River channel liquid level sensor data processing system and method

Abstract

The present application relates to the technical field of hydrological monitoring, in particular to a river channel liquid level sensor data processing system and method, the system comprises a sensor layer, a data acquisition and transmission layer, a data processing layer and an output layer, wherein the data processing layer is provided with a data preprocessing module, a data quality evaluation module, a self-adaptive fusion weight calculation module, a multi-source data fusion module and a sequence construction and repair module. The method steps include: collecting multi-sensor original data; performing outlier rejection, missing filling and time alignment; evaluating the real-time quality of each sensor from three dimensions of confidence, correlation and stability; dynamically calculating the self-adaptive fusion weight according to the quality evaluation result; obtaining the comprehensive liquid level value by weighted fusion; constructing and repairing the liquid level sequence and then outputting. The present application effectively suppresses abnormal data interference through multi-dimensional quality evaluation and self-adaptive weight adjustment, improves the accuracy and continuity of liquid level monitoring data, and is suitable for water level observation in complex hydrological environment.

Description

Technical Field

[0001] This invention relates to the field of hydrological monitoring technology, and in particular to a data processing system and method for a river channel liquid level sensor. Background Technology

[0002] River and canal water level monitoring is fundamental to hydrological observation, and the monitoring data serves as a crucial basis for flood warnings, water resource allocation, and aquatic ecosystem protection. With advancements in measurement technology and automation, various types of level sensors have been applied to hydrological monitoring stations, including ultrasonic level gauges, pressure level gauges, radar level gauges, float level gauges, capacitive level gauges, and laser level gauges. These devices acquire water level information based on different physical principles, such as sound wave reflection, hydrostatic pressure conversion, electromagnetic wave echo, buoyancy balance, capacitance changes, or light reflection.

[0003] In actual field operations, single-sensor measurement schemes face numerous interference factors. The water surface conditions in natural rivers and water conveyance channels are not always stable; phenomena such as wind and waves, precipitation, floating debris, changes in water sediment content, temperature stratification, and the accumulation of deposits on the sensor probe surface are common. Different types of sensors exhibit varying sensitivities to these factors: non-contact sensors may experience signal scattering or loss due to strong water surface fluctuations or obstruction by floating debris; contact sensors may be affected by siltation, water corrosion, or aquatic organism attachment, leading to response lag or reference drift. When only a single type of sensor is deployed at a monitoring section, if the device experiences a decrease in measurement accuracy or temporary failure due to environmental changes, the quality of the monitoring data will be directly compromised, and the reliability of subsequent hydrological analysis conclusions will also decrease.

[0004] To alleviate the limitations of single-sensor solutions, the industry has seen attempts at multi-sensor joint monitoring, typically deploying two or more level gauges at the same cross-section and obtaining a comprehensive water level value through data fusion. Common fusion methods include simple arithmetic averaging, weighted averaging, and state estimation based on Kalman filtering. However, existing fusion schemes still have room for improvement in the following aspects.

[0005] First, the weighting of fusion data often uses fixed allocation or is based on empirical presets. Fixed weights do not consider the dynamic differences exhibited by each sensor due to environmental interference during actual operation. When a sensor is subjected to instantaneous or continuous interference, its abnormal data still participates in the fusion calculation with the same weight, contaminating the final output. Second, there is a lack of systematic indicators for judging the quality of sensor measurements. Most schemes only identify anomalies based on whether a single measurement value exceeds the reasonable range or deviates from adjacent time values, failing to comprehensively evaluate the real-time operating condition of sensors from multiple dimensions such as the degree of conformity between the measured value and historical statistical characteristics, the consistency of the changing trends among multiple sensors, and the fluctuation state of the measured values ​​over continuous periods. Third, the methods for repairing fused data sequences are relatively simple. For sequence gaps caused by communication interruptions, temporary power outages, or synchronous failures of all sensors, the common practice is to leave them blank or use linear interpolation to complete them. The ability to repair continuous multi-point gaps is limited, and the repair results may deviate from the actual water level change patterns. Fourth, the automatic diagnostic function for sensor operating status is relatively weak. When sensors experience slow performance degradation or occasional anomalies, the system often fails to detect them in a timely manner and alert maintenance personnel, causing the problematic equipment to remain in a sub-healthy operating state for a long time, which cumulatively affects the overall quality of monitoring data.

[0006] The present invention aims to solve the technical problems existing in the prior art. To this end, a data processing system and method for a river channel liquid level sensor are proposed. Summary of the Invention

[0007] The purpose of this invention is to provide a data processing system and method for a river channel liquid level sensor, so as to solve the technical problems existing in the prior art.

[0008] By adopting the above technical solution, the present invention has the following beneficial effects:

[0009] This invention provides a data processing system for a river channel liquid level sensor.

[0010] The system comprises a sensor layer, a data acquisition and transmission layer, a data processing layer, and an output layer. Wherein:

[0011] The sensor layer is configured on the monitoring section of the river or channel, including at least two level sensors employing different measurement principles to acquire raw level measurement data of the same monitoring section. The different measurement principles include at least two of the following: ultrasonic, pressure, radar, capacitance, float, or laser methods. This redundant deployment of multiple sensor types provides a foundation for subsequent differentiated quality assessment and data fusion.

[0012] The data acquisition and transmission layer is communicatively connected to the sensor layer, and is used to receive the raw liquid level measurement data and transmit the raw liquid level measurement data to the data processing layer according to a preset sampling period.

[0013] The data processing layer is the core of the system and includes the following modules:

[0014] Data preprocessing module: Performs outlier removal, missing value filling and time alignment on the raw liquid level measurement data of each liquid level sensor, and outputs a standardized liquid level data sequence.

[0015] Data quality assessment module: For each level sensor at each sampling time, calculates multi-dimensional quality assessment parameters, including at least confidence assessment value, correlation assessment value, and stability assessment value, based on the standardized level data.

[0016] Adaptive fusion weight calculation module: Based on the multidimensional quality assessment parameters of each liquid level sensor, with the goal of minimizing the sum of squared weighted deviations between the fusion result and the measurement values ​​of each sensor, the module dynamically calculates the adaptive fusion weight of each liquid level sensor at each sampling time.

[0017] Multi-source data fusion module: Based on the adaptive fusion weights of each liquid level sensor, the standardized liquid level data of each liquid level sensor at the same sampling time are weighted and fused to obtain the fused liquid level value.

[0018] Sequence Construction and Repair Module: Constructs the fused liquid level values ​​at each sampling time into a fused liquid level sequence in chronological order, repairs missing sampling points in the fused liquid level sequence, and outputs the target liquid level sequence.

[0019] The output layer is used to output the target liquid level sequence to an external monitoring system or storage device.

[0020] Preferably, in the data quality assessment module, the confidence assessment value is determined based on the deviation between the current measurement value of the liquid level sensor and its historical mean; the correlation assessment value is determined based on the correlation coefficient between the current sensor measurement value and the trend of other sensor measurement values; and the stability assessment value is determined based on the variance or standard deviation of the sensor measurement values ​​at multiple consecutive times.

[0021] Preferably, in the adaptive fusion weight calculation module, the optimization problem aimed at minimizing the sum of squared weighted deviations between the fusion result and the measurements of each sensor is expressed as:

[0022]

[0023] The constraints are:

[0024]

[0025] in, Indicates the sampling time. Indicates the number of liquid level sensors. Indicates the sensor number, Indicates the first A liquid level sensor at time Adaptive fusion weights, Indicates time The fusion liquid level value, Indicates the first A liquid level sensor at time Standardized liquid level data.

[0026] The adaptive fusion weights of each sensor are calculated using a normalized comprehensive quality index:

[0027]

[0028] in, Indicates the first Each liquid level sensor at the sampling time Adaptive fusion weights, Indicates the number of liquid level sensors. Indicates the first Each liquid level sensor at the sampling time Comprehensive quality indicators;

[0029] Among them, the comprehensive quality indicators are:

[0030] in, , , They represent the first Each liquid level sensor at the sampling time The confidence score, correlation score, and stability score are as follows: , , The preset weighting coefficients are used, and they satisfy the following conditions: + + =1.

[0031] Preferably, in the multi-source data fusion module, the fused liquid level value is calculated as follows:

[0032]

[0033] in, Indicates the sampling time The fusion liquid level value, Indicates the number of liquid level sensors. Indicates the first Each liquid level sensor at the sampling time Adaptive fusion weights, Indicates the first Each liquid level sensor at the sampling time Standardized liquid level data.

[0034] Preferably, in the sequence construction and repair module, linear interpolation is used to repair internal missing sampling points; and historical trend curve fitting is used to repair multiple consecutive missing sampling points.

[0035] Furthermore, the data processing layer may also include a fault diagnosis and alarm module, which is used to determine that the sensor is abnormal and generate an alarm signal when the multidimensional quality assessment parameter of any liquid level sensor is continuously lower than a preset threshold.

[0036] A method for processing data from a river channel liquid level sensor

[0037] The method is applied to the above system and includes the following steps:

[0038] Step S1: Deploy at least two liquid level sensors using different measurement principles at the monitoring section of the river or channel to obtain the raw liquid level measurement data of the same monitoring section;

[0039] Step S2: Perform outlier removal, missing value imputation, and time alignment on the raw liquid level measurement data to output a standardized liquid level data sequence;

[0040] Step S3: For the standardized liquid level data of each liquid level sensor at each sampling time, calculate a multi-dimensional quality assessment parameter that includes at least the confidence assessment value, the correlation assessment value, and the stability assessment value.

[0041] Step S4: Based on the multidimensional quality assessment parameters of each liquid level sensor, with the goal of minimizing the weighted sum of squared deviations between the fusion result and the measurement values ​​of each sensor, dynamically calculate the adaptive fusion weight of each liquid level sensor at each sampling time.

[0042] Step S5: Based on the adaptive fusion weights of each liquid level sensor, perform weighted fusion calculation on the standardized liquid level data of each liquid level sensor at the same sampling time to obtain the fused liquid level value;

[0043] Step S6: Construct a fused liquid level sequence from the fused liquid level values ​​at each sampling time in chronological order, repair the missing sampling points in the fused liquid level sequence, and output the target liquid level sequence.

[0044] Preferably, the calculation method for each evaluation value in step S3 is as follows:

[0045] Calculate the confidence assessment value :

[0046]

[0047] in, For the first A liquid level sensor at time Standardized liquid level data, For the first The average measurement value of each liquid level sensor within a preset historical time period. This is the confidence decay coefficient;

[0048] Calculate the correlation assessment value :

[0049]

[0050] in, For the first The liquid level sensor and the first A liquid level sensor at time Pearson correlation coefficient between the trends of change in nearby measurements;

[0051] Calculate the stability evaluation value :

[0052]

[0053] in, For the first A liquid level sensor at time The standard deviation of the measurements over the previous k consecutive sampling times, This is the stability attenuation coefficient.

[0054] Preferably, the method further includes step S7: performing real-time fault diagnosis based on the multi-dimensional quality assessment parameters of each liquid level sensor; when the comprehensive quality index of any sensor is continuously lower than a preset threshold, an abnormality is determined and an alarm signal is generated, while automatically adjusting the fusion weight allocation of the remaining normal sensors.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] First, the determination of fusion weights no longer relies on a fixed allocation method, but is dynamically adjusted based on the actual measurement quality of each sensor at the current sampling moment. When a sensor's measurement value becomes abnormal due to water surface fluctuations, siltation, or floating debris interference, its corresponding confidence, correlation, and stability assessment results will automatically decrease, thereby reducing the weight of that sensor in the fusion calculation and minimizing the interference of abnormal data on the final liquid level value. This mechanism allows the fusion result to adaptively change with the operating status of each sensor, improving the robustness of the data output.

[0057] Secondly, by constructing a multi-dimensional quality assessment system that includes confidence level, correlation, and stability, the real-time status of the sensor can be judged from multiple perspectives. Simply relying on whether the measured value deviates from the historical average or is consistent with other sensors often fails to fully reflect the sensor's true operating condition. By comprehensively considering all three factors, it is possible to more accurately identify whether the sensor is being affected by environmental factors or experiencing performance degradation, providing a relatively reliable reference for subsequent weight allocation and fault diagnosis.

[0058] Third, in the process of constructing the fused liquid level sequence, different repair methods are adopted for the missing individual sampling points and the missing multiple consecutive points, so that the final output target liquid level sequence maintains the continuity of time while the repair result is more in line with the actual water level change pattern, which is beneficial to subsequent hydrological analysis, flood forecasting and water resource scheduling.

[0059] Fourth, by employing redundant sensor deployments based on different measurement principles, coupled with the aforementioned data processing methods, the impact of a single sensor failure or interference on the overall output of the monitoring system can be reduced. Even if a sensor temporarily fails under extreme conditions, the system can still rely on the remaining normally functioning sensors to output usable liquid level data, thereby improving the availability of the monitoring system under harsh hydrological conditions.

[0060] Fifth, by continuously monitoring the quality indicators of each sensor, the system can promptly generate alarm prompts when sensor performance shows a trend of decline or malfunctions, providing on-site maintenance personnel with clearer maintenance directions, which helps to shorten the time for fault detection and handling, and reduce the data quality risks caused by equipment being in a sub-healthy state for a long time. Attached Figure Description

[0061] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0062] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention.

[0063] Figure 2 This is a schematic diagram of the data processing flow of the method of the present invention.

[0064] Figure 3 This is a schematic diagram illustrating the principle of multi-sensor adaptive fusion weight calculation.

[0065] Figure 4 This is a diagram showing the relationship between the multidimensional evaluation parameters of the data quality assessment module.

[0066] Figure 5 A diagram illustrating the application deployment scenario for the system. Detailed Implementation

[0067] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0068] Example 1

[0069] This embodiment uses a river hydrological monitoring station as an application scenario to introduce the basic structure and workflow of the system of the present invention.

[0070] See Figure 1 The present invention provides a data processing system for a river channel liquid level sensor, comprising a sensor layer, a data acquisition and transmission layer, a data processing layer, and an output layer.

[0071] The sensor layer is mounted on a support frame at the river monitoring section. In this embodiment, three types of liquid level sensors using different measurement principles are arranged: ultrasonic liquid level sensors, pressure liquid level sensors, and radar liquid level sensors. Ultrasonic liquid level sensors calculate liquid level height using the time difference between ultrasonic wave emission and echo reception, offering advantages such as non-contact operation and easy installation. However, measurement accuracy is affected by severe water surface fluctuations or the presence of floating debris. Pressure liquid level sensors measure liquid level by converting the hydrostatic pressure at the bottom of the water body into liquid level height. The measurement results are unaffected by floating debris, but the sensors need to be submerged, and errors may occur when there is severe siltation. Radar liquid level sensors measure liquid level using the principle of electromagnetic wave reflection, offering high measurement accuracy and being unaffected by temperature changes. However, they have strict requirements for the installation angle, and the signal may attenuate under heavy rainfall conditions. The differences in the measurement principles of these three sensors result in different patterns of influence from various environmental factors, providing a physical basis for subsequent data quality assessment and adaptive fusion.

[0072] The data acquisition and transmission layer includes a data acquisition terminal installed at the monitoring site. This terminal is connected to each liquid level sensor through a standard analog signal interface or digital communication interface. It reads the measured values ​​of each sensor in real time according to a preset sampling period and transmits the data to the data processing layer deployed remotely or locally through a wireless communication module.

[0073] The data processing layer is the core part of this invention. After receiving multiple raw liquid level measurement data from the data acquisition and transmission layer, it sequentially performs the following processing steps.

[0074] First, the data preprocessing module performs outlier removal, missing value imputation, and time alignment on the received raw liquid level measurement data. The outlier removal stage analyzes the statistical characteristics of each sensor's data sequence to remove outliers that significantly exceed the reasonable range of variation. For example, if a sensor's measurement value experiences a jump far exceeding the rate of physical change that can be explained within a short period, that value is marked as an outlier and removed. The missing value imputation stage temporarily fills in data gaps caused by communication interruptions or sensor response failures using interpolation methods based on temporal proximity to ensure the continuity of subsequent processing. The time alignment stage unifies the data collected by different sensors onto the same time series benchmark, eliminating time deviations caused by minor differences in the sampling times of each sensor.

[0075] Next, the data quality assessment module calculates multidimensional quality assessment parameters for each liquid level sensor at each sampling time for the standardized liquid level data, including confidence assessment value, correlation assessment value, and stability assessment value.

[0076] Confidence rating Reflects the measurement value of the i-th sensor at the current sampling time t. The confidence level is relative to its historical statistical characteristics. When the deviation between the measured value and its historical mean is small, it indicates that the measured value conforms to the typical operating conditions of the sensor, and the confidence level is high; conversely, a large deviation reduces the confidence level. The formula for calculating the confidence level is:

[0077]

[0078] In the formula, Let be the average value of the measurements taken by the i-th sensor over a preset historical time period. This is the confidence decay coefficient, used to adjust the rate at which the bias decays the confidence level. This formula allows the confidence assessment value to vary between 0 and 1; the closer the value is to 1, the more reliable the measurement.

[0079] Correlation assessment value This reflects the degree of consistency between the trend of the measured value of the i-th sensor at the current sampling time t and the trends of other sensors. When different sensors consistently reflect the same trend of liquid level change, it indicates that all sensors are operating normally; if the trend of a certain sensor is significantly inconsistent with that of other sensors, that sensor may be malfunctioning. The formula for calculating the correlation evaluation value is:

[0080]

[0081] In the formula, For the i-th sensor and the i-th The Pearson correlation coefficient between the measurement trends of the i-th sensor and all other sensors around time t, where n is the total number of sensors. This formula calculates the average of the correlation coefficients of the i-th sensor with all other sensors to obtain its correlation assessment value.

[0082] Stability assessment value This reflects the degree of fluctuation in the measured values ​​of the i-th sensor over multiple consecutive sampling times. When the variance or standard deviation of the measured values ​​is small over a short period, it indicates that the sensor is operating stably; large fluctuations indicate that the sensor may be affected by environmental disturbances or is unstable. The formula for calculating the stability assessment value is:

[0083]

[0084] In the formula, Let be the standard deviation of the measurements of the i-th sensor over the k consecutive sampling times prior to time t. This is the stability decay coefficient. This formula allows the stability assessment value to vary between 0 and 1.

[0085] See Figure 4 The three evaluation values ​​described above characterize the sensor's measurement quality from different dimensions, and together they constitute a comprehensive description of the sensor's real-time state. Based on this, the three evaluation values ​​are weighted and summed to form a comprehensive quality index. :

[0086]

[0087] In the formula, , , The preset weighting coefficients represent the importance of the confidence assessment value, correlation assessment value, and stability assessment value in the overall quality evaluation, and satisfy the constraints. + + =1. In this embodiment, the values ​​of each coefficient can be adjusted according to the actual needs of different application scenarios. For example, in scenarios where the reliability of sensors varies greatly, the weight of the confidence assessment value can be appropriately increased; in scenarios with strong environmental interference, the weight of the stability assessment value can be appropriately increased.

[0088] See Figure 3 After obtaining the comprehensive quality index of each sensor, the adaptive fusion weight calculation module calculates the adaptive fusion weight of each sensor at the current sampling time according to the following formula:

[0089]

[0090] The physical meaning of this formula is: the proportion of each sensor's overall quality index to the sum of all sensor quality indices is used as the fusion weight of that sensor. Sensors with better quality have higher overall quality indices and are assigned a larger fusion weight; sensors with poorer quality have correspondingly lower weights. The theoretical basis of this adaptive weight allocation mechanism is to minimize the sum of squared weighted deviations between the fusion result and the measurements of each sensor, that is:

[0091]

[0092] The constraints are:

[0093] .

[0094] The optimal solution to the above optimization problem corresponds exactly to the weight allocation scheme expressed in the form of a comprehensive quality index ratio. Therefore, the weight calculation method adopted in this invention has a rigorous mathematical basis for optimality.

[0095] Subsequently, the multi-source data fusion module uses the aforementioned adaptive fusion weights to perform weighted fusion calculations on the standardized liquid level data from each sensor:

[0096]

[0097] The final result It is the fused liquid level value at sampling time t, which integrates measurement information from multiple sensors and has high reliability because the weights are dynamically allocated according to real-time quality.

[0098] The sequence construction and repair module organizes the fused liquid level values ​​from each sampling time into a fused liquid level sequence in chronological order. For cases where all sensors fail to provide valid data at a certain sampling time due to some reason, resulting in a missing sequence, this module repairs it in the following way: When the missing sampling point is located within the sequence, a linear interpolation method based on temporal proximity is used:

[0099]

[0100] When multiple consecutive sampling points are missing, the data is fitted and repaired based on the historical liquid level change trend curve of the monitoring section. After repair, the target liquid level sequence is sent to the data server of the remote monitoring center through the output layer for use in hydrological analysis, early warning and forecasting, and other applications.

[0101] In addition, this system may optionally include a fault diagnosis and alarm module. This module continuously monitors the multi-dimensional quality assessment parameters of each liquid level sensor, and when the comprehensive quality index of any sensor reaches a certain level... If the quality remains below a preset threshold for an extended period of time, the sensor is deemed to be in an abnormal operating state. An alarm signal is generated, and the fusion weight allocation of the remaining normally functioning sensors is automatically adjusted. (See also...) Figure 5 After receiving alarm information through the monitoring terminal, maintenance personnel can promptly go to the site to inspect and maintain the equipment.

[0102] See Figure 2 The above process constitutes the main steps of the method of the present invention: step S1 collects raw liquid level data from multiple sensors; step S2 performs data preprocessing; step S3 calculates multidimensional quality assessment parameters; step S4 dynamically calculates adaptive fusion weights; step S5 performs weighted fusion of multi-source data; step S6 constructs a fused liquid level sequence and repairs missing sampling points, and finally outputs the target liquid level sequence.

[0103] Example 2

[0104] This embodiment uses a water diversion channel as an application scenario to demonstrate the deployment method of the system of the present invention at multiple monitoring sections along a long-distance channel, as well as the dynamic adjustment process of adaptive fusion weights under different environmental conditions.

[0105] In this application scenario, several liquid level monitoring sections are set up along the water diversion channel, and each section is equipped with a combination of liquid level sensors using different measurement principles. This embodiment takes one of the monitoring sections as an example to illustrate the working process of the system. This section is equipped with both ultrasonic liquid level sensors and pressure liquid level sensors.

[0106] During normal channel operation, the water flow is stable with minimal surface fluctuations. Both ultrasonic and pressure-type level sensors operate in favorable conditions under these conditions. For ultrasonic level sensors, the calm water surface and stable echo signal result in a high confidence level. and stability assessment value All remained at a high level. For pressure-type liquid level sensors, due to less siltation and relatively stable water density, their various quality assessment values ​​were also high. The correlation assessment value between the two sensors... and The values ​​are also relatively high, indicating that the trends in the measured values ​​of both sensors are highly consistent. Under these conditions, the overall quality index of the two sensors is... and Numerical values ​​are similar, and each is assigned adaptive fusion weights. and They are roughly equal, and the fusion result is close to the arithmetic mean of the two.

[0107] The situation changes when there are many floating objects in the channel. When these objects pass below the monitoring section, they interfere with the normal reflection of the ultrasonic signal, causing the ultrasonic level sensor to produce abnormal readings. During this period, the ultrasonic level sensor readings show significant jumps, with a marked increase in deviation from their historical average, leading to a drop in confidence level. The value decreased rapidly; simultaneously, the fluctuation range of the measured value increased significantly, and the stability assessment value... It also decreases accordingly; the trend of its measured value no longer matches the trend of the pressure level sensor, and the correlation assessment value... The overall quality indicators of ultrasonic level sensors also declined. Significantly reduced. Pressure-type liquid level sensors, unaffected by floating debris, maintain stable quality assessment values ​​across the board, resulting in a higher overall quality index. It remains at a high level.

[0108] Based on the changes in the aforementioned quality indicators, the adaptive fusion weight calculation module automatically responds: the fusion weight of the ultrasonic level sensor. Significantly reduced fusion weights of pressure level sensors The corresponding increase. Fusion result. This system relies more heavily on measurements from pressure level sensors, effectively mitigating the impact of abnormal data from ultrasonic level sensors caused by floating debris on the final results. This process requires no manual intervention and is entirely automated, completed by the system based on real-time quality assessment results.

[0109] Once the floating debris passed the monitoring section and the water surface returned to calm, the ultrasonic level sensor readings returned to normal, various quality assessment values ​​gradually recovered, and the fusion weights also returned to normal levels. Throughout the process, the fused level sequence output by the system maintained good continuity and accuracy, without significant fluctuations or gaps due to sensor anomalies in localized periods.

[0110] This embodiment further illustrates that, through multi-dimensional quality assessment and adaptive weight allocation mechanisms, the present invention enables the system to automatically adjust its data processing strategy in the face of dynamically changing environmental disturbances, ensuring the reliability of the output results. In multi-section applications of long-distance channels, each section can independently run the above processing flow, and the processing results of each section can be summarized and collaboratively analyzed at a higher level, providing comprehensive and accurate liquid level data support for water volume scheduling and management throughout the channel.

[0111] Example 3

[0112] This embodiment uses flood season monitoring of a mountainous river as an application scenario to highlight the data processing capabilities and fault diagnosis functions of the present invention under extreme hydrological conditions.

[0113] Mountainous rivers are characterized by rapid flood rise and fall, large water level fluctuations, and dramatic changes in sediment content, placing high demands on the adaptability of liquid level monitoring systems. In this embodiment, three types of devices are configured at a monitoring section of the river: a radar liquid level sensor, a pressure liquid level sensor, and a float-type liquid level sensor. The radar liquid level sensor is a non-contact device installed high on the riverbank; the pressure liquid level sensor is placed at the bottom of the riverbed; and the float-type liquid level sensor is installed in a stilling well and connected to the river channel via a connecting pipe.

[0114] During the period of low water levels before the flood season, all three types of sensors operate normally. Due to the low water level, the hydrostatic pressure signal amplitude of the pressure-type liquid level sensor is small, resulting in a relatively low signal-to-noise ratio and a lower stability assessment value. Slightly lower than the other two sensors, but still within an acceptable range. The radar level sensor and float-type level sensor maintained high quality assessment values ​​across all categories. The overall quality indicators of the three sensors showed little difference, and the fusion weighting was relatively evenly distributed.

[0115] As rainfall begins, river levels start to rise. In the initial stages of this rise, increased water flow velocity triggers sediment buildup in the riverbed. At this time, the measuring end of a pressure-type liquid level sensor may be affected by sediment erosion or deposition, impacting its confidence level assessment. A certain degree of decline occurred; simultaneously, due to increased water flow turbulence, the water surface fluctuations faced by the radar level sensor intensified, leading to a decrease in its stability assessment value. The level decreases somewhat. However, because the float-type level sensor is installed in a still water well, which has wave-damping capabilities, its measurements are relatively stable, and its overall quality index remains at a high level. During this stage, the system's fusion weights automatically tilt towards the float-type level sensor, ensuring that the fusion result primarily reflects the water level changes in the still water well.

[0116] When floodwaters reach peak levels, the sediment load in rivers increases dramatically. High-sediment-laden flows can cause the following problems: First, the measuring port of pressure-type level sensors may become clogged with sediment, causing their readings to deviate significantly from the actual water level. Second, the connecting pipe of float-type level sensors may become blocked due to sediment deposition, leading to differences in water levels inside and outside the stilling well, resulting in measurement lag. Under these extreme conditions, radar level sensors, as non-contact devices, are not directly affected by sediment deposition, although their measurement accuracy is also affected by surface turbulence and rainfall attenuation. At this point, the overall quality indicators of the various sensors show significant differences: the overall quality indicator of pressure-type level sensors becomes very low due to a significant decrease in both confidence and correlation; the overall quality indicator of float-type level sensors is moderately low due to measurement lag and decreased correlation; and the overall quality indicator of radar level sensors is relatively the highest. Based on this, the adaptive fusion weight calculation module assigns the highest weight to the radar level sensor, the second highest weight to the float-type level sensor, while the weight of the pressure-type level sensor is significantly suppressed.

[0117] After the floodwaters recede, special attention needs to be paid to the recovery of the sensor's condition. If a pressure-type liquid level sensor is indeed blocked by silt, its measured values ​​will remain at abnormal levels for a considerable period, affecting its overall quality. The readings will remain below the normal threshold. The fault diagnosis and alarm module monitors this indicator and automatically generates an alarm signal when the conditions are met, prompting maintenance personnel to inspect and clean the pressure-type level sensor. Similarly, if the connecting pipe of a float-type level sensor is blocked, its measured value will lag significantly behind other sensors, and the correlation assessment value will remain low, also triggering a fault alarm. After maintenance personnel address the issue on-site based on the alarm information, all sensors return to normal operation, the overall quality index recovers, and the system automatically resumes its normal weight allocation scheme.

[0118] This embodiment fully demonstrates the adaptability and fault self-diagnosis capability of the present invention under extreme hydrological conditions. During floods, the system can dynamically adjust the fusion strategy based on the real-time measurement quality of each sensor, ensuring that the output liquid level data reflects the true water level changes to the greatest extent possible. After the flood, the system's fault diagnosis function can proactively detect potential sensor performance degradation issues, providing clear guidance for subsequent maintenance work and effectively reducing the data quality risks caused by equipment being in a sub-optimal state for a long time.

[0119] Furthermore, this embodiment reveals a significant advantage of the multi-sensor fusion scheme: even if one sensor completely fails under specific operating conditions, as long as the other sensors continue to function normally, the system can still continuously output highly reliable liquid level data, avoiding the data interruption problem caused by equipment failure common in single-sensor schemes. This characteristic has significant practical value for scenarios with extremely high requirements for data continuity, such as flood control during the flood season.

[0120] In summary, the river channel liquid level sensor data processing system and method provided by this invention effectively overcomes the shortcomings of existing technologies, such as fixed fusion weights, lack of systematic quality assessment, and insufficient sequence repair capabilities, by establishing a multi-dimensional quality assessment system, implementing an adaptive weight fusion mechanism, constructing a complete sequence repair process, and integrating fault diagnosis functions. This significantly improves the accuracy, continuity, and reliability of river channel liquid level monitoring data. The system is applicable to liquid level monitoring scenarios in various water bodies such as rivers, channels, lakes, and reservoirs, and has good prospects for promotion and practical value.

[0121] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0122] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A data processing system for a river channel liquid level sensor, characterized in that, include: The sensor layer, configured on the monitoring section of a river or channel, includes at least two liquid level sensors employing different measurement principles, used to acquire raw liquid level measurement data of the same monitoring section respectively. The different measurement principles include at least two of the following: ultrasonic method, pressure method, radar method, capacitance method, float method, or laser method. The data acquisition and transmission layer is communicatively connected to the sensor layer and is used to receive the raw liquid level measurement data and transmit the raw liquid level measurement data to the data processing layer according to a preset sampling period. A data processing layer is used to process the raw liquid level measurement data, and the data processing layer includes: The data preprocessing module is used to perform outlier removal, missing value filling and time alignment processing on the raw liquid level measurement data of each of the liquid level sensors, and output a standardized liquid level data sequence. The data quality assessment module is used to calculate, for each of the liquid level sensors at each sampling time, a multi-dimensional quality assessment parameter including at least a confidence assessment value, a correlation assessment value, and a stability assessment value for the standardized liquid level data. An adaptive fusion weight calculation module is used to dynamically calculate the adaptive fusion weight of each liquid level sensor at each sampling time based on the multidimensional quality assessment parameters of each liquid level sensor, with the goal of minimizing the sum of squared weighted deviations between the fusion result and the measured values ​​of each sensor. The multi-source data fusion module is used to perform weighted fusion calculation on the standardized liquid level data of each liquid level sensor at the same sampling time according to the adaptive fusion weight of each liquid level sensor to obtain the fused liquid level value; The sequence construction and repair module is used to construct the fused liquid level values ​​at each sampling time into a fused liquid level sequence in chronological order, repair the missing sampling points in the fused liquid level sequence, and output the target liquid level sequence. And an output layer for outputting the target liquid level sequence to an external monitoring system or storage device.

2. The system according to claim 1, characterized in that, In the data quality assessment module: The confidence assessment value is determined based on the deviation between the measured value of the liquid level sensor at the current sampling time and the average value of the sensor's measured values ​​over a preset historical time period. The smaller the deviation, the greater the confidence assessment value. The correlation evaluation value is determined based on the correlation coefficient between the trend of the liquid level sensor's measured value at the current sampling time and the trend of the measured value of other liquid level sensors at the same time. The larger the correlation coefficient, the larger the correlation evaluation value. The stability assessment value is determined based on the variance or standard deviation of the measurement values ​​of the liquid level sensor at multiple consecutive sampling times. The smaller the variance or standard deviation, the greater the stability assessment value.

3. The system according to claim 1, characterized in that, In the adaptive fusion weight calculation module, the optimization problem aimed at minimizing the sum of squared weighted deviations between the fusion result and the measurements of each sensor is expressed as: The constraints are: in, Indicates the sampling time. Indicates the number of liquid level sensors. Indicates the sensor number, Indicates the first A liquid level sensor at time Adaptive fusion weights, Indicates time The fusion liquid level value, Indicates the first A liquid level sensor at time Standardized liquid level data.

4. The system according to claim 3, characterized in that, In the adaptive fusion weight calculation module, the adaptive fusion weight of each liquid level sensor is calculated according to the following formula: in, Indicates the first Each liquid level sensor at the sampling time Adaptive fusion weights, Indicates the number of liquid level sensors. Indicates the first Each liquid level sensor at the sampling time The comprehensive quality index, the comprehensive quality index The calculation formula is: in, , , They represent the first Each liquid level sensor at the sampling time The confidence score, correlation score, and stability score are as follows: , , The preset weighting coefficients are used, and they satisfy the following conditions: + + =1.

5. The system according to claim 1, characterized in that, In the multi-source data fusion module, the fused liquid level value is calculated according to the following formula: in, Indicates the sampling time The fusion liquid level value, Indicates the number of liquid level sensors. Indicates the first Each liquid level sensor at the sampling time Adaptive fusion weights, Indicates the first Each liquid level sensor at the sampling time Standardized liquid level data.

6. The system according to claim 1, characterized in that, In the sequence construction and repair module, the method for repairing missing sampling points in the fused liquid level sequence is as follows: When the missing sampling point is located within the fused liquid level sequence, a linear interpolation method based on temporal proximity is used for repair. The repair formula is as follows: The sampling time corresponding to the missing data. This represents the most recent valid sampling time before the missing sampling point. This represents the most recent valid sampling time after the missing sampling point. In order to be in The fusion liquid level value after interpolation repair at any given time; The known effective fusion liquid level value; When multiple consecutive sampling points are missing, the missing points are fitted and repaired based on the historical liquid level change trend curve of the monitored section.

7. The system according to claim 1, characterized in that, The data processing layer also includes a fault diagnosis and alarm module, which is used to determine that the corresponding liquid level sensor has an abnormal working state and generate an alarm signal when at least one of the multi-dimensional quality assessment parameters of each liquid level sensor is continuously lower than a preset threshold.

8. A method for processing data from a river channel liquid level sensor, applied to the system described in any one of claims 1 to 7, characterized in that, Includes the following steps: Step S1: Deploy at least two liquid level sensors using different measurement principles at the monitoring section of the river or channel to obtain the original liquid level measurement data of the same monitoring section; Step S2: Perform outlier removal, missing value filling, and time alignment processing on the original liquid level measurement data to output a standardized liquid level data sequence; Step S3: For the standardized liquid level data of each liquid level sensor at each sampling time, calculate a multidimensional quality assessment parameter that includes at least a confidence assessment value, a correlation assessment value, and a stability assessment value. Step S4: Based on the multidimensional quality assessment parameters of each liquid level sensor, with the goal of minimizing the weighted sum of squared deviations between the fusion result and the measured values ​​of each sensor, dynamically calculate the adaptive fusion weight of each liquid level sensor at each sampling time. Step S5: Based on the adaptive fusion weights of each liquid level sensor, perform weighted fusion calculation on the standardized liquid level data of each liquid level sensor at the same sampling time to obtain the fused liquid level value; Step S6: Construct a fused liquid level sequence from the fused liquid level values ​​at each sampling time in chronological order, repair missing sampling points in the fused liquid level sequence, and output the target liquid level sequence.

9. The method according to claim 8, characterized in that, Step S3 specifically includes: Calculate the confidence assessment value : in, For the first A liquid level sensor at time Standardized liquid level data, For the first The average measurement value of each liquid level sensor within a preset historical time period. This is the confidence decay coefficient; Calculate the correlation assessment value : in, For the first The liquid level sensor and the first A liquid level sensor at time Pearson correlation coefficient between the trends of change in nearby measurements; Calculate the stability evaluation value : in, For the first A liquid level sensor at time The standard deviation of the measurements over the previous k consecutive sampling times, This is the stability attenuation coefficient.

10. The method according to claim 8 or 9, characterized in that, The method also includes step S7, which involves real-time fault diagnosis based on the multi-dimensional quality assessment parameters of each liquid level sensor, when the comprehensive quality index of any liquid level sensor... If the level sensor remains below the preset quality threshold for more than a preset time period, it is determined that the level sensor is in an abnormal working state, an alarm signal is generated, and the fusion weight allocation of the remaining normally functioning level sensors is automatically adjusted.

Images