Printing and dyeing intelligent manufacturing system based on bimodal monitoring network and monitoring method

Abstract

The invention discloses a monitoring method of a printing and dyeing intelligent manufacturing system based on a bimodal monitoring network, and the method comprises the steps: 1, data collection and preprocessing: a visual monitoring module and an auditory monitoring module work synchronously, and collect image and sound data in a cloth printing and dyeing process in real time; 2, data fusion and analysis: carrying out feature extraction on the image and sound data collected in the step 1, carrying out data fusion and comprehensive analysis, and generating a comprehensive evaluation report of the running state and the product quality of the printing and dyeing equipment; the dual-mode monitoring makes up the deficiency of single-mode monitoring, provides more comprehensive and accurate monitoring information, and improves the precision of fault detection and quality control.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial automation and intelligent manufacturing, and specifically relates to a printing and dyeing intelligent manufacturing system and a monitoring method based on a dual-mode monitoring network. Background Art

[0002] The printing and dyeing industry is an important part of the textile industry, mainly responsible for dyeing and printing patterns on textiles. The traditional printing and dyeing process usually relies on manual operation and experience judgment, which is easily affected by human factors, resulting in unstable quality, low efficiency and high production costs. With the development of science and technology, intelligent manufacturing technology has gradually been applied to the printing and dyeing industry, improving production efficiency and product quality through automated equipment and advanced monitoring methods.

[0003] Currently, the monitoring of printing and dyeing production lines mainly relies on a single visual monitoring or auditory monitoring method. Visual monitoring uses cameras and image processing technology to monitor the color and pattern of fabrics, and detects problems such as color difference and uneven printing and dyeing; auditory monitoring uses microphones and sound analysis technology to listen to the sound of equipment operation and detect equipment failures and abnormalities. However, a single mode of monitoring has certain limitations and cannot fully and accurately reflect problems in the production process. Summary of the invention

[0004] The inventors found that by combining visual monitoring and auditory monitoring to conduct dual-mode monitoring, a comprehensive analysis of the production process can be performed through data fusion technology to provide more comprehensive monitoring information. Visual monitoring and auditory monitoring complement each other, improving the accuracy and timeliness of fault diagnosis, thereby optimizing the production process, reducing downtime, and improving production efficiency and product quality.

[0005] In order to solve the above problems, the present invention proposes a printing and dyeing intelligent manufacturing system based on a dual-modal monitoring network.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows: a printing and dyeing intelligent manufacturing system based on a dual-modal monitoring network, comprising:

[0007] A visual monitoring module, which collects and analyzes the color, pattern, and surface quality of printed and dyed fabrics in real time, and identifies color differences, fabric stains, and printing and dyeing uniformity of printed and dyed fabrics;

[0008] An auditory monitoring module, which monitors the operating sound of the printing and dyeing equipment in real time and monitors abnormal sound information during the operation of the printing and dyeing equipment;

[0009] A data fusion module, which synchronously processes and comprehensively analyzes the image data collected by the visual monitoring module and the sound data collected by the auditory monitoring module to generate a comprehensive evaluation report on the operation status of the printing and dyeing equipment and the product quality;

[0010] A control unit, the control unit being used to control and adjust printing and dyeing parameters in real time according to the comprehensive evaluation report;

[0011] A fault diagnosis unit, which performs predictive maintenance and diagnosis of printing and dyeing equipment faults based on data collected by the visual monitoring module and the auditory monitoring module;

[0012] A safety monitoring module monitors the safety status of the workshop in real time through the visual monitoring module and the auditory monitoring module, and identifies and warns of potential safety hazards.

[0013] Furthermore, the visual monitoring module includes a high-resolution camera and an image processing unit.

[0014] Furthermore, the auditory monitoring module includes a high-sensitivity microphone and a sound analysis unit.

[0015] A monitoring method, including a printing and dyeing intelligent manufacturing system based on a dual-modal monitoring network, comprises the following steps:

[0016] Step 1: Data collection and preprocessing: The visual monitoring module and the auditory monitoring module work synchronously to collect image and sound data of the fabric printing and dyeing process in real time;

[0017] Step 2: Data fusion and analysis: Extract features from the image and sound data collected in step 1, perform data fusion and comprehensive analysis, and generate a comprehensive evaluation report on the operating status of the printing and dyeing equipment and product quality;

[0018] Step 3: Real-time control and adjustment: According to the comprehensive evaluation report in step 2, adjust the printing and dyeing parameters in real time;

[0019] Step 4: Fault prediction and diagnosis: Analyze according to step 3, predict possible faults of printing and dyeing equipment and give early warning;

[0020] Step 5. Safety monitoring and alarm: Monitor the workshop environment in real time, identify abnormal situations and issue alarms in time.

[0021] Furthermore, in step 1, the image processing unit processes the collected image; the image data I(t) is captured at time t, and the processing steps include:

[0022] S101, using a standard color card for comparison, and adjusting the image color to make it consistent with the actual fabric color;

[0023] S102, detecting the continuity and consistency of the pattern through a template matching algorithm, and identifying the defective pattern;

[0024] S103, using edge detection and morphological processing methods to identify stains and color differences on the cloth;

[0025] S104, set the original image to I raw (t), the preprocessed image is I prs (t), then:

[0026] I pre (t) = Preproces(I raw (t)).

[0027] Furthermore, in step 1, the sound analysis unit processes the collected sound signal; the sound signal S(t) is captured at time t, and the processing steps include:

[0028] S201, using a filter to remove environmental noise and non-critical sound signals;

[0029] S202, converting the sound signal into a spectrogram, analyzing the frequency components of the sound to detect abnormal frequencies;

[0030] S203, identifying normal and abnormal sound patterns through a machine learning model, and detecting equipment failure or abnormality;

[0031] S204, assuming that the original sound signal is S raw (t), the preprocessed sound signal is S pre (t), then:

[0032] S pre (t) = Preproces(S raw (t)).

[0033] Furthermore, in step 2, the feature extraction step includes:

[0034] S301, visual feature extraction, using YOLOv5 network to extract image features F I ;

[0035] Let the preprocessed image be I pre (t), the extracted image feature is F I (t), then:

[0036] F I (t) = YOLOv5(I pre (t));

[0037] S302, acoustic feature extraction, perform spectrum analysis on the sound signal, and extract sound feature F S ;

[0038] Use Fast Fourier Transform (FFT) to convert the sound signal from the time domain to the frequency domain:

[0039] F S (t) = FFT(S pre (t)).

[0040] Furthermore, in step 2, the data fusion step includes:

[0041] S401, feature level fusion, image feature F I (t) and sound characteristics F S (t) to generate comprehensive features F(t),

[0042] Fusion via weighted averaging:

[0043] F(t)=αF I (t)+βF S (t);

[0044] Among them, α and β are fusion weight coefficients, satisfying α+β=1;

[0045] S402, decision-level fusion, respectively identify the image features and sound features, and obtain their respective recognition results R I (t) and R S (t);

[0046] Suppose the image recognition model is M I , the sound recognition model is M S ,but:

[0047] R I (t) = M I (F I (t));

[0048] R S (t) = M S (F S (t));

[0049] The two recognition results are fused to generate the final recognition result R(t):

[0050] R(t)=R I (t)^R S (t).

[0051] Furthermore, in step 3, real-time analysis is performed based on the fusion feature F(t) to generate a control strategy C(t), and the steps include:

[0052] S501, assume that the control strategy model is M C ,but:

[0053] C(t)=M C (F(t));

[0054] S502, adjust the production parameter P(t) according to the control strategy C(t); set the current production parameter to P current (t), the adjusted production parameter is P new t), then:

[0055] P new (t)=Adjust(P current (t), C(t)).

[0056] Furthermore, in step 4, a machine learning algorithm is used to analyze historical data H and real-time data F(t) to extract key features T(t), and the steps include:

[0057] S601, assuming that the feature extraction function is Feature Extraction, then:

[0058] T(t)=Feature Extraction(H, F(t));

[0059] S602: training fault prediction model M P , real-time prediction of potential failures F P (t); Assume the fault prediction model is M P ,but:

[0060] F P (t) = M P (T(t));

[0061] In the step 5, the visual monitoring module and the auditory monitoring module are used to monitor the workshop environment and the operating status of the printing and dyeing equipment in real time, and the environmental features E(t) are extracted. The steps include:

[0062] S701: If the environment monitoring function is Environment Monitoring, then:

[0063] E(t)=EnvironmentMonitoring(I(t), S(t));

[0064] S702: Perform anomaly detection on environmental features E(t) to identify potential safety hazards H(t); the anomaly detection model is M A ,but:

[0065] H(t)=M A E(t);

[0066] S703, when an abnormal situation is detected, an alarm signal A(t) is issued;

[0067] If the alarm function is Alarm, then:

[0068] A(t)=Alarm(H(t)).

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] 1. Comprehensive monitoring of the printing and dyeing production process is achieved by combining visual and auditory monitoring methods. Visual monitoring can accurately detect the color, pattern and surface quality of the fabric, and auditory monitoring can promptly detect abnormal sound signals during equipment operation. This dual-mode monitoring makes up for the shortcomings of single-mode monitoring, provides more comprehensive and accurate monitoring information, and improves the accuracy of fault detection and quality control.

[0071] 2. Through data fusion technology, visual data and auditory data are processed comprehensively to generate a more accurate comprehensive evaluation report. The multi-sensor data fusion algorithm improves the reliability and accuracy of monitoring information and reduces the errors and omissions that may occur in a single sensor.

[0072] 3. Using machine learning algorithms and big data analysis technology, the operating status of equipment can be monitored and analyzed in real time, and equipment failures can be predicted in advance. By establishing an equipment health model, the equipment operating status can be evaluated in real time, potential failures can be discovered in time, and early warnings can be issued. This real-time fault prediction capability significantly improves the reliability of the production line and reduces downtime and maintenance costs caused by equipment failures.

[0073] 4. Combined with visual and auditory data for comprehensive analysis, through pattern recognition and feature extraction technology, it can accurately locate and diagnose equipment failures. The network's efficient performance in image target detection and recognition enables visual monitoring to quickly identify fabric surface defects and equipment anomalies. Accurate fault diagnosis capabilities improve maintenance efficiency, reduce production losses, and ensure the stable operation of the production line.

[0074] 5. The control unit automatically adjusts parameters such as temperature, pressure, dye concentration and speed during the printing and dyeing process through a closed-loop control system based on real-time monitoring data and comprehensive evaluation reports. The closed-loop control system can optimize production parameters in real time to ensure the best production results. Real-time control and optimization improve the automation and intelligence level of the production process, reduce human intervention and misoperation, and improve production efficiency and product quality.

[0075] 6. Through real-time monitoring and data analysis of the printing and dyeing process, the production parameters can be adjusted intelligently to ensure the consistency and stability of printing and dyeing quality. The automatic adjustment system can dynamically adjust the production parameters according to the different characteristics of the fabric and production requirements to meet different production needs. Intelligent adjustment of production parameters effectively reduces the uncertainties in the production process and improves the consistency and quality pass rate of products.

[0076] 7. The safety monitoring module combines visual and auditory monitoring methods to monitor the workshop environment and equipment operating status in real time, and promptly identify and deal with potential safety hazards. By detecting abnormal visual and auditory signals, it can effectively prevent accidents such as fires, leaks, and unsafe operations by personnel. Comprehensive safety monitoring improves the safety of the production process and ensures the safety of operators and equipment.

[0077] 8. The safety monitoring module can send out alarm signals in time when abnormal situations are detected, including local sound and light alarms and remote alarms. The alarm system can quickly notify on-site personnel and management personnel to take appropriate safety measures to prevent accidents from occurring and spreading. Timely alarm and response mechanisms improve the efficiency of handling safety incidents and reduce accident losses and personal injuries.

[0078] 9. By improving production efficiency, reducing equipment failures and downtime, and optimizing the production process, production costs can be significantly reduced. The improvement of automation and intelligence levels reduces dependence on manual operations and reduces labor costs. By improving product quality and consistency, the defective rate and rework costs are reduced, further reducing production costs.

[0079] 10. The efficient, accurate and safe production management system has improved the quality and production efficiency of printing and dyeing products, bringing higher economic benefits to the enterprise. Through the application of intelligent manufacturing technology, the innovation ability and market competitiveness of the enterprise have been improved. High-quality and efficient production capacity enables enterprises to better meet market demand, win more customers and orders, and enhance market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 This is the flow chart of the printing and dyeing intelligent manufacturing system based on the dual-modal monitoring network of this application document;

[0081] Figure 2 It is the visual feature extraction network model in this application document;

[0082] Figure 3 It is a schematic diagram of data fusion in this application document. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] like Figure 1-Figure 3As shown, the technical solution adopted by the present invention is as follows: This embodiment provides a printing and dyeing intelligent manufacturing system based on a dual-modal monitoring network,

[0085] It includes a visual monitoring module, which includes a high-resolution camera and an image processing unit for real-time acquisition and analysis of image data of printed and dyed fabrics. The image processing unit processes and analyzes the image data, including color correction, pattern matching, quality inspection, and identification of color difference, fabric stains, and printing and dyeing uniformity of printed and dyed fabrics.

[0086] It includes an auditory monitoring module, which includes a high-sensitivity microphone and a sound analysis unit for monitoring the operating sound of the printing and dyeing equipment. The sound analysis unit processes and analyzes the sound signal, including noise filtering, spectrum analysis and pattern recognition, and detects abnormal sound signals during the operation of the equipment to identify equipment failures or abnormal operation.

[0087] It includes a data fusion module, which is used to synchronously process and comprehensively analyze the image data collected by the visual monitoring module and the sound data collected by the auditory monitoring module. The multi-sensor data fusion algorithm combines visual and auditory data to generate a comprehensive evaluation report on the operating status of printing and dyeing equipment and product quality, improving the accuracy and reliability of monitoring.

[0088] It includes a control unit, which includes a closed-loop control system, which controls and adjusts printing and dyeing parameters, including temperature, pressure, dye concentration and speed, etc. in real time according to a comprehensive evaluation report.

[0089] It includes a fault diagnosis unit, which is based on a machine learning algorithm and uses the data collected by the visual monitoring module and the auditory monitoring module to analyze the historical operation data and real-time monitoring data of the equipment. It establishes an equipment health model, performs fault prediction and diagnosis, and warns of potential equipment failures in advance, thereby improving the reliability of the equipment and the operating efficiency of the production line.

[0090] It includes a safety monitoring module, which monitors the safety status of the workshop in real time through the visual monitoring module and the auditory monitoring module, identifies and alerts potential safety hazards, including fire, leakage, unsafe operation of personnel, etc., to ensure the safety of the production process.

[0091] Another embodiment further provides a monitoring method, the steps of which are as follows:

[0092] Step 1: Data collection and preprocessing: The visual monitoring module and the auditory monitoring module work synchronously to collect image and sound data of the fabric printing and dyeing process in real time;

[0093] Step 2: Data fusion and analysis: Extract features from the image and sound data collected in step 1, perform data fusion and comprehensive analysis, and generate a comprehensive evaluation report on the operating status of the printing and dyeing equipment and product quality;

[0094] Step 3: Real-time control and adjustment: According to the comprehensive evaluation report in step 2, adjust the printing and dyeing parameters in real time;

[0095] Step 4: Fault prediction and diagnosis: Analyze according to step 3, predict possible faults of printing and dyeing equipment and give early warning;

[0096] Step 5. Safety monitoring and alarm: Monitor the workshop environment in real time, identify abnormal situations and issue alarms in time.

[0097] In step 1, image data of the printed and dyed fabric is collected in real time by a high-resolution camera installed above the production line to capture detailed images of the fabric surface.

[0098] In step 1, the image processing unit processes the captured image, including color correction to eliminate the influence of lighting, pattern matching to detect the consistency of fabric patterns, and quality inspection to identify color differences, stains, and uneven printing and dyeing. Image data I(t) is captured at time t, and the processing steps include:

[0099] S101, using a standard color card for comparison, and adjusting the image color to make it consistent with the actual fabric color;

[0100] S102, detecting the continuity and consistency of the pattern through a template matching algorithm, and identifying the defective pattern;

[0101] S103, using edge detection and morphological processing methods to identify stains and color differences on the cloth;

[0102] S104, set the original image to I raw (t), the preprocessed image is I pre (t), then:

[0103] I pre (t) = Preproces(I raw (t)).

[0104] In step 1, the operation sound of the printing and dyeing equipment is monitored in real time through a high-sensitivity microphone, which is installed near key equipment to capture the sound signals generated during the operation of the equipment.

[0105] In step 1, the sound analysis unit processes the collected sound signal, including noise filtering, spectrum analysis and pattern recognition. The sound signal S(t) is captured at time t, and the processing steps include:

[0106] S201, using a filter to remove environmental noise and non-critical sound signals;

[0107] S202, converting the sound signal into a spectrogram, analyzing the frequency components of the sound to detect abnormal frequencies;

[0108] S203, identifying normal and abnormal sound patterns through a machine learning model, and detecting equipment failure or abnormality;

[0109] S204, assuming that the original sound signal is S raw (t), the preprocessed sound signal is S pre (t), then:

[0110] S pre (t) = Preproces(S raw (t)).

[0111] In step 2, the feature extraction step includes:

[0112] S301, visual feature extraction, using YOLOv5 network to extract image features F I ;

[0113] Let the preprocessed image be I pre (t), the extracted image feature is F I (t), then:

[0114] F I (t) = YOLOv5(I pre (t));

[0115] S302, acoustic feature extraction, perform spectrum analysis on the sound signal, and extract sound feature F S ;

[0116] Use Fast Fourier Transform (FFT) to convert the sound signal from the time domain to the frequency domain:

[0117] F S (t) = FFT(S pre (t)).

[0118] In step 2, the data fusion step includes:

[0119] S401, feature level fusion, image feature F I (t) and sound characteristics F S (t) to generate comprehensive features F(t),

[0120] Fusion via weighted averaging:

[0121] F(t)=αFI (t)+βF S (t);

[0122] Among them, α and β are fusion weight coefficients, satisfying α+β=1;

[0123] S402, decision-level fusion, respectively identify the image features and sound features, and obtain their respective recognition results R I (t) and R S (t);

[0124] Suppose the image recognition model is M I , the sound recognition model is M S ,but:

[0125] R I (t) = M I (F I (t));

[0126] R S (t) = M S (F S (t));

[0127] The two recognition results are fused to generate the final recognition result R(t):

[0128] R(t)=R I (t)^R S (t).

[0129] In step 3, real-time analysis is performed based on the fusion feature F(t) to generate a control strategy C(t), which includes:

[0130] S501, assume that the control strategy model is M C ,but:

[0131] C(t)=M C (F(t));

[0132] S502, adjust the production parameter P(t) according to the control strategy C(t); set the current production parameter to P current (t), the adjusted production parameter is P new (t), then:

[0133] P new (t)=Adjust(P current (t), C(t)).

[0134] In step 4, a machine learning algorithm is used to analyze historical data H and real-time data F(t) to extract key features T(t). The steps include:

[0135] S601, assuming that the feature extraction function is Feature Extraction, then:

[0136] T(t)=Feature Extraction(H, F(t));

[0137] S602: training fault prediction model M P , real-time prediction of potential failures F P (t); Assume the fault prediction model is M P ,but:

[0138] F P (t) = M P (T(t));

[0139] In step 5, the workshop environment and the operating status of the printing and dyeing equipment are monitored in real time through the visual monitoring module and the auditory monitoring module to extract the environmental features E(t). The steps include:

[0140] S701: If the environment monitoring function is Environment Monitoring, then:

[0141] E(t)=EnvironmentMonitoring(I(t), S(t));

[0142] S702: Perform anomaly detection on environmental features E(t) to identify potential safety hazards H(t); the anomaly detection model is M A ,but:

[0143] H(t)=M A E(t);

[0144] S703, when an abnormal situation is detected, an alarm signal A(t) is issued;

[0145] If the alarm function is Alarm, then:

[0146] A(t)=Alarm(H(t)).

[0147] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A printing and dyeing intelligent manufacturing system based on a dual-modal monitoring network, characterized in that: include: A visual monitoring module, which collects and analyzes the color, pattern, and surface quality of printed and dyed fabrics in real time, and identifies color differences, fabric stains, and printing and dyeing uniformity of printed and dyed fabrics; An auditory monitoring module, which monitors the operating sound of the printing and dyeing equipment in real time and monitors abnormal sound information during the operation of the printing and dyeing equipment; A data fusion module, which synchronously processes and comprehensively analyzes the image data collected by the visual monitoring module and the sound data collected by the auditory monitoring module to generate a comprehensive evaluation report on the operation status of the printing and dyeing equipment and the product quality; A control unit, the control unit being used to control and adjust printing and dyeing parameters in real time according to the comprehensive evaluation report; A fault diagnosis unit, which performs predictive maintenance and diagnosis of printing and dyeing equipment faults based on data collected by the visual monitoring module and the auditory monitoring module; A safety monitoring module monitors the safety status of the workshop in real time through the visual monitoring module and the auditory monitoring module, and identifies and warns of potential safety hazards.

2. The printing and dyeing intelligent manufacturing system based on the dual-modal monitoring network according to claim 1 is characterized in that: The visual monitoring module includes a high-resolution camera and an image processing unit.

3. The printing and dyeing intelligent manufacturing system based on the dual-modal monitoring network according to claim 1 is characterized in that: The auditory monitoring module includes a high-sensitivity microphone and a sound analysis unit.

4. A monitoring method, characterized in that: The printing and dyeing intelligent manufacturing system based on the dual-modal monitoring network according to claims 1 to 3 comprises the following steps: Step 1: Data collection and preprocessing: The visual monitoring module and the auditory monitoring module work synchronously to collect image and sound data of the fabric printing and dyeing process in real time; Step 2: Data fusion and analysis: Extract features from the image and sound data collected in step 1, perform data fusion and comprehensive analysis, and generate a comprehensive evaluation report on the operating status of the printing and dyeing equipment and product quality; Step 3: Real-time control and adjustment: According to the comprehensive evaluation report in step 2, adjust the printing and dyeing parameters in real time; Step 4: Fault prediction and diagnosis: Analyze according to step 3, predict possible faults of printing and dyeing equipment and give early warning; Step 5. Safety monitoring and alarm: Monitor the workshop environment in real time, identify abnormal situations and issue alarms in time.

5. The monitoring method according to claim 4, characterized in that: In step 1, the image processing unit processes the collected image; the image data I(t) is captured at time t, and the processing steps include: S101, using a standard color card for comparison, and adjusting the image color to make it consistent with the actual fabric color; S102, detecting the continuity and consistency of the pattern through a template matching algorithm, and identifying the defective pattern; S103, using edge detection and morphological processing methods to identify stains and color differences on the cloth; S104, set the original image to I raw (t), the preprocessed image is I pre (t), then: I pre (t)=Preproces(I raw (t))。 6. The monitoring method according to claim 4, characterized in that: In the step 1, the sound analysis unit processes the collected sound signal; The sound signal S(t) is captured at time t, and the processing steps include: S201, using a filter to remove environmental noise and non-critical sound signals; S202, converting the sound signal into a spectrogram, analyzing the frequency components of the sound to detect abnormal frequencies; S203, identifying normal and abnormal sound patterns through a machine learning model, and detecting equipment failure or abnormality; S204, assuming that the original sound signal is S raw (t), the preprocessed sound signal is S pre (t), then: S pre (t)=Preproces(S raw (t))。 7. The monitoring method according to claim 4, characterized in that: In the step 2, the feature extraction step includes: S301, visual feature extraction, using YOLOv5 network to extract image features F I ; Let the preprocessed image be I pre (t), the extracted image feature is F I (t), then: F I (t)=YOLOv5(I pre (t)); S302, acoustic feature extraction, perform spectrum analysis on the sound signal, and extract the sound feature F S ; Use Fast Fourier Transform (FFT) to convert the sound signal from the time domain to the frequency domain: F S (t)=FFT(S pre (t))。 8. The monitoring method according to claim 4, characterized in that: In the step 2, the data fusion step includes: S401, feature level fusion, image feature F I (t) and sound characteristics F S (t) to generate comprehensive features F(t), Fusion via weighted averaging: F(t)=αF l (t)+βF S (t); Among them, α and β are fusion weight coefficients, satisfying α+β=1; S402, decision-level fusion, respectively identify the image features and sound features, and obtain their respective recognition results R I (t) and R S (t); Suppose the image recognition model is M I , the sound recognition model is M S ,but: R I (t)=M I (F I (t)); R S (tt))=M S (F S (t)); The two recognition results are fused to generate the final recognition result R(t): R(t)=R I (t)^R S (t)。 9. The monitoring method according to claim 8, characterized in that: In step 3, real-time analysis is performed based on the fusion feature F(t) to generate a control strategy C(t), and the steps include: S501, assume that the control strategy model is M C ,but: C(t)=M C (F(t)); S502, adjust the production parameter P(t) according to the control strategy C(t); set the current production parameter to P current (t), the adjusted production parameter is P new (t), then: P new (t)=Adjust(P current (t),C(t))。 10. The monitoring method according to claim 8, characterized in that: In step 4, a machine learning algorithm is used to analyze historical data H and real-time data F(t) to extract key features T(t), and the steps include: S601, assuming that the feature extraction function is Feature Extraction, then: T(t)=Feature Extraction(H, F(t)); S602: training fault prediction model M P , real-time prediction of potential failures F P (t); Assume the fault prediction model is M P ,but: F P (t)=M P (T(t)); In the step 5, the visual monitoring module and the auditory monitoring module are used to monitor the workshop environment and the operating status of the printing and dyeing equipment in real time, and the environmental features E(t) are extracted. The steps include: S701: If the environment monitoring function is Environment Monitoring, then: E(t)=Environment Monitoring(I(t), S(t)); S702: Perform anomaly detection on environmental features E(t) to identify potential safety hazards H(t); the anomaly detection model is M A ,but: H(t)=M A E(t); S703, when an abnormal situation is detected, an alarm signal A(t) is issued; If the alarm function is Alarm, then: A(t)=Alarm(H(t)).

Images