials in real time. It obtains the spectral feature image of the raw materials through hyperspectral imaging, and uses the support vector machine combined with the improved kernel function algorithm to classify the spectral data and calculate the purity, obtaining the raw material quality parameter M and transmitting it to the central processing module. When the central processing module calculates, the raw material quality correction coefficient K is introduced M , and the objective function becomes F′ = αE + βQ - γC × K M , where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, y j is the standard raw material quality characteristic value, are the corresponding means respectively.

[0043] In the present invention, a device status monitoring module is further included. This module uses high-precision sensors to monitor key operating status parameters such as the motor current I, voltage U, and heating device power P of the production equipment in real time and accurately. Among them, the current sensor adopts the Hall effect principle and has high sensitivity and fast response characteristics, and can accurately capture the tiny fluctuations of the current; the voltage sensor is based on the voltage division principle and is combined with a high-speed analog-to-digital converter to achieve high-precision acquisition of voltage data; the heating device power is calculated through a power sensor in combination with the real-time voltage and current measurement values. In terms of fault prediction, an improved Long Short-Term Memory Network (LSTM) combined with an attention mechanism is used to construct a device fault prediction model P fault= f(I, U, P). The LSTM network can effectively process time - series data and capture the changing patterns of device operation state parameters over time. The incorporation of the attention mechanism enables the model to automatically focus on the state parameters that are more critical for fault prediction. For example, when the device load changes, it focuses on the changing trend of the current parameter. Through in - depth analysis of these parameters, the fault probability of the device can be accurately predicted. Once the fault probability exceeds the preset threshold, the module immediately issues a warning and feeds back the detailed device state data to the central processing module. After receiving the data, the central processing module combines it with the remaining useful life prediction model of the device based on degradation trajectory modeling and particle filtering algorithm. This model deeply mines and learns from the historical state data of the device, uses the particle filtering algorithm to estimate the current state of the device in real - time, and then accurately predicts the remaining available time of the device, so as to timely adjust production parameters and reasonably arrange the device maintenance plan.

[0044] In the present invention, the sensors in the parameter monitoring module adopt self - calibration technology. Regularly, according to the calibration formula X calibrated = X measured ×(1 + △X) for self - calibration. Specifically, through the Bayesian estimation method, comprehensively analyze the historical measurement data accumulated by the sensor for a long time, as well as the probability distribution of the standard value under different working conditions, and accurately calculate the calibration coefficient ΔX. With the help of this scientific calibration process, the errors introduced by factors such as sensor aging and environmental changes can be effectively eliminated, ensuring that the measurement data is always at a high - precision level, laying a solid foundation for the accurate regulation of subsequent production parameters.

[0045] In the present invention, the central processing module adopts a deep reinforcement learning - fuzzy logic hybrid algorithm to achieve real - time parameter adjustment. The deep reinforcement learning network is based on a multi - layer neural network architecture. By continuously interacting with the dynamic environment of the PVC steel wire hose production process, it continuously learns and explores long - term optimal production strategies. In this process, the network takes production parameters such as temperature, pressure, speed, and product quality feedback as state inputs, and takes the adjustment operations of these parameters as action outputs, and optimizes the strategy according to the reward mechanism (such as positive rewards like product quality compliance and production efficiency improvement, and negative rewards like product defects and equipment failures). At the same time, it combines fuzzy logic rules. Fuzzy logic can, based on expert experience, convert short - term fluctuations (such as slight changes in raw material quality and instantaneous fluctuations in environmental temperature) and uncertainty factors (such as sudden minor equipment failures) in the production process into fuzzy sets and rules. For example, when the temperature fluctuates within a small range, quickly judge whether to make a slight adjustment or a large - scale adjustment according to the fuzzy rules, achieving a rapid response to these sudden situations. Through this hybrid algorithm, the central processing module can dynamically and accurately adjust parameters such as temperature, pressure, and speed according to production data and quality feedback, greatly improving the stability and adaptability of the system.

[0046] In the present invention, the quality inspection module adopts an algorithm combining a generative adversarial network and transfer learning for hose quality evaluation. The generative adversarial network consists of a generator and a discriminator. The generator learns the feature distribution of existing quality inspection samples and generates more similar quality inspection samples to expand the dataset and alleviate the problem of insufficient samples. The discriminator is responsible for judging whether the input sample is a real sample or a forged sample generated by the generator. The two compete against each other, prompting the samples generated by the generator to become closer and closer to the real sample features. In terms of transfer learning, mature models (such as ResNet, VGG, etc.) that have been pre-trained in related fields such as image recognition are quickly adapted to the PVC steel wire hose quality inspection task by fine-tuning network parameters. Deep feature extraction and analysis are performed on the visible light and infrared light images of the hose obtained by machine vision to accurately identify appearance defects. Combining with data such as compressive strength and tensile strength obtained by high-precision physical property detection equipment, methods such as time series analysis are used to predict the long-term performance of the hose, providing more comprehensive and reliable feedback information for the central processing module.

[0047] In the present invention, the following steps are included:

[0048] Parameter monitoring step: Using high-precision intelligent sensors to real-time monitor parameters such as the plasticizing temperature T of PVC raw materials, the pressure P in the extruder, the wire winding speed V wire and the hose extrusion speed V hose . The sensor automatically adjusts the filtering parameters according to environmental interference and accurately transmits the data to the central processing unit.

[0049] Parameter calculation step: The central processing unit receives the parameter monitoring data and uses an improved multi-objective optimization algorithm to construct an objective function F = αE + βQ - γC with production efficiency E, product quality Q, and cost C, and introduces a dynamic weight adjustment mechanism to adjust the weight coefficients of α, β, and γ in real time. The simulated annealing-genetic hybrid algorithm is used for iterative calculation. First, the simulated annealing algorithm is used for global search to quickly approach the optimal solution region, and then the genetic algorithm is used for local fine search to obtain the optimal temperature T opt , the optimal pressure P opt , the optimal wire winding speed V wire-opt , and the optimal hose extrusion speed V hose-opt .

[0050] Parameter adjustment step: According to the optimal parameter values calculated by the parameters, the plasticizing temperature of the PVC raw materials is adjusted to T through a heating and cooling device with intelligent PID adjustment and feedforward compensation function opt ; the pressure in the extruder is controlled to P using a pressure regulating valve with adaptive variable structure control opt ; the wire winding speed is adjusted to V using a motor speed regulating device based on model predictive control wire-opt, adjust the extrusion speed of the hose to V hose-opt .

[0051] Quality inspection steps: Use a machine vision system based on multi-modal fusion deep learning to detect the appearance defects of the hose, fuse visible light and infrared light image information, and identify defects through an improved residual network combined with an attention mechanism; Test the compressive strength, tensile strength and other indicators of the hose through a physical performance detection device with high precision and automatic calibration and error correction functions, and feed back the detection results to the central processing unit.

[0052] Feedback adjustment steps: The central processing unit judges whether the product quality meets the standard according to the quality inspection feedback results. If it does not meet the standard, recalculate the optimal parameters and return to the parameter adjustment step for adjustment until it meets the standard.

[0053] In the present invention, before the parameter monitoring step, a raw material quality analysis step is further included. The hyperspectral imaging-machine learning fusion technology is used to detect the composition and purity of PVC raw materials in real time. The spectral characteristic image of the raw material is obtained through hyperspectral imaging, and the support vector machine combined with the improved kernel function algorithm is used to classify the spectral data and calculate the purity to obtain the raw material quality parameter M. The raw material quality correction coefficient K is introduced in the parameter calculation step M , and the objective function becomes F' = αE + βQ - γC × K M , where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, y j is the standard raw material quality characteristic value, are the corresponding means respectively.

[0054] In the present invention, while the parameter monitoring step is carried out, the equipment status monitoring step is carried out. Real-time monitor the operating state parameters such as the motor current I, voltage U, and heating device power P of the production equipment, and establish an equipment fault prediction model P using the improved long short-term memory network combined with an attention mechanism fault = f(I, U, P), and predict the fault probability. Give an early warning when it exceeds the preset threshold, and consider the equipment status factors to adjust the parameters in the parameter calculation step in combination with the equipment remaining life prediction model (based on degradation trajectory modeling and particle filter algorithm).

[0055] In the present invention, in the quality inspection step, an algorithm combining a generative adversarial network and transfer learning is used to evaluate the quality of the hose. Use the generative adversarial network to generate more quality inspection samples, and adapt the pre-trained model in the related field to the quality inspection task of the PVC steel wire hose through transfer learning, process the machine vision image, identify the appearance defects, and predict the long-term performance in combination with the physical performance detection data, providing a reliable basis for the feedback adjustment step.

[0056] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent replacements or changes, shall be covered by the protection scope of the present invention.

Claims

1. A system for multi-parameter collaborative control in the production of PVC steel wire hoses, characterized in that, It includes the following modules: Parameter monitoring module: An intelligent sensor is used to monitor the key parameters in the production of PVC raw materials in real time. This sensor has an adaptive filtering function, automatically adjusts the filtering parameters according to the interference in the production environment, reduces the influence of noise, and transmits the data to the central processing module; Central processing module: Receives the data from the parameter monitoring module, calculates the optimal collaborative values of each parameter using an improved multi-objective optimization algorithm, constructs the objective function F = αE + βQ - γC with production efficiency E, product quality Q, and cost C, and at the same time introduces a dynamic weight adjustment mechanism to adjust the α, β, and γ weight coefficients in real time according to the production stage and product requirements. Performs iterative calculations using the simulated annealing-genetic hybrid algorithm to obtain the optimal temperature T opt , the optimal pressure P opt , the optimal wire winding speed V wire-opt , the optimal hose extrusion speed V hose-opt ; Parameter adjustment module: Adjust the plasticizing temperature of PVC raw materials to T through a heating and cooling device with intelligent PID adjustment opt , this device has a feed-forward compensation function to compensate for temperature fluctuations; Use a pressure regulating valve with adaptive variable structure control to control the pressure in the extruder to P opt , automatically adjust the control structure according to the pressure change rate and deviation; With the help of a motor speed regulation device based on model predictive control, adjust the wire winding speed to V wire-opt , the hose extrusion speed to V hose-opt , predict future speed changes by establishing a motor dynamic model and adjust in advance; Quality inspection module: A machine vision system based on multi-modal fusion deep learning is used to detect the appearance defects of the hose. The physical performance of the hose is tested by physical performance testing equipment. The equipment has automatic calibration and error correction functions, and feeds back the inspection results to the central processing module.

2. The system for multi-parameter collaborative control in the production of PVC steel wire hoses according to claim 1, characterized in that, It also includes a raw material quality analysis module, which uses hyperspectral imaging - machine learning fusion technology to detect the composition and purity of PVC raw materials in real time. The spectral feature image of the raw materials is obtained through hyperspectral imaging, and the support vector machine combined with the improved kernel function algorithm is used to classify the spectral data and calculate the purity, obtaining the raw material quality parameter M and transmitting it to the central processing module. When the central processing module calculates, the raw material quality correction coefficient K is introduced M , and the objective function becomes F′ = αE + βQ - γC × K M , where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, and y j is the standard raw material quality characteristic value are the corresponding means respectively 3. The system for multi-parameter collaborative control in the production of PVC steel wire hoses according to claim 1, characterized in that, It also includes a device status monitoring module, which monitors the operating status parameters of production equipment in real time, and establishes a device fault prediction model P using an improved long short-term memory network combined with an attention mechanism fault = f(I, U, P). When the preset threshold is exceeded, it gives an early warning and feeds back the device status data to the central processing module, and the central processing module adjusts the parameters in combination with the device remaining life prediction model.

4. The system for multi-parameter collaborative control in the production of PVC steel wire hoses according to claim 1, characterized in that, The sensors in the parameter monitoring module adopt self-calibration technology and are periodically self-calibrated according to the calibration formula X xalibrated = X measured ×(1 + ΔX), where ΔX is calculated by the Bayesian estimation method, combining the probability distributions of the historical measurement data and the standard values of the sensors.

5. The system for multi-parameter collaborative control in the production of PVC steel wire hoses according to claim 1, characterized in that, The central processing module uses a deep reinforcement learning-fuzzy logic hybrid algorithm to adjust parameters. It uses a deep reinforcement learning network to learn the long-term optimal strategy of the production process, combines fuzzy logic rules to respond to short-term fluctuations, and dynamically adjusts parameters according to production data and quality feedback.

6. The system for multi-parameter collaborative control in the production of PVC steel wire hoses according to claim 1, characterized in that, The quality inspection module uses an algorithm combining generative adversarial networks and transfer learning to evaluate the quality of the hose. It uses generative adversarial networks to generate quality inspection samples, and adapts the pre-trained model in related fields to the quality inspection task of PVC steel wire hoses through transfer learning, processes the machine vision images, identifies appearance defects, and predicts the long-term performance in combination with physical performance test data.

7. A method for applying the multi-parameter collaborative control system in the production of PVC steel wire hoses according to any one of claims 1-6, characterized in that, It includes the following steps: Parameter monitoring steps: Use intelligent sensors to monitor the plasticizing temperature T of PVC raw materials, the pressure P in the extruder, and the wire winding speed V wire , and the hose extrusion speed V hose parameters in real time. The sensor automatically adjusts the filtering parameters according to environmental interference and transmits the data to the central processing unit; Parameter calculation steps: The central processing unit receives parameter monitoring data, applies an improved multi-objective optimization algorithm, constructs an objective function F = αE + βQ - γC using production efficiency E, product quality Q, and cost C, introduces a dynamic weight adjustment mechanism to adjust the α, β, and γ weight coefficients in real time, and performs iterative calculations using a simulated annealing-genetic hybrid algorithm. First, perform a global search through the simulated annealing algorithm, and then use the genetic algorithm for local search to obtain the optimal temperature T opt , the optimal pressure P opt , the optimal wire winding speed V wire-opt , the optimal hose extrusion speed V hose-opt ; Parameter adjustment steps: Adjust the plasticizing temperature of PVC raw materials to T through a heating and cooling device with intelligent PID adjustment and feedforward compensation function opt ; Control the pressure in the extruder to P using a pressure regulating valve with adaptive variable structure control opt ; Adjust the wire winding speed to V with the help of a motor speed regulating device based on model predictive control wire-opt and adjust the hose extrusion speed to V hose-opt ; Quality inspection step: A machine vision system based on multi-modal fusion deep learning is used to detect the appearance defects of the hose, and defects are identified through an improved residual network combined with an attention mechanism; the physical performance of the hose is tested by physical performance testing equipment with automatic calibration and error correction functions, and the inspection results are fed back to the central processing unit; Feedback adjustment step: The central processing unit judges whether the product quality meets the standard according to the quality inspection feedback result. If it does not meet the standard, it recalculates the optimal parameters and returns to the parameter adjustment step for adjustment until it meets the standard.

8. The method of the multi-parameter collaborative control system in the production of PVC steel wire hoses according to claim 7, characterized in that, Before the parameter monitoring step, it also includes a raw material quality analysis step. The hyperspectral imaging-machine learning fusion technology is used to detect the composition and purity of PVC raw materials in real time. The spectral feature image of the raw materials is obtained through hyperspectral imaging, and the support vector machine combined with the improved kernel function algorithm is used to classify and calculate the purity of the spectral data to obtain the raw material quality parameter M. The raw material quality correction coefficient K is introduced in the parameter calculation step M , and the objective function becomes F′ = αE + βQ - γC × K M , where K M is dynamically adjusted according to the similarity between the raw material quality and the preset standard. The similarity calculation formula is x i is the current raw material quality characteristic value, y j is the standard raw material quality characteristic value, are the corresponding means respectively 9. The method of the multi-parameter collaborative control system in the production of PVC steel wire hoses according to claim 7, characterized in that, While performing the parameter monitoring step, the device status monitoring step is carried out to monitor the motor current I, voltage U, and operating status parameters of the heating device power P of the production equipment in real time. An improved long short-term memory network combined with an attention mechanism is used to establish a device fault prediction model P fault = f(I, U, P), predict the fault probability, give an early warning when it exceeds the preset threshold, and combine the device remaining life prediction model in the parameter calculation step.

10. The method of the multi-parameter collaborative control system in the production of PVC steel wire hoses according to claim 7, characterized in that, In the quality inspection step, an algorithm combining generative adversarial networks and transfer learning is used to evaluate the quality of the hose. Generative adversarial networks are used to generate more quality inspection samples, and the pre-trained model in related fields is adapted to the quality inspection task of PVC steel wire hoses through transfer learning, processes the machine vision images, identifies appearance defects, and predicts the long-term performance in combination with physical performance test data.

Images