Magnetic levitation bearing multi-sensor signal fusion system

Abstract

The application provides a kind of magnetic suspension bearing multi-sensor signal fusion system, comprising: signal acquisition pre-processing module, for the synchronous sampling and data windowing of each signal output by front-end circuit;Signal quality index calculation module, for calculating the comprehensive signal quality index according to the signal after data windowing;Adaptive weighted fusion module, for generating fusion feedback signal according to the comprehensive signal quality index;Excitation parameter self-tuning module, for adjusting the excitation parameter of front-end circuit according to the comprehensive signal quality index;System output monitoring module, for outputting fusion feedback signal, determining and outputting sensor health status based on comprehensive signal quality index.By constructing the closed-loop mechanism of signal quality evaluation and adaptive fusion, front-end parameter self-tuning, the problem of insufficient output signal reliability caused by signal quality degradation in the prior art is solved, the stability and environmental adaptability of the system are improved, and quantitative monitoring of the sensor health status is realized.

Description

Technical Field

[0001] This invention relates to the field of magnetic levitation control technology, and more specifically, to a multi-sensor signal fusion system for magnetic levitation bearings. Background Technology

[0002] As a high-performance support technology, the stable operation of magnetic levitation bearings highly depends on accurate and reliable rotor position feedback signals. To improve the system's fault tolerance and reliability, multiple redundant displacement sensors are typically deployed at the same measurement location. After acquiring signals from multiple sensors, signal fusion technology is needed to synthesize them into a single feedback signal for use by the main controller. Currently, the commonly used technical solutions are to perform simple averaging of the multiple signals or to fuse them using fixed weighting coefficients.

[0003] However, this fixed-weight fusion method inherently assumes that all sensors maintain good and consistent performance throughout the entire operating cycle. In real-world conditions, individual sensors may experience a decline in output signal quality due to aging, drift, or external electromagnetic interference, temperature changes, and other factors. In such cases, the fixed fusion strategy cannot identify and reduce the weight of inferior signals, which directly negatively impact the fusion result, thereby reducing the accuracy and reliability of the final feedback signal.

[0004] Furthermore, the signal processing flow of existing systems is typically open-loop. Key physical parameters such as the excitation frequency and amplitude of the front-end sensor circuits are fixed after system debugging. When the interference spectrum in the working environment changes, the system cannot actively adjust the excitation frequency to avoid interference bands, nor can it adaptively adjust the excitation amplitude according to changes in signal amplitude to maintain the optimal signal-to-noise ratio. This limits the system's adaptability to complex and variable operating conditions.

[0005] Meanwhile, existing technologies lack online, quantitative assessment methods for the health status of sensors themselves. System users find it difficult to know the performance degradation trend of individual sensors in real time, and problems are usually only discovered after the sensor has completely failed and the signal is abnormal. This is not conducive to predictive maintenance and ensuring the long-term stable operation of the system. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-sensor signal fusion system for magnetic levitation bearings, which solves the problem of insufficient reliability of output signals due to signal quality degradation in the prior art, and improves the stability and environmental adaptability of the system.

[0007] In a first aspect, the present invention provides a multi-sensor signal fusion system for magnetic levitation bearings, the system comprising: The signal acquisition and preprocessing module is used to synchronously sample and window the signals output from the front-end circuits of the multi-sensor of the magnetic levitation bearing. A signal quality index calculation module, which is connected to the signal acquisition and preprocessing module, is used to calculate a comprehensive signal quality index based on the signal after data windowing. An adaptive weighted fusion module, which is connected to the signal quality index calculation module and the signal acquisition preprocessing module, is used to generate a fusion feedback signal based on the comprehensive signal quality index; An excitation parameter self-tuning module, which is connected to the signal quality index calculation module, is used to adjust the excitation parameters of the front-end circuit according to the comprehensive signal quality index. The system output monitoring module is connected to the adaptive weighted fusion module and the signal quality index calculation module. It is used to output the fused feedback signal, determine the sensor health status based on the comprehensive signal quality index, and output the health status.

[0008] In an optional implementation, the signal quality index calculation module includes: The short-term signal variance index calculation unit is used to calculate the arithmetic mean of the signal in each window after data windowing, calculate the variance value based on the arithmetic mean and the discrete sampled value of the signal in the window, and map the variance value to a set interval to obtain the short-term signal variance index. The spectral purity index calculation unit is used to perform spectral transformation on the signals in each window after data windowing to obtain the discrete Fourier transform coefficients of the single-sided spectrum. Based on the discrete Fourier transform coefficients, the energy in the defined effective signal frequency band, and the total signal energy, the spectral purity index is calculated. The redundancy pair correlation index calculation unit is used to calculate the Pearson correlation coefficient based on the signal of the corresponding window after the data of the preset redundancy pair sensor in the multi-sensor, and to obtain the redundancy pair correlation index based on the Pearson correlation coefficient. The cross-axis state coherence index calculation unit is used to calculate the power spectral density and cross-power spectral density based on the signal after data windowing, calculate the coherence function between the power spectral density and cross-power spectral density, and obtain the cross-axis state coherence index based on the function value of the coherence function at the rotor frequency. The integrated SQI generation unit is used to weight and summarize the short-term signal variance index, the spectral purity index, the redundancy pair correlation index, and the cross-axis state coherence index to generate a comprehensive signal quality index.

[0009] In an optional implementation, the integrated SQI generation unit is used to generate the integrated signal quality index in the following manner: The short-term signal variance index, the spectral purity index, the redundancy pair correlation index, and the cross-axis state coherence index are linearly weighted and summed.

[0010] In an optional implementation, the adaptive weighted fusion module includes: A dynamic weight calculation unit is used to generate real-time fusion weights for each signal based on the comprehensive signal quality index. The weighted fusion execution unit is used to perform weighted summation of each signal according to the real-time fusion weight to generate a fusion feedback signal.

[0011] In an optional implementation, the dynamic weight calculation unit is used to generate the real-time fusion weights in the following manner: Calculate the p-th power of the comprehensive signal quality index of each signal to obtain the calculation result, where p is a weighting index greater than or equal to 1; The calculation results of each signal are normalized to obtain the real-time fusion weight; In an optional implementation, the excitation parameter self-tuning module includes: An excitation frequency self-tuning unit is used to adjust the excitation frequency of the front-end circuit according to the spectral purity index output by the signal quality index calculation module. The excitation amplitude self-tuning unit is used to monitor the signal amplitude of each signal to adjust the excitation amplitude of the front-end circuit.

[0012] In an optional implementation, the excitation frequency self-tuning unit is used to adjust the excitation frequency of the front-end circuit in the following manner: The spectral purity index is compared with a preset frequency adjustment trigger threshold. When the spectral purity index is lower than the frequency adjustment trigger threshold, a new excitation frequency is generated based on the current frequency and a preset frequency step size. The digital control word corresponding to the new excitation frequency is sent to the digitally controllable front-end demodulation circuit through the control bus to adjust the excitation frequency of the front-end circuit.

[0013] In an optional implementation, the excitation amplitude self-tuning unit is used to monitor the signal amplitude of each signal to adjust the excitation amplitude of the front-end circuit in the following manner: The signal amplitude of each signal is monitored to obtain the peak value, and the peak value is compared with the preset upper amplitude threshold and lower amplitude threshold. When the peak value exceeds the upper limit threshold of the amplitude, the current excitation amplitude is reduced proportionally to obtain a new excitation amplitude; when the peak value is less than the lower limit threshold of the amplitude, the current excitation amplitude is increased proportionally to obtain a new excitation amplitude. The digital control word corresponding to the new excitation amplitude is sent to the digitally controllable front-end demodulation circuit to adjust the excitation amplitude of the front-end circuit.

[0014] In an optional implementation, the system output monitoring module includes: A feedback signal output unit is used to convert the fused feedback signal into a standard industrial signal; A health status monitoring unit is used to determine the sensor health status based on the comprehensive signal quality index and output the sensor health status.

[0015] In an optional implementation, the health status monitoring unit is used to determine the sensor health status based on the comprehensive signal quality index in the following manner: The comprehensive signal quality index is compared with preset warning thresholds and fault thresholds to obtain the comparison results; The sensor health status is determined based on the comparison results, and the health status is classified as healthy, warning, or faulty.

[0016] Compared with existing technologies, this invention provides a multi-sensor signal fusion system for magnetic levitation bearings. This fusion system calculates the comprehensive signal quality index of each signal and dynamically generates fusion weights based on the index for adaptive weighted fusion, thereby improving the reliability and stability of the fusion feedback signal. It also solves the technical problem in traditional fixed-weight fusion methods where the quality of some sensor signals degrades, thus polluting the final output signal.

[0017] Furthermore, by monitoring the comprehensive signal command indicators and adjusting the excitation parameters of the front-end circuit accordingly, this fusion system constructs a closed-loop self-tuning mechanism from signal quality assessment to front-end parameter optimization. This achieves the technical effect of improving signal acquisition quality from the source and enhancing the system's environmental adaptability, and solves the technical problem that existing systems cannot actively avoid interference or adapt to changes in operating conditions due to fixed acquisition parameters.

[0018] Furthermore, this fusion system outputs comprehensive signal quality indicators as data characterizing the health status of sensors, achieving the technical effect of quantitative monitoring and early warning of the health status of individual sensors. This solves the technical problems of traditional methods being unable to assess sensor health status online and providing data support for predictive maintenance. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the system architecture of the fusion system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the execution logic of each module in the fusion system provided in the embodiments of the present invention; Figure 3 This is a structural block diagram of the signal quality index calculation module provided in an embodiment of the present invention; Figure 4 This is a structural block diagram of the adaptive weighted fusion module provided in an embodiment of the present invention; Figure 5 This is a structural block diagram of the excitation parameter self-tuning module provided in an embodiment of the present invention; Figure 6 This is a structural block diagram of the system output monitoring module provided in an embodiment of the present invention.

[0021] Icons: 101 - Signal Acquisition and Preprocessing Module; 102 - Signal Quality Index Calculation Module; 1021 - Short-Term Signal Variance Index Calculation Unit; 1022 - Spectral Purity Index Calculation Unit; 1023 - Redundancy Pair Correlation Index Calculation Unit; 1024 - Cross-Axis State Coherence Index Calculation Unit; 1025 - Comprehensive SQI Generation Unit; 103 - Adaptive Weighted Fusion Module; 1031 - Dynamic Weight Calculation Unit; 1032 - Weighted Fusion Execution Unit; 104 - Excitation Parameter Self-Tuning Module; 1041 - Excitation Frequency Self-Tuning Unit; 1042 - Excitation Amplitude Self-Tuning Unit; 105 - System Output Monitoring Module; 1051 - Feedback Signal Output Unit; 1052 - Health Status Monitoring Unit. Detailed Implementation

[0022] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention.

[0023] Please see Figure 1 The diagram below shows the system architecture of the multi-sensor signal fusion system for magnetic levitation bearings provided in this embodiment of the invention. The fusion system includes a signal acquisition and preprocessing module 101, a signal quality index calculation module 102, an adaptive weighted fusion module 103, an excitation parameter self-tuning module 104, and a system output monitoring module 105.

[0024] The signal quality index calculation module 102 is connected to the signal acquisition and preprocessing module 101; the adaptive weighted fusion module 103 is connected to both the signal quality index calculation module 102 and the signal acquisition and preprocessing module 101; the excitation parameter self-tuning module 104 is connected to the signal quality index calculation module 102; and the system output monitoring module 105 is connected to both the adaptive weighted fusion module 103 and the signal quality index calculation module 102.

[0025] The signal acquisition and preprocessing module 101 is used to synchronously sample and window the signals output from the front-end circuit of the magnetic levitation bearing multi-sensor; the signal quality index calculation module 102 is used to calculate the comprehensive signal quality index based on the signal after data windowing; the adaptive weighted fusion module 103 is used to generate a fusion feedback signal based on the comprehensive signal quality index; the excitation parameter self-tuning module 104 is used to adjust the excitation parameters of the front-end circuit based on the comprehensive signal quality index; and the system output monitoring module 105 is used to output the fusion feedback signal, determine the sensor health status based on the comprehensive signal quality index, and output the health status.

[0026] Please refer to the following: Figure 2 The specific operating procedures of each of the above modules are as follows: S11, the signal acquisition and preprocessing module 101 performs synchronous sampling and data windowing on the signals output from the front-end circuit of the magnetic levitation bearing multi-sensor. S12, the signal quality index calculation module 102 calculates the comprehensive signal quality index based on the signal after data windowing; S13, the adaptive weighted fusion module 103 generates a fusion feedback signal based on the comprehensive signal quality index; S14, the excitation parameter self-tuning module 104 adjusts the excitation parameters of the front-end circuit according to the comprehensive signal quality index; S15, the system output monitoring module 105 outputs a fused feedback signal, determines the sensor health status based on the comprehensive signal quality index, and outputs the health status.

[0027] The above process is executed cyclically at predetermined intervals.

[0028] Specifically, in this embodiment, the signal acquisition and preprocessing module 101 is responsible for converting multiple analog signals from the digitally controllable front-end demodulation circuit into a data format suitable for subsequent digital signal processing. The signal acquisition and preprocessing module 101 synchronously samples and windowes the signals output from the front-end circuit of the magnetic levitation bearing multi-sensor, which can be achieved in the following ways: First, the multiple analog signals are synchronously digitized. Each signal from the redundant sensor probe group includes the original demodulated voltage signal, denoted as the analog signal. The analog signal is input to a multi-channel analog-to-digital converter (ADC). This ADC uses a common sampling clock to ensure that the sampling times of all channels are strictly aligned; this is synchronous sampling. Synchronous sampling maintains the original phase and time relationships between the different sensor signals, which is the basis for subsequent redundancy pair correlation and cross-axis state coherence analysis.

[0029] Sampling frequency The selection follows the Nyquist sampling theorem, and its value is set to the highest effective frequency component in the signal under test. More than twice the length of the analog signal to avoid signal aliasing. This digitization process converts the continuous analog signal... Convert to discrete-time signal sequence This process can be characterized as follows: ; In the formula, For the first Discrete sampled values ​​of the road sensor signal; For sampling point index; The sampling period is, and ; For the first The raw analog signal from the road sensor.

[0030] Then, the discrete-time signal sequence is windowed. Since the statistical characteristics of the sensor signal change with operating conditions, the system employs a short-time analysis method. The signal acquisition and preprocessing module 101 processes the continuous discrete signal sequence... The data is divided into fixed-length frames, also known as analysis windows. The length of each analysis window is... One sampling point.

[0031] Window length The choice of window is a key parameter, determining the frequency resolution of subsequent frequency domain analysis and the response speed of time domain analysis. A longer window provides higher frequency resolution but is less sensitive to instantaneous changes in signal characteristics; a shorter window, on the other hand, does not. The value is preset based on the dynamic response characteristics of the magnetic levitation bearing system.

[0032] To ensure continuous tracking of signal characteristic changes and avoid loss of critical transient information due to window boundary cutting, a length of [missing information] is set between adjacent analysis windows. The overlap of sampling points. The overlap rate can be set to, for example, around 50%. In this way, the signal acquisition preprocessing module 101 continuously outputs a series of data windows, each containing... The data from each sampling point is used as the standard input for the subsequent signal quality index calculation module 102.

[0033] In some embodiments, to suppress spectral leakage caused by data truncation, a window function, such as a Hanning window or a Hamming window, can be applied to each data window before subsequent spectral analysis. The selection and application of window functions are well-known techniques in the art and will not be elaborated upon here.

[0034] The signal quality index calculation module 102 receives the data window from the signal acquisition and preprocessing module 101 and calculates different quality dimensions of each signal.

[0035] Please see Figure 3 The signal quality index calculation module 102 includes: a short-term signal variance index calculation unit 1021, a spectral purity index calculation unit 1022, a redundancy pair correlation index calculation unit 1023, a cross-axis state coherence index calculation unit 1024, and a comprehensive SQI generation unit 1025. These units work in parallel, calculating different quality dimensions of each signal, and are weighted and summarized by the comprehensive SQI generation unit 1025 into a comprehensive signal quality index (SQI).

[0036] The signal quality index calculation module 102 calculates the intrinsic characteristic index of each signal in the data window through the short-term signal variance index calculation unit 1021 and the spectrum purity index calculation unit 1022.

[0037] The short-term signal variance index calculation unit 1021 is used to calculate the arithmetic mean of the signal in each window after data windowing. The variance value is calculated based on the arithmetic mean and the discrete sampled value of the signal in the window. The variance value is then mapped to a set interval to obtain the short-term signal variance.

[0038] The function of the short-term signal variance index calculation unit 1021 is to quantify the degree of random fluctuation of the signal within the current time window, which is directly related to the broadband random noise level in the signal.

[0039] Specifically, the short-term signal variance calculation unit 1021 obtains the short-term signal variance in the following manner: First, calculate the statistical variance of the signal window. For the first... Discrete signal sequence window of road sensor First, calculate its arithmetic mean. Then calculate the variance of the data within that window. The calculation formula is as follows: ; In the formula, For the first Variance of the road signal; The length of the analysis window; For the first Discrete sampled values ​​of the signal path; For signal window The mean.

[0040] Then, the variance values ​​are mapped to a normalized quality metric. Since a smaller variance value inherently indicates higher signal quality, to standardize the metric, it needs to be converted into an metric where a larger value represents higher quality. This embodiment uses an exponential mapping function to achieve this transformation, mapping the variance value to the interval (0, 1).

[0041] ; In the formula, For the first The short-term signal variance index of the road signal; It is a preset positive calibration coefficient used to adjust the sensitivity of the index to noise variance; For the first Variance of the road signal.

[0042] The spectral purity index calculation unit 1022 is used to perform spectral transformation on the signals in each window after data windowing to obtain the discrete Fourier transform coefficients of the single-sided spectrum. Based on the discrete Fourier transform coefficients, the energy in the defined effective signal frequency band, and the total signal energy, the spectral purity is calculated.

[0043] The function of the spectral purity index calculation unit 1022 is to evaluate the signal quality from the frequency domain perspective, especially to evaluate the energy ratio of the effective signal components, in order to characterize the degree to which the signal is affected by out-of-band noise and electromagnetic interference at specific frequencies.

[0044] Specifically, the spectral purity index calculation unit 1022 is used to obtain spectral purity in the following manner: First, perform a spectral transformation on the signal window. For the... The signal window of the road sensor Perform a Fast Fourier Transform (FFT) to obtain the Discrete Fourier Transform coefficients of its one-sided spectrum. The FFT algorithm is a well-known technique in this field and will not be elaborated upon here.

[0045] Then, the ratio of effective bandwidth energy to total energy is calculated. Based on the sensor's physical operating principle and signal modulation method, an effective signal bandwidth is predefined, with a frequency index range of [missing information]. This frequency band is typically a bandwidth centered on the excitation frequency. The ratio of energy within this frequency band to the total signal energy is calculated; this ratio is the spectral purity index. The calculation formula is as follows: ; In the formula, For the first The spectral purity index of the signal path; For signal The discrete Fourier transform coefficients; This is the starting frequency index of the effective signal bandwidth; This is the index of the termination frequency of the effective signal band; This represents the length of the analysis window. A value close to 1 indicates that the signal energy is mainly concentrated within the effective frequency band, resulting in high signal purity.

[0046] The signal quality index calculation module 102 further calculates indices based on multi-signal correlation through the redundant correlation index calculation unit 1023 and the cross-axis state coherence index calculation unit 1024. These indices evaluate the reliability of a single signal by comparing the relationships between different signals.

[0047] The redundant pair correlation index calculation unit 1023 is used to calculate the Pearson correlation coefficient based on the signal of the corresponding window after the data of the preset redundant pair sensor in the multi-sensor, and to obtain the redundant pair correlation based on the Pearson correlation coefficient.

[0048] The function of the redundancy correlation index calculation unit 1023 is to measure the consistency of the signal characteristics in the time domain between redundant sensor pairs configured at the same measurement point. High consistency indicates that the sensor pairs are in coordinated working state, while low consistency may indicate that one or more sensors are experiencing drift, failure, or other faults.

[0049] Specifically, the redundancy pair correlation index is obtained by the following method: First, calculate the Pearson correlation coefficient between redundant sensor pairs. For the first... The signal window of the road sensor and its preset redundancy pairs of sensors signal window Calculate the correlation coefficient between the two. .

[0050] ; In the formula, For sensors and Pearson correlation coefficient between signal windows; The length of the analysis window; and The first Road and the Discrete sampled values ​​of the signal path; and Signal windows and The mean.

[0051] Then, a redundancy pair correlation index is generated. This index... The absolute value of the correlation coefficient is directly used to make the index range fall within the [0, 1] interval. The larger the value, the stronger the correlation.

[0052] ; In the formula, For the first Redundancy of the signal path and correlation index; For sensors and Pearson correlation coefficient between signal windows.

[0053] In addition, the cross-axis state coherence index calculation unit 1024 is used to calculate the power spectral density and cross-power spectral density based on the signal after data windowing, calculate the coherence function between the power spectral density and cross-power spectral density, and obtain the cross-axis state coherence based on the function value of the coherence function at the rotor frequency.

[0054] The function of the cross-axis state coherence index calculation unit 1024 is to verify the physical self-consistency of the current signal by evaluating the degree of coupling of orthogonal axial signals at a specific frequency based on the rotor dynamics model. For rotating machinery, the vibrations in its orthogonal directions (such as the X-axis and Y-axis) have an inherent physical correlation at the rotational frequency.

[0055] Specifically, the cross-axis state coherence index calculation unit 1024 obtains the cross-axis state coherence in the following manner: First, calculate the power spectral density and cross-power spectral density of the signal. For the... Road signal window (e.g., from the X-axis), and a reference signal along an orthogonal axis (e.g., a fused signal along the Y-axis), calculate their power spectral densities. and and their cross-power spectral density The power spectral density can be estimated using methods such as the Welch method, which are well-known techniques in the field and will not be elaborated upon here.

[0056] Then, the coherence function is calculated. The coherence function characterizes the degree of linear correlation between two signals in the frequency domain.

[0057] ; In the formula, For signal The fused signal with the Y-axis at frequency The coherence function value at that location; Let be the cross-power spectral density between the two; and These are the power spectral densities of the two, respectively.

[0058] Finally, the cross-axis state coherence index is extracted. This index... The coherence function at rotor frequency The specific value at that location. This frequency... It is a known system operating parameter.

[0059] ; In the formula, For the first Cross-axis state coherence index of road signals; The coherence function at rotor frequency The value at that point. This value is close to 1, indicating that the current signal is highly consistent with the system's dynamic behavior at key physical frequencies.

[0060] Based on the above, the SQI generation unit 1025 receives the outputs from the aforementioned index calculation units, including the short-term signal variance index. Spectral purity index Redundancy in correlation indicators and cross-axis state coherence index The function of the integrated SQI generation unit 1025 is to combine these four normalized sub-indices, which evaluate signal quality from different dimensions, into a single index that can comprehensively characterize the signal quality. Comprehensive signal quality index of the reliability of the road sensor signal within the current data window .

[0061] The SQI generation unit 1025 is used to generate a comprehensive signal quality index in the following ways: The short-term signal variance, spectral purity, redundancy pair correlation, and cross-axis state coherence are linearly weighted and summed, specifically, this is achieved in the following way: This embodiment uses a linear weighting method to generate the comprehensive signal quality index. This method assigns a preset weight coefficient to each sub-index, and the magnitude of the weight coefficient reflects the importance of the corresponding sub-index in the comprehensive evaluation.

[0062] The formula for calculating the overall signal quality index is as follows: ; In the formula, For the first The overall signal quality index of the road signal; For the first The short-term signal variance of the road signal; For the first The spectral purity of the signal; For the first Redundancy of the road signal versus correlation; For the first Cross-axis state coherence of the road signal; , , , These are the weighting coefficients for the four corresponding sub-indicators.

[0063] Weighting coefficient , , , All are non-negative real numbers, and their sum is 1, that is... This constraint ensures that, since all sub-indicators have been normalized to the [0, 1] interval, the final generated comprehensive signal quality index is accurate. It will also fall within this range, which facilitates subsequent processing and threshold determination.

[0064] The weighting coefficients can be configured based on the specific application scenario of the magnetic levitation bearing system and the sensitivity requirements for different failure modes. For example, in applications with complex electromagnetic environments, the weighting coefficients can be appropriately increased based on spectral purity. Corresponding weight coefficients The value; and in scenarios where the gradual degradation of the sensor's own physical performance is a concern, it can improve the short-term signal variance. and redundancy on correlation Corresponding weight coefficients and The value of .

[0065] After the calculation is completed, the SQI generation unit 1025 will generate the comprehensive signal quality index of each signal. (in As a vector or dataset, the data is output to the adaptive weighted fusion module 103 and the excitation parameter self-tuning module 104.

[0066] The adaptive weighted fusion module 103 receives the comprehensive signal quality index of each signal from the signal quality index calculation module 102. and the corresponding raw signal data window from the signal acquisition and preprocessing module 101. The function of the adaptive weighted fusion module 103 is to dynamically allocate the contribution of each signal to the final output signal based on the real-time quality of each signal, thereby generating a fusion feedback signal with optimal quality.

[0067] Please see Figure 4The adaptive weighted fusion module 103 includes a dynamic weight calculation unit 1031 and a weighted fusion execution unit 1032. The dynamic weight calculation unit 1031 is used to generate real-time fusion weights for each signal based on the comprehensive signal quality index; the weighted fusion execution unit 1032 is used to perform weighted summation on each signal based on the real-time fusion weights to generate a fusion feedback signal.

[0068] The dynamic weight calculation unit 1031 is used to generate real-time fusion weights in the following manner: Calculate the p-th power of the comprehensive signal quality index of each signal to obtain the calculation result, where p is a weight exponent greater than or equal to 1; normalize the calculation results of each signal to obtain the real-time fusion weight.

[0069] Specifically, firstly, the dynamic weight calculation unit 1031 calculates the weight based on the comprehensive signal quality index of each signal. Calculate its normalized fusion weights Weight The value and The values ​​are positively correlated, meaning that the higher the signal quality, the greater its proportion in the fusion.

[0070] In this embodiment, to enhance the weight of high-quality signals and suppress the influence of low-quality signals, the weight calculation employs a normalization method with exponential adjustment: ; In the formula, For the first Normalized real-time fusion weights for road signals; For the first The overall signal quality index of the road signal; The weighting index is a preset constant greater than or equal to 1, used to adjust the non-linearity of weight allocation. The larger the value, the more significant the weight difference between high-quality and low-quality signals; This represents the total number of sensors participating in the fusion.

[0071] Based on this, the weighted fusion execution unit 1032 uses the real-time fusion weights generated by the dynamic weight calculation unit 1031. For each signal data window Perform weighted summation to generate the fused signal sequence. .

[0072] ; In the formula, The fused signal sequence (fused feedback signal); For the first Normalized fusion weights for road signals; For the first Discrete sampled values ​​of the signal path; This represents the total number of sensors participating in the fusion.

[0073] Through the above steps, the adaptive weighted fusion module 103 can integrate information from multiple sensors in real time. When a sensor signal's quality index is affected by interference, drift, or malfunction... When it decreases, its corresponding weight This signal will be automatically lowered to reduce the negative impact of the poor-quality signal on the final fusion result, ensuring the stability and reliability of the output signal. The final generated fused signal sequence... It is transmitted to the system output monitoring module 105.

[0074] In addition, the excitation parameter self-tuning module 104 receives specific sub-indices from the signal quality index calculation module 102. Its function is to construct a closed-loop control link from the a posteriori evaluation of signal quality to the feedforward optimization of front-end physical acquisition parameters, thereby actively improving the source quality of signal acquisition.

[0075] Please see Figure 5 The excitation parameter self-tuning module 104 includes an excitation frequency self-tuning unit 1041 and an excitation amplitude self-tuning unit 1042.

[0076] The excitation frequency self-tuning unit 1041 is used to adjust the excitation frequency of the front-end circuit according to the spectral purity index output by the signal quality index calculation module 102; the excitation amplitude self-tuning unit 1042 is used to monitor the signal amplitude of each signal to adjust the excitation amplitude of the front-end circuit.

[0077] The excitation frequency self-tuning unit 1041 aims to avoid strong electromagnetic interference bands in the environment by adjusting the excitation frequency of the front-end circuit, which can be achieved in the following ways: The spectral purity index is compared with the preset frequency adjustment trigger threshold. When the spectral purity index is lower than the frequency adjustment trigger threshold, a new excitation frequency is generated based on the current frequency and the preset frequency step size. The digital control word corresponding to the new excitation frequency is sent to the digital controllable front-end demodulation circuit through the control bus to adjust the excitation frequency of the front-end circuit.

[0078] Specifically, firstly, the excitation frequency self-tuning unit 1041 continuously monitors the spectral purity index of all channel signals. Adjust the trigger threshold of this indicator with a preset frequency. Compare them.

[0079] When the spectral purity index of any channel The value is below the threshold At that time, the system determines the current excitation frequency. External interference may have occurred. In this case, the excitation frequency self-tuning unit 1041 initiates the frequency adjustment program, generating a new candidate excitation frequency. In one specific embodiment, the new excitation frequency is generated by adding a preset frequency step size to the current frequency. To achieve this, that is .

[0080] The excitation frequency self-tuning unit 1041 will represent the new excitation frequency. The digital control word is sent to the digitally controllable front-end demodulation circuit via the control bus, so that it updates the frequency of the excitation signal.

[0081] The excitation amplitude self-tuning unit 1042 aims to maintain the amplitude of the sensor output signal within an optimal dynamic range by adjusting the excitation amplitude of the front-end circuit. This ensures a sufficient signal-to-noise ratio while avoiding saturation distortion of the analog-to-digital converter due to excessive signal strength. This can be achieved through the following methods: The signal amplitude of each signal is monitored to obtain the peak value. The peak value is compared with the preset upper and lower amplitude thresholds. When the peak value exceeds the upper amplitude threshold, the current excitation amplitude is reduced proportionally to obtain a new excitation amplitude. When the peak value is less than the lower amplitude threshold, the current excitation amplitude is increased proportionally to obtain a new excitation amplitude. The digital control word corresponding to the new excitation amplitude is sent to the digital controllable front-end demodulation circuit to adjust the excitation amplitude of the front-end circuit.

[0082] Specifically, the excitation amplitude self-tuning unit 1042 monitors each signal window. peak , or root mean square value.

[0083] The detected signal peak Compared with the preset amplitude upper limit threshold and amplitude lower limit threshold The two thresholds are compared. They define the optimal operating range for the ADC input.

[0084] like This indicates that the signal amplitude is too large, posing a risk of saturation, and the excitation amplitude needs to be reduced. If This indicates that the signal amplitude is too small, and the signal-to-noise ratio may be poor, requiring an increase in the excitation amplitude. New excitation amplitude The following proportions can be adjusted for calculation: ; In the formula, The new incentive amount has been adjusted. This represents the current incentive level; For a target peak voltage, its value is typically set at... and between; This represents the peak voltage measured within the current signal window.

[0085] After the calculation is completed, the excitation amplitude self-tuning unit 1042 generates a new excitation amplitude. The digital control word is sent to the digitally controllable front-end demodulation circuit to update the amplitude of the excitation signal.

[0086] Through the collaborative work of the two units, the system can proactively adjust the parameters of the front-end physical layer based on real-time signal quality feedback, thereby continuously optimizing the acquisition quality of the original signal under changing operating conditions and electromagnetic environments.

[0087] Please see Figure 6 The system output monitoring module 105 includes a feedback signal output unit 1051 and a health status monitoring unit 1052.

[0088] The feedback signal output unit 1051 is used to convert the fused feedback signal into a standard industrial signal; the health status monitoring unit 1052 is used to determine the sensor health status based on the comprehensive signal quality index and output the sensor health status.

[0089] Specifically, the feedback signal output unit 1051 in the system output monitoring module 105 receives the fused signal sequence from the adaptive weighted fusion module 103. The function of the feedback signal output unit 1051 is to convert this internal digital signal into a standard industrial signal that can be directly used by the external magnetic levitation bearing main controller.

[0090] In one embodiment, the output signal is a standard 4-20mA analog current signal. The feedback signal output unit 1051 is used for conversion processing in the following manner: Extract an instantaneous value representing the current state from the fused signal sequence. Because... It is a sequence within a data window, and the last sampling point of that window is usually selected. As the latest and most representative digital output value of the current physical quantity, it is denoted as... .

[0091] digital quantity Convert to corresponding physical displacement values This process is achieved through a predefined linear relationship.

[0092] ; In the formula, This represents the converted physical displacement value; Output the value for the selected digital quantity; This represents the sensitivity coefficient of the sensor. This is the zero-point offset for the sensor.

[0093] physical displacement value Mapped to a current value of 4-20mA This mapping relationship defines the effective measurement range of the sensor. Linearly corresponds to the current range of [4mA, 20mA].

[0094] ; In the formula, This is the final output current value, in milliamperes (mA). The physical displacement value to be converted; This represents the upper limit of the sensor's measurement range; This is the lower limit of the sensor's measurement range.

[0095] This current value is generated using a digital-to-analog converter (DAC) and a voltage-to-current (V / I) converter. The calculated target current value... First, the signal is converted into a corresponding digital control word, which is then input to the DAC to generate a proportional voltage signal. This voltage signal then drives a V / I conversion circuit, ultimately producing a stable and precise 4-20mA current at the output. The specific implementation of the DAC and V / I conversion circuit is well-known in the field and will not be described in detail here.

[0096] In another embodiment, the output signal is a digital communication protocol signal, such as a CAN bus signal or an EtherCAT message. In this case, the feedback signal output unit 1051 will output the calculated physical displacement value. The data is encapsulated according to the data frame format of a specific industrial bus protocol to form a data message, which is then sent to the main controller through the corresponding bus transceiver.

[0097] The health status monitoring unit 1052 in the system output monitoring module 105 receives real-time comprehensive signal quality indicators from various sensors in the signal quality indicator calculation module 102. The function of the health status monitoring unit 1052 is to assess and classify the health status of each sensor based on this quality index, and generate status information that can be used by the upper-level monitoring system, thereby providing data support for sensor fault diagnosis and predictive maintenance. The health status monitoring unit 1052 determines the sensor health status based on the comprehensive signal quality index in the following ways: The comprehensive signal quality index is compared with the preset warning threshold and fault threshold to obtain the comparison result; the sensor health status is determined based on the comparison result, and the health status is divided into healthy, warning or fault.

[0098] Specifically, the health status monitoring unit 1052 performs threshold judgments on the comprehensive signal quality indicators to classify health status. The health status monitoring unit 1052 internally presets at least two thresholds: a warning threshold... and a fault threshold ,and By real-time The value is compared with these two thresholds to classify the sensor’s health status into discrete levels, such as “healthy”, “warning”, and “faulty”.

[0099] like If the sensor is healthy, then its health status is determined to be "healthy".

[0100] like If the sensor health status is "warning", it indicates that the sensor signal quality has deteriorated and there may be signs of early failure.

[0101] like If the sensor's health status is "faulty," it indicates that the sensor signal quality has severely deteriorated and its data is unreliable.

[0102] The health status monitoring unit 1052 can also analyze the historical trends of comprehensive signal quality indicators to achieve early warning. To identify gradual degradation of sensors, the health status monitoring unit 1052 records data from each sensor in an internal memory. The historical sequence of values. This is achieved by analyzing data within a recent time window (e.g., the past few hundred data points). Historical data is analyzed to calculate its changing trends. In one specific embodiment, this can be achieved by calculating the slope of a linear regression of the data sequence within that time window. To quantify this trend, a consistently negative trend with an absolute value exceeding a preset trend threshold is defined. slope This indicates that the sensor's performance is continuously deteriorating, even with its current... The value is still in the "healthy" range, and the system can issue an early warning.

[0103] Finally, the health status information is encapsulated and output. The health status monitoring unit 1052 integrates the real-time signal quality indicators from various sensors. The determined health status level, and optional trend analysis results. Each quality sub-indicator is encapsulated into a structured data message. This message is sent to a host computer or equipment maintenance system via a standard communication interface (such as Modbus, CAN, or Ethernet). The format of this data message may include the sensor channel number, timestamp, health status code, and specific quality indicator values. After receiving this information, the upper-level system can perform real-time display, historical data recording, and alarm processing. The specific implementation of the communication protocol is well-known in the field and will not be elaborated here.

[0104] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0105] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0106] Furthermore, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0107] It should be noted that if the functionality is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] The above are merely embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-sensor signal fusion system for magnetic levitation bearings, characterized in that, The system includes: The signal acquisition and preprocessing module is used to synchronously sample and window the signals output from the front-end circuits of the multi-sensor of the magnetic levitation bearing. A signal quality index calculation module, which is connected to the signal acquisition and preprocessing module, is used to calculate a comprehensive signal quality index based on the signal after data windowing. An adaptive weighted fusion module, which is connected to the signal quality index calculation module and the signal acquisition preprocessing module, is used to generate a fusion feedback signal based on the comprehensive signal quality index; An excitation parameter self-tuning module, which is connected to the signal quality index calculation module, is used to adjust the excitation parameters of the front-end circuit according to the comprehensive signal quality index. The system output monitoring module is connected to the adaptive weighted fusion module and the signal quality index calculation module. It is used to output the fused feedback signal, determine the sensor health status based on the comprehensive signal quality index, and output the health status.

2. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 1, characterized in that, The signal quality index calculation module includes: The short-term signal variance index calculation unit is used to calculate the arithmetic mean of the signal in each window after data windowing, calculate the variance value based on the arithmetic mean and the discrete sampled value of the signal in the window, and map the variance value to a set interval to obtain the short-term signal variance index. The spectral purity index calculation unit is used to perform spectral transformation on the signals in each window after data windowing to obtain the discrete Fourier transform coefficients of the single-sided spectrum. Based on the discrete Fourier transform coefficients, the energy in the defined effective signal frequency band, and the total signal energy, the spectral purity index is calculated. The redundancy pair correlation index calculation unit is used to calculate the Pearson correlation coefficient based on the signal of the corresponding window after the data of the preset redundancy pair sensor in the multi-sensor, and to obtain the redundancy pair correlation index based on the Pearson correlation coefficient. The cross-axis state coherence index calculation unit is used to calculate the power spectral density and cross-power spectral density based on the signal after data windowing, calculate the coherence function between the power spectral density and cross-power spectral density, and obtain the cross-axis state coherence index based on the function value of the coherence function at the rotor frequency. The integrated SQI generation unit is used to weight and summarize the short-term signal variance, the spectral purity, the redundancy pair correlation, and the cross-axis state coherence to generate a comprehensive signal quality index.

3. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 2, characterized in that, The integrated SQI generation unit is used to generate the integrated signal quality index in the following manner: The short-term signal variance index, the spectral purity index, the redundancy pair correlation index, and the cross-axis state coherence index are linearly weighted and summed.

4. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 1, characterized in that, The adaptive weighted fusion module includes: A dynamic weight calculation unit is used to generate real-time fusion weights for each signal based on the comprehensive signal quality index. The weighted fusion execution unit is used to perform weighted summation of each signal according to the real-time fusion weight to generate a fusion feedback signal.

5. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 4, characterized in that, The dynamic weight calculation unit is used to generate the real-time fusion weight in the following manner: Calculate the p-th power of the comprehensive signal quality index of each signal to obtain the calculation result, where p is a weighting index greater than or equal to 1; The calculation results of each signal are normalized to obtain the real-time fusion weight.

6. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 1, characterized in that, The excitation parameter self-tuning module includes: An excitation frequency self-tuning unit is used to adjust the excitation frequency of the front-end circuit according to the spectral purity index output by the signal quality index calculation module. The excitation amplitude self-tuning unit is used to monitor the signal amplitude of each signal to adjust the excitation amplitude of the front-end circuit.

7. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 6, characterized in that, The excitation frequency self-tuning unit is used to adjust the excitation frequency of the front-end circuit in the following manner: The spectral purity index is compared with a preset frequency adjustment trigger threshold. When the spectral purity index is lower than the frequency adjustment trigger threshold, a new excitation frequency is generated based on the current frequency and a preset frequency step size. The digital control word corresponding to the new excitation frequency is sent to the digitally controllable front-end demodulation circuit through the control bus to adjust the excitation frequency of the front-end circuit.

8. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 6, characterized in that, The excitation amplitude self-tuning unit is used to monitor the signal amplitude of each signal to adjust the excitation amplitude of the front-end circuit in the following manner: The signal amplitude of each signal is monitored to obtain the peak value, and the peak value is compared with the preset upper amplitude threshold and lower amplitude threshold. When the peak value exceeds the upper limit threshold of the amplitude, the current excitation amplitude is reduced proportionally to obtain a new excitation amplitude; when the peak value is less than the lower limit threshold of the amplitude, the current excitation amplitude is increased proportionally to obtain a new excitation amplitude. The digital control word corresponding to the new excitation amplitude is sent to the digitally controllable front-end demodulation circuit to adjust the excitation amplitude of the front-end circuit.

9. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 1, characterized in that, The system output monitoring module includes: A feedback signal output unit is used to convert the fused feedback signal into a standard industrial signal; A health status monitoring unit is used to determine the sensor health status based on the comprehensive signal quality index and output the sensor health status.

10. The multi-sensor signal fusion system for magnetic levitation bearings according to claim 9, characterized in that, The health status monitoring unit is used to determine the sensor health status based on the comprehensive signal quality index in the following manner: The comprehensive signal quality index is compared with preset warning thresholds and fault thresholds to obtain the comparison results; The sensor health status is determined based on the comparison results, and the health status is classified as healthy, warning, or faulty.

Images