CN104386449A - Intelligent protection device for online detection of head/tail wheels of mining belt conveyor - Google Patents

Abstract

The invention relates to an intelligent protection device for on-line detection of head and tail wheels of a belt conveyor for mines, which includes a data acquisition module, an intelligent protection module, a data transmission module connected to the data acquisition module and the intelligent protection module, and an abnormal alarm system. Because: the data acquisition module is a multi-dimensional data acquisition module, the abnormal alarm system establishes a communication connection with the output end of the data transmission module through the Ethernet network or Wifi signal, the abnormal alarm system is managed by the remote service module, and the multi-dimensional data acquisition module is monitored by an abnormal discovery algorithm. The collected normal multi-dimensional data is collected and modeled, the future data is classified, and the abnormal data is sent out an alarm signal, which is composed of a big data-based diagnostic analysis module. It realizes the remote monitoring, fault alarm and remote control of the mining belt conveyor, all-round and all-weather diagnostic services, and timely detection of faults and timely shutdown through remote control to minimize losses.

Description

Intelligent protection device for on-line detection of head and tail wheels of mine belt conveyor

technical field

The invention relates to the monitoring technology of a belt conveyor for mines, in particular to an intelligent protection device for on-line detection of head and tail wheels of a belt conveyor for mines. the

Background technique

The mining belt conveyor has strong continuous transportation capacity, high operating efficiency, and easy automatic control, and has been widely used in the transportation of various bulk materials. The conveyor belt is an important part of the belt conveyor. The conveyor belt mainly includes ordinary canvas core belts, synthetic fiber core belts, and steel cord core belts. With the development of the direction, the steel cord belt is more and more widely used. The tensile strength of the steel cord conveyor belt has been greatly improved, but its longitudinal tear resistance has not been improved. It is only the strength of the rubber itself, so it is easy to cause longitudinal tearing. Mine belt conveyor is the main artery of production and transportation in factories and mines. Once an accident occurs, it will bring huge direct and indirect losses, especially for high-speed, long-distance, and large-inclination steel cord conveyor belts, which will cause even greater losses. According to statistics, the possibility of a longitudinal tear in a belt during its life cycle is about 20%. Conveyor belts worth millions of dollars or more may be completely destroyed in a short period of time once a longitudinal tear accident occurs. cause huge economic losses. Even if it can be repaired, it requires considerable manpower and time, which has a great impact on normal production. In recent years, the use of belt conveyors in China has been increasing, and its application range has become wider and wider. Accidents of belt longitudinal tearing have become more and more frequent. Most of the main reasons are the head and tail of mining belt conveyors. caused by wheel failure. the

At present, in order to monitor the running status of the belt conveyor, on-site PLC control is mostly used, which is displayed on the screen of the on-site PLC, and CAN bus technology is used to convert RS232 and RS485 signals to realize data exchange with other systems, but it has the following problems : First, the type of data collection is relatively simple. Some equipment mainly collects the vibration or noise index of the parts of the mining belt conveyor, and the other part of the equipment focuses on collecting the ambient temperature of the equipment. Data collection based on this method cannot be effective. Comprehensively display the multi-angle situation of the mining belt conveyor when it is working, so that the diagnosis system based on single-dimensional data is not accurate enough; second, there is no efficient fault diagnosis system for the mining belt conveyor based on multi-dimensional data. It is to diagnose the fault of the mining belt conveyor through the judgment of the threshold value and take corresponding control. This judgment method, in the case of large noise data, has a high degree of false sample, which usually leads to misjudgment of the fault. the

the

Contents of the invention

The purpose of the present invention is to provide a head and tail wheel for mining belt conveyor that solves the problems of the existing mining belt conveyor remote monitoring and fault diagnosis system, such as high cost, need for a dedicated network, low diagnostic accuracy, poor real-time performance, and lack of fault solutions. Online detection of intelligent protection devices. the

The technical scheme of the present invention is:

An intelligent protection device for on-line detection of head and tail wheels of a belt conveyor for mines, comprising a data acquisition module, an intelligent protection module, a data transmission module connected to the data acquisition module and the intelligent protection module, and an abnormal alarm system, characterized in that: The data acquisition module is a multi-dimensional data acquisition module, and the abnormal alarm system establishes a communication connection with the output end of the data transmission module through an Ethernet network or a Wifi signal,

The multi-dimensional data acquisition module is composed of shock sensors, temperature sensors and Hall sensors for measuring the three-dimensional vibration frequency of the head and tail wheels of the mine belt conveyor, the surrounding temperature and the speed of the head and tail wheels, and is electrically connected to the signal output terminals of each sensor. The built-in processor is composed of the built-in processor, and the output terminal of the built-in processor is electrically connected with the data transmission module,

Described data transmission module is made up of cloud processor, Ethernet interface, built-in storage module and Wifi communication module, is provided with the microsd card that realizes asynchronous transmission function in the described built-in storage module,

The abnormality alarm system is managed by a remote service for remotely managing the multidimensional data acquisition module and the data transmission module, remotely setting the sampling frequency of the multidimensional data acquisition module, sampling data type, on/off status, IP address, transmission network type and access password module, and a diagnostic analysis module based on big data that adopts an abnormality discovery algorithm to model the normal multi-dimensional data collected by the multi-dimensional data acquisition module, classifies future data, and sends an alarm signal to abnormal data.

The intelligent protection device is composed of an output control circuit that receives the control signal based on the abnormal alarm system shutdown protection transmitted by the data transmission device, a power supply circuit connected to the output control circuit, a drive control circuit and a relay protection that controls the mine belt drive motor Circuit composition.

The above-mentioned intelligent protection device for online detection of the head and tail wheels of the mining belt conveyor, the abnormality discovery algorithm of the big data-based diagnostic analysis module is specifically:

1) Feature extraction is performed on the multi-dimensional original time-series signals collected by the multi-dimensional data acquisition module. The feature extraction uses the Mel-frequency cepstral coefficients algorithm ( Mel-frequency cepstral coefficients , ( MFCCs )), and a set of discrete data points The signal segment of is represented by frequency features on the time axis;

2) Multiple sets of data are obtained by repeatedly sampling the timing signals of the normal state of the mine belt conveyor, and each set of data is extracted using the Mel frequency cepstral coefficient algorithm to form a feature sample space;

3) Use the Mel frequency cepstral coefficient algorithm to extract features from the data segment to be tested, and use K-Nearest Neighbors (KNN) to classify the data to be classified after feature extraction as normal or abnormal, that is, calculate the The distance between the classified data and the characteristic data of the sample space is calculated using the relative entropy/KL distance (Kullback–Leibler divergence). When the value of the KL distance exceeds the preset value, the data to be classified will be marked as abnormal, otherwise it is marked as normal and stored in the database of the abnormal alarm system;

4) The abnormal alarm system will scan the newly collected and categorized marked data points in its database system at fixed time intervals to discover the operating status of the mining belt conveyor within each time interval, and issue an alarm for the abnormally marked data Signal.

The above-mentioned intelligent protection device for on-line detection of the head and tail wheels of the mine belt conveyor, the specific steps of the Mel-frequency cepstral coefficient algorithm ( Mel-frequency cepstral coefficients , ( MFCCs )) are as follows: Cut into small signal segments on the axis; then do a periodic estimation of the spectral density for each signal segment; then use various filters provided by MFCCs to process the spectral density of each signal segment and for each applied filter Perform energy superposition, perform logarithmic operation on the superimposed energy of each filter, and finally perform inverse discrete cosine transform on each logarithmic operation result.

The above-mentioned intelligent protection device for online detection of the head and tail wheels of the mining belt conveyor divides the data of the continuous normal work of the mining belt conveyor into 24×7 samples per hour as the sample space for extraction and training. The sampling rate 6 data per minute, 360 data points per hour, and then MFCC s feature extraction is performed on 168 samples to form a new feature sample space.

In the above-mentioned intelligent protection device for on-line detection of head and tail wheels of the mine belt conveyor, the multi-dimensional data acquisition module samples according to the set frequency, and automatically uploads the collected data to the abnormal alarm system through the data transmission module, and automatically deletes the data collected by itself. Data retained. the

The above-mentioned intelligent protection device for online detection of the head and tail wheels of the mining belt conveyor, the alarm signal of the abnormal alarm system includes alarm prompts and shutdown protection. If it is shutdown protection, the control signal is sent to the data transmission device through the network, and the data transmission The device is transmitted to the intelligent protection device, and then outputs the corresponding control signal to control the mine belt conveyor to stop running. the

The beneficial effects of the present invention are:

1. The present invention is not judged by the threshold value, but by multi-dimensional data collection, and the diagnosis and analysis module processing based on big data. When the ambient temperature of the head and tail wheels of the mining belt is found to be abnormal, it can be directly inferred that the head and tail wheels are abnormal. The vibration frequency of the linkage part of the tail wheel is also abnormal, and the ambient temperature of the linkage part rises, indicating that the problem occurs in the linkage part rather than the directly measured part, and the fault judgment between the linkage part and the passive part is realized .

2. The present invention can be transmitted through a wireless network instead of the traditional CAN bus or ModBus. Through the data transmission device, the wireless communication between the multi-dimensional data acquisition module, the remote service management module and the diagnosis module can be realized, so as to avoid installation in harsh environments Difficulties bring great convenience to installation and debugging. the

3. The abnormal alarm system of mining belt head and tail wheel based on big data brings great convenience to users and managers. At the same time, through diagnosis and analysis of faults, and timely alarm and protection of faults, the damage of the belt is avoided. Prevent the occurrence of danger and the spread of accidents. the

Description of drawings

Fig. 1 is a block diagram of the present invention;

Fig. 2 is a block diagram of multidimensional data acquisition module in Fig. 1;

Fig. 3 is a block diagram of the data transmission module in Fig. 1;

Fig. 4 is a block diagram of the intelligent protection module in Fig. 1;

Fig. 5 is a block diagram of the abnormal alarm system in Fig. 1 .

Detailed ways

As shown in Figure 1, the intelligent protection device for on-line detection of the head and tail wheels of the mining belt conveyor includes a data acquisition module, an intelligent protection module, a data transmission module connected to the data acquisition module and the intelligent protection module, and an abnormal alarm system. The data acquisition module is a multi-dimensional data acquisition module, and the abnormal alarm system establishes a communication connection with the output end of the data transmission module through an Ethernet network or a Wifi signal. the

Wherein, as shown in Figure 2, the multidimensional data acquisition module is made up of a shock sensor, a temperature sensor, a Hall sensor and a built-in processor electrically connected to each sensor signal output end, wherein the shock sensor and the temperature sensor measure the mine belt synchronously in real time The three-dimensional vibration frequency and temperature of bearing pads on both sides of the nose and tail wheels of the transporter are measured by Hall sensors, and the output end of the built-in processor is electrically connected to the data transmission module. The multi-dimensional data acquisition module samples the three-dimensional vibration frequency, bearing bush temperature, and head and tail wheel speed according to the set frequency, and automatically uploads the collected data to the abnormal alarm system through the data transmission module, and automatically deletes the data retained by itself. the

Through the use of the above sensors, multiple conditions of the working state of the belt conveyor for mining can be monitored in real time. At the same time, by using the multi-dimensional data acquired by the sensor, the device can effectively collect the operation data for specific mining belt components, including the three-dimensional vibration frequency of the bearing pads on both sides of the head and tail wheels, the temperature of the bearing bushes, the speed of the head and tail wheels, etc. Because the key operating index of the mining belt is derived from the comprehensive inference of the measured multi-dimensional data, at the data collection end, it is necessary to standardize the data of each dimension on the time axis. That is to say, the collected multi-dimensional data can form a one-to-one correspondence on the time axis (but not necessarily based on the same sampling rate). Instead of judging by thresholds. For example: if the ambient temperature of a certain part of the mining belt is found to be abnormal, it cannot be directly inferred that the part is abnormal. If the vibration frequency of the linkage part of this part is also abnormal, and the ambient temperature of the linkage part rises, it means that the problem has occurred In the linkage component rather than the directly measured component. the

As shown in Figure 3, the data transmission module is composed of a cloud processor, an Ethernet interface, a built-in storage module and a Wifi communication module, and the built-in storage module is provided with a microsd card for realizing the asynchronous transmission function. The data transmission module transmits the multi-dimensional data collected to the abnormal alarm system through the Ethernet interface or the Wifi communication module. When there is a problem in the data transmission, the collected data can be stored in the built-in storage module, and the built-in storage module can pass the microsd card. It is read out in the form of asynchronous transmission to the abnormal alarm system. the

The transmission principle of the data transmission module: When the data acquisition module initiates a data transmission request, the data transmission device will check the currently connected network situation, if it finds that the Ethernet connection exists, it will use the Ethernet connection to complete the data transmission first, if the Ethernet connection is not available Under normal circumstances, the data transmission device will try to use the Wifi connection to send data. If the Wifi connection is not available, the data transmission module will notify the corresponding multi-dimensional data acquisition module to use the local storage to save the corresponding data. the

The abnormality alarm system is managed by a remote service for remotely managing the multidimensional data acquisition module and the data transmission module, remotely setting the sampling frequency of the multidimensional data acquisition module, sampling data type, on/off status, IP address, transmission network type and access password module, and a diagnostic analysis module based on big data that adopts an abnormality discovery algorithm to model the normal multi-dimensional data collected by the multi-dimensional data acquisition module, classifies future data, and sends an alarm signal to abnormal data. The abnormal alarm system stores a large amount of multi-dimensional data of various components of different mining belt conveyors, and establishes a data template for the normal operation of the mining belt conveyor based on a large amount of data, so as to use this template to compare the future data of the mining belt. According to the difference of the comparative analysis, the current operating status of the mining belt can be obtained. If there is an abnormality in the running state of the mining belt, the relevant personnel will be called to the police. the

The analysis and diagnosis module based on big data includes an abnormal discovery algorithm, and the data collected by the data acquisition module are all multi-dimensional time-series data (see Figure 5). After collecting and modeling normal multi-dimensional data, the machine of the present invention Learning algorithms that can efficiently classify (normal, abnormal) data over a period of time. The specific algorithm steps are as follows:

1) Feature extraction is performed on the multi-dimensional original time-series signals collected by the multi-dimensional data acquisition module. The feature extraction uses the Mel-frequency cepstral coefficients algorithm ( Mel-frequency cepstral coefficients , ( MFCCs )), and a set of discrete data points The signal segments of are represented by frequency features on the time axis. The function of this step is because the time series signal is composed of discrete data points, which cannot be effectively analyzed by computer. After processing through MFCCs, a set of signal segments of the original signal (for example, composed of 2000 sampling points) can be Represented by a few frequency features (such as the frequency period of the signal, displacement, etc.).

The specific steps for applying MFCCs to the present invention are as follows: First, the original signal is cut into small signal segments on the time axis (for the present invention, experiments have proved that the feature extraction effect is better with 100 sampling points as the length of the signal segment); Then do a periodic estimation of the spectral density for each signal segment; then use a variety of filters provided by the MFCCs algorithm to process the spectral density of each signal segment and perform energy superposition on each applied filter, and each filter The logarithmic operation is performed on the energy after superposition of the detectors, and the inverse discrete cosine transform is performed on each logarithmic operation result. the

2) Multiple sets of data are obtained by repeatedly sampling the time series signals of the normal state of the mine belt conveyor, and each set of data is extracted using the Mel frequency cepstral coefficient algorithm to form a feature sample space. In this embodiment, the data (sampling rate is 6 data per minute) of the mine belt conveyor working normally for one week is divided into 168 samples (24×7) per hour (360 data points) as the extraction and The sample space for training. Then MFCC feature extraction is performed on these 168 samples to form a new feature sample space (168 samples). The function of this step is to collect a large number of data samples and provide the learned sample space for step three. the

3) Use the Mel frequency cepstral coefficient algorithm to extract features from the data segment to be tested, and use K-Nearest Neighbors (KNN) to classify the data to be classified after feature extraction as normal or abnormal, that is, calculate the The distance between the classified data and the characteristic data of the sample space is calculated using the relative entropy/KL distance (Kullback–Leibler divergence). When the value of the KL distance exceeds the preset value, the data to be classified will be marked as Otherwise, it is marked as normal and stored in the database of the abnormal alarm system. The function of this step is that steps 1) and 2) have provided the characteristics of the data in the normal running state of the belt, and then cluster analysis can be performed on the characteristics of each small signal segment to find the belt running in normal state, through the MFCC feature. The commonality of the extracted features clusters the most common features to form a clustering model, so that in the actual production in the future, the model can be used to guide and find the abnormal state (because the data characteristics in the abnormal state and the data characteristics in the normal state are in the will not be divided into a cluster when clustering). the

4) The abnormal alarm system will scan the newly collected and classified data points in the database system at a fixed time interval (in this embodiment, the unit is hourly), to find the mining belt within a certain hour The operating status of the transport plane, and an alarm signal is sent out for abnormally marked data. The function of this step is to extract the characteristics of the data points collected every hour (through MFCCs), and then compare the characteristics of these data points with the existing clustering model to find out whether the characteristics of the new data points are normal or exception. See Figure 5, which is a three-dimensional vibration diagram of the equipment. The abscissa is the time, and the ordinate is the amplitude of the vibration. It can be seen from the figure that there is a vibration alarm from 10:55 to 10:56 in the morning, where the dotted line represents The X axis, the solid line represents the Y axis, and the dotted line represents the Z axis. the

As shown in Figure 4, the intelligent protection device is composed of an output control circuit based on the control signal of the abnormal alarm system shutdown protection transmitted by the receiving data transmission device, a power circuit connected to the output control circuit, a drive control circuit and a control mine belt Composition of relay protection circuit for drive motor. Based on the alarm signal of the mining belt abnormal alarm system based on big data, the system judges whether it is an alarm prompt or a shutdown protection. If it is a shutdown protection, the control signal is sent to the data transmission device through the network, and then transmitted to the intelligent protection through the data transmission device. The device, and then output the corresponding control signal to control the mine belt conveyor.

To sum up, the multi-dimensional data acquisition module in the device collects the state information data of the belt conveyor used in mines, and through the data transmission module, the data is remotely transmitted to the abnormal alarm system through the network, and the collected data is processed to obtain the processed The results are displayed to the relevant management personnel to realize the remote monitoring and fault alarm of the mining belt. The detection system of the present invention can realize the fault diagnosis and monitoring of any Wifi signal or the mining belt conveyor in the area covered by the Ethernet network, realize the seamless connection of the wireless and wired network of the mining belt conveyor fault diagnosis system, and is equipped with The mining belt conveyor of the fault diagnosis system provides a full range of round-the-clock diagnostic services. the

Claims (6)

1. An intelligent protection device for on-line detection of head and tail wheels of a belt conveyor for mines, comprising a data acquisition module, an intelligent protection module, a data transmission module and an abnormal alarm system respectively connected to the data acquisition module and the intelligent protection module, characterized in that : the data acquisition module is a multi-dimensional data acquisition module, and the abnormal alarm system establishes a communication connection with the output end of the data transmission module through an Ethernet network or a Wifi signal, The multi-dimensional data acquisition module is composed of shock sensors, temperature sensors and Hall sensors for measuring the three-dimensional vibration frequency of the head and tail wheels of the mine belt conveyor, the surrounding temperature and the speed of the head and tail wheels, and is electrically connected to the signal output terminals of each sensor. The built-in processor is composed of the built-in processor, and the output terminal of the built-in processor is electrically connected with the data transmission module, Described data transmission module is made up of cloud processor, Ethernet interface, built-in storage module and Wifi communication module, is provided with the microsd card that realizes asynchronous transmission function in the described built-in storage module, The abnormality alarm system is managed by a remote service for remotely managing the multidimensional data acquisition module and the data transmission module, remotely setting the sampling frequency of the multidimensional data acquisition module, sampling data type, on/off status, IP address, transmission network type and access password module, and a diagnostic analysis module based on big data that adopts an abnormality discovery algorithm to model the normal multi-dimensional data collected by the multi-dimensional data acquisition module, classifies future data, and sends an alarm signal to abnormal data. The intelligent protection device is composed of an output control circuit that receives the control signal based on the abnormal alarm system shutdown protection transmitted by the data transmission device, a power supply circuit connected to the output control circuit, a drive control circuit and a relay protection that controls the mine belt drive motor Circuit composition. 2. according to claim 1, be used for mine belt conveyer head and tail wheel online detection intelligent protection device, it is characterized in that, the abnormal discovery algorithm of the diagnostic analysis module based on big data is specifically: 1) Feature extraction is performed on the multi-dimensional original time-series signals collected by the multi-dimensional data acquisition module. The feature extraction uses the Mel-frequency cepstral coefficients algorithm ( Mel-frequency cepstral coefficients , ( MFCCs )), and a set of discrete data points The signal segment of is represented by frequency features on the time axis; 2) Multiple sets of data are obtained by repeatedly sampling the timing signals of the normal state of the mine belt conveyor, and each set of data is extracted using the Mel frequency cepstral coefficient algorithm to form a feature sample space; 3) Use the Mel frequency cepstral coefficient algorithm to extract features from the data segment to be tested, and use K-Nearest Neighbors (KNN) to classify the data to be classified after feature extraction as normal or abnormal, that is, calculate the The distance between the classified data and the characteristic data of the sample space is calculated using the relative entropy/KL distance (Kullback–Leibler divergence). When the value of the KL distance exceeds the preset value, the data to be classified will be marked as abnormal, otherwise it is marked as normal and stored in the database of the abnormal alarm system; 4) The abnormal alarm system will scan the newly collected and categorized marked data points in its database system at fixed time intervals to discover the operating status of the mining belt conveyor within each time interval, and issue an alarm for the abnormally marked data Signal. 3. according to claim 2, be used for mine belt conveyor head and tail wheel on-line detection intelligent protection device, it is characterized in that, the concrete of described Mel-frequency cepstral coefficient algorithm (Mel-frequency cepstral coefficients, (MFCCs)) The steps are as follows: first cut the signal segment to be processed into small signal segments on the time axis; then perform periodic estimation of the spectral density for each signal segment; then use various filters provided by MFCCs to process the spectrum of each signal segment Density and energy superposition of each applied filter, logarithmic operation on the superimposed energy of each filter, and finally inverse discrete cosine transform on the result of each logarithmic operation. 4. The intelligent protection device for on-line detection of the head and tail wheels of the mining belt conveyor according to claim 2, characterized in that: the data of the continuous normal work of the mining belt conveyor for a week is divided into 24×7 pieces per hour The sample is used as the sample space for extraction and training, the sampling rate is 6 data per minute, 360 data points per hour, and then MFCC s feature extraction is performed on 168 samples to form a new feature sample space. 5. The intelligent protection device for on-line detection of the head and tail wheels of the mine belt conveyor according to claim 1, characterized in that: the multi-dimensional data acquisition module samples according to the set frequency, and automatically uploads the collected data through the data transmission module To the abnormal alarm system, and automatically delete the data it retains. 6. The intelligent protection device for on-line detection of the head and tail wheels of the mine belt conveyor according to claim 1, wherein the alarm signal of the abnormal alarm system includes an alarm prompt and a shutdown protection, if it is a shutdown protection, the network will The control signal is sent to the data transmission device, which is transmitted to the intelligent protection device through the data transmission device, and then the corresponding control signal is output to control the mine belt conveyor to stop running.