A remote metering monitoring system based on heterogeneous sensor cooperative detection

Abstract

The application discloses a remote metering monitoring system based on heterogeneous sensor cooperative detection and belongs to the technical field of sensor monitoring. The system comprises a heterogeneous sensing layer, a data processing layer and a decision control layer. The heterogeneous sensing layer comprises a high-precision sensor group, a high-stability sensor group and an environmental parameter acquisition unit, is used for acquiring original data in a target area and constructing an environmental coupling matrix. The data processing layer comprises a data alignment unit and an adaptive calibration engine unit, is used for performing data alignment on the original data collected by the heterogeneous sensing layer, performing adaptive calibration on the aligned original data by using the environmental coupling matrix, and obtaining target data. The decision control layer is used for performing multi-scale anomaly detection on the target data according to a multi-modal anomaly detection algorithm and outputting corresponding alarm signals according to the detection results. The application solves the problem of low reliability of the single-sensor system used in the existing environmental monitoring field.

Description

Technical Field

[0001] This invention belongs to the field of sensor monitoring technology, specifically relating to a remote metering and monitoring system based on collaborative detection of heterogeneous sensors. Background Technology

[0002] In the field of environmental monitoring, photochemical sensors are widely used in atmospheric composition analysis and water pollutant detection due to their high-precision substance concentration detection capabilities. Based on this characteristic, existing single-sensor systems can directly contact the measured environment through the sensor probe, converting physical or chemical signals into electrical signals, which are then processed by analog-to-digital conversion and algorithms to output monitoring results. However, in practical applications, photochemical sensors generally exhibit both high accuracy and low stability. When affected by environmental factors, existing single-sensor systems cannot accurately determine whether the current measurement fluctuations are caused by equipment failure or environmental interference. Therefore, it is difficult to achieve the measurement accuracy under ideal operating conditions during monitoring. This necessitates periodic manual calibration of the single-sensor system, increasing labor costs and requiring monitoring interruptions during calibration, thus failing to meet the continuous online monitoring requirements of some environments. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a remote metering and monitoring system based on heterogeneous sensor collaborative detection to solve the above-mentioned technical problems.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A remote metering and monitoring system based on heterogeneous sensor collaborative detection includes:

[0006] Heterogeneous sensing layer, data processing layer, and decision control layer;

[0007] The heterogeneous sensing layer includes a high-precision sensor group, a high-stability sensor group, and an environmental parameter acquisition unit, which are used to acquire raw data within the target area and construct an environmental coupling matrix; the raw data includes raw data from the high-precision sensor group and raw data from the high-stability sensor group.

[0008] The data processing layer includes a data alignment unit and an adaptive calibration engine unit, which are used to align the raw data collected by the heterogeneous sensing layer and adaptively calibrate the aligned raw data using the environment coupling matrix to obtain the target data.

[0009] The decision control layer is configured to perform multi-scale anomaly detection on the target data according to the multi-modal anomaly detection algorithm, and output corresponding alarm signals according to the detection results; wherein the multi-scale anomaly detection includes time domain residual detection, frequency domain energy entropy detection and trend similarity detection.

[0010] Further, the high-precision sensor group includes a photochemical sensor and a quantum sensor, with a precision of 0.5%-1%FS, for collecting high-precision sensor group original data and realizing ppb-level trace substance detection.

[0011] The high-stability sensor group includes an electrochemical sensor and a semiconductor sensor, with a precision of 3%-5%FS, for collecting high-stability sensor group original data and providing a reference signal.

[0012] The environmental parameter acquisition unit includes a deployed temperature and humidity and barometric pressure sensor network, for constructing an environmental coupling matrix and realizing differentiated compensation support requirements.

[0013] Further, the data alignment unit is configured to solve the problem of non-uniformity of the time and space reference of the high-precision sensor group and the high-stability sensor group in the heterogeneous sensor layer, including time synchronization and space mapping; wherein the PTP protocol is used to realize The level alignment solves the time synchronization problem, and the three-dimensional coordinate compensation mechanism of the sensor layout position is used to solve the space mapping problem.

[0014] Further, the adaptive calibration operation includes:

[0015] The aligned original data is obtained, denoted as the to-be-calibrated data;

[0016] The to-be-calibrated data is analyzed for environmental coupling degree using the environmental coupling matrix, to obtain a target environmental coupling degree.

[0017] If the target environmental coupling degree is not greater than a preset coupling threshold, the to-be-calibrated data is subjected to adaptive calibration operation through a bidirectional calibration model, to obtain target data.

[0018] Further, the bidirectional calibration model is:

[0019]

[0020] Wherein, and respectively represent the target data of the calibrated high-precision sensor group and the high-stability sensor group, and respectively represent the dynamic confidence coefficients of the corresponding target data, and and represents an environmental compensation function for improving short-term monitoring accuracy, represents a drift compensation function for correcting errors caused by long-term drift, represents an environmental parameter;

[0021]

[0022] wherein, represents the difference response coefficient of the high-stability sensor group to the class of environmental parameters, represents the number of classes of environmental parameters, represents the offset of the class of environmental parameters; , represents a drift function constructed based on historical environmental parameter data, represents a drift rate;

[0023]

[0024] wherein, is a natural constant, and respectively represent the sensor type adjustment factor of the high-precision sensor group and the high-stability sensor group, represents a real-time comprehensive reliability index, represents the number of evaluation dimensions, represents the weight coefficient of the evaluation dimension, represents the real-time evaluation value of the evaluation dimension.

[0025] Further, the number of evaluation dimensions is three, and three-dimensional evaluation dimension indicators are established, including: time consistency, environmental coupling degree, and historical confidence;

[0026] wherein, the real-time evaluation value corresponding to the time consistency evaluation dimension is determined by the sliding window variance ratio ;

[0027] The real-time evaluation value corresponding to the environmental coupling degree evaluation dimension is determined by the sensitivity coefficient difference rate of the environmental coupling degree;

[0028] The real-time evaluation value corresponding to the historical confidence evaluation dimension is determined by the reliability decay value of the historical environmental parameter data.

[0029] Further, according to the detection result, a corresponding alarm signal is output, including:

[0030] obtaining a detection result; wherein the detection result includes a time domain residual detection result, a frequency domain energy entropy detection result, and a trend similarity detection result;

[0031] The time domain residual detection result is determined by a sliding window residual test, the frequency domain energy entropy detection result is determined by a wavelet packet energy entropy detection, and the trend similarity detection result is determined by a dynamic time warping similarity calculation.

[0032] If any two of the detection results meet the triggering condition, an alarm signal is output.

[0033] Further, the decision control layer further comprises a multi-sensor group cooperative detection unit, wherein the multi-sensor group cooperative detection unit is configured to perform operations comprising:

[0034] a target region is set; wherein the target region is a measurement monitoring range for the same pollution source, and the environmental elements in the target region need to meet the spatio-temporal consistency;

[0035] a second sensor network is constructed according to target data collected by all high-precision sensor groups and high-stability sensor groups in the target region;

[0036] Every preset time interval, all target data at the current time in the second sensor network are acquired, and each group of target data is spatio-temporally labeled to obtain multiple groups of labeled data; wherein all labeled data meet the time consistency requirement;

[0037] The multiple groups of labeled data are respectively divided into a precision data set and a stability data set according to the sensor categories, and meanwhile, the labeled data in the precision data set and the stability data set are respectively sorted according to the distance relationship between the layout positions of the corresponding sensors and the pollution source positions, to obtain a precision sorting set and a stability sorting set;

[0038] Pollutant data types of the pollution source are acquired, which are used to determine the association rules of the pollutants in the target region; wherein the association rules include the physical migration and diffusion rules of the pollutants;

[0039] It is respectively judged whether the labeled data in the precision sorting set and the stability sorting set meet the association rules, to obtain a first judgment result;

[0040] If the first judgment result is that the association rules are met, it is judged whether the error between the diffusion difference rate of the precision data of any layout position in the precision sorting set and other precision data before and after the layout position, and the second diffusion difference rate of the stability data of the corresponding layout position in the stability sorting set and other stability data before and after the layout position, is less than a diffusion error threshold, to obtain a second judgment result;

[0041] If the second judgment result is not less than, it is determined that the high-precision sensor group and the high-stability sensor group at the current layout position are abnormal.

[0042] The beneficial effects of the present application are:

[0043] This invention provides a remote metrological monitoring system based on heterogeneous sensor collaborative detection. By using high-stability and high-precision sensors for collaborative detection and constructing an environmental coupling matrix to analyze interference factors in real time, the system achieves dynamic calibration of sensor performance without the need for manual calibration, thus meeting the continuous online monitoring requirements of the monitoring environment.

[0044] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0045] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0046] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0047] Figure 1 This is a schematic diagram of a remote metering and monitoring system based on heterogeneous sensor collaborative detection in an embodiment of the present invention;

[0048] Figure 2 This is a flowchart illustrating the execution process of an adaptive calibration engine unit in a remote metrology and monitoring system based on heterogeneous sensor collaborative detection, as described in an embodiment of the present invention.

[0049] Figure 3 This is a flowchart illustrating the execution process of a multi-sensor group collaborative detection unit in a remote metering and monitoring system based on heterogeneous sensor collaborative detection, as described in an embodiment of the present invention. Detailed Implementation

[0050] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0051] like Figure 1 As shown, this invention proposes a remote metering and monitoring system based on heterogeneous sensor collaborative detection, comprising: a heterogeneous sensing layer, a data processing layer, and a decision control layer;

[0052] The heterogeneous sensing layer includes a high-precision sensor group, a high-stability sensor group, and an environmental parameter acquisition unit, which are used to acquire raw data within the target area and construct an environmental coupling matrix; the raw data includes raw data from the high-precision sensor group and raw data from the high-stability sensor group.

[0053] The data processing layer comprises a data alignment unit and an adaptive calibration engine unit, which are used for data alignment of raw data collected by the heterogeneous sensing layer, and adaptive calibration of the aligned raw data by using an environment coupling matrix to obtain target data;

[0054] The decision control layer is used for multi-scale anomaly detection of the target data according to a multi-modal anomaly detection algorithm, and outputs a corresponding alarm signal according to the detection result; wherein the multi-scale anomaly detection comprises time domain residual detection, frequency domain energy entropy detection and trend similarity detection;

[0055] The working principle of the above technical solution is that in order to break through the one-way calibration limitation of the single sensor system used in the prior art and realize complementary optimization of precision-stability, the present application proposes a heterogeneous sensor cooperative detection method for solving the problems existing in the prior art; wherein when facing fixed pollution source monitoring, for example, high temperature and high humidity flue gas discharged from a chimney will cause sensor response delay, the existing single sensor system cannot correct the environmental parameters in real time to complete adaptive calibration; or when mobile pollution source monitoring is performed, vibration noise during driving of a mobile carrier is mixed with sensor signals, and the single sensor system is difficult to separate effective data features, resulting in that the collection precision is difficult to achieve the expected effect;

[0056] The above-mentioned heterogeneous sensor cooperative detection method in actual application comprises a heterogeneous sensing layer and a data processing layer, and in order to meet the application of the method in a monitoring environment, the present application proposes a decision control layer combined with data anomaly monitoring to form an environmental monitoring system suitable for mobile and fixed source pollutants, which comprises a heterogeneous sensing layer, a data processing layer and a decision control layer;

[0057] Specifically, the heterogeneous sensing layer comprises a high-precision sensor group, a high-stability sensor group and an environmental parameter acquisition unit.

[0058] The high-precision sensor group collects concentration data of the target pollutant at a high frequency (for example, 10Hz-100Hz) to form high-precision sensor group raw data to be output, which has high sensitivity but is easily disturbed by temperature and humidity.

[0059] The high-stability sensor group synchronously collects homologous data with the high-precision sensor group at a relatively low frequency (for example, 1Hz-10Hz) to form high-stability sensor group raw data to be output, which is used as a stable reference signal for subsequent bidirectional calibration.

[0060] The environmental parameter acquisition unit arranges a certain number of environmental data sensors at predetermined positions to form a corresponding environmental parameter sensor network, which is used to subsequently construct an environmental coupling matrix, in combination with a predetermined difference sensitivity coefficient, to provide a calibration basis for the collected original data, thereby meeting the differentiated compensation requirements for the original data;

[0061] The data processing layer includes a data alignment unit and an adaptive calibration engine unit.

[0062] The data alignment unit is responsible for the temporal and spatial synchronization of the two types of original data collected, facilitating subsequent calibration processing.

[0063] The adaptive calibration engine unit analyzes the environmental coupling degree of the original data processed by the data alignment unit using the environmental coupling matrix and, in combination with a predetermined difference sensitivity coefficient judgment standard, performs adaptive calibration on the aligned original data. It is worth noting that the difference sensitivity coefficient is determined by the actual situation of the monitoring environment and is a measure of the stability of the coupling relationship between environmental factors and data characteristics as a whole. It is obtained by accumulating or statistically analyzing the fluctuations of the sensitivity coefficient (such as the absolute value of the coefficient of the environmental factor affecting the data in a regression model). The greater the fluctuations, the more unstable the environmental factors affect the data.

[0064] The decision control layer is responsible for multi-scale detection of the calibrated data and proposes a three-level fusion decision mechanism to judge the detection results, realizing the intelligent alarm function of measurement monitoring.

[0065] The beneficial effects of the above technical solution are as follows: through the above technical solution, the synergistic detection of high-stability sensors and high-precision sensors is utilized, and an environmental coupling matrix is constructed to analyze interference factors in real time, realizing dynamic calibration of sensor performance without the need for manual calibration, which can meet the continuous online monitoring needs of the monitoring environment. Through multi-scale anomaly detection of the calibrated target data, the accuracy of environmental monitoring is improved.

[0066] In one embodiment, the high-precision sensor group includes photochemical sensors and quantum sensors with a precision of 0.5%-1% FS, which are used to collect high-precision sensor group original data and realize ppb-level trace substance detection.

[0067] The high-stability sensor group includes electrochemical sensors and semiconductor sensors with a precision of 3%-5% FS, which are used to collect high-stability sensor group original data and provide a reference signal.

[0068] The environmental parameter acquisition unit includes a deployed temperature and humidity sensor network and a barometric pressure sensor network, which are used to construct an environmental coupling matrix to realize differentiated compensation support requirements.

[0069] The working principle of the above technical solution is that the heterogeneous sensing layer includes three types of acquisition units, which are a high-precision sensor group, a high-stability sensor group, and an environmental parameter acquisition unit. The high-precision sensor group includes multiple high-precision sensors, including but not limited to photochemical sensors and quantum sensors. The specific number of sensors is determined according to the range of on-site monitoring. It is worth noting that the accuracy of any sensor in the high-precision sensor group is preferably within 0.5% FS-1% FS (full scale), which is used to collect raw data of the high-precision sensor group to meet the ppb (parts per billion) level trace substance detection requirement. The characteristics of this type of sensor are usually high sensitivity but susceptible to temperature and humidity interference, and the drift rate is greater than 5% per week.

[0070] The high-stability sensor group also includes multiple high-stability sensors, and the specific number of sensors is preferably the same as the number of high-precision sensors, and the layout position is adjacent, which is convenient for subsequent bidirectional calibration. The high-stability sensor group includes but is not limited to electrochemical sensors and semiconductor sensors, and the accuracy of any sensor is preferably within 3% FS-5% FS, which is used to collect raw data of the high-stability sensor group to provide a reference signal for subsequent bidirectional calibration. The characteristics of this type of sensor are usually strong anti-interference and the monthly drift rate is less than 2%.

[0071] The environmental parameter acquisition unit includes multiple environmental parameter related acquisition sensors, such as temperature and humidity sensors, barometric pressure sensors, etc. It is used to be arranged at the adjacent position of the high-precision sensor group and the high-stability sensor group, and to connect all environmental data of the environmental parameter related acquisition sensors to construct an environmental parameter sensor network. According to the environmental parameters in the network, an environmental coupling matrix of the surrounding environment of the high-precision sensor and the high-stability sensor is constructed, which is used to realize the differentiated compensation requirement of the raw data during subsequent bidirectional calibration.

[0072] The beneficial effect of the above technical solution is that the heterogeneous sensing layer constructed by the above technical solution provides reliable data support for subsequent bidirectional calibration operations.

[0073] In one embodiment, the data alignment unit is used to solve the problem of non-uniformity of time and space reference of the high-precision sensor group and the high-stability sensor group in the heterogeneous sensing layer, including time synchronization and space mapping problems. The PTP protocol is used to realize level alignment solves the time synchronization problem, and the three-dimensional coordinate compensation mechanism of the sensor layout position is used to solve the space mapping problem.

[0074] The working principle of the above technical solution is that in order to ensure the accuracy of subsequent bidirectional calibration, the time and space reference of two kinds of original data with different accuracies needs to be unified before calibration; wherein, the PTP (Precision Time Protocol) protocol is preferably adopted to realize time synchronization, and the PTP is a network protocol specially designed for high-precision time synchronization, which can realize synchronization accuracy of sub-microsecond to microsecond; for the space mapping problem, a three-dimensional coordinate compensation mechanism of sensor layout position is preferably used to solve, and the accuracy compensation is realized by constructing a "position-error" mapping model, it is worth noting that the error of position should be less than 0.1 mm, and the specific space accuracy compensation technology is relatively mature in the existing environmental monitoring field, which will not be described here.

[0075] The beneficial effects of the above technical solution are that for the problem of non-uniformity of time and space reference of high-precision sensor group and high-stability sensor group in the heterogeneous sensor layer, the time synchronization and space mapping are combined to further improve the accuracy of the collected original data, which is beneficial to provide reliable data support for subsequent bidirectional calibration operation.

[0076] As shown in Figure 2 , in one embodiment, the adaptive calibration operation includes:

[0077] S101, acquiring aligned original data, denoted as to-be-calibrated data;

[0078] S102, performing environmental coupling degree analysis on the to-be-calibrated data by using an environmental coupling matrix to obtain a target environmental coupling degree;

[0079] S103, if the target environmental coupling degree is not greater than a preset coupling threshold, performing adaptive calibration operation on the to-be-calibrated data by using a bidirectional calibration model to obtain target data;

[0080] It is worth noting that if the target environmental coupling degree is less than the preset coupling threshold, it indicates that the current coupling relationship is unstable, and the to-be-calibrated data is continuously calibrated by using a pre-constructed environmental compensation model to obtain target data;

[0081] The specific environmental compensation model is generated by pre-acquiring related environmental data when the sensor is laid out;

[0082] The beneficial effects of the above technical solution are that by using the environmental coupling matrix to analyze the interference factors of the to-be-detected data of the high-stability sensor and the high-precision sensor in real time, the target environmental coupling degree is determined, and the compensation mechanism is selected according to the target environmental coupling degree, so as to realize the dynamic calibration of the sensor performance in all aspects, without manual calibration, which can meet the continuous online monitoring demand of the monitoring environment.

[0083] In one embodiment, the bidirectional calibration model is:

[0084]

[0085] wherein, and respectively represent the calibrated high-precision sensor group target data and the high-stability sensor group target data, and respectively represent the dynamic confidence coefficients of the corresponding target data, and respectively represent the aligned high-precision sensor group original data and the high-stability sensor group original data, represents an environmental compensation function for improving short-term monitoring accuracy, represents a drift compensation function for correcting errors caused by long-term drift, represents an environmental parameter;

[0086]

[0087] wherein, represents the difference response coefficient of the high-stability sensor group to the environmental parameter, represents the number of categories of environmental parameters, represents the offset of the environmental parameter; , represents a drift function constructed based on historical environmental parameter data, represents a drift rate;

[0088]

[0089] wherein, is a natural constant, and respectively represent the sensor type adjustment factors of the high-precision sensor group and the high-stability sensor group, represents a real-time comprehensive reliability index, represents the number of evaluation dimensions, represents the weight coefficient of the evaluation dimension, represents the real-time evaluation value of the evaluation dimension;

[0090] The beneficial effects of the above technical solution are: compared with the two-way calibration model in the prior art which only relies on a single dimension parameter, the two-way calibration model proposed in the application realizes the dual improvement of the environment interference and equipment failure distinguishing ability and the calibration precision by introducing a difference response coefficient and a dynamic confidence coefficient adjustment mechanism. First, the difference response coefficient as a core identification parameter quantifies the response difference of the sensor to different environmental parameters (such as the sensitivity coefficient absolute value of environmental factors such as temperature and humidity), and constructs a multi-dimensional interference-failure feature space. For example, when the difference response coefficient of a certain environmental parameter suddenly changes (such as the vibration sensitivity coefficient suddenly rising from 0.1% / g to 0.8% / g), in the subsequent fault diagnosis process combined with the frequency domain energy entropy detection result, the difference between the sensor mechanical connection loosening (equipment failure) and normal environmental vibration (interference) can be accurately identified, which is beneficial to improve the distinguishing accuracy of the traditional two-way calibration model, and solves the core problem of "interference and failure feature confusion" in the prior art.

[0091] In addition, the two-way calibration model in the application dynamically optimizes the confidence coefficient through a multi-dimensional evaluation system, replaces the traditional fixed value setting, and in the face of complex working conditions, automatically adjusts the weight distribution according to the real-time monitored evaluation value, so that the self-adaptation ability of the calibration model is improved, and the dynamic precision and environmental adaptability of the two-way calibration are significantly improved.

[0092] In one embodiment, the number of evaluation dimensions is three, and three-dimensional evaluation dimension indicators are established, including time consistency, environmental coupling degree and historical confidence;

[0093] The real-time evaluation value corresponding to the time consistency evaluation dimension is determined by the sliding window variance ratio .

[0094] The real-time evaluation value corresponding to the environmental coupling degree evaluation dimension is determined by the sensitivity coefficient difference rate of the environmental coupling degree;

[0095] The real-time evaluation value corresponding to the historical confidence evaluation dimension is determined by the reliability decay value of the historical environmental parameter data;

[0096] The working principle and beneficial effects of the above technical solution are: in a conventional state, to meet the daily monitoring requirements, the multi-dimensional evaluation system for the dynamic confidence coefficient proposed in the application optimally establishes three-dimensional evaluation dimension indicators according to time consistency, environmental coupling degree and historical confidence; wherein the real-time evaluation value corresponding to the time consistency evaluation dimension is determined by the sliding window variance ratio The weight coefficient is preferably 0.4; the real-time evaluation value corresponding to the environmental coupling degree evaluation dimension is determined by the sensitivity coefficient difference rate of the environmental coupling degree, and the weight coefficient is preferably 0.3; the real-time evaluation value corresponding to the historical confidence evaluation dimension is determined by the reliability attenuation value of the historical environmental parameter data, and the weight coefficient is preferably 0.3.

[0097] In one embodiment, a corresponding alarm signal is output according to the detection result, including:

[0098] Obtaining a detection result; wherein the detection result includes a time domain residual detection result, a frequency domain energy entropy detection result, and a trend similarity detection result;

[0099] The time domain residual detection result is determined by a sliding window residual test, the frequency domain energy entropy detection result is determined by a wavelet packet energy entropy detection, and the trend similarity detection result is determined by a dynamic time warping similarity calculation;

[0100] If any two detection results in the detection result meet the trigger condition, an alarm signal is output;

[0101] The working principle of the above technical solution is as follows: at present, due to the need for manual calibration of a single sensor system during monitoring, the corresponding remote monitoring platform does not set a corresponding sensor performance real-time evaluation means, and usually relies on periodic inspection during manual calibration to find abnormal data, which leads to late discovery of abnormal data and weak reliability and real-time performance. To this end, the present application proposes a monitoring method, which uses a multi-modal anomaly detection algorithm to monitor detection data in real time, and uses a three-level fusion decision mechanism to fuse and judge the monitoring results, thereby reducing the misjudgment rate. Specifically, it includes:

[0102] Obtaining a time domain residual detection result, a frequency domain energy entropy detection result, and a trend similarity detection result;

[0103] The time domain residual detection result is determined by a sliding window residual test:

[0104] The time domain residual detection includes:

[0105] An ARIMA prediction model is constructed to calculate the time domain residual of the target data If the time domain residual (three-sigma criterion), indicating that the sensor may have a step fault or sudden environmental disturbance, and it is determined that this detection result meets the trigger condition.

[0106] The ARIMA prediction model is:

[0107]

[0108] wherein, This represents the high-precision target data or the high-stability target data of the sensor group at time b. This indicates the autoregressive order in the ARIMA prediction model. This represents the autoregressive coefficient at time b. This represents the standard deviation of the residual sequence;

[0109] The frequency domain energy entropy detection result is determined by wavelet packet energy entropy detection:

[0110] The energy entropy value of the c-th layer after wavelet packet decomposition at the current time is calculated. ;

[0111] like This indicates that the frequency domain characteristics at the current moment have changed significantly, which may indicate an anomaly. Therefore, this detection result is determined to meet the triggering conditions. This represents the baseline value of the energy entropy of the c-th layer after wavelet packet decomposition at the current moment;

[0112] Trend similarity detection results are determined by dynamic time-warped similarity calculation:

[0113] The minimum distance of the calibrated data sequence of a high-precision and high-stability sensor is calculated using the Dynamic Time Warping (DTW) algorithm. Anomaly thresholds were set based on the box plot method. ; For history The upper quartile, This corresponds to the interquartile range;

[0114] when When this occurs, it indicates a systematic deviation in the high-precision and high-stability sensor, and the result of this test is deemed to meet the triggering condition.

[0115] Finally, if any two of the test results meet the triggering conditions, an alarm signal is output; it is worth noting that if all three meet the triggering conditions, a higher-level alarm signal is output.

[0116] The beneficial effects of the above technical solution are as follows: the above technical solution provides a reliable remote monitoring method to realize real-time judgment of sensor performance, thereby improving the reliability of monitoring data and the level of system automation in complex environments.

[0117] like Figure 3 As shown, in one embodiment, the decision control layer further includes a multi-sensor group collaborative detection unit, wherein the multi-sensor group collaborative detection unit is used to perform operations including:

[0118] S201. Set the target area; whereby the target area is the measurement and monitoring range for the same pollution source, and the environmental elements within the target area must meet the requirements of spatiotemporal consistency.

[0119] S202, constructing a second sensor network according to target data collected by all high-precision sensor groups and high-stability sensor groups in the target region;

[0120] S203, acquiring all target data at the current time in the second sensor network every preset time period, and performing space-time marking on each group of target data to obtain multiple groups of marked data; wherein all marked data meet the time consistency requirement;

[0121] S204, dividing the multiple groups of marked data into precision data sets and stability data sets according to the sensor categories, and simultaneously, sorting the marked data in the precision data sets and the stability data sets according to the distance relationship between the layout positions of the corresponding sensors and the positions of the pollution sources to obtain precision sorting sets and stability sorting sets;

[0122] S205, acquiring the pollutant data type of the pollution source, which is used to determine the association rule of the pollutants in the target region; wherein the association rule includes the physical migration and diffusion rule of the pollutants;

[0123] S206, respectively judging whether the marked data in the precision sorting sets and the stability sorting sets meet the association rule to obtain a first judgment result;

[0124] S207, if the first judgment result meets the association rule, acquiring the diffusion difference rate between the precision data of any layout position in the precision sorting sets and other precision data of the positions before and after the layout position, and the second diffusion difference rate between the stability data of the corresponding layout position in the stability sorting sets and other stability data of the positions before and after the layout position, and determining whether the error between the diffusion difference rate and the second diffusion difference rate is less than a diffusion error threshold to obtain a second judgment result;

[0125] S208, if the second judgment result is not less than, determining that the high-precision sensor group and the high-stability sensor group at the current layout position exist abnormities;

[0126] The working principle of the above technical solution is that in order to improve the accurate judgment of the sensor performance, the present application further proposes a multi-sensor group cooperative detection unit. Compared with the aforementioned single pair of heterogeneous sensor group multi-scale anomaly detection in the decision control layer, the multi-sensor group cooperative detection unit proposed in this technical solution can judge whether the sensor performance of each pair of heterogeneous sensor group exists abnormity from a more macroscopic perspective. Specifically, the multi-sensor group cooperative detection unit needs to perform operations including the following operations:

[0127] Firstly, a target area needs to be set, all sensor groups in the target area are used for collaborative detection, it is worth noting that the setting rule of the target area is to determine the measurement monitoring range of the same pollution source, at the same time, the environmental elements in the target area need to meet the spatio-temporal consistency, that is, in the target area, the environmental elements (such as meteorological parameters, pollutant concentration, etc.) need to present regularity, continuity and collaboration in spatial distribution and time change, for example, under normal meteorological parameter state, the pollutant concentration around the chemical plant presents radial attenuation with the pollution source as the center in space, and presents regular fluctuation in time with the production cycle of the chemical plant; for subsequent regular detection of sensor parameters;

[0128] It is worth noting that the process of setting the target area is usually set before the system runs, combined with the data input by the staff, in the actual operation of the multi-sensor group collaborative detection unit, the range parameters of the stored target area are usually directly called, and the target data collected by all high-precision sensor groups and high-stability sensor groups in the target area are directly called according to the range parameters, and then the second sensor network is constructed; and every pre-set time period, from all the target data of the second sensor network at the current time, and mark each group of target data in space and time to obtain multiple groups of marked data, wherein the pre-set time period is preferably set according to the regular production cycle of the monitored pollution source, for example, the production cycle of the pollutant in the chemical plant is preferably used to set the pre-set time period; it should be noted that when acquiring data from the second sensor network, all the acquired data need to meet the time consistency requirement, that is, the acquisition time of the target data needs to be unified;

[0129] Due to the slight difference between the high-precision sensor and the high-stability sensor after bidirectional calibration, in order to reduce the influence of the slight difference on the detection result in the subsequent collaborative detection process, after obtaining multiple groups of marked data, the multiple groups of marked data are respectively divided into precision data sets and stability data sets for separate processing according to the sensor type, at the same time, based on the spatial position mark in the aforementioned multiple groups of marked data and the distance relationship with the position of the target pollution source, the marked data in the precision data set and the stability data set are sorted again, this process is to facilitate the subsequent difference detection of the target data of adjacent positions according to the diffusion law, to obtain a precision sorting set and a stability sorting set; wherein the closer the layout position of the sensor corresponding to the marked data in the precision sorting set and the stability sorting set to the position of the pollution source, the closer the sorting position of the corresponding marked data.

[0130] Then the target pollution source is obtained, the type of data to be monitored, such as gaseous pollutants and heavy metal pollutants diffused with water flow, etc. produced by the chemical plant, and since the aforementioned environment elements are determined to meet the spatio-temporal consistency when the target area is defined, the physical migration and diffusion law of the pollutants in the current target area can be determined according to the type of data of the pollutants and in combination with the environment elements of the target area, that is, the attenuation and diffusion law of the pollutants. It is worth noting that the method for determining the attenuation and diffusion law of the pollutants includes collecting research data of the target area, the research data at least including corresponding data corresponding to the type of data of the pollutants, taking heavy metal pollutants as an example, the research data at least including river network data, water quality data and water power data, and constructing a river network topology based on the river network data, and finally constructing a corresponding physical model and a data model of the target area based on the research data, and then simulating and calculating the attenuation and diffusion law of the pollutants through the digital model and verifying through the physical model. The specific determination steps exist in the corresponding technical solutions in the prior art, and will not be described here.

[0131] Then it is judged whether the marked data in the precision sorting set and the stable sorting set meets the association rule, and a first judgment result is obtained, that is, whether the marked data of each spatial position meets the attenuation and diffusion law of the pollutants with the marked data collected by the sensors at the spatial positions before and after it, and if not, the sensors corresponding to the marked data that do not meet the rule are marked as abnormal;

[0132] If it meets, further obtain the diffusion difference rate of the precision data of any layout position in the precision sorting set and other precision data of the positions before and after it, and the second diffusion difference rate of the stable data of the corresponding layout position in the stable sorting set and other stable data of the positions before and after it, and determine whether the error between them is less than the diffusion error threshold to obtain a second judgment result; wherein the diffusion difference rate is the concentration difference, and the diffusion error threshold is preferably less than 0.5%;

[0133] If the second judgment result is not less than, it is determined that the high-precision sensor group and the high-stability sensor group of the current layout position are abnormal, and are marked as abnormal; finally, all the sensors with abnormal marks are output with corresponding alarm signals; if the second judgment result is less than, the next detection is performed after waiting for a preset time interval;

[0134] The beneficial effects of the above technical solution are: through the above technical solution, on the basis of independent judgment of each group of heterogeneous sensor groups (high-precision sensor group and high-stability sensor group), further collaborative judgment is performed on all heterogeneous sensor groups in the target area. Compared with independent judgment, each group of heterogeneous sensor groups only performs multi-scale anomaly detection on its own data, resulting in a lack of information interaction between different heterogeneous sensor groups, so that the data presents an "island" state, and the accuracy of the anomaly detection result may be affected. Based on this, the application provides a multi-sensor group collaborative detection unit, which constructs a global collaborative detection sensor network covering the target area, combines the physical migration and diffusion law of pollutants, and performs cross-group correlation analysis on the target data of all sensor groups in the target area, thereby avoiding missed detection of a single group of heterogeneous sensor groups due to information limitation, and further improving the anomaly detection accuracy of the sensor performance.

[0135] Finally, it should be pointed out that the above preferred embodiments are only used to illustrate the technical solutions of the present application and not to limit it. Although the present application has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present application.

Claims

1. A remote metering monitoring system based on cooperative detection of heterogeneous sensors, characterized in that, The heterogeneous sensing layer, the data processing layer, and the decision control layer are included. The heterogeneous sensing layer includes a high-precision sensor group, a high-stability sensor group, and an environmental parameter acquisition unit, which are used to collect original data in a target area and construct an environmental coupling matrix; the original data includes high-precision sensor group original data and high-stability sensor group original data. The data processing layer includes a data alignment unit and an adaptive calibration engine unit, which are used to perform data alignment on the original data collected by the heterogeneous sensing layer, and perform adaptive calibration on the aligned original data using the environmental coupling matrix to obtain target data. The adaptive calibration operation includes: Obtaining the aligned original data, denoted as the to-be-calibrated data; Performing environmental coupling degree analysis on the to-be-calibrated data using the environmental coupling matrix to obtain a target environmental coupling degree; If the target environmental coupling degree is not greater than a preset coupling threshold, performing adaptive calibration on the to-be-calibrated data through a bidirectional calibration model to obtain the target data; The bidirectional calibration model is: wherein, and respectively represent the calibrated high-precision sensor group target data and the high-stability sensor group target data, and respectively represent the dynamic confidence coefficients of the corresponding target data, and respectively represent the aligned high-precision sensor group original data and the high-stability sensor group original data, represents an environmental compensation function for improving short-term monitoring accuracy, represents a drift compensation function for correcting errors caused by long-term drift, represents an environmental parameter; wherein, represents a difference response coefficient of the high-stability sensor group to the class of environmental parameters, represents the number of classes of environmental parameters, represents an offset amount of the class of environmental parameters; , represents a drift function constructed based on historical environmental parameter data, represents a drift rate; wherein, is a natural constant, and respectively represent sensor type adjustment factors of the high-precision sensor group and the high-stability sensor group, represents a real-time comprehensive reliability index, represents the number of evaluation dimensions, represents a weight coefficient of the th evaluation dimension, represents a real-time evaluation value of the th evaluation dimension; The decision control layer is used to perform multi-scale anomaly detection on the target data according to a multi-modal anomaly detection algorithm, and output corresponding alarm signals according to the detection results; the multi-scale anomaly detection includes time domain residual detection, frequency domain energy entropy detection, and trend similarity detection.

2. The remote meter monitoring system based on cooperative detection of heterogeneous sensors according to claim 1, characterized in that, The high-precision sensor group includes photochemical sensors and quantum sensors, with a precision of 0.5%-1% FS, which are used to collect high-precision sensor group original data and realize ppb-level trace substance detection; The high-stability sensor group includes electrochemical sensors and semiconductor sensors, with a precision of 3%-5% FS, which are used to collect high-stability sensor group original data and provide reference signals; The environmental parameter acquisition unit includes a network of deployed temperature and humidity and air pressure sensors, which is used to construct an environmental coupling matrix and realize differentiated compensation support requirements.

3. The remote meter monitoring system based on cooperative detection of heterogeneous sensors according to claim 1, characterized in that, The data alignment unit is used for solving the problem of non-uniformity of time and space reference of high-precision sensor groups and high-stability sensor groups in a heterogeneous sensing layer, including time synchronization and space mapping; wherein, the time synchronization is realized based on a PTP protocol The level alignment solves the time synchronization problem, and the space mapping problem is solved by using a three-dimensional coordinate compensation mechanism of the sensor layout position.

4. The remote meter monitoring system based on cooperative detection of heterogeneous sensors according to claim 1, characterized in that, The number of evaluation dimensions is three, and three-dimensional evaluation dimension indicators are established, including time consistency, environmental coupling degree, and historical confidence; The real-time evaluation value corresponding to the time consistency evaluation dimension is determined by the sliding window variance ratio; The real-time evaluation value corresponding to the environmental coupling degree evaluation dimension is determined by the sensitivity coefficient difference rate of the environmental coupling degree; The real-time evaluation value corresponding to the historical confidence evaluation dimension is determined by the reliability decay value of the historical environmental parameter data.

5. The remote meter monitoring system based on cooperative detection of heterogeneous sensors according to claim 1, characterized in that, According to the detection results, corresponding alarm signals are output, including: Obtaining the detection results; the detection results include time domain residual detection results, frequency domain energy entropy detection results, and trend similarity detection results; The time domain residual detection results are determined by sliding window residual inspection, the frequency domain energy entropy detection results are determined by wavelet packet energy entropy detection, and the trend similarity detection results are determined by dynamic time warping similarity calculation; If any two of the detection results meet the trigger condition, an alarm signal is output.

6. The remote meter monitoring system based on cooperative detection of heterogeneous sensors according to claim 1, characterized in that, The decision control layer also includes a multi-sensor group cooperative detection unit, which is used to perform the following operations: Setting a target area; the target area is a measurement monitoring range for the same pollution source, and the environmental elements in the target area need to meet the spatiotemporal consistency; According to target data collected by all high-precision sensor groups and high-stability sensor groups in the target region, a second sensor network is constructed; Every preset time period, all target data at the current time in the second sensor network is acquired, and each group of target data is marked in time and space to obtain multiple groups of marked data; all marked data meet time consistency requirements; According to sensor categories, the multiple groups of marked data are respectively divided into precision data sets and stability data sets, and meanwhile, according to the distance relationship between the layout positions of corresponding sensors and the pollution source position, the marked data in the precision data sets and the stability data sets are respectively sorted to obtain precision sorting sets and stability sorting sets; Pollutant data types of the pollution source are acquired, which are used to determine the association rules of pollutants in the target region; the association rules include the physical migration and diffusion rules of pollutants; Whether the marked data in the precision sorting sets and the stability sorting sets meet the association rules is respectively judged to obtain first judgment results; If the first judgment results meet the association rules, whether the error between the diffusion difference rate of the precision data of any layout position in the precision sorting sets and other precision data of the front and rear positions and the second diffusion difference rate of the stability data of the corresponding layout position in the stability sorting sets and other stability data of the front and rear positions is less than a diffusion error threshold is acquired to obtain second judgment results; If the second judgment results are not less than, it is determined that the high-precision sensor group and the high-stability sensor group at the current layout position are abnormal.

Images