Textile processing risk perception intelligent identification analysis system

Abstract

The invention relates to the technical field of textile manufacturing safety monitoring, in particular to a textile processing risk perception intelligent identification analysis system which comprises a video analysis module, an equipment prediction module, a product tracing module, a risk fusion module and an early warning response module. Wherein the video analysis module is used for outputting personnel violation signals; the equipment prediction module is used for carrying out fault probability calculation by fusing violation signals and time sequence data and outputting an equipment fault probability; the product tracing module is used for outputting defect source information; the risk fusion module is used for constructing a comprehensive risk index; and the early warning response module is used for triggering a multi-stage early warning instruction according to the threshold interval of the comprehensive risk index. According to the invention, through fusion analysis and graded early warning of multi-source risk factors, intelligent collaborative management and control of personnel, equipment and product defects in the textile production process are realized, and the accuracy of risk identification and the timeliness of safety response are effectively improved.

Claims

1. A textile processing risk perception intelligent identification and analysis system, characterized by: It includes video analysis module, equipment prediction module, product traceability module, risk fusion module and early warning response module; among them: Video analysis module: used to collect real-time video streams from the production workshop, generate compensation parameters based on the equipment vibration spectrum, identify illegal operations by personnel through pixel displacement compensation algorithm, and output personnel violation signals; Equipment prediction module: This module receives personnel violation signals and time series data on equipment temperature, current, and speed. It calculates the failure probability by integrating the violation signals and time series data, and outputs the equipment failure probability. Product traceability module: This module is used to obtain personnel violation signals and equipment failure probabilities, combine process path data recorded by RFID tags and semi-finished product surface image features, and locate the defect source process by correlating defect features with process parameters, and output defect source information; Risk fusion module: used to dynamically weight personnel violation signals, equipment failure probability, and defect source information to construct a comprehensive risk index; Early warning response module: used to trigger multi-level early warning instructions according to the threshold range of the comprehensive risk index, and generate emergency response operations to the execution terminal.

2. The textile processing risk perception intelligent identification and analysis system according to claim 1 is characterized in that: The video analysis module includes a video acquisition unit, a vibration spectrum extraction unit, a pixel compensation calculation unit and a violation identification unit; wherein: The video acquisition unit is used to obtain high-definition video streams in the workshop in real time at preset time intervals through industrial camera equipment deployed in the production workshop; Vibration spectrum extraction unit: used to obtain vibration acceleration signals from the device vibration sensor corresponding to the video acquisition position, convert the time domain signal into frequency domain data through fast Fourier transform, extract the main frequency component and construct the spectrum amplitude sequence of the corresponding time period; Pixel compensation calculation unit: used to establish the corresponding pixel displacement mapping function based on the spectrum amplitude sequence, and perform frame-by-frame position correction on the jitter area in the original video frame caused by device vibration; Violation recognition unit: Based on the corrected video frame sequence, it identifies the relative position relationship between the trajectory of the person's body movements and the boundary of the operating area. When there are behavioral characteristics such as crossing the dangerous boundary or not wearing protective equipment, it outputs a person violation signal.

3. The textile processing risk perception intelligent identification and analysis system according to claim 2 is characterized in that: The violation identification unit includes: Human skeleton modeling subunit: used to perform posture estimation algorithm on each frame of pixel-compensated video image, extract the two-dimensional spatial coordinates of several key points of the human body, build a skeleton structure model, and form the human motion trajectory data; Region Boundary Mapping Subunit: This subunit is used to load the operation region boundary template and align it with the video image coordinate system, so that each frame of the image has an accurate boundary reference for subsequent behavior judgment. Trajectory matching judgment subunit: used to compare the relative relationship between the person's skeleton trajectory and the boundary area, and determine whether there is a violation based on specific conditions; Violation judgment condition 1: In any frame, if the hand key point falls outside the region boundary, it is judged as an out-of-bounds operation; Violation judgment condition 2: In consecutive frames, the height of the knee key point is continuously lower than the set warning height, which is judged as non-standing operation; Violation signal output subunit: used to generate a personnel violation signal when any judgment condition is met, and output it to the equipment prediction module and product traceability module.

4. The textile processing risk perception intelligent identification and analysis system according to claim 1 is characterized in that: The equipment prediction module includes a state data preprocessing unit, a violation coupling modeling unit, a fault probability calculation unit and a probability output unit; wherein: State data preprocessing unit: used to receive equipment operation sequence data from the production site, including continuous sampling sequences of temperature, current and speed, and perform unified time base alignment, missing value filling and outlier removal operations; Violation coupling modeling unit: used to fuse personnel violation signals with time series data at the feature level and construct a fusion input vector set; Fault probability calculation unit: Based on the Bayesian conditional probability calculation formula, estimate the probability value P of the device failure at the current moment f (t); Probability output unit: used to output the failure probability value P f (t) is encapsulated as a structured data packet with the sampling time.

5. The textile processing risk perception intelligent identification and analysis system according to claim 1 is characterized in that: The product tracing module includes a path information extraction unit, an image feature analysis unit, a defect feature attribution unit, and a defect source output unit; wherein: Path information extraction unit: used to parse the historical process path information recorded in the RFID electronic tag bound to the semi-finished product, and perform index matching with the process database based on the unique identifier embedded in the tag, to obtain the equipment number, processing time period and process parameter records corresponding to each process that the semi-finished product passed through in the processing flow, forming a process path data set arranged in chronological order; Image feature analysis unit: used to extract the defect area features in the semi-finished product surface image, and generate the image feature vector of the current defect through edge segmentation and texture encoding; Defect feature attribution unit: used to match defect image features with the process parameters of each process included in the path, identify the process node most relevant to the defect, and determine the processing link where the defect occurred; Defect source output unit: used to generate defect source information including defect type, traceability process number and equipment identification based on the attribution results.

6. The textile processing risk perception intelligent identification and analysis system according to claim 5 is characterized in that: The image feature analysis unit includes: Defect area detection subunit: used to perform brightness normalization and high-pass filtering operations on the semi-finished product surface image, detect local abnormal areas in combination with background modeling, and extract the initial defect mask by threshold segmentation; Edge feature extraction subunit: used to perform Canny edge detection in the defect mask area, extract defect contour information, and calculate edge density E d ; Texture coding generation subunit: used to construct the gray level co-occurrence matrix in the defect area and calculate the contrast C and energy E when the direction is 0° as the texture description component; Feature vector output subunit: used to convert edge density E d , texture contrast C, texture energy E indicators are spliced ​​into a one-dimensional image feature vector F = [E d , C, E].

7. The textile processing risk perception intelligent identification and analysis system according to claim 6 is characterized in that: The defect feature attribution unit includes: Process parameter standardization subunit: This unit is used to extract the process parameter sets of each process recorded by the RFID path, including the corresponding tension value, temperature setting, and speed level, and perform minimum-maximum normalization processing to unify the different parameter dimensions; Feature matching scoring subunit: used to convert the current defect image feature vector F = [E d , C, E] and the standard defect feature template set corresponding to each process are used to calculate the Euclidean distance to obtain the matching score S for each process. i ; Related process determination subunit: used to determine all scoring values ​​S i Sort them and take the process with the minimum score as the suspected defect source node; Responsibility node identification subunit: used to determine whether the defect occurs in the early, middle or end links based on the processing time period of the selected process in the process path.

8. The textile processing risk perception intelligent identification and analysis system according to claim 1 is characterized in that: The risk fusion module includes a signal normalization unit and a risk index calculation unit; wherein: Signal normalization unit: used to receive personnel violation signals, equipment failure probability and defect source information, and convert the three types of heterogeneous indicators into normalized values ​​between 0 and 1, respectively denoted as V n 、P n 、D n ; Risk index calculation unit: used to calculate the comprehensive risk index at the current moment. The formula is: R s =w v ·V n +w p ·P n +w d ·D n , where R s is the comprehensive risk index; w v ,w p ,w d are the fusion weights of the three factors of personnel violation, equipment failure and defect liability respectively; and satisfy w v +w p +w d =1.

9. The textile processing risk perception intelligent identification and analysis system according to claim 8 is characterized in that: The early warning response module includes a threshold interval determination unit; wherein: Threshold interval determination unit: configured to receive the comprehensive risk index R output from the risk fusion module s , and compare it with a preset multi-level risk threshold interval to determine the current risk level; the threshold interval is divided into three levels, namely: low risk 0 ≤ R s < T1; medium risk T1 ≤ R s < T2; high risk R s ≥ T2; where T1 and T2 are fixed risk level boundary values, satisfying 0 < T1 < T2 < 1.

10. The textile processing risk perception intelligent identification and analysis system according to claim 9, characterized in that: The early warning response module also includes an early warning instruction generation unit and a response task scheduling unit; wherein: Warning instruction generation unit: used to generate warning instructions of corresponding levels according to the threshold interval judgment results; Response task scheduling unit: used to assign different response strategies according to the warning level, specifically: When the risk level is low, only log and monitor; When the risk level is medium, a local sound and light alarm or interface prompt will be triggered; When the risk level is high, a shutdown command or a forced manual review signal is immediately sent to the workshop execution terminal.

Images