An industrial equipment perception information processing system

Abstract

The present invention proposes an industrial equipment perception information processing system, which belongs to the field of equipment perception technology and includes: a temperature sensing module, a humidity sensing module, a sound sensing module, an equipment perception module, and an information processing module. The temperature sensing module is used to collect temperature information such as the external temperature and internal temperature of the equipment, the humidity sensing module is used to collect humidity information such as the external humidity and internal humidity of the equipment, the sound sensing module is used to collect sound data of the equipment working status, the equipment perception module is used to obtain equipment data, and the equipment data includes equipment internal status data. The information processing module is used to process the temperature information, humidity information, sound data and equipment data to obtain processed data and store it, so as to realize the processing of the daily working status perception information of the industrial equipment, and can predict the failure of the equipment, locate the fault location and diagnose the fault type, and assist personnel to eliminate the fault in time.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment perception, and in particular relates to an industrial equipment perception information processing system. Background Art

[0002] At present, there are problems such as an imperfect intelligent technology system, core technologies that need to be broken through, and low levels of industrial application of intelligent technologies. The integration of new-generation artificial intelligence and manufacturing technologies can quickly improve the level of intelligent manufacturing and accelerate my country's progress from a manufacturing power to a manufacturing power. At present, technologies such as big data, chips, and GPUs are developing rapidly, and the conditions and environment for the development of artificial intelligence (AI) have undergone tremendous changes. Artificial intelligence is transforming from academic traction to demand traction, and is developing towards an intelligent system that integrates cloud, edge, and end.

[0003] Acoustic testing is a non-destructive testing method that collects and analyzes sound signals containing information such as the state, process, size, and defects of the object to be tested to achieve fault identification; therefore, acoustic testing has broad development prospects in industrial fault diagnosis. Some faults of refrigeration products, five-axis CNC tool grinder equipment parts, automatic assembly lines, etc. are mainly detected and troubleshooted by human hearing. However, the accuracy, efficiency, consistency and reliability of human hearing detection are relatively low. How to rely on audio-visual intelligent perception technology to solve the above problems is currently a pressing issue. Summary of the invention

[0004] The present invention aims at the problems under the prior art. The technical solution adopted in this application is: an industrial equipment perception information processing system, including: a temperature sensing module, a humidity sensing module, a sound sensing module, a device perception module, and an information processing module. The temperature sensing module is used to collect temperature information such as the external temperature and internal temperature of the equipment. The humidity sensing module is used to collect humidity information such as the external humidity and internal humidity of the equipment. The sound sensing module is used to collect sound data of the equipment working status. The equipment perception module is used to obtain equipment data, and the equipment data includes equipment internal status data. The information processing module is used to process the temperature information, humidity information, sound data and equipment data to obtain processed data and store it, so as to realize the processing of the daily working status perception information of the industrial equipment.

[0005] Furthermore, the temperature sensing module includes a temperature sensor, which is installed in the internal detection area of ​​the device. The temperature sensor also needs to be installed outside the device to detect the external ambient temperature; the humidity sensing module includes a humidity sensor, which is also installed in the internal detection area of ​​the device and outside the device.

[0006] Furthermore, the sound sensing module adopts a distributed optical fiber sound sensor, uses the phase of coherent Rayleigh scattered light rather than the light intensity to detect sound or vibration signals within the audio range, uses a narrow linewidth laser seed source to generate a single-frequency pulse laser source, and the input end of the acousto-optic frequency shifting loop is connected to the output end of the narrow linewidth laser seed source to convert the single-frequency pulse laser source into a continuous multi-frequency pulse laser source, which is sent to the sensing optical fiber, receives the echo signal transmitted back by the sensing optical fiber, and sends the echo signal to the receiving component; the receiving component receives the echo signal and converts the echo signal into an electrical signal, and performs data analysis on the electrical signal to obtain sound data.

[0007] Furthermore, the equipment perception module obtains the operating power, load status and internal operating parameters of the industrial equipment during the operating period from the equipment control center through the 4G or 5G network.

[0008] Furthermore, the information processing module inputs the temperature information, humidity information, sound data and equipment data into the trained convolutional neural network model, extracts the target operation feature data, uses the DBSCAN clustering algorithm to classify the target operation feature data of the equipment, and inputs the classified target operation feature data and the equipment's historical fault category data into the Bayesian risk decision model for calculation, and outputs different equipment fault categories and conditional probabilities of various types of faults corresponding to the real-time operating parameters of the equipment.

[0009] Compared with the prior art, the advantages and positive effects of the present invention are: The industrial equipment perception information processing system adopted by the present invention includes: a temperature sensor module, a humidity sensor module, a sound sensor module, an equipment perception module, and an information processing module, which realize the processing of the daily working status perception information of the industrial equipment, and can predict the failure of the equipment, and locate the fault location and diagnose the fault type, and assist personnel to eliminate the fault in time, and provide auxiliary role for the daily maintenance and long-term effective operation of the industrial equipment. It can effectively reduce the current daily maintenance costs of industrial equipment, reduce equipment failure rate and maintenance time, thereby effectively improving industrial output, and therefore has significant significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0011] Figure 1 It is a structural schematic diagram of the industrial equipment perception information processing system of the present invention; Figure 2 is a structural schematic diagram of a sound sensor module of the present invention; DETAILED DESCRIPTION In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0012] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0013] Embodiment 1, as Figure 1 As shown, the present application provides an industrial equipment perception information processing system, including: an industrial equipment perception information processing system, which uses a distributed optical fiber sound sensor to collect sound data of industrial equipment, wherein compared with the traditional method of directly obtaining sound data through single-frequency pulse laser source processing, it will be interfered by strong environmental factors, resulting in reduced accuracy of sound data collection. The present application uses the fast-axis light and slow-axis light transmitted by the improved narrow-band polarization-maintaining fiber grating to modulate to obtain two light waves with a wavelength difference of five picometers, and adds a sound modulation processing process in light processing. By setting a first polarization controller, a first tunable differential group delay device, a second polarization controller and a first analyzer, when QPSK fast-axis light and QPSK slow-axis light are input, two 16-QAM signal lights with a wavelength difference of five picometers are output, and the two light waves are mixed through a concentrator to generate a single-frequency pulse. The single-frequency pulse is used as a single-frequency pulse laser source to improve the accuracy of the light source, thereby obtaining high-precision industrial equipment sound data and improving the information perception capability of the industrial equipment.

[0014] The industrial equipment perception information processing system inputs temperature information, humidity information, sound data and equipment data into the trained convolutional neural network model. The traditional model output structure is directly used for risk assessment, but this method will result in low accuracy of the assessment data and cannot meet the daily maintenance needs of industrial equipment. After the convolutional neural network model extracts the target operation feature data, this application uses the DBSCAN clustering algorithm to classify the target operation feature data of the equipment, effectively reducing the low classification accuracy problem in the subsequent processing process, and can optimize the clustering quality, enhance scalability, and obtain better noise processing data. The risk assessment data obtained by classification calculation is more accurate and meets the perception needs of industrial equipment.

[0015] The traditional DBSCAN clustering algorithm is to give a radius range R, and then for all samples, calculate the number of samples in the neighborhood of r as the local density, and calculate the minimum distance of each point with higher sample density as the screening criterion. The fault categories divided in this way are prone to misjudgment. Since there may be combinations of fault categories, such as motor faults and circuit faults occurring at the same time, if this method is used, they will only be classified into one category, which will affect the accurate perception needs of industrial equipment. For this, technical improvements have been made. The threshold of the number of neighbors is defined by a parameter MinPts. If the number of points in the neighborhood of a point is greater than or equal to MinPts, then the point is marked as a core point; the method of connecting a point with multiple core points density can effectively and accurately divide various fault types. Adding feature fusion + clustering can achieve that if multiple faults occur simultaneously in the device, the classification results formed by feature fusion + clustering include multiple categories, thereby obtaining accurate device information. Adding preprocessing of sound data can improve the accuracy and efficiency of sound data classification, which is conducive to obtaining accurate device perception results.

[0016] Specifically, it includes: temperature sensing module, humidity sensing module, sound sensing module, equipment perception module, and information processing module. The temperature sensing module is used to collect temperature information such as external temperature and internal temperature of the equipment. The humidity sensing module is used to collect humidity information such as external humidity and internal humidity of the equipment. The sound sensing module is used to collect sound data of the equipment working status. The equipment perception module is used to obtain equipment data, and the equipment data includes equipment internal status data. The information processing module is used to process the temperature information, humidity information, sound data and equipment data to obtain processed data and store it, so as to realize the processing of daily working status perception information of industrial equipment.

[0017] The temperature sensing module includes a temperature sensor, which is installed in the internal detection area of ​​the device. The temperature sensor also needs to be installed outside the device to detect the external ambient temperature; the humidity sensing module includes a humidity sensor, which is also installed in the internal detection area of ​​the device and outside the device.

[0018] The sound sensing module uses a distributed optical fiber sound sensor, the structure diagram is as follows Figure 2As shown, the phase of coherent Rayleigh scattered light is used instead of the light intensity to detect the sound or vibration signal within the audio range, and a narrow linewidth laser seed source is used to generate a single-frequency pulse laser source, wherein the narrow linewidth laser seed source modulates the fast-axis light and the slow-axis light transmitted by the narrowband polarization-maintaining fiber grating through the stress birefringence effect of the acousto-optic modulator to obtain two light waves with a wavelength difference of the order of five picometers, and the two light waves are mixed through a concentrator to generate a single-frequency pulse, and the single-frequency pulse is used as a single-frequency pulse laser source to improve the light source accuracy, and the input end of the acousto-optic frequency shifting ring is connected to the output end of the narrow linewidth laser seed source to convert the single-frequency pulse laser source into a continuous multi-frequency pulse laser source, and the multi-frequency pulse laser source is sent to the sensing optical fiber, and the echo signal transmitted by the sensing optical fiber is received, and the echo signal is sent to the receiving component; the receiving component receives the echo signal and converts the echo signal into an electrical signal, and performs data analysis on the electrical signal to obtain sound data.

[0019] The equipment perception module obtains the operating power, load status and internal operating parameters of the industrial equipment during the operating period from the equipment control center through the 4G or 5G network.

[0020] The information processing module inputs temperature information, humidity information, sound data and equipment data into the trained convolutional neural network model, extracts target operation feature data, uses the DBSCAN clustering algorithm to classify the target operation feature data of the equipment, and inputs the classified target operation feature data and equipment historical fault category data into the Bayesian risk decision model for calculation, and outputs different equipment fault categories and conditional probabilities of various types of faults corresponding to the real-time operating parameters of the equipment.

[0021] The convolutional neural network model inserts the SPPFS module into the traditional convolutional neural network model, converts the output vector of the backbone network into a high-dimensional vector, connects the high-dimensional vector to the transformer encoder, and connects the output of the transformer encoder to the multi-layer perceptron. The target operation feature data is extracted through the multi-layer perceptron, and the high-dimensional feature extraction function of the SPPFS module is fully utilized to extract the sound state information from the sound data. The SPPFS module consists of Conv and MaxPooling modules.

[0022] The convolutional neural network model is trained by acquiring historical data, and the historical data of temperature information, humidity information, sound data and equipment data are used to construct training sets and validation sets. The training set is first input into the convolutional neural network for training, and then the validation set is input into the convolutional neural network, and finally the convolutional neural network model is saved; finally, the newly collected operation data is input into the trained convolutional neural network model, and the convolutional neural network model can extract the target operation feature data from the data. For example, after pre-training the temperature information, humidity information, sound data and equipment data corresponding to the fault, when the same data parameters as the motor fault appear, the convolutional neural network model will extract the corresponding target operation feature data, and classify the fault through the DBSCAN clustering algorithm, matching the corresponding fault category, and further using the Bayesian risk decision model for calculation to obtain the final conditional probability of the fault.

[0023] Among them, the DBSCAN clustering algorithm calculates the E neighborhood of all target running feature data, and for each point P in the target running feature data, calculates how many neighbors it has in its E neighborhood. The threshold of the number of neighbors is defined by a parameter MinPts. If the number of points in the neighborhood of a point is greater than or equal to MinPts, then the point is marked as a core point; for each core point, find all points that are densely connected to it. If point P is in the neighborhood of point O, and O is a core point, then P is a point that is densely connected to O; mark noise points and boundary points: points that are not marked as core points are marked as noise points. Points that are densely connected to a core point but are not core points are marked as boundary points; assign an independent cluster label to each core point or point that is densely connected to it: assign an independent cluster label to each core point or point that is densely connected to it. If a point is densely connected to multiple core points, it will be assigned the cluster label of the first core point found; all noise points form an independent cluster to complete the classification, and the classified target operation feature data is divided into motor failure, line short circuit, insulation degradation, loose parts, flow pressure imbalance, etc.

[0024] The Bayesian risk decision model directly calculates the probability of occurrence of a fault event T under the combination of n fault factors Xi (i = 1, 2, ..., n) by inputting equipment data and joint probability distribution: P(T = t), P(T = t) = ΣP(T = t|X1 = x1, X2 = x2, ..., Xn = xn) × P(X1 = x1, X2 = x2, ..., Xn = xn), t = {t1, t2, ...., tp}, xi = {xi1, xi2, ..., xi Qi},i=1,2,...,nwhere t represents the state of the leaf node, and there are a total of {t1,t2,....,tp}p states; xi represents the state of the root node, and there are a total of {xi1,xi2,...,xiQi}Qi states; ΣP(T=t|X1=x1,X2=x2,...,Xn=xn) represents the conditional probability table of BN forward conduction; P(X1=x1,X2=x2,...,Xn=xn) represents the joint probability of each risk factor.

[0025]

[0026] For industrial equipment such as large-scale processing machinery in production workshops, transport vehicles and other types of production equipment, experienced technicians can always determine the fault location and type of fault by the sound of the equipment running when the equipment fails, such as bolt falling off, circuit aging and short circuit, etc. Therefore, it is necessary to carry out low-cost monitoring in the daily operation of current industrial equipment. The sound monitoring process is not only low-cost, but also can effectively and quickly determine the fault location and type during the fault, which can greatly save maintenance time and has positive significance for industrial equipment maintenance. The equipment of this embodiment can not only process perception information in real time, but also when maintenance personnel need to perform regular maintenance and repair on the equipment, they can use remote communication equipment such as mobile phones to send the perception request to the base station. The base station transmits the perception request information to the information processing module of the industrial equipment perception information processing system, and then the information processing module processes the perception request, summarizes the perception information and transmits it to the maintenance personnel's remote communication equipment.

[0027] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An industrial equipment perception information processing system, characterized in that: include: Temperature sensing module, humidity sensing module, sound sensing module, equipment perception module, information processing module, the temperature sensing module is used to collect temperature information such as external temperature and internal temperature of the equipment, the humidity sensing module is used to collect humidity information such as external humidity and internal humidity of the equipment, the sound sensing module is used to collect sound data of the equipment working status, the equipment perception module is used to obtain equipment data, the equipment data includes equipment internal status data, the information processing module is used to process the temperature information, humidity information, sound data and equipment data to obtain processed data and store it, so as to realize the processing of daily working status perception information of industrial equipment.

2. The industrial equipment perception information processing system according to claim 1, characterized in that: The temperature sensing module includes a temperature sensor, which is installed in the internal detection area of ​​the device. The temperature sensor also needs to be installed outside the device to detect the external ambient temperature; the humidity sensing module includes a humidity sensor, which is also installed in the internal detection area of ​​the device and outside the device.

3. The industrial equipment perception information processing system according to claim 1, characterized in that: The sound sensing module adopts a distributed optical fiber sound sensor, uses the phase of coherent Rayleigh scattered light rather than the light intensity to detect sound or vibration signals within the audio range, uses a narrow linewidth laser seed source to generate a single-frequency pulse laser source, and the input end of the acousto-optic frequency shifting loop is connected to the output end of the narrow linewidth laser seed source to convert the single-frequency pulse laser source into a continuous multi-frequency pulse laser source, which is sent to the sensing optical fiber, receives the echo signal transmitted back by the sensing optical fiber, and sends the echo signal to a receiving component; the receiving component receives the echo signal and converts the echo signal into an electrical signal, and performs data analysis on the electrical signal to obtain sound data.

4. The industrial equipment perception information processing system according to claim 1, characterized in that: The equipment perception module obtains the operating power, load status and internal operating parameters of the industrial equipment during the operating period from the equipment control center through the 4G or 5G network.

5. The industrial equipment perception information processing system according to claim 1, characterized in that: The information processing module inputs temperature information, humidity information, sound data and equipment data into the trained convolutional neural network model, extracts target operation feature data, uses the DBSCAN clustering algorithm to classify the target operation feature data of the equipment, and inputs the classified target operation feature data and equipment historical fault category data into the Bayesian risk decision model for calculation, and outputs different equipment fault categories and conditional probabilities of various types of faults corresponding to the real-time operating parameters of the equipment.

Images