Crystal oscillator monitoring method and system based on multi-sensor network

Abstract

The invention provides a crystal oscillator monitoring method and system based on a multi-sensor network. The method comprises the following steps: acquiring operation environment data of the quartz crystal oscillator, monitoring an output signal of the quartz crystal oscillator, and acquiring output frequency and time for measuring the crystal oscillator in each batch; calculating the daily frequency change rate of the quartz crystal oscillator, and establishing an aging linear regression equation of the quartz crystal oscillator through a least square method based on the operation environment data, the output frequency, the daily frequency change rate and the time; and predicting an output frequency prediction value of the quartz crystal oscillator based on the aging linear regression equation, and performing frequency compensation on the quartz crystal oscillator according to the difference between the output frequency prediction value and a target frequency prediction value. Through the method and the corresponding system, the frequency stability and the anti-interference capability of long-term operation of the quartz crystal oscillator system and the overall operation reliability of the terminal equipment can be greatly improved.

Description

Technical Field

[0001] The invention proposes a crystal oscillator monitoring method and system based on a multi-sensor network, and relates to the field of crystal oscillator monitoring. Background Art

[0002] With the development of the Internet of Things (IoT) and distributed control systems, online monitoring of key electronic components (such as quartz crystal oscillators, or "crystals") using multi-sensor networks has become an important means of ensuring system stability and high-precision operation. Multi-sensor networks utilize a distributed deployment of multiple nodes, each equipped with sensors for various environmental and electrical parameters, including temperature, humidity, and voltage. These nodes collect real-time data on the crystal oscillator's operating status and environmental conditions, which are then transmitted to a central node for analysis and processing. While multi-sensor networks improve data coverage and real-time data collection, existing monitoring methods still face significant challenges and limitations in practical applications. Most sensor nodes are low-power embedded systems with limited computing power, typically only suitable for simple data processing or linear analysis. These systems struggle to capture the highly nonlinear or complex coupling between environmental changes and crystal oscillator performance, resulting in insufficient prediction accuracy for monitoring models and limited ability to identify sudden anomalies and subtle performance degradations. Summary of the Invention

[0003] The present invention provides a crystal oscillator monitoring method and system based on a multi-sensor network to solve the above-mentioned problems: The present invention proposes a crystal oscillator monitoring method based on a multi-sensor network, the method comprising: Obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of each batch of crystal oscillator measurements; Calculate the daily frequency change rate of the quartz crystal oscillator and establish the aging linear regression equation of the quartz crystal oscillator using the least squares method based on the operating environment data, output frequency, daily frequency change rate, and time; An output frequency prediction value of a quartz crystal oscillator is predicted based on an aging linear regression equation, and frequency compensation is performed on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

[0004] Furthermore, the operating environment data of the quartz crystal oscillator is obtained, the output signal of the quartz crystal oscillator is monitored, and the output frequency and time of each batch of crystal oscillator measurements are obtained, including: Arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the data on the operating environment to a local microprocessor unit; The crystal oscillator output signal is measured by a frequency meter to obtain the output frequency and time of the crystal oscillator measurement for each batch.

[0005] Furthermore, the daily frequency change rate of the quartz crystal oscillator is calculated, and the aging linear regression equation of the quartz crystal oscillator is established by the least squares method based on the operating environment data, output frequency, daily frequency change rate and time, including: Calculate the daily rate of change of frequency of a quartz crystal oscillator:

[0006] in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ; The aging linear regression equation of the quartz crystal oscillator is established by the least squares method based on the operating environment data, output frequency, daily frequency change rate and time:

[0007] in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

[0008] Furthermore, the residual is obtained through machine learning prediction, including: Using the aging linear regression equation, the predicted value of the crystal oscillator output frequency is obtained based on the operating environment data and output signal of the quartz crystal oscillator; Subtract the predicted value from the actual value of the crystal oscillator output frequency to calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set. Select the machine learning algorithm Random Forest and train it on a feature-labeled dataset. This allows the model to learn the nonlinear relationships that the aging linear regression equation fails to capture and outputs predicted values ​​for the residuals. The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

[0009] Furthermore, predicting an output frequency prediction value of the quartz crystal oscillator based on the aging linear regression equation, and performing frequency compensation on the quartz crystal oscillator according to a difference between the output frequency prediction value and the target frequency prediction value, including: Calculate the difference between the predicted frequency and the target frequency, and design the predicted frequency offset as the compensation control amount.

[0010] The compensation control amount is injected into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation.

[0011] The present invention proposes a crystal oscillator monitoring system based on a multi-sensor network, the system comprising: The module for obtaining operating environment data is used to obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of the crystal oscillator for each batch of measurements; Establish a regression equation module to calculate the daily frequency change rate of the quartz crystal oscillator. Use the least squares method to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate, and time. The compensation module is used to predict an output frequency prediction value of the quartz crystal oscillator based on an aging linear regression equation, and perform frequency compensation on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

[0012] Furthermore, the module for obtaining operating environment data includes: An environment data measurement module is used to arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and to arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the operating environment data to a local microprocessor unit; The output frequency measurement module is used to measure the crystal oscillator output signal through a frequency meter to obtain the output frequency and time of the crystal oscillator measurement in each batch.

[0013] Furthermore, the regression equation establishment module includes: The module for calculating the daily frequency change rate is used to calculate the daily frequency change rate of the quartz crystal oscillator:

[0014] in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ; Modeling module, used to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate and time through the least squares method:

[0015] in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

[0016] Furthermore, the modeling module includes: A prediction module, configured to obtain a predicted value of a crystal oscillator output frequency based on operating environment data and an output signal of the quartz crystal oscillator by using an aging linear regression equation; Generate a training data module, which is used to subtract the predicted value from the actual value of the crystal oscillator output frequency and calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set; The prediction residual module is used to select the machine learning algorithm Random Forest and train it on a feature label dataset to enable the model to learn the nonlinear relationship that the aging linear regression equation fails to capture and output the predicted value of the residual; The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

[0017] Furthermore, the compensation module includes: The compensation calculation module calculates the difference between the predicted frequency and the target frequency, and designs the predicted frequency offset as the compensation control amount.

[0018] The compensation control module is used to inject the compensation control amount into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation.

[0019] The present invention has the following beneficial effects: The quartz crystal oscillator monitoring system first collects operating environment data (such as temperature, humidity, and voltage) and output signals in real time, recording the output frequency and corresponding time for each batch of crystal oscillators. Subsequently, based on data from different times and batches, the system calculates the daily frequency change rate of the quartz crystal oscillator, thereby obtaining dynamic characteristics reflecting the aging of the crystal oscillator and changes in its operating state. After collecting sufficient historical frequency, environment, and change rate data, a least squares method is used to establish an aging linear regression equation with operating environment, output frequency, daily change rate, and time as input variables, enabling accurate prediction of the future output frequency of the quartz crystal oscillator. Specifically, the linear regression equation can fit and reveal the influence of environmental changes, aging trends, and time on frequency deviation. After obtaining the predicted output frequency, the system compares and analyzes it with the target frequency (the expected frequency under ideal conditions). If the predicted frequency deviates, compensation logic proactively adjusts the quartz crystal oscillator's control parameters (such as bias voltage or compensation circuit parameters) to correct the output frequency in real time, bringing it close to or returning to the target frequency. This dynamically compensates for crystal oscillator aging and environmental influences, and provides frequency stabilization. Systematically collecting environmental and output data and using linear regression modeling, the system effectively characterizes and tracks the quartz crystal oscillator's frequency aging process, providing a scientific basis for subsequent maintenance and replacement, significantly reducing manual monitoring costs. Incorporating multidimensional environmental factors, historical frequency change data, and temporal dynamics into the modeling process, the system more comprehensively describes the complex mechanisms affecting frequency variation compared to traditional methods that only consider a single parameter, improving the accuracy of frequency prediction and monitoring. Real-time frequency compensation enhances system stability and reliability. By analyzing the difference between the predicted and target frequencies and dynamically compensating for them, the system significantly improves the long-term frequency stability and interference immunity of the quartz crystal oscillator system, as well as the overall operational reliability of the terminal device. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a schematic diagram of a crystal oscillator monitoring method based on a multi-sensor network according to the present invention. DETAILED DESCRIPTION

[0021] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein may be combined with each other.

[0022] The following description sets forth numerous specific details to facilitate a thorough understanding of the present invention. The embodiments described are merely a portion of the embodiments of the present invention, rather than all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0024] One embodiment of the present invention provides a crystal oscillator monitoring method based on a multi-sensor network, the method comprising: Obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of each batch of crystal oscillator measurements; Calculate the daily frequency change rate of the quartz crystal oscillator and establish the aging linear regression equation of the quartz crystal oscillator using the least squares method based on the operating environment data, output frequency, daily frequency change rate, and time; An output frequency prediction value of a quartz crystal oscillator is predicted based on an aging linear regression equation, and frequency compensation is performed on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

[0025] The working principle and effect of the above technical solution are as follows: By deploying multiple environmental sensors at the quartz crystal oscillator's operating site, the system collects real-time operating environment data (such as temperature, humidity, and supply voltage) as well as its output signal. The system periodically records the output frequency of each batch of crystal oscillators and the corresponding time points. The system then calculates the daily frequency change rate of the quartz crystal oscillator to quantify the dynamic characteristics of the frequency as a function of aging and environmental changes. After data preprocessing, the solution uses the least squares method to construct a linear regression equation for the quartz crystal oscillator's aging process using environmental data, output frequency, daily frequency change rate, and time as independent variables, thereby establishing a quantitative prediction model. During online operation, the system inputs the regression equation based on real-time and historical data to obtain a predicted output frequency value. By comparing the predicted frequency value with the target ideal frequency in real time, the technical solution automatically adjusts the compensation circuit or software parameters to accurately compensate for frequency offset, thereby correcting for frequency drift caused by aging or environmental changes. This technical solution enables dynamic, quantitative, and intelligent management of the crystal oscillator's aging process. It not only monitors the oscillator's output frequency and its temporal trends in real time, but also comprehensively considers environmental disturbances and aging effects, enabling more accurate frequency predictions. Through online analysis and compensation using a regression model, it automatically corrects for frequency drift caused by aging and environmental factors, improving the frequency stability of quartz crystal oscillators over long-term operation. Furthermore, this solution's adaptive model and closed-loop compensation capabilities effectively delay equipment maintenance and replacement cycles, reducing maintenance costs and improving overall system reliability and operational safety.

[0026] One embodiment of the present invention obtains operating environment data of a quartz crystal oscillator, monitors the output signal of the quartz crystal oscillator, and obtains the output frequency and time of each batch of crystal oscillator measurements, including: Arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the data on the operating environment to a local microprocessor unit; The crystal oscillator output signal is measured by a frequency meter to obtain the output frequency and time of the crystal oscillator measurement for each batch.

[0027] The working principle and effect of the above technical solution are as follows: By scientifically deploying multiple types of sensors around the core device and surrounding environment of the quartz crystal oscillator, a global environmental perception network is constructed. Specifically, temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors are deployed around the quartz crystal oscillator, and voltage sensors are placed at the input pins of the quartz crystal oscillator. These sensors respectively collect parameters such as ambient temperature, humidity, mechanical stress, ambient vibration, and power supply voltage in real time, comprehensively reflecting the environmental conditions during the actual operation of the quartz crystal oscillator. The collected sensor data is uploaded to a local microprocessor via a data acquisition channel, which aggregates, stores, and preprocesses all environmental parameters. Simultaneously, a frequency meter monitors and records the quartz crystal oscillator's output signal in real time, obtaining the output frequency and test time corresponding to each measurement batch. This enables synchronized and dynamic monitoring of the quartz crystal oscillator's operating status and its surrounding environment. This technical solution enables the simultaneous collection and archiving of multi-dimensional environmental data and output performance data of the quartz crystal oscillator, providing high-quality and rich basic data support for subsequent frequency stability analysis, aging trend modeling, and frequency compensation decision-making. Through the global environmental perception network and high-precision output signal synchronization measurement, the breadth and depth of quartz crystal oscillator health monitoring have been significantly improved. The multi-type, multi-point deployment of sensors enables more comprehensive and accurate environmental data collection, capable of detecting various external factors that affect crystal oscillator performance, and improving the credibility and interpretability of monitoring data. The frequency meter works in conjunction with the environmental sensor to promptly capture minute frequency fluctuations caused by factors such as temperature, humidity, stress, vibration, and voltage, providing a rich data basis for analyzing and tracing various frequency anomalies. It achieves highly correlated collection of operating environment and key output indicators, providing key data support for establishing frequency stability models, fault diagnosis models, and intelligent compensation algorithms under environmental influences. It improves equipment reliability and automation level, and effectively supports quartz crystal oscillator life management, early warning, and adaptive fault processing through online and real-time detection of key environmental parameters and output frequency, significantly improving the automation level and reliability of equipment operation.

[0028] One embodiment of the present invention provides a crystal oscillator monitoring method based on a multi-sensor network, which calculates the daily frequency change rate of a quartz crystal oscillator and establishes an aging linear regression equation for the quartz crystal oscillator using the least squares method based on operating environment data, output frequency, daily frequency change rate, and time, including: Calculate the daily rate of change of frequency of a quartz crystal oscillator:

[0029] in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ; The aging linear regression equation of the quartz crystal oscillator is established by the least squares method based on the operating environment data, output frequency, daily frequency change rate and time:

[0030] in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

[0031] The working principle and effect of the above technical solution are as follows: First, by placing a variety of environmental sensors around the quartz crystal oscillator, multi-dimensional real-time operating environment data including temperature, humidity, mechanical stress, vibration, voltage, etc. is obtained, and the output frequency and sampling time of each batch of quartz crystal oscillators are measured and recorded using a local frequency meter. In the data preprocessing stage, this solution normalizes and statistically processes the frequency and time data within each sampling period, calculates the daily frequency change rate of the crystal oscillator, and effectively reveals the dynamic trend of frequency changes over time (i.e., under the combined effects of aging and environmental disturbances); further, the operating environment data, output frequency, daily frequency change rate, and sampling time are modeled using the least squares method for multivariate linear regression, and a dynamic regression equation for the aging process of the quartz crystal oscillator is established: Through the above model, the system can accurately predict the future frequency change trend of the quartz crystal oscillator based on historical and real-time sensor big data, providing a theoretical basis and decision support for subsequent frequency compensation, early warning, and life management. High-precision aging trend analysis and prediction, deep fusion calculation of multi-frequency, multi-point environmental and performance data, effectively and quantitatively separates and reveals the combined impact of environmental disturbances and aging factors on quartz crystal oscillator frequency changes, enabling accurate characterization and analysis of frequency change trends. The dynamic model has strong adaptive capabilities, optimizing residual predictions through machine learning, further improving the regression model's adaptability to nonlinear errors and unknown complex influences, making frequency predictions closer to actual operating conditions and enhancing the model's versatility and robustness. Intelligent frequency compensation and reliability assurance, based on output frequency change trends and environmental data, can detect early signs of abnormal aging and performance degradation of quartz crystal oscillators in advance, supporting automated frequency compensation and fault warnings, and improving the long-term reliability, safety, and intelligent maintenance level of the overall system. Through continuous data collection and modeling through a multi-dimensional sensor network, it supports health status monitoring and management of quartz crystal oscillators throughout their life cycle, extending their service life and reducing maintenance and replacement costs, with significant engineering application value.

[0032] In one embodiment of the present invention, a crystal oscillator monitoring method based on a multi-sensor network is provided, wherein the residual is obtained by machine learning prediction, comprising: Using the aging linear regression equation, the predicted value of the crystal oscillator output frequency is obtained based on the operating environment data and output signal of the quartz crystal oscillator; Subtract the predicted value from the actual value of the crystal oscillator output frequency to calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set. Select the machine learning algorithm Random Forest and train it on a feature-labeled dataset. This allows the model to learn the nonlinear relationships that the aging linear regression equation fails to capture and outputs predicted values ​​for the residuals. The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

[0033] The working principle and effect of the above technical solution are as follows: frequency prediction and residual calculation. First, based on the collected operating environment data and output signal of the quartz crystal oscillator, the aforementioned aging linear regression equation is used to calculate the predicted value of the crystal oscillator's output frequency under given conditions. Then, a residual sample dataset is constructed. The actual observed value of the crystal oscillator's output frequency is subtracted from the predicted value of the regression equation to obtain the residual corresponding to each data set. The residual reflects the complex changes that the linear model fails to fit. Using the actual collected operating environment parameters as input features and the residuals as target labels, a feature-label dataset is constructed. Nonlinear machine learning algorithms such as random forests are used to train the dataset, enabling the model to learn nonlinear relationships that the aging linear regression equation cannot capture, thereby gaining the ability to predict residuals. During the online use phase, the output of the residual prediction model is added to the prediction result of the aging linear regression equation to obtain a more accurate revised prediction value for the quartz crystal oscillator's output frequency. By continuously evaluating the prediction accuracy and stability of the combined model, newly acquired data is continuously used to iteratively optimize the model parameters, achieving dynamic performance improvement with feedback. Nonlinear error compensation capabilities are enhanced. Machine learning models such as random forests can effectively learn from complex environmental influences and signal variations, effectively compensating for frequency anomalies and drift that are not accounted for by traditional linear regression models, significantly improving overall prediction accuracy. System robustness and generalization capabilities are enhanced. By integrating linear and machine learning models, this approach leverages existing physical mechanisms while introducing data-driven adaptive capabilities, making the prediction system more adaptable to complex and changing real-world operating environments. Closed-loop optimization and dynamic iteration support model feedback and continuous iterative updates, adapting to various operating conditions and aging stages of quartz crystal oscillators throughout their lifecycle, ensuring that the prediction and compensation models are always optimal or relatively optimal. This significantly improves quartz crystal oscillator frequency stability and system reliability, reduces maintenance and replacement costs, and enhances system intelligence and automation. It is suitable for scenarios with extremely high requirements for time and frequency stability, such as communications, navigation, and high-precision measurement.

[0034] It realizes the organic integration of linear modeling and nonlinear compensation, fully explores the complex laws between the operating environment and the output signal, and provides an efficient and reliable technical means for high-reliability prediction and health management of quartz crystal oscillators.

[0035] One embodiment of the present invention provides a crystal oscillator monitoring method based on a multi-sensor network, which predicts an output frequency prediction value of a quartz crystal oscillator based on an aging linear regression equation, and performs frequency compensation on the quartz crystal oscillator based on the difference between the output frequency prediction value and a target frequency prediction value, including: Calculate the difference between the predicted frequency and the target frequency, and design the predicted frequency offset as the compensation control amount.

[0036] The compensation control amount is injected into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation.

[0037] The working principle and effect of the above technical solution are as follows: Intelligent frequency prediction uses a network of environmental sensors to collect multi-dimensional environmental data from the quartz crystal oscillator, including temperature, humidity, stress, vibration, and voltage, in real time. This data, combined with the output signal, is input into an aging linear regression equation to achieve high-precision prediction of the quartz crystal oscillator's output frequency. The current predicted output frequency is compared with the preset target ideal frequency to calculate the frequency offset between the two—the difference between the predicted and target frequencies. This frequency offset is analyzed and used as a compensation control variable. Based on the calculated compensation control variable, the compensation channel is dynamically adjusted. The compensation variable can be applied to the quartz crystal oscillator in real time through various techniques, including hardware parameter fine-tuning (e.g., voltage-controlled circuits), software algorithm modification (e.g., digital signal processing), or signal inverse modulation, to precisely correct the output frequency and maintain it stable near the target frequency. This system accurately compensates for frequency offsets generated by the quartz crystal oscillator during operation in real time, ensuring long-term stability at the target output frequency, effectively overcoming the adverse effects of environmental disturbances and device aging. Support for multiple compensation strategies (hardware, software, and signal modulation) enhances the solution's adaptability to different quartz oscillator types and applications, enhancing system scalability and applicability. The dynamic, automatic frequency compensation mechanism significantly reduces the need for maintenance and replacement of quartz crystal oscillators due to aging and environmental fluctuations, extending their stable service life and reducing lifecycle operational costs. This upgrades the crystal oscillator system from passive monitoring to active regulation, providing strong technical support for the extremely high frequency stability requirements of high-reliability systems (such as satellite communications, navigation timing, and industrial control).

[0038] One embodiment of the present invention provides a crystal oscillator monitoring system based on a multi-sensor network, the system comprising: The module for obtaining operating environment data is used to obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of the crystal oscillator for each batch of measurements; Establish a regression equation module to calculate the daily frequency change rate of the quartz crystal oscillator. Use the least squares method to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate, and time. The compensation module is used to predict an output frequency prediction value of the quartz crystal oscillator based on an aging linear regression equation, and perform frequency compensation on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

[0039] In one embodiment of the present invention, a crystal oscillator monitoring system based on a multi-sensor network is provided, wherein the module for acquiring operating environment data includes: An environment data measurement module is used to arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and to arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the operating environment data to a local microprocessor unit; The output frequency measurement module is used to measure the crystal oscillator output signal through a frequency meter to obtain the output frequency and time of the crystal oscillator measurement in each batch.

[0040] In one embodiment of the present invention, a crystal oscillator monitoring system based on a multi-sensor network is provided, wherein the module for establishing a regression equation includes: The module for calculating the daily frequency change rate is used to calculate the daily frequency change rate of the quartz crystal oscillator:

[0041] in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ; Modeling module, used to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate and time through the least squares method:

[0042] in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

[0043] In one embodiment of the present invention, a crystal oscillator monitoring system based on a multi-sensor network is provided, wherein the modeling module includes: A prediction module, configured to obtain a predicted value of a crystal oscillator output frequency based on operating environment data and an output signal of the quartz crystal oscillator by using an aging linear regression equation; Generate a training data module, which is used to subtract the predicted value from the actual value of the crystal oscillator output frequency and calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set; The prediction residual module is used to select the machine learning algorithm Random Forest and train it on a feature label dataset to enable the model to learn the nonlinear relationship that the aging linear regression equation fails to capture and output the predicted value of the residual; The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

[0044] In one embodiment of the present invention, a crystal oscillator monitoring system based on a multi-sensor network is provided, wherein the compensation module includes: The compensation calculation module calculates the difference between the predicted frequency and the target frequency, and designs the predicted frequency offset as the compensation control amount.

[0045] The compensation control module is used to inject the compensation control amount into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A crystal oscillator monitoring method based on a multi-sensor network, characterized in that: The method comprises: Obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of each batch of crystal oscillator measurements; Calculate the daily frequency change rate of the quartz crystal oscillator and establish the aging linear regression equation of the quartz crystal oscillator using the least squares method based on the operating environment data, output frequency, daily frequency change rate, and time; An output frequency prediction value of a quartz crystal oscillator is predicted based on an aging linear regression equation, and frequency compensation is performed on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

2. The crystal oscillator monitoring method based on a multi-sensor network according to claim 1, characterized in that: Obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of each batch of crystal oscillator measurements, including: Arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the data on the operating environment to a local microprocessor unit; The crystal oscillator output signal is measured by a frequency meter to obtain the output frequency and time of the crystal oscillator measurement for each batch.

3. The crystal oscillator monitoring method based on a multi-sensor network according to claim 1, characterized in that: Calculate the daily frequency change rate of the quartz crystal oscillator. Use the least squares method to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate, and time. This includes: Calculate the daily rate of change of frequency of a quartz crystal oscillator: ； in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ； The aging linear regression equation of the quartz crystal oscillator is established by the least squares method based on the operating environment data, output frequency, daily frequency change rate and time: ； in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

4. The crystal oscillator monitoring method based on a multi-sensor network according to claim 1, characterized in that: The residuals are predicted through machine learning and include: Using the aging linear regression equation, the predicted value of the crystal oscillator output frequency is obtained based on the operating environment data and output signal of the quartz crystal oscillator; Subtract the predicted value from the actual value of the crystal oscillator output frequency to calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set. Select the machine learning algorithm Random Forest and train it on a feature-labeled dataset. This allows the model to learn the nonlinear relationships that the aging linear regression equation fails to capture and outputs predicted values ​​for the residuals. The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

5. The crystal oscillator monitoring method based on a multi-sensor network according to claim 1, characterized in that: The method includes predicting an output frequency prediction value of a quartz crystal oscillator based on an aging linear regression equation, and performing frequency compensation on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value, including: Calculate the difference between the predicted frequency and the target frequency, and design the predicted frequency offset as the compensation control amount; The compensation control amount is injected into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation.

6. A crystal oscillator monitoring system based on a multi-sensor network, characterized in that: The system comprises: The module for obtaining operating environment data is used to obtain the operating environment data of the quartz crystal oscillator, monitor the output signal of the quartz crystal oscillator, and obtain the output frequency and time of the crystal oscillator for each batch of measurements; Establish a regression equation module to calculate the daily frequency change rate of the quartz crystal oscillator. Use the least squares method to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate, and time. The compensation module is used to predict an output frequency prediction value of the quartz crystal oscillator based on an aging linear regression equation, and perform frequency compensation on the quartz crystal oscillator according to a difference between the output frequency prediction value and a target frequency prediction value.

7. The crystal oscillator monitoring system based on a multi-sensor network according to claim 6, characterized in that: The module for obtaining operating environment data includes: An environment data measurement module is used to arrange temperature sensors, humidity sensors, mechanical stress sensors, and vibration sensors around the quartz crystal oscillator, and to arrange voltage sensors at the input pins of the quartz crystal oscillator to form a global environmental perception network for the quartz crystal oscillator, collect data on the operating environment of the quartz crystal oscillator, and upload the operating environment data to a local microprocessor unit; The output frequency measurement module is used to measure the crystal oscillator output signal through a frequency meter to obtain the output frequency and time of the crystal oscillator measurement in each batch.

8. The crystal oscillator monitoring system based on a multi-sensor network according to claim 6, characterized in that: The regression equation module includes: The module for calculating the daily frequency change rate is used to calculate the daily frequency change rate of the quartz crystal oscillator: ； in, represents the daily frequency change rate, q represents the total number of samples in each batch, M represents the number of sampling points, represents the frequency value of the jth sampling, represents the average frequency within the selected sampling period, Indicates the standard frequency value, represents the sampling time of the jth sampling, Indicates the average sampling time of this batch; ； Modeling module, used to establish the aging linear regression equation of the quartz crystal oscillator based on the operating environment data, output frequency, daily frequency change rate and time through the least squares method: ； in, represents the predicted value of the crystal oscillator output frequency at t after least squares fitting, represents the regression coefficient, represents the actual measured value of the jth influencing factor in the operating environment data at time t, represents the rate of environmental change, and ϵ represents the residual, which is obtained through machine learning prediction.

9. The crystal oscillator monitoring system based on a multi-sensor network according to claim 6, characterized in that: The modeling module includes: A prediction module, configured to obtain a predicted value of a crystal oscillator output frequency based on operating environment data and an output signal of the quartz crystal oscillator by using an aging linear regression equation; Generate a training data module, which is used to subtract the predicted value from the actual value of the crystal oscillator output frequency and calculate the residual. The calculated residual is used as the new target variable and combined with the operating environment data to generate a feature label data set; The prediction residual module is used to select the machine learning algorithm Random Forest and train it on a feature label dataset to enable the model to learn the nonlinear relationship that the aging linear regression equation fails to capture and output the predicted value of the residual; The prediction results of the residual model are added to the output of the aging linear regression equation to form a new prediction value, and the performance of the combined aging linear regression equation is evaluated. After the evaluation, iterative optimization can continue.

10. The crystal oscillator monitoring system based on a multi-sensor network according to claim 6, characterized in that: The compensation module includes: The compensation calculation module calculates the difference between the predicted frequency and the target frequency, and designs the predicted frequency offset as the compensation control amount; The compensation control module is used to inject the compensation control amount into the quartz crystal oscillator through hardware parameter adjustment, software algorithm correction and signal inverse modulation.

Images