Gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment

Abstract

The invention relates to the technical field of gas steel cylinder monitoring, in particular to a gas steel cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment, and the method realizes comprehensive monitoring of internal and external environments of a gas steel cylinder through cooperative work of various sensors. Meanwhile, a dimension reduction algorithm and a time sequence analysis technology are adopted, key feature vectors are effectively extracted, and the efficiency and accuracy of data analysis are improved. In addition, a dynamic weight adjustment mechanism is introduced, the importance of data of each sensor can be automatically adjusted according to changes of environmental conditions, and the monitoring accuracy and reliability are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas cylinder monitoring, and in particular to a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment. Background Art

[0002] At present, gas cylinders are widely used in various fields, such as medical, chemical, welding, etc. To ensure safe use, real-time monitoring of the internal pressure of gas cylinders is essential. The traditional method of monitoring the pressure of gas cylinders mainly relies on a single pressure sensor, which cannot fully reflect the actual status of the cylinder. This method has the following defects: First, a single sensor is easily interfered by environmental factors, resulting in inaccurate measurement results; second, it is impossible to consider the impact of external environmental factors such as temperature and humidity on the internal pressure of the cylinder, which may lead to misjudgment; third, the lack of comprehensive analysis of multiple data makes it difficult to detect potential safety hazards in a timely manner.

[0003] To solve these problems, some improved methods have been proposed. For example, sensors are installed on cylinders. However, although a variety of data are collected, there is a lack of effective data fusion algorithms, and the information of each sensor cannot be fully utilized. At the same time, there is a lack of dynamic adjustment mechanism for the importance of each sensor data under different environmental conditions, which makes it difficult to adapt to the complex and changing use environment. In addition, the data processing methods in the prior art are often too simple to extract truly valuable information from massive data. For example, only calculating the average or maximum value ignores the time series characteristics of the data and makes it difficult to capture the trend and law of pressure changes. At the same time, there is a lack of effective dimensionality reduction algorithms. Faced with high-dimensional sensor data, the processing efficiency is low and it is difficult to achieve real-time monitoring.

[0004] Therefore, a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment is needed to solve the above problems. Summary of the invention

[0005] The present invention provides a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment. The method realizes comprehensive monitoring of the internal and external environment of the gas cylinder through the collaborative work of multiple sensors. At the same time, the dimension reduction algorithm and time series analysis technology are used to effectively extract key feature vectors, thereby improving the efficiency and accuracy of data analysis. In addition, a dynamic weight adjustment mechanism is introduced, which can automatically adjust the importance of each sensor data according to changes in environmental conditions, further improving the accuracy and reliability of monitoring.

[0006] The technical solution adopted by the present invention to solve the above technical problems is: a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment, comprising the following steps: S1: Install cameras and environmental monitoring sensors on the gas cylinders, and install electronic pressure sensors and temperature sensors inside the gas cylinders; S2: Through the integrated data acquisition system, the data streams from all sensors are collected synchronously in real time, and the collected raw data are preliminarily cleaned to remove noise and outliers, providing a data set for subsequent analysis; S3: Use dimensionality reduction algorithms to reduce the number of features from the dataset; S4: Apply time series analysis algorithm to extract key feature vectors from the reduced dimension data set. The key feature vectors include the average pressure change rate and temperature fluctuation amplitude inside the cylinder; S5: Apply time series analysis algorithms to extract representative feature vectors from the data set, including mean, variance, and peak value; S6: By real-time monitoring of changes in the external environmental conditions of the gas cylinder, the importance weight of each sensor data is automatically adjusted. If the environmental humidity exceeds the standard, the weight of the humidity sensor is increased; S7: Set the alarm threshold. Once the monitoring data exceeds the set range, the alarm will be triggered to remind the operator to take necessary maintenance measures.

[0007] Furthermore, the PCA principal component analysis method is used in S3 to reduce the number of features from the data set and the data dimension, including: S3-1: Standardize or normalize all features in the data set, including the pressure inside the gas cylinder, the internal and external temperature of the gas cylinder, and the ambient humidity of the gas cylinder; S3-2: Construct a covariance matrix to describe the linear relationship between each feature. Let D be an n×m data matrix, where n represents the number of data points collected from all sensors and m represents the number of collected features. The covariance matrix C is calculated as follows: ; S3-3: Compute the eigenvalues ​​and corresponding eigenvectors of the covariance matrix C using numerical linear algebra tools; S3-4: Sort the eigenvectors according to the size of the eigenvalue of the covariance matrix C. The larger the eigenvalue, the more data variance the corresponding eigenvector explains; S3-5: Select the first k largest eigenvalues ​​and their corresponding eigenvectors as principal components, project the original data set into the coordinate system composed of the selected k principal components, reduce the data dimension, and generate a new low-dimensional data set.

[0008] Furthermore, the environmental monitoring sensor includes a temperature sensor and a humidity sensor.

[0009] Furthermore, the mean in S5 includes: internal pressure: the average value of the internal pressure measurement values ​​of the gas cylinder within every 5 minutes; temperature: the average value of the internal or external temperature measurement values ​​of the gas cylinder within every hour; the variance includes: internal pressure: the degree of change of the internal pressure measurement values ​​of the gas cylinder around its mean within every 5 minutes; temperature: the dispersion of the internal and external temperature measurement values ​​of the gas cylinder around its mean within every hour; the peak includes: internal pressure: the maximum value reached by the internal pressure of the gas cylinder within 5 minutes; temperature: the highest or lowest temperature inside and outside the gas cylinder within every hour.

[0010] Furthermore, the time series analysis algorithm is applied in S4 to extract key feature vectors from the reduced-dimensional data set, including: S4-1: Import the reduced-dimensionality dataset into the data processing tool; S4-2: Set the timestamp column as the index of the data table; S4-3: Use the sorting function in the data processing tool to arrange the data in the dataset in ascending order according to the timestamp.

[0011] Furthermore, the time series analysis algorithm is applied in S5 to extract representative feature vectors from the data set, including: S5-1: Initialize the feature vector set V, which is used to store the key feature vectors extracted from the reduced-dimensional data set; S5-2: Calculate and record the mean internal pressure μ of the gas cylinder p , variance σ² p and peak value p max For internal pressure, take every 5 minutes as a cycle and calculate the average value of all measurement points in the cycle as the mean value μ p Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² p , and record the maximum pressure value in each cycle as the peak value p max ; S5-3: Calculate and record the mean temperature μ t , variance σ² t and peak value t extreme For the internal and external temperatures of the gas cylinder, take one hour as a cycle and calculate the average value of all measurement points in the cycle as the mean μ t Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² t , and record the highest or lowest temperature in each cycle as the extreme temperature t extreme ; S5-4: Add the mean, variance and peak features calculated above to the feature vector set V; S5-5: Output the feature vector set V for subsequent analysis.

[0012] Furthermore, in S6, by real-time monitoring of changes in the external environmental conditions of the gas cylinder, the importance weight of each sensor data is automatically adjusted. If the ambient humidity exceeds the standard, the weight of the humidity sensor is strengthened, including: S6-1: Initialize the weight vector W and assign an initial weight W to each sensor data i , where i represents the i-th sensor, and the sum of the weight values ​​is equal to 1; S6-2: Setting environmental condition threshold T eh , T et , T eh is the warning threshold of ambient humidity, T et It is the warning threshold of the internal and external temperature of the gas cylinder; S6-3: Obtain data streams from all environmental monitoring sensors in real time and compare them with the corresponding environmental condition thresholds T eh , T et Compare and if any environmental parameter exceeds its set threshold, proceed to the next step; otherwise, maintain the current weight unchanged; S6-4: for an environmental parameter that exceeds an environmental condition threshold, determining a sensor whose weight needs to be adjusted; S6-5: Calculate a new weight adjustment coefficient α; S6-6: Update the corresponding weight W in the weight vector W i , the updated weight vector W is applied to the data acquisition system to ensure that the latest weights are used for weighted analysis in subsequent data processing; S6-7: Record weight adjustment events and reasons to form a log file for troubleshooting.

[0013] The advantages of the present invention are: The present invention provides a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment. The method realizes comprehensive monitoring of the internal and external environment of the gas cylinder through the collaborative work of multiple sensors. At the same time, the dimension reduction algorithm and time series analysis technology are used to effectively extract key feature vectors, thereby improving the efficiency and accuracy of data analysis. In addition, a dynamic weight adjustment mechanism is introduced, which can automatically adjust the importance of each sensor data according to changes in environmental conditions, further improving the accuracy and reliability of monitoring. DETAILED DESCRIPTION

[0014] The technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] Embodiment 1: The present invention provides a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment. In the field of gas cylinder pressure monitoring, traditional methods mainly rely on a single sensor for data collection and analysis. Although this method is simple and direct, it has many limitations in practical applications. Specifically, a single sensor cannot fully capture the complex environmental changes inside and outside the gas cylinder, resulting in the accuracy and reliability of the monitoring data being questioned. For example, relying only on the internal pressure sensor may ignore the impact of external temperature changes on the pressure reading, resulting in misjudgment. In addition, a single data source is susceptible to environmental interference or sensor failure, increasing the instability and false alarm rate of the system. These problems directly affect the core performance indicators of the gas cylinder pressure monitoring system, including monitoring accuracy, reliability and real-time performance. In the gas storage area of ​​a large chemical plant, hundreds of high-pressure gas cylinders are used to store various industrial gases. These cylinders are distributed under different environmental conditions, some are located in indoor constant temperature areas, and some are exposed outdoors and are affected by weather factors such as sunshine, rain and snow. The traditional monitoring system only installs a pressure sensor on each cylinder, and records pressure data every 30 minutes. However, this simple monitoring method cannot cope with complex actual conditions. For example, on a sunny summer afternoon, the pressure reading of an outdoor cylinder suddenly increased, but the system could not determine whether it was due to a gas leak or a simple temperature increase. At the same time, due to the lack of real-time environmental data, the system could not effectively compare and analyze cylinders in different locations. In this case, technicians often need to conduct manual inspections, which is not only inefficient but also may face safety risks. If these technical problems cannot be effectively solved, it will have a serious impact on the entire gas storage and use system. First, inaccurate monitoring data may lead to wrong operational decisions, such as triggering emergency discharge procedures in unnecessary situations, causing economic losses and environmental pollution. Secondly, due to the inability to detect potential leaks or pressure anomalies in time, serious safety accidents may occur. Furthermore, the system's inefficiency and high false alarm rate will increase labor costs and management difficulties, and reduce the operating efficiency of the entire chemical plant. In the long run, these problems may also hinder the further development of gas storage technology and limit its promotion in a wider range and more complex application scenarios. Therefore, it is particularly important and urgent to develop a new method that can comprehensively, accurately and in real time monitor the pressure of gas cylinders.

[0016] In view of the limitations of the traditional single sensor monitoring method, this application proposes a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment. In the process of exploring solutions, the first consideration is to increase the number and types of sensors. By installing cameras and environmental monitoring sensors on gas cylinders, and installing electronic pressure sensors and temperature sensors in the cylinders, the environmental changes inside and outside the cylinders can be fully captured. However, simply increasing the number of sensors is not enough to solve the problem. How to effectively process and analyze large amounts of data from multiple sensors becomes the next challenge. To this end, this application proposes the use of an integrated data acquisition system to synchronously collect data streams from all sensors in real time, and to perform preliminary cleaning of the raw data to remove noise and outliers. Next, considering that the data dimension may be too high, affecting processing efficiency and analysis accuracy, this application introduces a dimensionality reduction algorithm (such as PCA principal component analysis) to reduce the number and dimension of features of the data set. This step helps to extract the most critical information and simplify the subsequent analysis process. In order to extract meaningful information from the reduced-dimensional data, this application uses a time series analysis algorithm. This method can effectively capture key feature vectors such as the average pressure change rate and temperature fluctuation amplitude inside the cylinder, providing more accurate and dynamic data support for the monitoring system.

[0017] In addition, in order to further improve the accuracy and reliability of monitoring, this application also proposes to extract representative feature vectors from the data set, including mean, variance and peak value. These statistical features can more comprehensively describe the state changes of gas cylinders. Taking into account the impact of different environmental conditions on the monitoring results, this application designs a dynamic weight adjustment mechanism. By monitoring the changes in the external environmental conditions of the gas cylinders in real time, the system can automatically adjust the importance weight of each sensor data. For example, when the ambient humidity exceeds the standard, the system will automatically strengthen the weight of the humidity sensor to ensure the accuracy of the monitoring results.

[0018] In order to detect potential risks in a timely manner, this application adds an alarm mechanism to the solution. By setting the alarm threshold, once the monitoring data exceeds the set range, the system will trigger an alarm to remind the operator to take necessary maintenance measures. Therefore, the present application proposes a gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment, which includes the following steps: installing a camera and an environmental monitoring sensor on the gas cylinder, and installing an electronic pressure sensor and a temperature sensor in the gas cylinder; through an integrated data acquisition system, real-time synchronous acquisition of data streams from all sensors, preliminary cleaning of the collected raw data, removal of noise and outliers, and providing a data set for subsequent analysis; using a dimensionality reduction algorithm (such as PCA principal component analysis) to reduce the number of features from the data set Data dimension; applying a time series analysis algorithm to extract key feature vectors from the reduced-dimensional data set, the key feature vectors include the average pressure change rate and temperature fluctuation amplitude inside the cylinder; applying a time series analysis algorithm to extract representative feature vectors from the data set, including mean, variance, and peak value; by real-time monitoring of changes in the external environmental conditions of the gas cylinder, automatically adjust the importance weight of each sensor data, if the environmental humidity exceeds the standard, then strengthen the weight of the humidity sensor; set an alarm threshold, once the monitoring data exceeds the set range, an alarm is triggered to remind the operator to take necessary maintenance measures. Among them, the camera refers to an image acquisition device used to capture the external state of the gas cylinder, which can be implemented by a high-resolution digital camera. Among them, the environmental monitoring sensor refers to the equipment used to detect the environmental parameters around the gas cylinder, which can be implemented by temperature sensors and humidity sensors. Among them, the electronic pressure sensor refers to the equipment used to measure the internal pressure of the gas cylinder, which can be implemented by piezoelectric or capacitive pressure sensors. Among them, the temperature sensor refers to the equipment used to measure the internal temperature of the gas cylinder, which can be implemented by thermocouples or thermistors.

[0019] The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment proposed in this application realizes comprehensive monitoring of the internal and external environment of the gas cylinder through the collaborative work of multiple sensors. At the same time, the dimension reduction algorithm and time series analysis technology are used to effectively extract key feature vectors, thereby improving the efficiency and accuracy of data analysis. In addition, a dynamic weight adjustment mechanism is introduced, which can automatically adjust the importance of each sensor data according to changes in environmental conditions, further improving the accuracy and reliability of monitoring.

[0020] The working principle of this application can be described in detail as follows: First, multiple sensors are installed on the gas cylinder. The camera is installed on the outside of the cylinder to capture visual information on the surface of the cylinder. Environmental monitoring sensors, including temperature sensors and humidity sensors, are installed around the cylinder to monitor external environmental conditions. Electronic pressure sensors and temperature sensors are installed inside the cylinder to measure internal pressure and temperature, respectively. Next, the data streams of all sensors are collected synchronously in real time through an integrated data acquisition system. The system performs preliminary cleaning on the collected raw data to remove noise and outliers to provide a reliable data set for subsequent analysis. Then, a dimensionality reduction algorithm (such as PCA principal component analysis) is used to reduce the number of features and dimensions of the data set. This step helps to extract the most critical information, simplify the subsequent analysis process, and improve processing efficiency. After that, a time series analysis algorithm is applied to extract key feature vectors from the reduced dimensionality data set, including the average pressure change rate and temperature fluctuation amplitude inside the cylinder. These feature vectors can reflect the dynamic changes in the state of the gas cylinder. Furthermore, the system extracts representative feature vectors from the data set, including mean, variance, and peak. These statistical features can more comprehensively describe the state changes of the gas cylinder. At the same time, the system automatically adjusts the importance weight of each sensor data by monitoring the changes in the external environmental conditions of the gas cylinder in real time. For example, when the ambient humidity exceeds the standard, the system will automatically strengthen the weight of the humidity sensor to ensure the accuracy of the monitoring results. Finally, the system sets the alarm threshold. Once the monitoring data exceeds the set range, the alarm is triggered to remind the operator to take necessary maintenance measures. This method of multi-sensor fusion and weight adjustment can fully capture the complex environmental changes inside and outside the gas cylinder and improve the accuracy and reliability of the monitoring data. Through dimensionality reduction and feature extraction, the system can efficiently process large amounts of data and extract key information. The dynamic weight adjustment mechanism further improves the system's adaptability to different environmental conditions. As a preferred embodiment, the present application can be implemented in the gas storage area of ​​a large chemical plant. Specifically, a high-pressure nitrogen cylinder with a capacity of 50L can be selected as the monitoring object. On the outside of the cylinder, a 1080p high-definition camera is installed to capture the state changes on the surface of the cylinder in real time. Environmental monitoring sensors include a PT100 platinum resistance temperature sensor (measuring range -50℃ to 150℃, accuracy ±0.1℃) and a capacitive humidity sensor (measuring range 0-100%RH, accuracy ±2%RH). Inside the cylinder, a piezoelectric pressure sensor (measuring range 0-25MPa, accuracy ±0.1%) and a K-type thermocouple temperature sensor (measuring range -200℃ to 1300℃, accuracy ±1.5℃) are installed. All sensors are connected to the central control unit through an industrial-grade data acquisition module, and the sampling frequency is set to 1 time per second. The data acquisition system is developed in the Python programming language and uses the Pandas library for data processing.The PCA dimensionality reduction algorithm is implemented using the scikit-learn library, and 95% of the variance information is set to be retained. The time series analysis uses the statsmodels library, and the ARIMA model is used to extract key eigenvectors. The weight adjustment mechanism sets the initial weights as follows: 0.4 for pressure sensor, 0.3 for internal temperature sensor, 0.15 for external temperature sensor, and 0.15 for humidity sensor. When the ambient humidity exceeds 80%RH, the weight of the humidity sensor is automatically adjusted to 0.3, and the weights of other sensors are reduced accordingly. The alarm thresholds are set as follows: the internal pressure change rate exceeds 0.1MPa / min, the internal temperature change rate exceeds 2℃ / min, the external temperature exceeds 40℃ or is lower than -10℃, and the humidity exceeds 90%RH. Once the alarm is triggered, the system sounds an alarm through an audible and visual alarm, and sends a text message notification to the management personnel. Through this implementation method, comprehensive, accurate and real-time monitoring of the gas cylinder pressure can be achieved, greatly improving the reliability and safety of the monitoring system.

[0021] In the process of implementation of this application, there is still the problem that the data dimension is too high, resulting in increased computational complexity and reduced processing efficiency. In this regard, the present application further proposes a technical solution using PCA principal component analysis to reduce the number of features from the data set. The technical solution of this application uses PCA principal component analysis to reduce the number of features and data dimensions of the data set, thereby effectively solving the problem of increased computational complexity and reduced processing efficiency caused by excessive data dimensions. The method first standardizes or normalizes all features in the data set, including the pressure inside the gas cylinder, the internal and external temperature, and the ambient humidity. This step can avoid the dominant influence of certain features on the PCA results due to their large values. Next, a covariance matrix is ​​constructed to describe the linear relationship between each feature. Suppose D is an n×m data matrix, where n represents the number of data points collected from all sensors, and m represents the number of features collected, then the covariance matrix C is calculated in a specific way. Then, the eigenvalues ​​and corresponding eigenvectors of the covariance matrix C are calculated using numerical linear algebra tools.

[0022] The eigenvectors are sorted according to the size of the eigenvalue of the covariance matrix C. The larger the eigenvalue, the more data variance the corresponding eigenvector explains. Finally, the first k largest eigenvalues ​​and their corresponding eigenvectors are selected as principal components, and the original data set is projected into a coordinate system consisting of the selected k principal components to achieve a reduction in data dimension and generate a new low-dimensional data set. In practical applications, the value of k can be determined according to specific needs and data characteristics. For example, the first k principal components with a cumulative explained variance of 85% or 90% can be selected. This can significantly reduce the data dimension while retaining most of the useful information. By using the PCA principal component analysis method, the technical solution of the present application can effectively reduce the data dimension, thereby reducing the computational complexity and improving the data processing efficiency. This is of great significance for real-time monitoring and analysis of gas cylinder pressure, especially when a large amount of sensor data needs to be processed. In addition, the PCA method can also help identify the main change patterns in the data, remove redundant information, and highlight key features. This not only helps to improve the accuracy of subsequent analysis, but also reduces storage requirements and optimizes system resource utilization. As a preferred embodiment, the incremental PCA algorithm can be considered. This algorithm allows the PCA model to be gradually updated as new data arrives without recalculating the entire data set. This is particularly useful for processing continuously flowing in sensor data, which can further improve the real-time performance of the system. Furthermore, other dimensionality reduction techniques, such as t-SNE or UMAP, can be combined to handle nonlinear relationships. These methods can be used as a supplement to PCA to visualize high-dimensional data or process complex nonlinear feature relationships. It can be seen that the technical solution of using PCA principal component analysis to reduce data dimensions proposed in this application not only solves the problem of excessive data dimensions, but also provides a more efficient and accurate data processing method for gas cylinder pressure monitoring systems. The application of this method enables the system to better cope with the challenges of large-scale data analysis and improves overall monitoring efficiency and accuracy.

[0023] The technical solution of the present application solves the problem of unclear types of environmental monitoring sensors by clarifying that environmental monitoring sensors include temperature sensors and humidity sensors. This design can more comprehensively monitor the environmental conditions around gas cylinders and provide more accurate data support for gas cylinder pressure monitoring. Specifically, the temperature sensor is used to monitor temperature changes around gas cylinders, and the humidity sensor is used to monitor environmental humidity. The combination of these two sensors can provide more comprehensive environmental data, which helps to more accurately evaluate the impact of environmental factors on gas cylinder pressure. The role of the temperature sensor is to monitor the temperature changes around the gas cylinder in real time. Temperature is an important factor affecting gas pressure because gas pressure increases with increasing temperature. By continuously monitoring the temperature, temperature anomalies can be detected in time to prevent pressure anomalies caused by temperature changes. The humidity sensor is used to monitor environmental humidity. Humidity may affect the storage of certain gases, especially for gases that react easily with moisture. By monitoring humidity, corrosion of gas cylinders or other safety hazards caused by excessive environmental humidity can be prevented. The data of the temperature sensor and the humidity sensor can be fused and analyzed with the data of other sensors. For example, when a sharp rise in temperature is detected, the data of the pressure sensor can be combined to determine whether there is a risk of pressure anomalies. Similarly, when the humidity sensor detects abnormal ambient humidity, it can be combined with the camera's image data to determine whether there are visible signs of condensation or corrosion.

[0024] As a preferred embodiment, the temperature sensor can be a thermocouple or a thermistor, with a measurement range of -40°C to 100°C and an accuracy of ±0.5°C. The humidity sensor can be a capacitive humidity sensor, with a measurement range of 0%RH to 100%RH and an accuracy of ±2%RH. These sensors can collect data synchronously with other sensors through a data acquisition system to achieve real-time monitoring. By using temperature sensors and humidity sensors, the technical solution of the present application can more comprehensively monitor the storage environment of gas cylinders, thereby improving the accuracy and comprehensiveness of monitoring. This design can not only better prevent pressure anomalies caused by environmental factors, but also provide more valuable data support for the safe storage of gas cylinders, thereby improving the reliability and safety of the entire monitoring system. Compared with the prior art, the present application provides a more specific and comprehensive environmental monitoring solution by clearly pointing out that the environmental monitoring sensor includes a temperature sensor and a humidity sensor. Compared with the solution using only a single type of sensor, this design can more comprehensively evaluate the impact of environmental factors on the pressure of gas cylinders, thereby improving the accuracy and reliability of monitoring.

[0025] The present application further proposes an improved feature extraction method. This method uses different time windows and statistical indicators for different types of sensor data to more accurately capture key information in the gas cylinder pressure monitoring system. Specifically, the improved method proposed in the present application includes the following aspects: First, for internal pressure data, 5 minutes is used as the time window. In each 5-minute time window, the average value of the internal pressure measurement value of the gas cylinder is calculated as the pressure mean value of the time period. This method can effectively smooth short-term fluctuations and reflect the overall trend of pressure. Secondly, for temperature data, 1 hour is used as the time window. In each hour, the average values ​​of the internal and external temperature measurements of the gas cylinder are calculated respectively. This longer time window can better reflect the slow change characteristics of temperature and avoid being affected by instantaneous fluctuations. Further, the present application also introduces variance as an important statistical indicator. For internal pressure, the degree of change of the pressure measurement value around its mean value within every 5 minutes is calculated. This indicator can reflect the stability of pressure and help to detect abnormal fluctuations in time. Similarly, for temperature data, the dispersion of the temperature measurement value around its mean value within each hour is calculated to evaluate the fluctuation amplitude of temperature. In addition, the present application also focuses on the collection of peak data. For internal pressure, record the maximum pressure value reached every 5 minutes. For temperature, record the highest or lowest temperature every hour. These extreme value data are of great significance for identifying potential safety hazards. Through this improved feature extraction method, key information in the gas cylinder pressure monitoring system can be captured more comprehensively and accurately. The mean reflects the overall trend, the variance reflects the stability of the data, and the peak value helps to discover potential abnormalities. The combination of these three statistical indicators provides a more reliable basis for subsequent data analysis and decision-making.

[0026] In practical application, this method can be implemented by following the steps below: First, the raw data is preprocessed, including removing outliers and noise. Then, the data is divided into corresponding time windows according to different sensor types and data characteristics. For pressure data, a data segment is formed every 5 minutes; for temperature data, a data segment is formed every hour. Next, a statistical algorithm is applied to each data segment. When calculating the mean, a simple arithmetic mean method can be used. When calculating the variance, a standard variance calculation formula can be used. For peak extraction, it is necessary to traverse each data segment to find the maximum or minimum value. These calculations can be automated through programming. For example, Python's NumPy library can be used to efficiently perform statistical analysis of large-scale data. The calculation results can be stored in a database in real time for subsequent analysis. In this way, the original high-frequency, high-dimensional sensor data can be converted into more refined and meaningful feature vectors. This not only greatly reduces the burden of data storage and processing, but also improves the efficiency and accuracy of subsequent analysis.

[0027] For example, in a specific embodiment, assuming there is a gas cylinder, the internal pressure sensor collects data once per second, and the temperature sensor collects data once per minute. In a one-hour monitoring cycle, 3600 pressure data points and 60 temperature data points will be generated.

[0028] Applying the method of this application, the pressure data is first divided into 12 5-minute time windows. For each of the 300 data points in each window, the mean, variance, and maximum are calculated. For example, the pressure mean of the first 5-minute window may be 10.5 MPa, the variance is 0.02 MPa², and the maximum is 10.8 MPa.

[0029] For temperature data, these 60 data points are directly used to calculate the mean, variance, and extreme value within one hour. Assume that the results are: the internal temperature mean is 25.5℃, the variance is 0.5℃², and the maximum temperature is 26.2℃; the external temperature mean is 23.8℃, the variance is 0.3℃², and the minimum temperature is 23.2℃. In this way, the raw data of one hour is condensed into several key features: 12 groups of pressure features (each group contains mean, variance, and peak value) and 2 groups of temperature features (internal and external, each group contains mean, variance, and extreme value). These features can effectively reflect the state changes of gas cylinders and provide the possibility of timely detection of abnormalities.

[0030] Compared with traditional methods, the technical solution of the present application has the following advantages: First, by adopting different time windows, it better adapts to the characteristics of different types of data. The 5-minute pressure data window can reflect pressure changes in a timely manner, while the 1-hour temperature data window can smooth short-term fluctuations and reflect the overall trend. Secondly, the introduction of variance as a statistical indicator provides important information on data stability. This is often overlooked in traditional methods, but it is of great significance for predicting potential failures. Furthermore, by recording peak data, this solution can capture extreme situations that may be masked by the average value, which is particularly important for safety monitoring. This method greatly reduces the amount of data while retaining key information and improving the efficiency and accuracy of subsequent analysis. This not only saves storage and computing resources, but also provides the possibility for real-time monitoring and rapid decision-making.

[0031] The present application further proposes a method for extracting key feature vectors from a reduced-dimensionality dataset. The method includes importing the reduced-dimensionality dataset into a data processing tool, setting the timestamp column as an index of a data table, and using the sorting function in the data processing tool to arrange the data in the dataset in ascending order according to the timestamp. This method can organize and analyze time series data more efficiently by using a data processing tool (such as Pandas DataFrame) to process the reduced-dimensionality dataset. Setting the timestamp as an index can facilitate time-related operations and queries, while arranging the data in ascending order by timestamp ensures the temporal continuity of the data, which is conducive to subsequent time series analysis. Specifically, the following steps are included: first, importing the reduced-dimensionality dataset into a data processing tool. This step can use the DataFrame structure in the Pandas library, which provides powerful data operations and analysis functions. By importing the data into the DataFrame, subsequent data processing and analysis operations can be easily performed. Next, the timestamp column is set as an index of the data table. The importance of this step is that it makes the dataset a time series data structure. By setting the timestamp as an index, it is easier to slice, resample, or perform other time-related operations on data by time. This is especially important for analyzing the trend of gas cylinder pressure changes over time. Finally, use the sort function in the data processing tool to arrange the data in the data set in ascending order by timestamp. This step ensures that the data is arranged in chronological order, which helps to observe the time trends and patterns of the data. For gas cylinder pressure monitoring, this chronological arrangement can help identify the law of pressure changes and make it easier to find abnormal situations.

[0032] This method can more effectively extract key feature vectors from the reduced-dimensional data set. For example, it is easier to calculate the average pressure change rate and temperature fluctuation amplitude inside the cylinder. The average pressure change rate can be obtained by calculating the pressure difference between adjacent time points and dividing it by the time interval, while the temperature fluctuation amplitude can be expressed by calculating the temperature standard deviation within a certain time window. As a preferred implementation, Python's Pandas library can be used to implement this method. The specific steps are as follows: Import the required libraries: import pandas as pd import numpy as np Import the reduced-dimensional dataset into DataFrame: df = pd.read_csv('dimensionality reduced dataset.csv') Set the timestamp column as index: df['timestamp'] = pd.to_datetime(df['timestamp']) df.set_index('timestamp', inplace=True) Sort the data by timestamp in ascending order: df.sort_index(inplace=True) Extract key feature vectors: # Calculate the average pressure change rate df['pressure_change_rate'] = df['pressure'].diff() / df.index.to_series().diff().dt.total_seconds() # Calculate temperature fluctuations (using a rolling window) df['temperature_fluctuation'] = df['temperature'].rolling(window='1H').std() # Extract key feature vectors key_features = df[['pressure_change_rate', 'temperature_fluctuation']].resample('5T').mean() Through this implementation, key feature vectors can be efficiently extracted from the reduced-dimensional data set, including the average pressure change rate and temperature fluctuation amplitude inside the cylinder. These feature vectors can provide important inputs for subsequent gas cylinder pressure monitoring and anomaly detection. Compared with the prior art, the method of the present application has the following advantages: First, by using professional data processing tools, the efficiency and accuracy of data processing are improved. Secondly, setting the timestamp as an index makes the data easier to perform time series analysis. Finally, by arranging the data in chronological order, it is easier to find time-related patterns and trends in the data. These advantages make the method of the present application have better application prospects in the field of gas cylinder pressure monitoring, and can identify potential safety hazards more quickly and accurately.

[0033] The present application further proposes a method for extracting representative feature vectors from a data set using a time series analysis algorithm. The method first initializes a feature vector set V to store key feature vectors extracted from the reduced-dimensional data set. Then, the mean value μ of the internal pressure of the gas cylinder is calculated and recorded. p , variance σ² p and peak value p max For internal pressure, take every 5 minutes as a cycle and calculate the average value of all measurement points in the cycle as the mean value μ p Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² p , and record the maximum pressure value in each cycle as the peak value p max .

[0034] Next, calculate and record the mean temperature μ t , variance σ² t For the internal and external temperatures of the gas cylinder, take one hour as a cycle and calculate the average value of all measurement points in the cycle as the mean value μ t Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² t, and record the highest or lowest temperature in each cycle as the extreme temperature textreme. Finally, the mean, variance and peak features calculated above are added to the feature vector set V, and the feature vector set V is output for subsequent analysis. This method extracts representative feature vectors by periodic statistical analysis of pressure and temperature data. These feature vectors can effectively reflect the changing trends and fluctuations of the pressure and temperature inside the gas cylinder. Specifically, the mean reflects the average level of pressure or temperature in a certain period of time, the variance reflects the discreteness of the data, and the peak reflects the extreme situation. These three indicators together constitute a comprehensive feature vector that can accurately describe the state of the gas cylinder. By using different time periods (5 minutes for pressure and 1 hour for temperature) to calculate these features, the present application can capture changes on different time scales. This method takes into account the fact that pressure changes are usually faster than temperature changes, thereby improving the accuracy and representativeness of feature extraction. Storing these feature vectors in set V provides a rich data basis for subsequent analysis and monitoring. This method can not only reflect the state of the gas cylinder in real time, but also discover potential anomalies or trend changes by comparing feature vectors in different time periods.

[0035] For example, in practical applications, a specific embodiment can be set. Assume that the internal pressure data of a gas cylinder within one hour is as follows (one data point every 5 minutes, in MPa): [10.2, 10.3, 10.1, 10.4, 10.2, 10.3, 10.5, 10.4, 10.3, 10.2, 10.1, 10.3]. At the same time, the internal temperature data within the hour is as follows (one data point every 5 minutes, in ° C): [25.1, 25.3, 25.2, 25.4, 25.3, 25.5, 25.6, 25.4, 25.3, 25.2, 25.1, 25.3].

[0036] For pressure data, the mean μ can be calculated: p =10.275 MPa Variance σ² p ≈0.0156 MPa² Peak value p max =10.5 MPa. For the temperature data, we can calculate: Mean μ t =25.308℃, variance σ² t ≈0.0226℃², extreme temperature textreme=25.6℃ (maximum temperature), these calculated eigenvalues ​​will be added to the eigenvector set V to form a complete eigenvector. In this way, the application can effectively extract and represent the key status information of the gas cylinder, providing a reliable data basis for subsequent monitoring and analysis.

[0037] Compared with the prior art, the method of the present application has the following advantages: First, by using different time periods to calculate the characteristics of pressure and temperature, the changing characteristics of different parameters can be better captured. Second, by considering the mean, variance and peak value at the same time, a more comprehensive state description is provided, rather than relying solely on a single indicator. Finally, this method can adapt to different types of gas cylinders and different usage environments, and has strong versatility and adaptability. These advantages enable the present application to monitor the status of gas cylinders more accurately and comprehensively, thereby improving safety and management efficiency.

[0038] The present application also proposes a technical solution that automatically adjusts the importance weight of each sensor data by real-time monitoring of changes in the external environmental conditions of the gas cylinder. If the ambient humidity exceeds the standard, the weight of the humidity sensor is increased. This technical solution can more accurately reflect the impact of current environmental conditions on gas cylinder pressure monitoring by dynamically adjusting the weight of sensor data, thereby improving the accuracy and reliability of monitoring. Specifically, when the ambient humidity exceeds the preset threshold, increasing the weight of the humidity sensor data can better capture the potential impact of humidity changes on the gas cylinder pressure, so that corresponding adjustments and warnings can be made in a timely manner. The technical solution of the present application includes the following steps: First, initialize the weight vector W and assign an initial weight W to the data of each sensor. i , where i represents the i-th sensor and the sum of the weight values ​​is equal to 1. This ensures that all sensor data have reasonable weight distribution in the initial state. Next, set the environmental condition threshold T eh and T et , where T eh is the warning threshold of ambient humidity, T et are the warning thresholds for the internal and external temperatures of the gas cylinder. These thresholds are set based on the requirements for safe operation of the gas cylinder and empirical data. The system will obtain data streams from all environmental monitoring sensors in real time and compare them with the corresponding environmental condition thresholds T eh 、T et If any environmental parameter exceeds its set threshold, it proceeds to the next step; otherwise, the current weight remains unchanged. This real-time monitoring and comparison mechanism can capture changes in environmental conditions in a timely manner.

[0039] For environmental parameters that exceed the environmental condition threshold, determine the sensors that need to adjust the weight. For example, when the ambient humidity exceeds T eh When the humidity sensor weight is adjusted, the humidity sensor weight will be adjusted. Then, the new weight adjustment coefficient α is calculated. The value of α can be determined according to the degree to which the environmental parameter exceeds the threshold, for example, it can be calculated using a linear or nonlinear function. Update the corresponding weight W in the weight vector W i. The updated weight vector W is applied to the data acquisition system to ensure that the latest weights are used for weighted analysis in subsequent data processing. This dynamic adjustment mechanism enables the system to respond quickly to environmental changes. Finally, the weight adjustment events and reasons are recorded to form a log file for troubleshooting. This step helps with system maintenance and performance optimization. Through this dynamic weight adjustment mechanism, the present application can more flexibly respond to gas cylinder pressure monitoring needs under different environmental conditions. For example, in a high humidity environment, by increasing the weight of the humidity sensor, the system can more sensitively capture pressure fluctuations that may be caused by humidity changes, thereby improving the accuracy and reliability of monitoring. As a preferred implementation, it is possible to consider using a machine learning algorithm to optimize the weight adjustment process. Specifically, the following steps can be adopted: Data Collection: Continuously collect environmental parameters (such as humidity, temperature) and corresponding gas cylinder pressure data.

[0040] Model training: Use the collected historical data to train a machine learning model (such as a random forest or neural network) that can predict the optimal sensor weights based on environmental parameters.

[0041] Real-time prediction: During system operation, real-time environmental parameters are input into the trained model to obtain the optimal weight for prediction.

[0042] Weight update: The weights predicted by the model are compared with the current system weights, and if the difference exceeds a preset threshold, the system weights are updated.

[0043] Model updating: Regularly retrain the model with newly collected data to adapt to possible long-term environmental changes.

[0044] In this way, the system can adjust the sensor weights more intelligently, not only to cope with sudden environmental changes, but also to gradually learn and adapt to long-term environmental trends, further improving the accuracy and reliability of monitoring. Compared with the prior art, the dynamic weight adjustment mechanism of the present application has significant advantages. The traditional fixed weight method cannot adapt to complex and changeable environmental conditions, which may lead to inaccurate monitoring results in some cases. By adjusting the weights in real time, the present application can better reflect the impact of the current environment on the gas cylinder pressure, thereby improving the accuracy and reliability of monitoring. In addition, the method of the present application is also scalable, and new environmental parameters and corresponding sensors can be added as needed to further improve the adaptability and performance of the system.

[0045] The present application further proposes a weight adjustment method, including setting an upper limit and a lower limit for weight adjustment, and introducing a smoothing factor to control the speed of weight adjustment. This improved weight adjustment method is intended to solve the problems of instability and over-adjustment that may occur during the weight adjustment process. By setting the upper and lower limits of the weight adjustment, the weight of a certain sensor can be prevented from excessively increasing or decreasing, thereby maintaining the overall balance of the system. The introduction of a smoothing factor can control the speed of weight adjustment and avoid frequent adjustments caused by short-term fluctuations in environmental conditions. Specifically, this improved weight adjustment method includes the following steps: First, set the upper and lower limits of the weight adjustment for each sensor. These limits can be determined based on the specific needs of the system and the importance of each sensor. For example, the lower limit of the weight of the pressure sensor can be set to 0.3 and the upper limit to 0.6 to ensure that the pressure data always maintains a certain importance.

[0046] Secondly, a smoothing factor β (0<β<1) is introduced to control the speed of weight adjustment. The smaller the value of the smoothing factor, the slower the weight adjustment speed and the smoother the system's response to environmental changes. The selection of the smoothing factor requires a trade-off between system response speed and stability. During the weight adjustment process, the new weight calculation formula is: w_new =β*w_calculated+(1-β)*w_old. Among them, w_new is the new weight finally used, w_calculated is the theoretical new weight calculated according to environmental conditions, and w_old is the old weight. Finally, the calculated new weight is checked to ensure that it does not exceed the preset upper and lower limits. If it exceeds the range, the weight is set to the corresponding upper or lower limit value.

[0047] This improved weight adjustment method can be used in conjunction with the previously described weight adjustment process. After calculating the new weight adjustment coefficient α in step S6-5, smoothing factors and weight limits can be applied to further optimize the weight adjustment process. In this way, a more stable and controllable weight adjustment process can be achieved. Even when environmental conditions change rapidly, the system can maintain relatively stable performance and avoid unstable monitoring results caused by drastic fluctuations in weights. At the same time, by limiting the range of variation of each sensor weight, it is ensured that the system always maintains effective monitoring of all key parameters.

[0048] As a preferred implementation, appropriate smoothing factors and weight limits can be selected according to the specific needs of the system and the operating environment. For example, in indoor use scenarios where the environment is relatively stable, a smaller smoothing factor (such as β = 0.1) can be selected to achieve a smoother weight adjustment. In outdoor or environmentally variable scenarios, a larger smoothing factor (such as β = 0.3) can be selected to increase the system's response speed to environmental changes. At the same time, the upper and lower limits of the weight can be set according to the importance of each sensor in different scenarios. For example, in a high temperature environment, the upper limit of the weight of the temperature sensor can be increased to better reflect the impact of temperature changes on the pressure of the gas cylinder.

[0049] The specific implementation example is as follows: Assume that a gas cylinder pressure monitoring system includes a pressure sensor, an internal temperature sensor, an external temperature sensor, and a humidity sensor. Initially, the weights of these four sensors are 0.4, 0.2, 0.2, and 0.2, respectively. The system sets the following parameters: Weight adjustment upper and lower limits: Pressure sensor: lower limit 0.3, upper limit 0.6; Internal temperature sensor: lower limit 0.1, upper limit 0.3; External temperature sensor: lower limit 0.1, upper limit 0.3; Humidity sensor: lower limit 0.1, upper limit 0.4; Smoothing factor β = 0.2; Environmental Condition Thresholds: Ambient humidity warning threshold T eh =80%; Temperature warning threshold T et = 35℃; When the system detects that the ambient humidity reaches 85%, exceeding the preset threshold, the system calculates the theoretical new weight w_calculated of the humidity sensor as 0.35 according to the previous weight adjustment method. Apply the improved weight adjustment method: 1. Calculate the new weight using the smoothing factor: w_new = 0.2 * 0.35 + (1 - 0.2) * 0.2 = 0.23 2. Check if it is out of the preset range: 0.23 is within the weight range of the humidity sensor (0.1 to 0.4) and no adjustment is required.

[0050] 3. Update the weight of the humidity sensor to 0.23.

[0051] 4. Adjust the weights of other sensors accordingly, make sure the sum of all weights is 1, and check if they are within their respective allowed ranges.

[0052] In this way, the system is able to moderately increase the weight of the humidity sensor when the ambient humidity changes suddenly, while avoiding drastic changes in the weight. This smooth weight adjustment process helps maintain the stability of the system while still being able to respond appropriately to environmental changes.

[0053] Compared with the prior art, the improved weight adjustment method proposed in this application has the following advantages: 1. Enhanced system stability: By introducing smoothing factors and weight limits, frequent weight adjustments caused by short-term environmental fluctuations are effectively avoided, making system performance more stable.

[0054] 2. Improved adaptability: Although the weight adjustment is smoother, the system can still appropriately adjust the importance of each sensor according to environmental changes, maintaining sensitivity to environmental changes.

[0055] 3. Better controllability: By setting upper and lower limits of weights, it ensures that the system can maintain effective monitoring of all key parameters even in extreme cases, avoiding the situation where a certain sensor completely dominates the monitoring results.

[0056] 4. Higher flexibility: The smoothing factor and weight limit can be flexibly adjusted according to different application scenarios and requirements, so that the system can better adapt to various usage environments.

[0057] This improved weight adjustment method not only solves the problem of unstable weight adjustment that may occur in the original technology, but also improves the overall performance and reliability of the system, allowing the gas cylinder pressure monitoring system to operate more accurately and stably under various complex environmental conditions.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment, characterized in that: The following steps are involved: S1: Install cameras and environmental monitoring sensors on the gas cylinders, and install electronic pressure sensors and temperature sensors inside the gas cylinders; S2: Through the integrated data acquisition system, the data streams from all sensors are collected synchronously in real time, and the collected raw data are preliminarily cleaned to remove noise and outliers, providing a data set for subsequent analysis; S3: Use dimensionality reduction algorithms to reduce the number of features from the dataset; S4: Apply time series analysis algorithm to extract key feature vectors from the reduced dimension data set. The key feature vectors include the average pressure change rate and temperature fluctuation amplitude inside the cylinder; S5: Apply time series analysis algorithms to extract representative feature vectors from the data set, including mean, variance, and peak value; S6: By real-time monitoring of changes in the external environmental conditions of the gas cylinder, the importance weight of each sensor data is automatically adjusted. If the environmental humidity exceeds the standard, the weight of the humidity sensor is increased; S7: Set the alarm threshold. Once the monitoring data exceeds the set range, the alarm will be triggered to remind the operator to take necessary maintenance measures.

2. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: The PCA principal component analysis method is used in S3 to reduce the number of features from the data set, including: S3-1: Standardize or normalize all features in the data set, including the pressure inside the gas cylinder, the internal and external temperature of the gas cylinder, and the ambient humidity of the gas cylinder; S3-2: Construct a covariance matrix to describe the linear relationship between each feature. Let D be an n×m data matrix, where n represents the number of data points collected from all sensors and m represents the number of collected features. The covariance matrix C is calculated as follows: ； S3-3: Compute the eigenvalues ​​and corresponding eigenvectors of the covariance matrix C using numerical linear algebra tools; S3-4: Sort the eigenvectors according to the size of the eigenvalue of the covariance matrix C. The larger the eigenvalue, the more data variance the corresponding eigenvector explains; S3-5: Select the first k largest eigenvalues ​​and their corresponding eigenvectors as principal components, project the original data set into the coordinate system composed of the selected k principal components, reduce the data dimension, and generate a new low-dimensional data set.

3. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: The environment monitoring sensor includes a temperature sensor and a humidity sensor.

4. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: The mean in S5 includes: internal pressure: the average value of the internal pressure measurement values ​​of the gas cylinder within every 5 minutes; temperature: the average value of the internal or external temperature measurement values ​​of the gas cylinder within every hour; the variance includes: internal pressure: the degree of change of the internal pressure measurement values ​​of the gas cylinder around its mean within every 5 minutes; temperature: the dispersion of the internal and external temperature measurement values ​​of the gas cylinder around its mean within every hour; the peak includes: internal pressure: the maximum value reached by the internal pressure of the gas cylinder within 5 minutes; temperature: the highest or lowest temperature inside and outside the gas cylinder within every hour.

5. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: In S4, a time series analysis algorithm is applied to extract key feature vectors from the reduced-dimensional data set, including: S4-1: Import the reduced-dimensionality dataset into the data processing tool; S4-2: Set the timestamp column as the index of the data table; S4-3: Use the sorting function in the data processing tool to arrange the data in the dataset in ascending order according to the timestamp.

6. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: The representative feature vectors extracted from the data set by applying the time series analysis algorithm in S5 include: S5-1: Initialize the feature vector set V, which is used to store the key feature vectors extracted from the reduced-dimensional data set; S5-2: Calculate and record the mean internal pressure μ of the gas cylinder p , variance σ² p and peak value p max For internal pressure, take every 5 minutes as a cycle and calculate the average value of all measurement points in the cycle as the mean value μ p Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² p , and record the maximum pressure value in each cycle as the peak value p max ; S5-3: Calculate and record the mean temperature μ t , variance σ² t and peak value t extreme For the internal and external temperatures of the gas cylinder, take one hour as a cycle and calculate the average value of all measurement points in the cycle as the mean μ t Calculate the average of the squared deviations of each measurement point from the mean in each cycle as the variance σ² t , and record the highest or lowest temperature in each cycle as the extreme temperature t extreme ; S5-4: Add the mean, variance and peak features calculated above to the feature vector set V; S5-5: Output the feature vector set V for subsequent analysis.

7. The gas cylinder pressure monitoring method based on multi-sensor fusion and weight adjustment according to claim 1 is characterized in that: In S6, by real-time monitoring of changes in the external environmental conditions of the gas cylinder, the importance weight of each sensor data is automatically adjusted. If the ambient humidity exceeds the standard, the weight of the humidity sensor is strengthened, including: S6-1: Initialize the weight vector W and assign an initial weight W to each sensor data i , where i represents the i-th sensor, and the sum of the weight values ​​is equal to 1; S6-2: Setting environmental condition threshold T eh , T et , T eh is the warning threshold of ambient humidity, T et It is the warning threshold of the internal and external temperature of the gas cylinder; S6-3: Obtain data streams from all environmental monitoring sensors in real time and compare them with the corresponding environmental condition thresholds T eh , T et Compare and if any environmental parameter exceeds its set threshold, proceed to the next step; otherwise, maintain the current weight unchanged; S6-4: for an environmental parameter that exceeds an environmental condition threshold, determining a sensor whose weight needs to be adjusted; S6-5: Calculate a new weight adjustment coefficient α; S6-6: Update the corresponding weight W in the weight vector W i , the updated weight vector W is applied to the data acquisition system to ensure that the latest weights are used for weighted analysis in subsequent data processing; S6-7: Record weight adjustment events and reasons to form a log file for troubleshooting.

Images