Search

CN-122020481-A - Rapier loom fault diagnosis system

CN122020481ACN 122020481 ACN122020481 ACN 122020481ACN-122020481-A

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis system of a rapier loom; the system fuses fault tree analysis and a probabilistic neural network based on whale algorithm optimization, and the fault diagnosis steps comprise a first step of establishing a rapier loom fault tree diagnosis model, a second step of optimizing fault tree model parameters by using the whale algorithm, a third step of constructing a sample data set and normalizing, and a fourth step of probabilistic neural network training and fault diagnosis. Based on a diagnosis model of the fault tree of the rapier loom, the diagnosis model based on fusion optimization of the probability neural network and the fault tree algorithm introduces real-time newly-added state monitoring data of the monitoring system, converts the data set into sample feature vectors, performs normalization processing, introduces the sample feature vectors into the probability neural network algorithm model for training, calculates and outputs fault symptom probability, and realizes real-time, complete and rapid fault identification and positioning of the fault diagnosis system of the rapier loom.

Inventors

  • XIAO YANJUN
  • HAN ZIJIAN

Assignees

  • 河北工业大学

Dates

Publication Date
20260512
Application Date
20260204

Claims (9)

  1. 1. The rapier loom fault diagnosis system is characterized by integrating fault tree analysis and a probabilistic neural network optimized based on whale algorithm, and comprises the following steps: step one, establishing a fault tree diagnosis model of the rapier loom: Based on a failure mechanism and a process flow of the rapier loom, splitting a failure tree model into a failure tree model of a starting inspection subsystem, a weaving subsystem and a warp-feeding reeling subsystem, and carrying out qualitative and quantitative diagnosis through Boolean operation and bottom event probability importance analysis; Optimizing fault tree model parameters by using whale algorithm: the rule weight, the precondition attribute weight and the output confidence coefficient of the fault symptom are optimized through a whale algorithm, so that the accuracy and the decision capability of fault diagnosis are improved; Step three, constructing a sample data set and carrying out normalization treatment: integrating the optimized fault tree model data and real-time monitoring data, generating a sample data set containing 11 fault symptom characteristics, and carrying out normalization processing; training the probability neural network and diagnosing faults: And inputting the normalized data set into a probabilistic neural network for training, outputting fault symptom probability, calculating the importance of fault reasons, and generating fault positioning and checking sequence suggestions.
  2. 2. The system for diagnosing faults of a rapier loom according to claim 1, wherein in the first step, the qualitative analysis determines the weak links of the system through the minimum cut set, and the quantitative analysis calculates the criticality of the fault cause through the probability importance of the bottom event.
  3. 3. The system for diagnosing faults of a rapier loom according to claim 1, wherein the whale algorithm in the second step optimizes weight parameters of all subsystems in a fault tree model by simulating random search and spiral line mechanisms of whale predation behaviors of a whale; The whale algorithm adopts the following flow: Initializing whale population; calculating the fitness of each whale; updating the optimal whale individuals; updating whale positions; If the P is less than 0.5, judging whether A is more than or equal to 1, if so, searching for a prey and capturing the prey, and recording an optimal target value; If P <0.5 is not satisfied, surrounding the prey, and recording an optimal target value; Judging whether the optimal target value meets the termination condition, if so, outputting the optimal target value and ending; If the target value is not met, updating the whale position, and sequentially carrying out the steps until the optimal target value is output after the termination condition is met.
  4. 4. The rapier loom fault diagnosis system according to claim 1, wherein in the step three, the normalization process adopts a linear transformation method to map the sample characteristic value to the (0, 1) interval, so as to eliminate the influence of the dimension difference on the classification result.
  5. 5. The rapier loom fault diagnosis system of claim 1, wherein the structure of the probabilistic neural network in the fourth step comprises an input layer of 11 neurons, a mode layer of 1430 neurons, a summation layer of 11 neurons and an output layer of 11 neurons, and the output layer judges the fault occurrence state by a probability threshold.
  6. 6. The rapier loom fault diagnosis system of claim 5, wherein the number of neurons in the pattern layer is consistent with the number of training samples, each class layer corresponds to one fault symptom type, and 100% classification recognition accuracy is achieved through the training set.
  7. 7. The rapier loom fault diagnosis system of claim 1, wherein the system recognizes abnormal states of the rapier loom in real time, and the abnormal states include abnormal oil level of an oil tank, overrun of spindle rotation speed, deviation of rotation speed of a servo motor and excessive temperature.
  8. 8. The rapier loom fault diagnosis system according to claim 1, wherein the importance of the fault causes is ordered by probability values output by a probability neural network, fault troubleshooting priority suggestions are provided for maintenance personnel, abnormal data samples exceeding a normal numerical range are automatically removed in the diagnosis process, and fault causes with high probability importance are preferentially processed.
  9. 9. The rapier loom fault diagnosis system of claim 1, wherein the sample data set comprises 160 sets of data, wherein the training set is 1430 samples, the test set is 330 samples, and the fault sign classification recognition accuracy of the test set is 99.39%.

Description

Rapier loom fault diagnosis system Technical Field The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis system of a rapier loom. Background Along with the continuous improvement of the intelligent level of rapier loom equipment, the equipment control system is increasingly complex. In order to meet the requirements of users developing in the new era of rapier looms, the safety and reliability of the rapier loom equipment in the running process are ensured to be the current research hot spot. Because the mechanical structure of loom equipment and weaving process are comparatively complicated, the probability of failure occurrence is higher, and once failure causes the shut down, light then causes economic loss, and serious then endangers the life safety of production staff. Therefore, effective fault diagnosis is an important measure for ensuring the normal operation of the rapier loom device, and the loom fault diagnosis system has a vital significance for the reliable operation of the loom. The existing research focuses on the diagnosis of local faults of the rapier loom, and lacks a comprehensive diagnosis system for complete faults of the whole equipment, and the method for diagnosing the local faults leads to insufficient grasp of the whole health state of the equipment, so that the high-efficiency whole fault early warning and maintenance are difficult to realize. Due to the lack of a global view, a plurality of possible fault points often need to be checked one by one in the diagnosis process, and the progressive diagnosis mode is long in time consumption, so that the diagnosis efficiency is affected. In the face of a highly-automated control system, the existing fault diagnosis system has insufficient functions in terms of intelligence, and cannot meet the rapid and accurate diagnosis requirements under a complex control environment. Therefore, the application provides a fault diagnosis system of the rapier loom. Disclosure of Invention In order to make up the defects of the prior art and solve the technical problems in the background art, the invention provides a rapier loom fault diagnosis system. The invention is realized by the following technical scheme: A rapier loom fault diagnosis system fuses fault tree analysis and a probabilistic neural network optimized based on whale algorithm, and comprises the following steps: step one, establishing a fault tree diagnosis model of the rapier loom: based on the failure mechanism and the technological process of the rapier loom, the failure tree model is split into failure tree models of a starting inspection subsystem, a weaving subsystem and a warp-feeding reeling subsystem, and qualitative and quantitative diagnosis is carried out through Boolean operation and bottom event probability importance analysis. Optimizing fault tree model parameters by using whale algorithm: the rule weight, the precondition attribute weight and the output confidence coefficient of the fault symptom are optimized through a whale algorithm, so that the accuracy and the decision capability of fault diagnosis are improved; Step three, constructing a sample data set and carrying out normalization treatment: integrating the optimized fault tree model data and real-time monitoring data, generating a sample data set containing 11 fault symptom characteristics, and carrying out normalization processing; training the probability neural network and diagnosing faults: And inputting the normalized data set into a probabilistic neural network for training, outputting fault symptom probability, calculating the importance of fault reasons, and generating fault positioning and checking sequence suggestions. Preferably, in the first step, the weak links of the system are determined through the minimum cut set by qualitative analysis, and the criticality of the fault cause is calculated through the probability importance of the bottom event by quantitative analysis. Preferably, in the second step, the whale algorithm optimizes the weight parameters of all subsystems in the fault tree model by simulating the random search and spiral line mechanism of the whale predation behavior of the whale; The whale algorithm adopts the following flow: Initializing whale population; calculating the fitness of each whale; updating the optimal whale individuals; updating whale positions; If the P is less than 0.5, judging whether A is more than or equal to 1, if so, searching for a prey and capturing the prey, and recording an optimal target value; If P <0.5 is not satisfied, surrounding the prey, and recording an optimal target value; Judging whether the optimal target value meets the termination condition, if so, outputting the optimal target value and ending; If the target value is not met, updating the whale position, and sequentially carrying out the steps until the optimal target value is output after the termination condition is met. Preferably, in the th