A wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration

Abstract

The present invention discloses a wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration, which belongs to the technical field of rolling mill vibration monitoring. The steps of the present invention are: 1. Install vibration monitoring devices at multiple locations of the rolling mill, and dynamically generate a connection routing table according to the network density and battery power to realize intelligent wireless networking; 2. The main control chip communicates with the vibration sensor to collect the rolling mill vibration acceleration signal; 3. Correct the collected rolling mill vibration acceleration signal to remove the trend item; 4. Segmented windowing is performed on the corrected vibration acceleration signal, and then frequency domain integration and filtering are performed to obtain the speed and displacement signals after segmented integration; 5. The real speed and displacement signals are synthesized by the overlap-addition method; 6. The vibration data is processed to obtain the vibration intensity, dominant frequency, peak value, and peak-to-peak value, and the rolling mill vibration is monitored based on the obtained parameters. The present invention prolongs the service life of the vibration monitoring device and reduces spectrum leakage.

Description

Technical Field

[0001] The present invention relates to the technical field of rolling mill vibration monitoring, and more particularly to a rolling mill vibration wireless monitoring method based on segmented windowed frequency domain integration. Background Art

[0002] During rolling mill operation, vibration is a particularly serious problem, accompanied by significant noise. Furthermore, rolling mill production lines typically consist of multiple rolling mills and other ancillary equipment, necessitating the deployment of vibration sensors at multiple locations for vibration monitoring. Comprehensively evaluating vibration information from these multiple locations allows for timely assessment of rolling mill status and ultimately increases mill lifespan.

[0003] Multi-point vibration monitoring for rolling mills requires extensive on-site cable routing, resulting in operational difficulties, high maintenance costs, and a high risk of damage. Current wireless networking methods utilize batteries to power vibration sensor nodes, which reduces operational complexity but can lead to network hotspots and short network lifespans due to insufficient battery power.

[0004] Currently, there are three metrics for evaluating vibration information: acceleration, velocity, and displacement. However, in practical engineering applications, velocity and displacement are difficult to acquire, making them unsuitable for multi-point vibration measurement in rolling mills. Commonly used acceleration integration algorithms include time-domain integration and frequency-domain integration. However, due to noise and bias in the acceleration data, time-domain integration produces trend terms, leading to cumulative errors in velocity and displacement. Signal processing methods such as Fourier transforms or wavelet transforms can be used to convert acceleration data to the frequency domain. The frequency domain data is then integrated twice to obtain frequency-domain representations of velocity and displacement. A commonly used frequency-domain integration algorithm is the Omega algorithm, which leverages the relationship between the power spectral density and autocorrelation function of the acceleration signal. By integrating the autocorrelation function, the autocorrelation functions of velocity and displacement are obtained. Then, through Fourier transforms, the power spectral density of velocity and displacement is obtained, allowing the time-domain velocity and displacement to be restored. However, spectrum leakage can occur, leading to errors. Summary of the Invention

[0005] 1. Technical problem to be solved by the invention

[0006] In view of the existing solutions for rolling mill vibration monitoring in the prior art, the problems of spectrum leakage and poor accuracy during vibration acceleration integration, and short life of wireless sensor networks, the present invention provides a wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration; the present invention uses an improved low-power Bluetooth mesh network (BLE MESH) for networking, which extends the service life of the vibration monitoring device; and adopts an improved segmented windowed frequency domain integration algorithm to reduce spectrum leakage and improve the accuracy of vibration data after integration.

[0007] 2. Technical solution

[0008] In order to achieve the above object, the technical solution provided by the present invention is:

[0009] The present invention provides a wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration, comprising the following steps:

[0010] Step 1: Install vibration monitoring devices at multiple locations of the rolling mill, and dynamically generate a connection routing table based on network density and battery power to achieve intelligent wireless networking;

[0011] Step 2: The main control chip communicates with the vibration sensor to collect the vibration acceleration signal of the rolling mill;

[0012] Step 3: Correct the collected rolling mill vibration acceleration signal to remove the trend item;

[0013] Step 4: perform segmented windowing on the corrected vibration acceleration signal, and then perform frequency domain integration to obtain segmented integrated velocity and displacement signals;

[0014] Step 5: Synthesize the real velocity and displacement signals through overlap-addition method;

[0015] Step 6: Process the vibration data to obtain vibration intensity, dominant frequency, peak value, and peak-to-peak value, and monitor the rolling mill vibration based on the obtained parameters.

[0016] Furthermore, the vibration monitoring device adopts a magnetic structure to be adsorbed on the surface of multiple locations of the rolling mill.

[0017] Furthermore, the vibration monitoring devices in step 1 are connected to each other through the improved BLE MESH networking protocol. The specific process is as follows:

[0018] 1) The vibration monitoring device node opens the broadcast channel to scan the surrounding nodes and records the number of scanned surrounding nodes N i And the MAC address of each node and save it, and so on, each node can obtain the number of device nodes N that can be connected to the surrounding i And the MAC address of each node;

[0019] 2) Assume that a device node N1 needs to send a connection request to a device node N2. It is necessary to compare the priorities of each node device. The priority calculation formula is I = α (N1.N i +cN2.N i )+βBAT, where BAT is the battery capacity of the device, N1.N i Represents the number of device nodes around N1, N2.N i represents the number of device nodes around N2, α and β are coefficients, their sum is 1, and the coefficient c is 2;

[0020] 3) After the priority of each node is determined, the nodes are connected to each other according to the priority to form a network and routing table;

[0021] 4) Based on the BLE MESH networking protocol, according to the routing table recorded by each node, the message is transmitted to the Bluetooth gateway through multiple transfers and uploaded to the cloud.

[0022] Furthermore, in step 2, the vibration acceleration signals of the rolling mill in three directions are collected and recorded as a x , a y , a z , the sampling frequency is f s , the number of sampling points is N; Step 3 First, perform high-pass filtering on the N data of the vibration acceleration signal in the three directions of the rolling mill to filter out the frequency below the critical value f c The low-frequency components of the signal are used to filter out the DC component of the gravitational acceleration.

[0023] Furthermore, in step 3, the least square method is used to remove the trend term and correct the acceleration, wherein the corrected acceleration a in the Z direction of the rolling mill vibration is z 'The following formula is obtained:

[0024]

[0025] Where, is a linear function, a0 and a1 are the coefficients of the polynomial function; N is the number of sampling points; the corrected accelerations of the rolling mill vibration in the X and Y directions can be obtained similarly.

[0026] Furthermore, in step 4, the corrected X-, Y-, and Z-axis time series data containing N vibration data points are divided into three segments with a 50% overlap, each segment including N / 2 vibration data points. The three segments of data are converted to the frequency domain through fast Fourier transform, and Hamming window and integration operations are added. After frequency domain integration, the frequency domain sequences of velocity and displacement are obtained, and then the frequency domain sequences are returned to the time domain by IFFT operation to obtain the velocity and displacement signals after three-segment integration.

[0027] Furthermore, in step 5, the results obtained by the integration are overlap-added at 50%, and after the overlap-addition, the middle N / 2 data are taken as the real velocity and displacement signals.

[0028] Furthermore, a MEMS three-axis accelerometer ADXL357 is used as a vibration sensor to collect the vibration signal of the rolling mill. The vibration sensor is connected to the main control chip through SPI communication. The main control chip is a high-performance, low-power SoC of the nrf52840 model.

[0029] 3. Beneficial effects

[0030] Compared with the existing known technologies, the technical solution provided by the present invention has the following significant effects:

[0031] (1) The present invention provides a wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration, which collects vibration acceleration data and removes interference through an algorithm, then segments the acceleration data, and after segmentation, performs a fast Fourier transform (FFT) to convert the time domain signal into a frequency domain signal, then performs a windowed integration algorithm on the frequency domain signal of the acceleration to obtain the frequency domain signals of the velocity and displacement, and then uses an inverse fast Fourier transform (IFFT) to obtain the time domain signal, and after superposition, obtains effective and accurate velocity and displacement signals, and finally calculates the vibration intensity, peak value, peak-to-peak value and vibration intensity based on the above time-frequency domain signals and uploads the data; the improved segmented windowed frequency domain integration algorithm reduces spectrum leakage and improves the accuracy of the vibration data after integration.

[0032] (2) The present invention provides a wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration. It uses low-power devices and an improved Bluetooth mesh network (BLE MESH) for networking, which solves the problem of simultaneous collection of vibrations at multiple locations in the rolling mill and extends the service life of the vibration monitoring device. The vibration monitoring device adopts a magnetic structure and uses wireless instead of wired, which reduces the difficulty of installation. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Schematic diagram of the structure of the vibration monitoring device of the present invention;

[0034] Figure 2 This is a structural block diagram of the rolling mill vibration wireless monitoring system of the present invention;

[0035] Figure 3 This is a hardware structure diagram of the rolling mill vibration wireless monitoring system of the present invention;

[0036] Figure 4 This is the main flow chart of the wireless monitoring of rolling mill vibration of the present invention;

[0037] Figure 5 This is a flow chart of the segmented windowed frequency domain integration in the present invention;

[0038] Figure 6 (a) is the acceleration time domain diagram; Figure 6 (b) is a comparison diagram between theoretical displacement integral and actual integral; Figure 6 (c) in the figure is the actual spectrum diagram.

[0039] Explanation of the numbers in the schematic diagram:

[0040] 1. Top cover; 2. Bushing; 3. Base; 4. Magnet; 5. Through hole; 6. Threaded hole; 7. Stud. DETAILED DESCRIPTION

[0041] In order to further understand the content of the present invention, the present invention is described in detail with reference to the accompanying drawings and embodiments.

[0042] Example 1

[0043] Combine Figure 2 and Figure 3 The rolling mill vibration wireless monitoring system of this embodiment mainly includes a power module, a main control chip, a three-axis MEMS vibration sensor, a radio frequency module, a communication circuit, etc.

[0044] Considering the mill's on-site vibration conditions and the device's standby duration, the vibration sensor uses the ADXL357 MEMS triaxial accelerometer. This chip has a power supply range of 2.25V to 3.6V, consumes 200uA in measurement mode and 20uA in standby mode, and offers optional measurement ranges of ±10g, ±20g, and ±40g. The sampling frequency is adjustable between 4 and 4000Hz, providing greater flexibility. To meet the data rate requirements of the high sampling frequency, the sensor circuit connects to the main control chip via SPI communication. During PCB design, independent power and ground lines were used to minimize interference from the on-site environment and ensure electromagnetic compatibility.

[0045] The power module in this embodiment is powered by a lithium battery. However, the operating voltage range of a lithium battery is 3.0V to 4.2V, which does not meet the power supply requirements of other chips. Therefore, this embodiment adds a voltage regulator circuit to the lithium battery voltage. The ME6211C33 chip is used to convert the lithium battery voltage to 3.3V to power the vibration sensor and main control chip.

[0046] In addition, since the lithium battery voltage is too high and the ADC module of the main control chip can only collect voltage analog quantities below 3.3V, this embodiment adds a voltage divider circuit to reduce the voltage. This part of the circuit is used to collect battery power for use in the networking stage and battery power warning prompts.

[0047] For the sake of low power consumption, high performance and strong wireless connection capability, the main control chip of this embodiment is selected as nrf52840. nrf52840 is a high-performance and low-power SoC launched by Nordic Semiconductor. It has a processor speed of up to 64MHz and 1MB of Flash memory, which can support complex applications and algorithms. The main control chip adopts advanced power optimization technology, including programmable power management unit (PPMU), power mode control and low-power clock, which can greatly extend battery life without sacrificing performance.

[0048] The peripheral circuits of the main control chip include a crystal oscillator circuit, a download circuit, and a power supply circuit. Considering the low-power Bluetooth wireless transmission rate, an antenna matching circuit was added. This is achieved by adding a matching network between the antenna and the chip. The matching network generally consists of an inductor, capacitor, and resistor. The antenna selected is an external 2.4GHz copper rod antenna, which has an ideal signal coverage range.

[0049] Combine Figure 1 The vibration monitoring device of this embodiment adopts a magnetic structure. Specifically, the vibration monitoring device includes a circuit board, a top cover 1, a bushing 2, a base 3 and a magnet 4. A threaded hole 6 is provided on the base 3, and the designed circuit board is fixed to the base 3 with screws. A threaded connection is used between the base 3 and the bushing 2, and between the top cover 1 and the bushing 2 to form a closed space. A through hole 5 is provided on the bushing 2, and the through hole 5 is a reserved hole for the external antenna interface. The device is powered by a lithium battery, and the battery is also placed in the bushing 2. A stud 7 is provided at the bottom of the base 3, and the magnet 4 is screwed into the stud 7 to achieve fixation with the base 3. In this way, the vibration monitoring device can be adsorbed on the surface of the equipment using the magnet 4 to ensure that the sensor is in full contact with the surface of the equipment so as to accurately measure the vibration signal. In addition, in order to obtain more accurate vibration data, sensors are installed at different positions of the rolling mill equipment for monitoring, and the appropriate number of installations is selected according to actual conditions.

[0050] In this embodiment, vibration monitoring devices are installed at multiple locations on the rolling mill, including the gearbox, motor, rollers, and support base. Furthermore, the vibration monitoring device dynamically generates a connection routing table based on network density and battery charge to implement intelligent wireless networking. The device uses an NRF52840 processor to read the three-axis acceleration signals (vibration data) collected by the ADX1357.

[0051] Combine Figure 4 This embodiment proposes a segmented windowed frequency domain integration algorithm for processing vibration data. First, the algorithm corrects the acceleration signal. Second, to reduce spectrum leakage, the acceleration signal is segmented and windowed before frequency domain integration to obtain segmented velocity and displacement signals. Finally, the true velocity and displacement signals are synthesized using the overlap-add method. The vibration data is then processed to obtain vibration intensity, dominant frequency, peak value, and peak-to-peak value, which are then uploaded to the cloud via a wireless sensor network.

[0052] The specific steps of vibration monitoring in this embodiment are:

[0053] After the node vibration monitoring device is installed, it forms a dynamic network with surrounding node vibration monitoring devices to achieve multi-point monitoring of rolling mill vibration. The following describes the collection and processing of vibration data using the Z axis as an example.

[0054] Step 1: Bluetooth Low Energy Mesh (BLE MESH) networking phase, specifically:

[0055] The nrf52840 microcontroller integrates an ARM Cortex-M4F processor and a BLE wireless module. The networking phase involves initializing the device's Bluetooth protocol stack, scanning, adding, and configuring other vibration monitoring devices, establishing and maintaining the network topology, and transmitting vibration data between devices and between devices and the Bluetooth gateway. Nodes use battery power, the number of surrounding nodes, and the number of nearby nodes they record to determine priority and connection relationships, generating routing tables and mesh networks.

[0056] Rolling mill production lines typically consist of multiple rolling mills and other ancillary equipment, requiring vibration sensors to be deployed at numerous locations for monitoring. Traditional solutions use a flooding protocol for communication. Because flooding broadcasts packets, even nodes that don't need them will receive them. This wastes bandwidth and energy, leading to a sharp increase in the number of packets in the network, causing congestion and performance degradation.

[0057] Based on this, the flooding protocol was changed to a dynamic routing protocol, generating a tree-structured routing table to reduce network traffic and unnecessary data transmission and reception. Devices were divided into central nodes and ordinary nodes. Because central nodes need to forward large amounts of data, battery life was conserved by converting the nodes to a balanced mesh network. This dynamically updates the central node based on battery power and network density, balancing the battery power of each device and thus extending the life of the entire wireless network.

[0058] Vibration sensor devices are interconnected via an improved BLE MESH networking protocol, enabling wireless monitoring of rolling mill vibration data. This networking algorithm is primarily designed to increase network aggregation and reduce the number of central nodes, thereby extending the life of the wireless network.

[0059] The specific process is as follows: 1) The device node opens its broadcast channel to scan the surrounding nodes and records the number of surrounding nodes that can be scanned N i And the MAC address of each node and save it, and so on, each node can obtain the number of device nodes N that can be connected to the surrounding i 2) Assume that a device node N1 needs to send a connection request to a device node N2, and needs to compare the priorities of each node device. The priority calculation formula is I = α (N1.N i +cN2.N i )+βBAT, where BAT is the battery capacity of the device, N1.N i Represents the number of device nodes around N1, N2.N irepresents the number of device nodes around N2, α and β are coefficients whose sum is 1, and c is 2. 3) Each node has a determined priority, and nodes are connected to each other based on their priority to form a network and routing table. This network is dynamically updated over time to achieve battery power balance. 4) Message transmission: Based on the BLE MESH, according to the routing table recorded by each node, messages are transmitted through multiple relays to the Bluetooth gateway and uploaded to the cloud.

[0060] Step 2: Collect vibration signals, specifically:

[0061] There are two ways to trigger the collection of vibration signals: one is to trigger the collection by the interrupt of the microcontroller timer; the other is to send the collection instruction by the host computer. Before collecting data, the data output format, data rate (i.e. sampling frequency) and measurement range of the vibration sensor need to be set by the microcontroller. The sampling frequency of this device is f s The frequency is 2KHZ and the measuring range is ±20g, which can be adjusted by the user. After the configuration is completed, the vibration data in the three axes (X, Y, Z) is read from the vibration sensor data buffer through SPI communication and saved in the internal RAM of the microcontroller. The number of sampling points N = 2048. The data in the three vibration directions are recorded as a x , a y , a z .

[0062] Step 3: Use the algorithm to correct the acceleration signal, specifically:

[0063] The temperature coefficients and linear expansion coefficients of sensor component materials vary. When the temperature changes, the resistance of the components changes by varying amounts, leading to imbalanced sensor output and zero drift. Furthermore, environmental interference can cause the measured data to deviate from the baseline, creating a trend term in the measurement signal. This trend term can affect the acceleration signal, particularly after the acceleration is integrated. The influence of the trend term is further amplified, causing the valid signal to be submerged in the trend term. For this reason, this embodiment removes the trend term before integrating the acceleration data.

[0064] This embodiment firstly z These 2048 data are high-pass filtered to remove frequencies below the critical value f c =5HZ low-frequency components to filter out the DC component of gravity acceleration. In addition, this embodiment uses the least squares method to remove the trend term to correct the acceleration. For the sake of simplicity, the polynomial function is set is a linear function; the corrected acceleration is a z '

[0065]

[0066]

[0067]

[0068] Where: a0 and a1 are the coefficients of the polynomial function; N is the number of sampling points; a z ' is the data after removing the trend item.

[0069] Step 4: The acceleration signal collected by the vibration monitoring device is non-periodic truncation. This non-periodic truncation of the signal causes its spectrum to tail within the frequency band, a phenomenon known as spectral leakage. By using different window functions to truncate the signal, the truncated signal is approximated to a periodic signal, which can reduce spectral leakage. However, windowing also attenuates the signal energy, especially at the edges. Therefore, the signal energy obtained after frequency domain integration of the windowed signal is less than the energy of the theoretical integrated signal.

[0070] In order to solve this problem, a segmented windowed frequency domain integration is designed. Segmented windowing is to process the signal in segments to reduce the energy loss caused by truncation, so as to retain the energy of the signal as much as possible. In this way, spectrum leakage can be reduced while retaining the energy of the signal as much as possible, thereby improving the accuracy and reliability of the acceleration integration into velocity and displacement. In this embodiment, the 2048 sets of data collected by the acceleration sensor are first segmented and windowed before frequency domain integration. Finally, the data after segmented integration are overlapped and added, and the middle segment is selected as the final integration result. Specifically:

[0071] a z 'For the Z-axis time series data containing 2048 vibration data points after correction, these 2048 data points are divided into 3 segments of 1024 data points, namely a z1 'a z2 'a z3 ', where a z1 'The value is a z 'The first 1024 data (that is, the data sequence from 0 to 1024); a z2 'The value is a z 'The middle 1024 data (that is, the data sequence from 512 to 1536); a z3 'The value is a z 'The last 1024 data (that is, the data sequence from 1024 to 2048). These three segments of data are converted to the frequency domain through fast Fourier transform (FFT), and Hamming window and integration operations are added. The Hamming window is represented in the time domain as With a z1 'For example: First, z1 ' and w z Perform FFT to get A z1 '(ω) and Wz (ω), A z1 ' is a complex sequence with real part Re(A z1 ') and the imaginary part Im(A z1 ').

[0072] After frequency domain integration, the frequency domain sequences of velocity and displacement are V z1 '、S z1 ',in:

[0073]

[0074]

[0075] In the formula

[0076] Then the frequency domain sequence V z1 '、S z1 'Perform IFFT operation to return to the time domain to obtain v z1 '、s z1 '. Similarly, the velocity and displacement signals after three-stage integration can be obtained.

[0077] Step 5: Synthesize the true velocity and displacement signals through overlap-addition method, specifically:

[0078] Assume that the true velocity and displacement time series are v z 、s z The integrated results are overlap-added with a 50% accuracy. The middle 1024 data points are then taken after the overlap-addition process. The number of valid sequence points decreases from 2048 to 1024. The velocity and displacement are calculated as follows.

[0079] v z =v z1 '+v z2 '+v z3 ',512≤z≤1536

[0080] s z =s z1 '+s z2 '+s z3 ',512≤z≤1536

[0081] Step 6: Process the vibration data to obtain vibration intensity, dominant frequency, peak value, and peak-to-peak value. Specifically:

[0082] The dominant frequency can be obtained by calculating the frequency domain sequence of the acceleration, velocity and displacement after FFT, and selecting the sequence point n with the highest amplitude spectrum. The point with the highest amplitude represents the main frequency component of the rolling mill. where f sis the known sampling frequency, n is the sequence number of the maximum amplitude point, and N is the known number of sampling points selected for the transformation operation. Based on n, we find the frequencies of the 2n, 3n, 4n, and 5n points, which are the double to quintupled frequencies. The dominant frequency is composed of the main frequency and its multiples.

[0083] Vibration severity refers to the root mean square value of the vibration velocity at a specified point on a mechanical device. It can measure the vibration intensity of the machine and can be obtained from the velocity time domain series obtained after integration. Add up the velocity values ​​of all time series, divide by the number of valid points, and then take the square root.

[0084] Peak value and peak-to-peak value refer to the maximum value and the difference between the maximum value and the minimum value in a signal cycle, respectively. Take the acceleration of the z-axis as an example to solve:

[0085] a zp =max(a z )

[0086] a zp-p =max(a z )-min(a z )

[0087] After the vibration characteristics are obtained, the data is uploaded to the cloud via wireless communication.

[0088] To verify the effectiveness and accuracy of the proposed segmented windowed integration algorithm, a simulation of the rolling mill acceleration with random noise was used for verification analysis. The simulated acceleration signal contains frequency components of 10Hz, 15Hz, and 25Hz, and the expression is: a(t) = 10πsin(10πt) + 15sin(15πt) + 25πsin(25πt) + rand. The simulation results are shown in Figure 2. Figure 6 As shown. Figure 6 (a) is the time domain diagram of the acceleration signal. Figure 6 The box in (b) shows the displacement data after piecewise windowed integral restoration. Compared with the theoretical value, it can be seen that the integration effect of the algorithm meets the actual needs, with a mean square error of 0.0140 and an error of about 4%. Figure 6 (c) shows the frequency spectrum. The dominant frequencies (9.7656, 15.1367, and 24.9023) are consistent with the original dominant frequencies. This algorithm accurately converts acceleration into velocity and displacement signals, providing more comprehensive vibration information for the rolling mill and providing more effective data for fault analysis.

[0089] The above is a schematic description of the present invention and its embodiments, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs a structure and embodiment similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A wireless monitoring method for rolling mill vibration based on segmented windowed frequency domain integration, characterized in that: The steps are: Step 1: Install vibration monitoring devices at multiple locations on the rolling mill, dynamically generate a connection routing table based on network density and battery power to implement intelligent wireless networking; the vibration monitoring devices are connected to each other through the improved BLE MESH networking protocol. The specific process is as follows: 1) The vibration monitoring device node opens the broadcast channel to scan the surrounding nodes and records the number of scanned surrounding nodes N i And the MAC address of each node and save it, and so on, each node can obtain the number of device nodes N that can be connected to the surrounding i And the MAC address of each node; 2) Assume that a device node N1 needs to send a connection request to a device node N2. It is necessary to compare the priorities of each node device. The priority calculation formula is I = α (N1.N i +cN2.N i )+βBAT, where BAT is the battery capacity of the device, N1.N i Represents the number of device nodes around N1, N2.N i represents the number of device nodes around N2, α and β are coefficients, their sum is 1, and the coefficient c is 2; 3) After the priority of each node is determined, the nodes are connected to each other according to the priority to form a network and routing table; 4) Based on the BLE MESH networking protocol, according to the routing table recorded by each node, the message is transferred to the Bluetooth gateway through multiple relays and uploaded to the cloud; Step 2: The main control chip communicates with the vibration sensor to collect the vibration acceleration signal of the rolling mill; Step 3: Correct the collected rolling mill vibration acceleration signal to remove the trend item; Step 4: perform segmented windowing on the corrected vibration acceleration signal, and then perform frequency domain integration to obtain segmented integrated velocity and displacement signals; Step 5: Synthesize the real velocity and displacement signals through overlap-addition method; Step 6: Process the vibration data to obtain vibration intensity, dominant frequency, peak value, and peak-to-peak value, and monitor the rolling mill vibration based on the obtained parameters.

2. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 1, characterized in that: The vibration monitoring device adopts a magnetic structure to be adsorbed on the surface of multiple locations of the rolling mill.

3. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 2, characterized in that: Step 2: Collect the vibration acceleration signals of the rolling mill in three directions and record them as a x , a y , a z , the sampling frequency is f s , the number of sampling points is N; Step 3 First, perform high-pass filtering on the N data of the vibration acceleration signal in the three directions of the rolling mill to filter out the frequency below the critical value f c The low-frequency components of the signal are used to filter out the DC component of the gravitational acceleration.

4. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 3, characterized in that: Step 3: Use the least square method to remove the trend term and correct the acceleration, where the corrected acceleration a in the Z direction of the mill vibration is z 'The following formula is obtained: Where, is a linear function, a0 and a1 are the coefficients of the polynomial function; N is the number of sampling points; the corrected accelerations of the rolling mill vibration in the X and Y directions can be obtained similarly.

5. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 4, characterized in that: In step 4, the corrected X-, Y-, and Z-axis time series data containing N vibration data points are divided into three segments with a 50% overlap, each segment including N / 2 vibration data points. The three segments of data are converted to the frequency domain through fast Fourier transform, and Hamming window and integration operations are added. After frequency domain integration, the frequency domain sequences of velocity and displacement are obtained. The frequency domain sequences are then returned to the time domain through IFFT operation to obtain the velocity and displacement signals after three-segment integration.

6. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 5, characterized in that: In step 5, the integrated results are overlapped and added with a 50% overlap, and after the overlap-addition, the middle N / 2 data are taken as the true velocity and displacement signals.

7. The method for wireless monitoring of rolling mill vibration based on segmented windowed frequency domain integration according to claim 6, characterized in that: The MEMS three-axis accelerometer ADXL357 is used as a vibration sensor to collect the vibration signal of the rolling mill. The vibration sensor is connected to the main control chip through SPI communication. The main control chip is the high-performance and low-power SoC of the nrf52840 model.

Images