Composite three-axis vibration temperature sensor based on direct-current carrier bus

Abstract

The invention discloses a composite three-axis vibration temperature sensor based on a direct current carrier bus, and relates to the technical field of industrial equipment state monitoring. The sensor comprises a vibration temperature body, a data acquisition and processing board, a power supply and communication interface board and an external connector, the data acquisition and processing board and the power supply and communication interface board comprise a sensing unit, a signal conditioning unit, a central processing and storage unit and a direct current carrier bus interface unit; a two-wire system direct-current carrier bus technology is adopted, power supply and data communication are achieved through the same pair of cables at the same time, in addition, a single-chip microcontroller integrated with a large-capacity SRAM is adopted, intelligent processing such as multi-channel synchronous data collection, temperature compensation, time-frequency domain analysis and envelope analysis is independently completed at the edge end, and the intelligent performance of the system is improved. The refined characteristic data are uploaded through a direct current carrier bus; according to the invention, high-precision, high-reliability and low-cost equipment state monitoring is realized, and excellent industrial applicability and environmental adaptability are achieved.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation, equipment condition monitoring and predictive maintenance technology, and particularly relates to a composite triaxial vibration temperature sensor based on a DC carrier bus. Background Technology

[0002] In modern industrial production systems, rotating machinery such as motors, pumps, fans, and gearboxes are crucial for ensuring the continuous and stable operation of production lines. Unplanned downtime of these critical equipment can lead to significant economic losses and potential safety risks. This has made Predictive Maintenance (PdM) a key technology for ensuring reliable equipment operation, gradually replacing traditional "periodic maintenance" or "failure-based maintenance" models. The core of predictive maintenance technology lies in continuously collecting physical parameters reflecting the health status of equipment in real time or periodically through a Condition Monitoring System (CMS), and then performing trend analysis and fault diagnosis based on these parameters. Vibration and temperature are widely recognized as two of the most direct and effective key indicators. Different fault types typically exhibit specific frequency characteristics in the vibration signal spectrum, revealing various mechanical faults in equipment, such as rotor imbalance, shaft misalignment, bearing wear, gear tooth breakage, and foundation loosening. Therefore, accurate, wide-bandwidth measurement and analysis of vibration signals are fundamental to early fault diagnosis. Temperature changes are usually closely related to equipment load, lubrication status, and friction conditions. Abnormal temperature increases are often direct signs of equipment overload, lubrication failure, or exacerbated faults. Therefore, developing high-precision, high-reliability, and easily deployable vibration and temperature sensors is crucial for building efficient and intelligent industrial equipment condition monitoring systems.

[0003] Currently, a mainstream technical solution widely used in industrial settings typically employs a system architecture of "discrete sensors + multiple cables + centralized data acquisition (DAQ)". A typical implementation is as follows: multiple discrete sensors are installed in parallel at key measurement points requiring broadband monitoring (such as bearing housings of high-speed rotating shafts) to cover the full frequency band detection needs. For example, a triaxial MEMS accelerometer is installed to measure the overall low- and mid-frequency vibrations and static tilt angle of the equipment; simultaneously, a single-axis high-frequency piezoelectric (IEPE / PZT) accelerometer is installed to specifically capture high-frequency impact signals generated by early failures in components such as bearings and gears; additionally, a separate temperature sensor, such as a Pt100 resistance temperature detector (RTD) or a K-type thermocouple, is installed near the same location to monitor the equipment temperature. Each sensor requires a dedicated cable leading from the equipment measurement point, passing through cable trays or cable management systems, and connecting to the field control cabinet or relay box respectively. A multi-channel data acquisition card (DAQ) is installed inside the control cabinet. This card needs to have input modules compatible with different sensor types, such as a voltage input module for receiving MEMS signals, an IEPE input module with a built-in constant current source, and a temperature acquisition module for RTDs / thermocouples. The acquisition card digitizes the raw and analog waveform signals from all sensors and uploads massive amounts of raw data packets to a host computer, PLC, or cloud server via industrial Ethernet (such as Modbus-TCP, OPC UA). Finally, dedicated analysis software on the server performs FFT spectrum analysis, feature extraction, and fault diagnosis calculations on the received massive amounts of data.

[0004] While the aforementioned existing technical solutions can achieve equipment status monitoring to a certain extent, their inherent architecture and technical limitations have the following significant drawbacks: (1) Complex wiring, high deployment cost and poor system scalability: Traditional industrial sensors generally use multi-wire connection, that is, multiple sensors need to be deployed at each measuring point, and each sensor needs multiple independent multi-core cables, resulting in complex on-site wiring and huge workload. The comprehensive cost of cables, cable trays, labor, etc. is very high. Moreover, the numerous connection points also greatly increase the probability of system failure. In addition, when monitoring points need to be added later, it is almost equivalent to a small engineering project. The system expansion is extremely inconvenient and difficult to adapt to the needs of large-scale, distributed monitoring.

[0005] (2) Fragmented monitoring schemes and difficulty in data fusion: Since vibration and temperature are collected by different types of independent sensors at slightly different locations and times, the spatiotemporal synchronization of multi-physical quantity data at the source is poor, making it difficult to achieve accurate multi-physical quantity correlation analysis. This results in inherent deficiencies in subsequent accurate vibration-temperature correlation analysis. Furthermore, the existing sensor types are limited (such as only MEMS or only piezoelectric), which cannot effectively take into account both low-frequency (such as equipment start-up and shutdown, static tilt angle) and high-frequency (such as early fault characteristics of bearings and gears) vibration signals, resulting in limited monitoring bandwidth and inability to achieve comprehensive, wide-bandgap status monitoring of equipment.

[0006] (3) Centralized data processing, heavy system burden and poor real-time performance: The remote centralized processing of massive amounts of unprocessed raw waveform data creates a "data fire hydrant" effect, which occupies huge network bandwidth and increases the server's computing pressure. Moreover, when the number of monitoring points is large, the system's data transmission delay and processing delay will increase significantly, affecting the real-time alarm and response capabilities to sudden faults.

[0007] (4) Data accuracy is easily affected by environmental interference and lacks intelligent compensation: The sensitive element of the sensor is easily affected by the temperature change of the working environment and drifts. However, most existing sensors lack effective and real-time onboard temperature compensation algorithms and do not have local computing capabilities. They cannot dynamically and accurately compensate for the sensitivity of vibration measurement based on real-time temperature at the data acquisition source, resulting in a decrease in the accuracy and reliability of measurement data in complex industrial environments. Summary of the Invention

[0008] The purpose of this invention is to address the numerous shortcomings of existing technologies, such as complex wiring, high deployment costs, limited monitoring dimensions, low data fusion, heavy system data processing burden, and susceptibility to environmental influences on measurement accuracy, by providing a composite triaxial vibration temperature sensor based on a DC carrier bus. The objective of this invention is achieved through the following technical solution: A composite triaxial vibration temperature sensor based on a DC carrier bus includes: Vibration temperature body 1: Made of metal material, located at the bottom of the sensor, with mounting threads at the bottom for fixing to the device under test; Data acquisition and processing board 2 and power supply and communication interface board 3: Both are installed in the upper cavity of the sensor vibration temperature body 1, and include a sensing unit, a signal conditioning unit, a central processing and storage unit, and a DC carrier bus interface unit. The sensing unit is located inside the vibration temperature body 1 and is the source of multi-physical quantity signal acquisition. The signal conditioning unit is integrated on the data acquisition and processing board 2 and is responsible for filtering and amplifying the raw and weak sensor signal output by the sensing unit. The central processing and storage unit is installed on the data acquisition and processing board 2 and adopts an integrated architecture of a single microcontroller. The DC carrier bus interface unit is arranged on a dedicated power supply circuit board and is used to connect a two-wire DC carrier bus to obtain power from the bus and perform bidirectional data communication. External connector 4, installed at the top of the sensor, is used to connect a two-wire DC carrier bus for power supply and communication functions.

[0009] The sensing unit includes: a Z-axis piezoelectric sensing element, which is a piezoelectric ceramic accelerometer used to collect high-frequency impact vibration signals; an XY-axis sensing element, which is a microelectromechanical system accelerometer integrating X and Y axes used to collect low-frequency vibration and static tilt angle signals; and a temperature sensing element, which is a digital temperature sensor installed near the metal base of the sensor used to collect the surface temperature of the device.

[0010] The signal conditioning unit includes: an X / Y axis signal conditioning path, which receives signals output from MEMS sensing elements and performs filtering and gain adjustment; and a Z axis signal conditioning path, which is used to condition signals output from piezoelectric elements.

[0011] The microcontroller of the central processing and storage unit integrates a floating-point arithmetic unit, DSP instruction set, high-precision analog-to-digital converter, UART interface, single-bus interface and on-chip SRAM; the central processing unit opens a data buffer through the on-chip SRAM to cache the raw vibration data acquired synchronously from multiple channels, and performs signal preprocessing, feature extraction and frequency domain analysis locally, without the need for external storage chips.

[0012] The DC carrier bus interface unit includes: a bus physical interface for non-polarity connection, equipped with a fuse and TVS diode for overcurrent and surge protection respectively; a multi-stage power processing subunit; a communication modulation and demodulation subunit for signal coupling with the DC bus via a two-wire carrier communication chip; a multi-stage power processing subunit including a first-stage buck circuit that uses a DC-DC buck converter to convert the DC voltage input to the bus into an intermediate voltage; and a second-stage voltage regulator circuit including at least two independent low-dropout linear regulators to generate digital and analog power supplies respectively, achieving isolation between the analog and digital power supplies.

[0013] A control method for a composite triaxial vibration temperature sensor based on a DC carrier bus includes the following steps: Step 1: System initialization; After the system is powered on, configure the analog-to-digital converter, hardware timer, UART communication interface and single-bus interface, and read the user configuration parameters from the internal memory; Step 2: Multi-channel synchronous data acquisition. Based on the signal output of the hardware timer, the triaxial vibration analog signal is synchronously acquired and stored in the SRAM buffer after analog-to-digital conversion. At the same time, the temperature value is read through the single bus interface. Step 3: In-situ data preprocessing and compensation. The original vibration data is temperature compensated using the current temperature value, and the compensated data is windowed. Step 4: Edge intelligent computing and feature extraction, performing hierarchical feature extraction at the edge, including: For the vibration data of the X, Y, and Z axes, time-domain characteristics are calculated, including mean, effective value, peak-to-peak value, and kurtosis; fast Fourier transform is performed on the Z-axis vibration data to extract the dominant frequency, dominant frequency amplitude, and second harmonic amplitude; envelope analysis is performed on the Z-axis vibration data to extract the envelope effective value and envelope peak value. Step 5: Pack the extracted feature values ​​into data frames according to a predetermined protocol, send them to the host computer via a DC carrier bus, and respond to the host computer's remote configuration and query commands.

[0014] In step three, temperature compensation is used to perform in-place numerical correction on the triaxial vibration data in the buffer based on the temperature value synchronously acquired by the temperature sensing element after data acquisition, using a preset calibration function to eliminate temperature drift error.

[0015] The formula for calculating the mean in step four is: Where N represents the total number of sampling points, and n represents the discrete-time index; The formula for calculating the effective value is: Where N represents the total number of sampling points, and n represents the discrete-time index; The formula for calculating peak-to-peak value is: The formula for calculating kurtosis is: Where N represents the number of waveform points, and μ represents the mean; The formula for calculating the Fast Fourier Transform is: Where N represents the number of waveform points, Represents the imaginary unit; The formulas for calculating the clock speed and peak clock speed are as follows: Where N represents the number of waveform points, Indicates the sampling frequency. This indicates the spectrum index corresponding to the peak frequency. The formula for calculating the second harmonic amplitude is: Where N represents the number of waveform points, Indicates the device's reference frequency. Indicates the sampling frequency; The formulas for calculating the effective value and peak value of the envelope are as follows: Where N represents the number of waveform points, This represents the Hilbert transform.

[0016] Envelope analysis specifically includes: performing digital bandpass filtering on the Z-axis vibration data to obtain a high-frequency signal; calculating the absolute value of the high-frequency signal and extracting the envelope signal through low-pass filtering; and calculating the effective value and peak value of the envelope signal.

[0017] Data frames include: synchronization and addressing fields, instruction and function fields, data payload fields, and data integrity verification fields.

[0018] The beneficial effects of this invention, through its integrated sensor system architecture and built-in edge intelligence, are mainly reflected in the following four aspects: (1) It fundamentally simplifies system deployment, significantly reduces overall costs and improves networking flexibility.

[0019] To address the issues of complex cabling and expansion difficulties caused by multiple cables in traditional solutions, this invention employs a two-wire DC carrier bus technology, requiring only a single two-core cable to replace multiple power and signal lines in traditional solutions. This design not only fundamentally simplifies on-site cabling, significantly reducing overall costs associated with cables, cable trays, and labor, but also minimizes system failure points. Furthermore, it supports convenient daisy-chain networking, enabling the system to exhibit excellent scalability and flexibility.

[0020] (2) Achieving precise synchronous monitoring across the entire frequency band and multiple dimensions, solving the data fusion problem. Unlike existing technologies that use multiple separate sensors in parallel, resulting in poor data synchronization and high costs, this invention integrates a high-frequency response Z-axis piezoelectric sensing element (covering high-frequency impact) and a low-frequency characteristic XY-axis MEMS sensing element (covering static tilt angle and low-frequency vibration) into a single device, combined with a temperature sensing unit. This integrated design ensures that the broadband vibration signal from static tilt angle to high-frequency impact and the temperature signal have strict spatiotemporal consistency at the source, providing a high-quality data foundation for subsequent multidimensional data fusion and precise diagnosis.

[0021] (3) Realize the transformation from "raw data acquisition" to "local intelligent analysis" and improve the real-time performance of the system. This invention utilizes a high-performance embedded processor, enabling the sensor itself to directly perform complex signal processing algorithms such as FFT spectrum analysis and feature extraction at the data acquisition front end. This edge computing mode not only greatly reduces the communication and computing burden on the back-end network bus and the host computer, but also avoids the "data fire hydrant" effect, significantly improving the system's real-time response capability and operating efficiency to sudden failures.

[0022] (4) Through dynamic compensation of hardware and software collaboration, the measurement reliability in complex environments is significantly improved.

[0023] To address the challenges of complex and variable industrial environments, this invention leverages a built-in temperature sensing unit and powerful onboard computing capabilities to run a real-time temperature compensation algorithm, dynamically correcting measurement errors caused by temperature drift in vibration-sensitive components. This mechanism effectively overcomes the lack of local intelligent compensation in traditional sensors, significantly improving measurement accuracy and data reliability in complex and changing industrial environments. Attached Figure Description

[0024] Figure 1 It is an exploded view of the product, a schematic diagram of the sensor's physical structure, shape, and internal component layout; Figure 2 This is the overall hardware functional block diagram of the sensor; Figure 3 This is the circuit schematic of the signal conditioning unit; Figure 4 This is a flowchart of the main working process of the sensor; Among them, 1-vibration temperature body, 2-data acquisition and processing board, 3-power supply and communication interface board, and 4-external connector. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] This invention provides a composite triaxial vibration temperature sensor based on a DC carrier bus. A schematic diagram of the sensor's physical structure, external shape, and internal component layout is shown below. Figure 1 As shown, the sensor mainly consists of four parts: a vibration and temperature body 1, a data acquisition and processing board 2, a power supply and communication interface board 3, and an external connector 4. Vibration Temperature Body 1: Made of high-strength metal materials such as stainless steel, it features standard mounting threads at the bottom for secure installation onto the housing of the device under test. The body integrates the core sensing unit and active functional components of the front-end analog circuitry. This integrated design minimizes the transmission path of the weak analog signal output from the core sensing unit and places it in an optimal electromagnetic shielding environment, fundamentally and significantly improving the signal-to-noise ratio and electromagnetic interference immunity of signal acquisition.

[0027] Data acquisition and processing board 2 and power and communication interface board 3: These are the core electronic components installed in the upper cavity of the vibration and temperature body 1. They innovatively employ a vertically stacked dual-circuit board structure, consisting of data acquisition and processing board 2 and power and communication interface board 3 from bottom to top. This physically separated design achieves strict functional zoning: the lower-layer data acquisition and processing board 2 integrates high-precision analog-to-digital conversion, microprocessor, and other digital processing circuits, focusing on achieving accurate digitization and preliminary analysis of vibration and temperature signals; the upper-layer power and communication interface board 3 is responsible for power supply and bus communication circuits. This architecture effectively isolates the highly sensitive analog / digital mixed-signal circuits from the power and communication circuits, which may generate significant electrical noise, thus creating a crucial barrier within the system and effectively preventing noise from the power and communication buses from interfering with the high-precision data acquisition process.

[0028] External connector 4: Installed at the very top of the sensor body, it provides a robust and reliable physical interface for connecting the sensor to external systems and is used to connect a two-wire DC carrier bus to enable power supply and communication functions.

[0029] Below we will combine Figure 2 The core technical units that realize the functions of this invention are described in detail. These functional units are physically distributed across three functional areas: the vibration temperature body 1, the data acquisition and processing board 2, and the power supply and communication interface board 3.

[0030] (1) Sensing unit The sensing unit is the source for acquiring multi-physical quantity signals. This unit is located inside the vibration temperature body 1 and includes the following three core components: Z-axis piezoelectric sensing element: Piezoelectric ceramic (PZT) accelerometer is preferably installed along the Z-axis of the sensor (usually in the same direction as the main shaft of the device). With its extremely high stiffness and excellent high-frequency response characteristics (up to 10kHz and above), this element is specially designed to accurately capture high-frequency impact and resonance signals generated by components such as bearings and gears in the early stage of failure.

[0031] XY-axis MEMS sensing element: It adopts a micro-electro-mechanical system (MEMS) accelerometer chip that integrates sensing structures in two orthogonal directions of X and Y axes. The advantage of MEMS sensors lies in their high sensitivity to low-frequency and DC signals. They can effectively measure the overall vibration of equipment, low-frequency shaking during start-up and shutdown, and static tilt caused by foundation settlement or changes in installation angle, perfectly making up for the shortcomings of piezoelectric elements in measuring static and extremely low-frequency signals.

[0032] Temperature sensing element: A high-precision digital temperature sensor with a single-bus interface. This element is mounted close to the metal base of the sensor to ensure that it can reflect the surface temperature of the measured device most accurately, quickly, and precisely.

[0033] Through the above-mentioned composite design, the sensing unit achieves synchronous and point-to-point measurement of the four most critical parameters characterizing the equipment status (Z-axis high-frequency vibration, X / Y-axis low-frequency vibration, and temperature) within a single device, providing high-quality source data for subsequent multi-dimensional data fusion and accurate diagnosis.

[0034] (2) Signal conditioning unit The signal conditioning unit circuitry is integrated on the data acquisition and processing board 2, responsible for filtering and amplifying the weak raw sensor signal output from the sensing unit. For example... Figure 3 As shown, this unit contains the following two independent and optimized conditioning pathways: X / Y axis signal conditioning path: The X / Y axis signal conditioning path receives the X_OUT and Y_OUT signals output from the MEMS sensing element, and then enters a two-stage active low-pass filter composed of operational amplifiers U4.1 to U4.4. This filter is precisely configured with resistors and capacitors, and its main function is to filter out high-frequency noise that exceeds the effective frequency band and to perform appropriate gain adjustment on the signal, ultimately outputting clear and stable analog signals X_ADC and Y_ADC for subsequent analog-to-digital conversion.

[0035] Z-axis signal conditioning path: The Z-axis signal conditioning path is for the Z_OUT signal from the PZT piezoelectric element. The Z_OUT signal enters the active bandpass or low-pass filter circuit composed of operational amplifier U5. This path is specially designed for the high frequency and high dynamic range signal characteristics of PZT. The signal is precisely conditioned by a combination of resistors and capacitors and finally output as Z_ADC.

[0036] (3) Central processing and storage unit This unit is installed on the data acquisition and processing board 2. This invention innovatively adopts an integrated architecture of a single microcontroller, realizing complex functions that previously required multiple chip combinations (such as "MCU+DSP" or "MCU+external RAM") to complete. Thus, while ensuring high performance, it achieves a comprehensive technical effect of significantly reducing sensor cost, greatly reducing physical size, and simultaneously reducing system power consumption and failure rate.

[0037] In this preferred embodiment, the solution implements the single-chip integration using a STMicroelectronics STM32H723 microcontroller. The key to selecting this chip lies in its unique combination of internal resources, which enables it to perfectly support the integrated architecture proposed in this invention. Fusion of computing power (MCU+DSP Fusion): This chip integrates a high-performance Cortex-M7 core (including FPU and DSP instruction set), which enables it to independently perform high-speed FFT operations on multiple vibration signals locally without the need for an external dedicated digital signal processing chip (DSP); at the same time, it efficiently handles routine control tasks such as device management, bus communication protocol stack, and temperature compensation algorithms; this fusion of computing power is the primary prerequisite for achieving low cost and small size.

[0038] Internalization of storage resources (Elimination of External RAM): This solution utilizes the large-capacity 564KB SRAM integrated within the STM32H723 to cache raw data. This "on-chip RAM" resource is sufficient to create a local acquisition buffer and algorithm working area for multi-channel, high-precision ADC data, thereby completely avoiding the use of any external RAM chips. This reduces PCB area, cost, and wiring complexity, which is another key to achieving sensor miniaturization and low cost.

[0039] High integration of peripherals: The chip also integrates the high-precision ADC required by this solution, the single-bus interface for processing digital temperature sensors, and the UART interface for communicating with the Powerbus module, which minimizes the number of peripheral circuit components required and improves the integration and reliability of the entire system.

[0040] (4) DC carrier bus interface unit The DC carrier bus interface unit is the core module for realizing two-wire power supply and communication for sensors. Its circuit is mainly laid out on a dedicated power supply circuit board to achieve physical isolation from the sensitive analog signals on the data acquisition and processing board, ensuring an extremely high signal-to-noise ratio.

[0041] It should be noted that the principles of the DC bus-based carrier communication and power supply circuit involved in this unit are common knowledge familiar to those skilled in the art. Its core functions can be implemented using existing commercial carrier communication chips and mature power management chips. Those skilled in the art can purchase relevant chips through commercial channels and build peripheral circuits based on their general datasheets. Therefore, this embodiment will not elaborate on its specific transistor-level circuit connections, but only describes its system-level architecture and functional logic in this invention. This unit mainly consists of the following three functional sub-modules: Bus Physical Interface and Protection Module: The sensor connects to a two-wire bus via its input endpoint. To adapt to the complex installation environment of industrial sites, a full-bridge rectifier circuit is first connected to the input, achieving a non-polarity connection and improving the fault tolerance of field wiring. In addition, the interface circuit is equipped with overcurrent protection components (such as self-resetting fuses) and transient voltage suppressor diodes (TVS) for electrostatic discharge and surge protection, thereby enhancing the robustness of the interface under harsh operating conditions.

[0042] Communication Modulation and Demodulation Module: The core of this module is a two-wire carrier communication chip (Modem). It is connected to the rectified DC bus via capacitive coupling and other methods, and is responsible for the physical layer transmission of signals. When the central processing unit needs to send data, this chip modulates the digital logic signal from the MCU serial communication interface (such as UART_TX) into a high-frequency carrier signal and superimposes it onto the DC bus; conversely, it demodulates the carrier signal on the bus into a digital signal and transmits it to the MCU.

[0043] Multi-stage power supply module: To provide a clean and stable power supply to the sensor, this invention adopts a refined power supply architecture of "DC-DC + LDO" with two-stage buck and multi-channel isolation. First-stage main buck: The high-voltage DC power from the bus first passes through a high-efficiency Buck circuit composed of a synchronous buck converter, power inductor, and peripheral components, efficiently reducing it to a lower intermediate voltage. Second-stage fine regulation and isolation: The intermediate voltage generated in the first stage is then shunted to two independent low-dropout linear regulators (LDOs) for secondary bucking and noise isolation. The first LDO generates a clean digital power supply (e.g., 3.3V_MCU) specifically for the central processing unit and other digital logic circuits; the second LDO generates an isolated, ultra-low-noise analog power supply (e.g., 3.3V_ADC) specifically for all analog circuits on the data acquisition and processing board, including the operational amplifiers and ADC module of the signal conditioning unit.

[0044] This design, which physically isolates analog and digital power supplies at the source through independent LDOs, can prevent digital noise generated during high-speed MCU operation from coupling and contaminating the weak analog vibration signal at the front end through the power path, thereby greatly ensuring the accuracy and reliability of data acquisition.

[0045] The main working process diagram of the sensor of the present invention is as follows: Figure 4 As shown, the core workflow after the sensor is powered on is implemented by the firmware program running in the central processing unit, and its control method specifically includes the following steps: Step 1: System initialization.

[0046] After the system is powered on, it first performs a comprehensive hardware and parameter initialization.

[0047] This stage configures and starts all key functional modules: initializes the ADC and sets it to multi-channel synchronous sampling mode; configures the hardware timer (TIM) to generate accurate sampling trigger signals; sets the baud rate and protocol format of the UART and PB331 communication chip to establish a reliable bus connection; and initializes the single-wire interface for reading temperature sensors for temperature sensor communication.

[0048] Then, the user configuration parameters, including the sensor's bus address, sampling frequency, number of FFT analysis points, and preset alarm thresholds, are read from the internal flash memory to complete the sensor's initialization and startup.

[0049] Step 2: Multi-channel synchronous data acquisition.

[0050] After initialization, the system enters a continuous data acquisition loop.

[0051] The hardware timer (TIM) periodically generates a trigger signal at a set frequency, driving the high-precision ADC inside the central processing unit to simultaneously initiate analog-to-digital conversion of the three analog signals (X_ADC, Y_ADC, Z_ADC) output by the signal conditioning unit. The digital sampling points obtained after conversion are immediately stored efficiently in the raw data buffer pre-allocated in a large-capacity SRAM via DMA (direct memory access).

[0052] Meanwhile, the MCU periodically sends temperature conversion commands to the temperature sensor via a single-bus interface and reads its precise digital temperature value. This data acquisition process continues until the SRAM buffer is filled with a complete frame of data.

[0053] Step 3: In-situ data preprocessing and compensation.

[0054] After a frame of data is acquired, the central processing unit performs in-place preprocessing on the data in SRAM before using it for FFT calculation to optimize subsequent analysis.

[0055] This process first reads the temperature value corresponding to the current frame synchronously, and then corrects each sampling point of the triaxial vibration data in the buffer point by point according to the calibration function to eliminate the measurement error caused by the temperature drift of the sensor itself. Subsequently, in order to suppress the spectral leakage caused by the subsequent FFT operation, the MCU applies a window function to the temperature-compensated time-domain data to obtain the original, complex data waveform.

[0056] Step 4: Edge intelligent computing and feature extraction.

[0057] This invention innovatively constructs a hierarchical, multi-dimensional feature extraction engine on the sensor terminal. This engine is entirely independently operated by the central processing unit using its internal DSP instruction set and hardware floating-point unit (FDU), refining the raw, complex data waveforms into a set of concise yet highly condensed device status feature values. This process mainly consists of the following three parallel analysis layers: 1. Time-domain analysis of foundation vibration (for X, Y, and Z axes): For each signal x[n] of the X, Y, and Z axes, perform the following calculations to calculate the mean, effective value, peak-to-peak value, and kurtosis index. To save computing resources, the calculation of multiple indexes can be completed by traversing the data buffer once or twice.

[0058] The mean (Mean, μ) is calculated using the following formula: Where N represents the total number of sampling points, and n represents the discrete-time index. The amplitude of all sampling points is accumulated in one loop, and after the loop ends, it is divided by the total number of points N.

[0059] The root mean square (RMS) is calculated using the following formula: Where N represents the total number of sampling points, and n represents the discrete-time index. In each iteration of the loop, the square of the amplitude of each sampling point is accumulated. After the loop ends, the sum is divided by N, and finally, the optimized floating-point square root function arm_sqrt_f32() from the ARM CMSIS-DSP library is called to obtain the final result.

[0060] Peak-to-peak (P2P) value is calculated using the following formula: When iterating through the data, the MCU sets two variables, max_val and min_val, and compares and updates them in real time. After the loop ends, the two variables are subtracted.

[0061] Kurtosis is calculated using the following formula: Where N represents the number of waveform points and μ represents the mean.

[0062] The first step is to calculate the mean μ. The second step is to iterate through the data again, accumulating (x[n] - μ) to the fourth and second powers, and finally performing the calculation according to the formula. This process makes full use of the MCU's hardware floating-point unit (FPU) to accelerate exponentiation and division.

[0063] 2. Key frequency domain feature analysis (mainly focusing on the Z-axis) Because the Z-axis PZT sensor is highly sensitive to high-frequency shocks, its data is best suited for diagnosing faults in rotating components. The MCU performs an FFT operation on the Z-axis data to extract the following key frequency domain features: The formula for calculating the Fast Fourier Transform (FFT) is as follows: Where N represents the number of waveform points, It represents the imaginary unit.

[0064] The peak frequency (peakFreq) and peak frequency (peakAmp) are calculated using the following formulas: Where N represents the number of waveform points, Indicates the sampling frequency. This indicates the spectrum index corresponding to the peak frequency.

[0065] Iterate through the amplitude spectrum array Mag[k] and find its maximum value Mag[k_max].

[0066] The second harmonic amplitude (amp2x) is calculated using the following formula: Where N represents the number of waveform points, Indicates the device's reference frequency. Indicates the sampling frequency.

[0067] Based on the preset device reference frequency f_base, calculate its index position k_2x in the spectrum.

[0068] 3. High-frequency resonance demodulation analysis (envelope analysis, for the Z-axis) This is an advanced analytical technique for diagnosing early, subtle bearing faults. When a bearing develops a tiny crack, it generates periodic high-frequency impacts that excite high-frequency resonances in sensors or structures. The purpose of envelope analysis is to demodulate implicit, low-frequency fault characteristic information from these high-frequency resonance signals.

[0069] Digital bandpass filtering: First, apply a digital bandpass filter (IIR Butterworth filter) to z[n] to obtain the filtered high-frequency signal z_filtered[n].

[0070] Envelope Analysis: Subsequently, the "absolute value-low-pass filtering" method is used to calculate the mean of the filtered signal z_filtered[n] and subtract the mean to eliminate DC bias. The absolute value of the centered signal is taken to obtain z_filtered[n] - μ|. This step is functionally equivalent to detection. A low-pass filter is applied again to the rectified signal to filter out residual high-frequency components, finally obtaining a smooth envelope signal, denoted as envelope[n]. The MCU regards the obtained envelope signal envelope[n] as a new time-domain signal with more concentrated information and reapplies the above RMS and Peak calculation methods to it to obtain two final diagnostic indicators: Envelope RMS and Envelope Peak. The calculation formulas are as follows: Where x(t) represents the original time-domain signal, t represents the time variable, τ represents the integral variable, z(t) represents the complex analytic signal, x_h(t) represents the signal after Hilbert transform, H{x(t)} represents the Hilbert transform, j represents the imaginary unit, and E(t) represents the envelope signal.

[0071] Step 5: Data Packaging and Bus Communication.

[0072] After all the calculations and extractions are completed, the sensor enters the data packaging and bus communication stage, reporting valuable "information" rather than massive amounts of "data". The core of this step is to organize discrete feature values ​​into a structured data frame suitable for transmission on the bus and send it out through the bus interface unit.

[0073] Furthermore, it doesn't just perform one-way data reporting; instead, through a fully functional two-way communication protocol, it transforms the sensor into an intelligent network node that can be remotely configured, diagnosed, and managed by a host computer. The MCU dynamically constructs data frames based on a custom communication protocol optimized for this application scenario. The data frames defined by this protocol logically include, but are not limited to, the following functional fields: Synchronization and Address Field: Used to identify the start of a data frame and contains the unique bus address of this sensor, ensuring that the host computer can accurately communicate with the specific sensor in the network.

[0074] Function Code Field: Defines the type of data frame. Examples include sensor address setting, feature value reporting, raw waveform upload, and remote parameter configuration.

[0075] Payload Field: This is the core of the data frame. It is designed as a variable-length or structured data container, the content and structure of which are determined by the function code. For example, the payload can contain all key characteristic values ​​such as kurtosis, spectral frequency, and envelope RMS, or it can contain configurable "parameter IDs" and "parameter values" such as sampling frequency and FFT points, or it can contain a piece of unprocessed, high-resolution raw ADC sampled data.

[0076] Error Check Field: This field ensures the accuracy of data during bus transmission. Its value is calculated based on all other bytes within the frame using a pre-defined verification algorithm.

Claims

1. A composite triaxial vibration temperature sensor based on a DC carrier bus, characterized in that, include: The vibration temperature body (1) is made of metal material and is located at the bottom of the sensor. It has mounting threads at the bottom for fixing to the device under test. The data acquisition and processing board (2) and the power and communication interface board (3) are installed in the upper cavity of the sensor vibration temperature body (1). The data acquisition and processing board (2) includes a sensing unit, a signal conditioning unit, and a central processing and storage unit. The power and communication interface board (3) is provided with a DC carrier bus interface unit. The central processing and storage unit is a microcontroller that controls the sensing unit to acquire sensor signals and for the signal conditioning unit to filter and amplify them. The DC carrier bus interface unit is used to connect to a two-wire DC carrier bus, obtain power from the bus, and perform bidirectional data communication. An external connector (4) is installed at the top of the sensor and connected to the DC carrier bus interface unit to connect a two-wire DC carrier bus that enables power supply and communication functions.

2. The sensor according to claim 1, characterized in that: The sensing unit includes: The Z-axis piezoelectric sensing element, using a piezoelectric ceramic accelerometer, is used to collect high-frequency impact vibration signals. The XY axis sensing element adopts a microelectromechanical system accelerometer integrating the X and Y axes to collect low-frequency vibration and static tilt angle signals; The temperature sensing element, employing a digital temperature sensor, is installed near the metal base of the sensor to collect the surface temperature of the device.

3. The sensor according to claim 1, characterized in that, The signal conditioning unit includes an X / Y axis signal conditioning path and a Z axis signal conditioning path; the X / Y axis signal conditioning path receives the signal output by the MEMS sensing element and performs filtering and gain adjustment; the Z axis signal conditioning path is used to condition the signal output by the piezoelectric element.

4. The sensor according to claim 1, characterized in that, The microcontroller of the central processing and storage unit integrates a floating-point arithmetic unit, a DSP instruction set, a high-precision analog-to-digital converter, a UART interface, a single-bus interface, and on-chip SRAM. The central processing unit opens a data buffer through the on-chip SRAM to cache the raw vibration data acquired synchronously from multiple channels, and performs signal preprocessing, feature extraction, and frequency domain analysis locally without the need for external storage chips.

5. The sensor according to claim 1, characterized in that, The DC carrier bus interface unit includes: The bus physical interface enables non-polarity connection and is equipped with fuses and TVS diodes to provide overcurrent protection and surge protection, respectively. It is a multi-level power processing subunit. The communication modulation and demodulation subunit is coupled to the DC bus via a two-wire carrier communication chip. The multi-stage power processing subunit includes a first-stage buck circuit that uses a DC-DC buck converter to convert the DC voltage input from the bus into an intermediate voltage; and a second-stage voltage regulation circuit that includes at least two independent low-dropout linear regulators to generate digital power and analog power respectively, thereby achieving isolation between the analog and digital power supplies.

6. A control method for a composite triaxial vibration temperature sensor based on a DC carrier bus as described in claim 1, characterized in that, Includes the following steps: Step 1: System initialization; After the system is powered on, configure the analog-to-digital converter, hardware timer, UART communication interface and single-bus interface, and read the user configuration parameters from the internal memory; Step 2: Multi-channel synchronous data acquisition. Based on the signal output of the hardware timer, the triaxial vibration analog signal is synchronously acquired and stored in the SRAM buffer after analog-to-digital conversion. At the same time, the temperature value is read through the single bus interface. Step 3: In-situ data preprocessing and compensation. The original vibration data is temperature compensated using the current temperature value, and the compensated data is windowed. Step 4: Edge intelligent computing and feature extraction, performing hierarchical feature extraction at the edge, including: For the vibration data of the X, Y, and Z axes, time-domain characteristics are calculated, including mean, effective value, peak-to-peak value, and kurtosis; fast Fourier transform is performed on the Z-axis vibration data to extract the dominant frequency, dominant frequency amplitude, and second harmonic amplitude; envelope analysis is performed on the Z-axis vibration data to extract the envelope effective value and envelope peak value. Step 5: Pack the extracted feature values ​​into data frames according to a predetermined protocol, send them to the host computer via a DC carrier bus, and respond to the host computer's remote configuration and query commands.

7. The method according to claim 6, characterized in that, The temperature compensation in step three is used to perform in-situ numerical correction on the triaxial vibration data in the buffer based on the temperature value synchronously acquired by the temperature sensing element after data acquisition is completed, using a preset calibration function to eliminate temperature drift error.

8. The method according to claim 6, characterized in that, The formula for calculating the mean in step four is as follows: Where N represents the total number of sampling points, and n represents the discrete-time index; The formula for calculating the effective value is: Where N represents the total number of sampling points, and n represents the discrete-time index; The formula for calculating the peak-to-peak value is: The kurtosis calculation formula is as follows: Where N represents the number of waveform points, and μ represents the mean; The formula for calculating the Fast Fourier Transform is as follows: Where N represents the number of waveform points, Represents the imaginary unit; The formulas for calculating the main frequency and the peak main frequency are as follows: Where N represents the number of waveform points, Indicates the sampling frequency. This indicates the spectrum index corresponding to the peak frequency. The formula for calculating the second harmonic amplitude is as follows: Where N represents the number of waveform points, Indicates the device's reference frequency. Indicates the sampling frequency; The formulas for calculating the effective value and peak value of the envelope are as follows: in, Represents the original time-domain signal. Represents a time variable. Represents the integral variable. Represents a complex analytic signal. This represents the signal after the Hilbert transform. Represents the Hilbert transform. Represents the imaginary unit This represents the envelope signal.

9. The method according to claim 6, characterized in that, The envelope analysis in step four specifically includes: performing digital bandpass filtering on the Z-axis vibration data to obtain a high-frequency signal; calculating the absolute value of the high-frequency signal and extracting the envelope signal through low-pass filtering; and calculating the effective value and peak value of the envelope signal.

10. The method according to claim 6, characterized in that, The data frame in step five includes a synchronization and addressing field, an instruction and function field, a data payload field, and a data integrity verification field.

Images