Integrated multi-sensor fusion system for monitoring state of rotating mechanical part

Abstract

The invention provides an integrated multi-sensing fusion system for monitoring the state of a rotating mechanical part, which comprises a multi-source sensing module, a signal processing module and a fusion analysis module, and is characterized in that the multi-source sensing module comprises a plurality of sensing units and is used for collecting multi-dimensional data in the operation process of rotating mechanical equipment; the signal processing module is used for performing preprocessing and analog-to-digital conversion on analog signals acquired by each sensing unit in the multi-source sensing module, the signal processing module comprises a main control unit, and the fusion analysis module comprises a fusion algorithm. The main control unit performs comprehensive analysis and information fusion on data output by each sensing unit through the fusion algorithm, and realizes real-time transmission of a processing result through a communication interface; by integrating a plurality of sensing units, a signal processing module and a fusion algorithm into a whole, deep fusion of hardware and software is realized, and the real-time performance and accuracy of state monitoring of the rotating mechanical part are ensured.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment health monitoring technology, and more specifically, to an integrated multi-sensor fusion system for monitoring the condition of rotating machinery components. Background Technology

[0002] Rotating machinery plays a crucial role in industrial production, with a wide range of applications encompassing numerous key pieces of equipment such as motors, speed reducers, fans, generators, air compressors, centrifuges, and water pumps. The stable operation of these devices is a core element ensuring the efficient and safe operation of the entire production system. Failures can not only lead to a significant drop in production efficiency but also potentially trigger serious safety accidents, causing substantial economic losses. Therefore, real-time status monitoring of critical components is essential to ensure the reliable operation of rotating machinery and to promptly detect potential faults. Real-time monitoring allows for timely understanding of the equipment's operating status, early detection of signs of malfunctions, and the implementation of appropriate maintenance measures to prevent further deterioration.

[0003] With the continuous improvement of industrial automation and intelligence, condition monitoring technology is also constantly evolving. Traditional periodic maintenance methods have many drawbacks, such as fixed maintenance cycles that are difficult to adjust flexibly according to the actual operating conditions of equipment, easily leading to over-maintenance or under-maintenance. Data-driven predictive maintenance, on the other hand, can accurately predict the time and type of equipment failure based on real-time analysis of equipment operating data, thereby enabling targeted maintenance and greatly improving maintenance efficiency and equipment reliability. Therefore, condition monitoring technology is gradually shifting from traditional periodic maintenance to data-driven predictive maintenance. In the development of condition monitoring technology, multi-sensor fusion technology has been introduced to improve monitoring accuracy and reliability. Multi-sensor fusion technology can comprehensively utilize information from different types of sensors to obtain more comprehensive and accurate equipment operating status data, thereby improving the accuracy of fault diagnosis.

[0004] However, in practical applications, multi-sensor fusion technology still faces some bottlenecks. Currently, most condition monitoring systems adopt a split architecture. In this architecture, each sensor is installed and processes data independently, resulting in a system composed of multiple modules. This not only makes the system bulky but also complicates the installation process, making it difficult to adapt to space-constrained application scenarios. Furthermore, this split architecture lacks an effective collaborative analysis mechanism. The data collected by each sensor lacks effective correlation and integration, limiting the deep fusion and comprehensive judgment capabilities of multi-source data and failing to fully leverage the advantages of multi-sensor fusion technology.

[0005] Furthermore, in the process of comprehensive condition assessment of rotating machinery, multi-sensor fusion technology often needs to acquire data on multiple parameters simultaneously, such as temperature, acceleration, and sound. Different types of sensors have different response times and sampling frequencies, which brings significant technical challenges to the synchronous acquisition of multi-source data. Poor data synchronization may lead to time discrepancies in data collected by different sensors, affecting the accuracy of data analysis and reducing the reliability of fault diagnosis. Traditional data fusion methods typically rely on complex computational models when processing multi-source data. These models not only increase the computational burden on the system, slowing down its operation, but may also generate high processing latency, making it difficult to meet real-time requirements. In industrial production, real-time performance is crucial for fault diagnosis; failure to detect faults promptly and accurately may lead to further escalation and more serious consequences. Meanwhile, existing systems also have significant shortcomings in hardware circuit design. Most sensor interfaces are not properly conditioned, making signals susceptible to noise interference during transmission and affecting signal quality. In analog signal acquisition, the lack of appropriate filtering and amplification design results in a low signal-to-noise ratio, severely impacting subsequent feature extraction and model recognition, further reducing the accuracy of fault diagnosis. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide an integrated multi-sensor fusion system for monitoring the condition of rotating machinery components. This system solves problems such as low integration, poor data synchronization, and weak intelligent analysis capabilities in existing monitoring systems. The present invention achieves simultaneous acquisition of multiple parameters, including temperature, acceleration, and sound, and combines them with a lightweight intelligent model oriented towards edge computing for efficient analysis and deployment. This significantly enhances the real-time performance and accuracy of equipment condition assessment, providing a more effective solution for the condition monitoring and fault diagnosis of rotating machinery.

[0007] To solve the above problems, the technical solution of the present invention is as follows:

[0008] An integrated multi-sensor fusion system for monitoring the condition of rotating machinery components includes a multi-source sensing module, a signal processing module, and a fusion analysis module. The multi-source sensing module includes multiple sensing units for collecting multi-dimensional data during the operation of the rotating machinery. The signal processing module preprocesses and performs analog-to-digital conversion on the analog signals collected by each sensing unit in the multi-source sensing module. The signal processing module includes a main control unit. The fusion analysis module includes a fusion algorithm. The main control unit performs comprehensive analysis and information fusion on the data output by each sensing unit through the fusion algorithm, and transmits the processing results in real time via a communication interface. By integrating multiple sensing units, the signal processing module, and the fusion algorithm into one system, deep hardware and software integration is achieved, ensuring the real-time performance and accuracy of the condition monitoring of rotating machinery components.

[0009] Preferably, the multi-source sensing module includes a temperature sensing unit, an acceleration sensing unit, and a sound sensing unit to collect multi-source signals from rotating mechanical components, and adopts a chiplet packaging form to achieve a highly integrated and miniaturized design.

[0010] Preferably, the signal processing module further includes a reference voltage setting unit, a DC bias circuit, a signal conditioning circuit, a synchronous acquisition control unit, a high-speed ADC converter, and a communication interface, used to preprocess, digitize, and perform data fusion analysis on the signals acquired by the multi-source sensing module, and to realize real-time data transmission.

[0011] Preferably, the reference voltage setting unit includes a high-precision voltage regulator and an operational amplifier. The high-precision voltage regulator receives a 5V power input and generates a stable initial reference voltage based on this power input. The operational amplifier is connected to the high-precision voltage regulator and is used to adjust and amplify the initial reference voltage to output a stable and adjustable reference voltage V. REF .

[0012] Preferably, the DC bias circuit includes an operational amplifier, two resistors R0 and R1, and a +2.5V DC voltage source. The original analog signal is connected to the inverting input of the operational amplifier through resistor R0, while the +2.5V DC voltage source is directly connected to the non-inverting input of the operational amplifier to provide the required DC bias level for the input signal. The feedback resistor R1 is connected between the output and inverting input of the operational amplifier to form a negative feedback path, so as to ensure stable operation of the circuit and perform appropriate gain adjustment on the signal.

[0013] Preferably, the signal conditioning circuit includes a differential amplifier and an external RC network, with its non-inverting input terminal connected to the positive input signal V through resistor R1 and capacitor C1. in+ The negative input signal V is connected through resistor R3. in-Simultaneously connect the reference voltage source V REF The inverting input terminal is connected to the output terminal V through a feedback network consisting of resistor R2 and capacitor C2. out Connected; the key parameters of this circuit satisfy the following formula: differential gain Cutoff frequency

[0014] Preferably, the synchronous acquisition and control unit uses PWM signals to achieve hardware-level time synchronization, and combines it with the real-time scheduling mechanism of the embedded operating system to ensure that data from different sources can be accurately aligned and processed synchronously.

[0015] Preferably, the main control unit uses an STM32H743 microcontroller to realize the synchronous acquisition of multi-channel signals, control the high-speed ADC converter to perform data conversion, and run a fusion algorithm to comprehensively analyze and process the acquired data, thereby making a comprehensive judgment and health assessment of the rotating machinery.

[0016] Preferably, the fusion analysis module is used to deploy the fusion algorithm in the main control unit, which can identify abnormal features in high-frequency signals, thereby realizing accurate monitoring and intelligent diagnosis of the equipment's operating status, and transmitting data in real time through the communication interface.

[0017] Preferably, the fusion analysis module employs a data fusion algorithm based on depthwise separable convolution and attention mechanism to perform feature extraction, information fusion, and pattern recognition on multi-source sensor data, thereby achieving high-precision judgment of the health status of the device.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] 1. This invention integrates a temperature sensing unit, an acceleration sensing unit, and a sound sensing unit based on MEMS technology and adopts a chiplet packaging form. In the face of complex working conditions, it can realize multi-dimensional and all-round evaluation of the device's operating status. At the same time, chiplet packaging not only helps to reduce the size of the device and reduce power consumption, but also improves the system integration and performance.

[0020] 2. This invention, through a rational circuit design, enables the reference voltage setting unit, DC bias circuit, and signal conditioning circuit to work collaboratively, jointly ensuring the accuracy and stability of signal acquisition. Specifically, the reference voltage setting unit provides a stable and reliable reference voltage to the system, ensuring that all functional modules operate under a unified reference; the DC bias circuit adjusts the input AC signal to a suitable level range for subsequent circuit processing, preventing saturation or distortion in the analog-to-digital converter (ADC); and the signal conditioning circuit effectively suppresses noise interference and improves the signal-to-noise ratio and signal integrity by filtering and amplifying the input signal. These three circuit components work closely together to form a complete signal preprocessing chain, providing strong support for achieving high-precision and high-reliability signal acquisition.

[0021] 3. This invention addresses the shortcomings of traditional condition monitoring systems in multi-parameter synchronous acquisition by utilizing PWM (Pulse Width Modulation) signals to achieve hardware-level time synchronization, ensuring accurate alignment and synchronized processing of data from different sensors. This solves the common timing inconsistency problem in multi-source data acquisition and improves the accuracy of data analysis.

[0022] 4. This invention utilizes a data fusion algorithm based on a depthwise separable convolutional module and an attention mechanism to significantly improve the fusion efficiency and feature extraction accuracy of multi-source sensor data while ensuring the model's lightweight nature. The depthwise separable convolutional module effectively reduces computational complexity, making it suitable for resource-constrained embedded scenarios. The attention mechanism dynamically adjusts the weights of different sensor features to achieve multi-source data fusion, thereby enhancing the ability to discriminate device status and ultimately achieving lower power consumption, higher diagnostic accuracy, and stronger adaptability to complex operating conditions.

[0023] 5. This invention integrates sensing acquisition, signal processing, and data analysis functions into a unified platform through an integrated hardware and software design. This not only improves the system's response speed and operating efficiency but also enhances the real-time performance and accuracy of multi-source data fusion, thereby enabling efficient monitoring and intelligent diagnosis of equipment status. Attached Figure Description

[0024] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0025] Figure 1 This is a module architecture diagram of the integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to the present invention;

[0026] Figure 2 This is a chiplet encapsulation structure diagram of a multi-source sensing module;

[0027] Figure 3 Set the unit circuit diagram for the reference voltage;

[0028] Figure 4 This is a diagram of a DC bias circuit.

[0029] Figure 5 This is a circuit diagram for signal conditioning.

[0030] Figure 6 Here is a flowchart of the ADC conversion process;

[0031] Figure 7 This is a structural diagram of a data fusion algorithm based on depthwise separable convolutional modules and attention mechanisms. Detailed Implementation

[0032] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0033] Specifically, this invention provides an integrated multi-sensor fusion system for monitoring the condition of rotating machinery components. During the operation of rotating machinery, this system can monitor its health status in real time. For example... Figure 1 As shown, the system includes a multi-source sensing module, a signal processing module, and a fusion analysis module. The multi-source sensing module includes multiple sensing units for collecting multi-dimensional data during the operation of rotating machinery. The signal processing module preprocesses and performs analog-to-digital conversion on the analog signals collected by each sensing unit in the multi-source sensing module. The signal processing module includes a main control unit, which performs comprehensive analysis and information fusion on the data output by each sensing unit through a fusion algorithm, and transmits the processing results in real time via a communication interface. By integrating multiple sensing units, the signal processing module, and the fusion algorithm into one system, deep hardware and software integration is achieved, ensuring stable system operation and the real-time performance and accuracy of rotating machinery component condition monitoring.

[0034] The multi-source sensing module includes a temperature sensing unit, an acceleration sensing unit, and a sound sensing unit, all of which are manufactured using MEMS technology. These sensor units are characterized by small size, light weight, low power consumption, and high reliability, minimizing the overall system size and making them suitable for applications in space-constrained and resource-limited environments. Figure 2As shown, in this embodiment, the temperature sensing unit, acceleration sensing unit, and sound sensing unit are integrated onto the same silicon substrate 4 using chiplet packaging technology, and electrical interconnection between the functional chips is achieved through microbumps 1. This structure effectively reduces wiring delay in traditional multi-chip packaging and improves the overall integration and space utilization of the module.

[0035] A through-silicon via (TSV)2 structure is integrated within the silicon substrate to enable vertical electrical signal transmission, improving communication efficiency and signal integrity between sensor chips and the underlying packaging substrate. Simultaneously, a redistribution layer (RDL) is disposed on the surface of the silicon substrate for flexible layout of lateral interconnects, further optimizing the electrical connections between functional modules and meeting complex signal routing requirements. The entire multi-source sensing module is interconnected with external circuits or the packaging substrate via the bottom C4 solder ball (Controlled Collapse Chip Connection)3, fulfilling functions such as power supply, data transmission, and mechanical support. This design constructs a compact, high-performance, and highly integrated multi-source sensing module suitable for miniaturized applications of high-density, multi-functional sensor systems, significantly improving the integration level and reliability of system-in-package.

[0036] The signal processing module includes a reference voltage setting unit, a DC bias circuit, a signal conditioning circuit, a synchronous acquisition control unit, a high-speed ADC converter, a main control unit, and a communication interface. It is used to preprocess, digitize, and perform data fusion analysis on the signals acquired by the multi-source sensing module, and to realize real-time data transmission.

[0037] The signal processing module, through a well-designed circuit, integrates a reference voltage setting unit, a DC bias circuit, and a signal conditioning circuit into a single high-precision, low-noise front-end processing unit. This unit filters, amplifies, adjusts the bias, and adapts the levels of the raw signals acquired by the multi-source sensing modules, effectively suppressing noise interference and improving the accuracy and stability of signal acquisition. Furthermore, the signal conditioning circuit includes a low-pass filter network and uses an operational amplifier structure to achieve controllable gain adjustment and DC bias setting, ensuring the output signal matches the input range of the subsequent high-speed analog-to-digital converter (ADC). The system also employs PWM control technology to achieve synchronous acquisition of signals from multiple sensors, ensuring consistency of data across all channels in the time dimension. In addition, a high-speed ADC converter is configured to efficiently and accurately convert the conditioned analog signals into digital signals, which are then transmitted to the main control unit via an SPI interface for feature extraction, status recognition, and fault diagnosis by subsequent algorithm modules.

[0038] like Figure 3 As shown, the reference voltage setting unit includes a high-precision voltage regulator and an operational amplifier. The input of the high-precision voltage regulator is connected to an external 5V power supply to receive this power input and generate a low-noise, high-stability initial reference voltage (+2.5V). The operational amplifier is connected to the output of the high-precision voltage regulator, forming a voltage follower or non-inverting amplifier circuit. This circuit buffers, adjusts, and amplifies the initial reference voltage, thereby outputting a stable and adjustable reference voltage signal V. REF .

[0039] like Figure 4 As shown, the DC bias circuit includes an operational amplifier, two resistors (R0 and R1), and a +2.5V DC voltage source. The original analog signal is connected to the inverting input of the operational amplifier through resistor R0, while the +2.5V DC voltage source is directly connected to the non-inverting input of the operational amplifier, providing the required DC bias level for the input signal. The feedback resistor R1 is connected between the output and inverting input of the operational amplifier, forming a negative feedback path to ensure stable circuit operation and appropriate signal gain adjustment. Through this structure, the circuit can shift the center level of the AC input signal to +2.5V, adapting it to the input voltage range of subsequent analog-to-digital converters or processing units.

[0040] like Figure 5 As shown, the signal conditioning circuit includes a differential amplifier and an external RC network. Its non-inverting input terminal is connected to the positive input signal V through resistor R1 and capacitor C1. in+ The negative input signal V is connected through resistor R3. in- Simultaneously connect the reference voltage source V REF The inverting input terminal is connected to the output terminal V through a feedback network consisting of resistor R2 and capacitor C2. out Connected; the key parameters of this circuit satisfy the following formula: differential gain Cutoff frequency

[0041]

[0042] The aforementioned synchronous acquisition and control unit utilizes PWM (Pulse Width Modulation) signals to achieve hardware-level time synchronization, and combines this with the real-time scheduling mechanism of the embedded operating system to ensure that data from different sources can be accurately aligned and processed synchronously.

[0043] Specifically, the time synchronization based on PWM signals includes the following process:

[0044] Frequency setting: The frequency of the PWM signal is kept consistent with the ADC sampling rate to ensure that only one complete PWM cycle is contained in each sampling period.

[0045] Duty cycle control: Set to a standard square wave format of 50%, which makes the rising and falling edges symmetrical, facilitating accurate time synchronization and measurement.

[0046] Phase alignment: Multiple PWM signals achieve strict phase alignment through a hardware synchronization mechanism, with channel jitter less than 1ns, ensuring high consistency of each channel in the time dimension.

[0047] like Figure 6 As shown, the high-speed ADC converter is used to convert synchronously acquired analog signals into digital signals, specifically including the following steps:

[0048] Initialization configuration phase: The ADC is set to external trigger operating mode, and its sampling rate is determined by the PWM signal frequency. Simultaneously, a programmable gain amplifier is configured according to the input signal range (±10V) to achieve dynamic range matching.

[0049] Trigger Response and Sampling Start Phase: The ADC uses the rising edge of the PWM signal as the conversion start command to ensure that each sampling occurs at a preset time point. After receiving the same trigger signal, all channels of the multi-channel ADC begin sampling in parallel, thereby eliminating timing deviations between channels and achieving true synchronous acquisition.

[0050] Signal Conversion and Data Output Stage: After the trigger signal starts, the ADC enters the conversion stage, completing the precise digitization of the analog signal. Upon completion of the conversion, the ADC outputs a data ready signal pulse to notify the main control unit to read the conversion result and perform subsequent data processing or transmission.

[0051] The main control unit uses an STM32H743 microcontroller to perform the following functions: synchronous acquisition of multi-channel signals, control of high-speed ADC converter for data conversion, and running fusion algorithm to comprehensively analyze and process the acquired data, thereby making a comprehensive judgment and health assessment of the rotating machinery's state.

[0052] The fusion analysis module is used to deploy the fusion algorithm in the main control unit, which can identify abnormal features in high-frequency signals, thereby achieving accurate monitoring and intelligent diagnosis of equipment operating status, and transmitting data in real time through a communication interface. The communication interface includes a Modbus-TCP interface and an RS485 interface, supporting real-time data transmission with the integrated multi-sensor fusion system.

[0053] like Figure 7As shown, the fusion analysis module uses a lightweight model with a depthwise separable convolution module and a self-attention mechanism for data fusion analysis. The model receives multi-source sensing data, uses FFT to extract high-frequency components in the data preprocessing stage, then performs local feature extraction, and combines the self-attention mechanism for global weighted fusion, finally outputting the operating status results of the rotating mechanical components.

[0054] The fusion analysis module of this invention deploys a trained multi-source data fusion model in the main control unit, serving as the core processing unit for intelligent analysis and state recognition. This model, trained based on historical sensor data, possesses the ability to efficiently identify the operating state of rotating machinery components. In specific implementation, the fusion analysis module extracts features from the digital signals received from the signal processing module, including time-domain statistical features, frequency-domain energy distribution, and time-frequency domain characteristic parameters. Subsequently, a multi-source data fusion algorithm is used to fuse information from data from multiple sensor channels, such as temperature, vibration, and acoustic emission, improving the accuracy and robustness of state recognition.

[0055] Multi-source data fusion is a key component in constructing the algorithm model. Feature extraction is performed on the data collected by the multi-source sensing module from three sensing units. For the temperature sensing unit: mean μ... A variance σ 2 A Accelerometer unit: peak frequency p B Frequency band energy E B Sound sensing unit: sound pressure level L c Spectral entropy H c Because the multi-source sensing module has 3 sensing units, N=3, and each sensing unit provides a feature vector f of different dimensions. i , where i = A, B, C are the temperature sensing unit, acceleration sensing unit, and sound sensing unit, respectively.

[0056] The eigenvector of the temperature sensing unit: f A (t)=[μ A (t),σ 2 A (t)]

[0057] eigenvector of the accelerometer unit: f B (t)=[p B (t),E B (t)]

[0058] Feature vector of the sound sensing unit: f C (t)=[L C (t),H C (t)]

[0059] The fused feature vector F(t) can then be expressed as:

[0060] F(t)=[μ A (t),σ 2 A (t),p B (t),E B (t),L C (t),H C (t)] T

[0061] In this invention, after deployment, the system is installed in key parts of rotating machinery (such as near bearings, shafts, or gearboxes) to ensure that each sensor can accurately collect multi-source sensing data such as temperature, vibration acceleration, and sound signals generated during operation. After power-on, the multi-source sensing modules begin to operate collaboratively. Each sensing unit is interconnected with the silicon substrate via micro-bumps to achieve high-precision signal acquisition. Simultaneously, the signal processing module performs noise suppression, amplification gain adjustment, and DC bias setting on the received raw analog signal, and provides a unified reference voltage through a reference voltage circuit to improve signal acquisition accuracy and stability.

[0062] After conditioning, the analog signal is digitized by a high-speed ADC converter and then controlled by a PWM pulse width modulation signal to achieve synchronous acquisition of multi-channel data, ensuring the consistency of data from each sensor in the time dimension. Once the digital signal is transmitted to the main control unit, the fusion analysis module employs a data fusion algorithm based on depthwise separable convolution and attention mechanisms to efficiently extract features, fuse information, and recognize patterns from the multi-source sensor data, thereby achieving a high-precision assessment of the equipment's health status. Finally, the diagnostic results are uploaded in real-time to a host computer or cloud platform via a communication interface, providing data support for equipment fault early warning and maintenance decisions.

[0063] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An integrated multi-sensor fusion system for monitoring the condition of rotating machinery components, characterized in that, The system includes a multi-source sensing module, a signal processing module, and a fusion analysis module. The multi-source sensing module comprises multiple sensing units for collecting multi-dimensional data during the operation of rotating machinery. The signal processing module preprocesses and performs analog-to-digital conversion on the analog signals collected by each sensing unit in the multi-source sensing module. The signal processing module includes a main control unit. The fusion analysis module includes a fusion algorithm. The main control unit performs comprehensive analysis and information fusion on the data output by each sensing unit through the fusion algorithm, and transmits the processing results in real time via a communication interface. By integrating multiple sensing units, the signal processing module, and the fusion algorithm into one system, deep hardware and software integration is achieved, ensuring the real-time performance and accuracy of the condition monitoring of rotating machinery components.

2. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 1, characterized in that, The multi-source sensing module includes a temperature sensing unit, an acceleration sensing unit, and a sound sensing unit. It collects multi-source signals from rotating mechanical components and uses a chiplet package to achieve a highly integrated and miniaturized design.

3. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 1, characterized in that, The signal processing module also includes a reference voltage setting unit, a DC bias circuit, a signal conditioning circuit, a synchronous acquisition control unit, a high-speed ADC converter, and a communication interface, which are used to preprocess, digitize, and perform data fusion analysis on the signals acquired by the multi-source sensing module, and realize real-time data transmission.

4. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 3, characterized in that, The reference voltage setting unit includes a high-precision voltage regulator and an operational amplifier. The high-precision voltage regulator receives a 5V power input and generates a stable initial reference voltage based on this input. The operational amplifier is connected to the high-precision voltage regulator and adjusts and amplifies the initial reference voltage to output a stable and adjustable reference voltage V. REF .

5. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 3, characterized in that, The DC bias circuit includes an operational amplifier, two resistors R0 and R1, and a +2.5V DC voltage source. The original analog signal is connected to the inverting input of the operational amplifier through resistor R0, while the +2.5V DC voltage source is directly connected to the non-inverting input of the operational amplifier to provide the required DC bias level for the input signal. The feedback resistor R1 is connected between the output and inverting input of the operational amplifier to form a negative feedback path, so as to ensure stable operation of the circuit and perform appropriate gain adjustment on the signal.

6. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 3, characterized in that, The signal conditioning circuit includes a differential amplifier and an external RC network. Its non-inverting input terminal is connected to the positive input signal V through resistor R1 and capacitor C1. in+ The negative input signal V is connected through resistor R3. in- Simultaneously connect the reference voltage source V REF The inverting input terminal is connected to the output terminal V through a feedback network consisting of resistor R2 and capacitor C2. out Connected; the key parameters of this circuit satisfy the following formula: differential gain Cutoff frequency 7. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 3, characterized in that, The aforementioned synchronous acquisition and control unit utilizes PWM signals to achieve hardware-level time synchronization, and combines this with the real-time scheduling mechanism of the embedded operating system to ensure that data from different sources can be accurately aligned and processed synchronously.

8. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 1, characterized in that, The main control unit uses an STM32H743 microcontroller to realize the synchronous acquisition of multi-channel signals, control the high-speed ADC converter to perform data conversion, and run a fusion algorithm to comprehensively analyze and process the acquired data, thereby making a comprehensive judgment and health assessment of the rotating machinery.

9. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 1, characterized in that, The fusion analysis module is used to deploy the fusion algorithm in the main control unit, which can identify abnormal features in high-frequency signals, thereby realizing accurate monitoring and intelligent diagnosis of the equipment's operating status, and transmitting data in real time through the communication interface.

10. The integrated multi-sensor fusion system for monitoring the condition of rotating machinery components according to claim 9, characterized in that, The fusion analysis module employs a data fusion algorithm based on depthwise separable convolution and attention mechanism to perform feature extraction, information fusion, and pattern recognition on multi-source sensor data, thereby achieving high-precision judgment of the health status of the device.

Images