Intelligent water meter anomaly detection system based on multi-modal sensor fusion

Abstract

The present invention discloses a multi-modal sensing fusion intelligent water meter anomaly detection system, which relates to the technical field of industrial production water monitoring. The system comprises: an equipment composition acquisition module interactively acquires the target production line composition. A production log acquisition module acquires a set of associated indicators. A sensor equipment deployment module deploys multi-modal sensing equipment for production equipment. A water use associated data set acquisition module collects data to obtain a data set. A water use prediction edge node construction module constructs a node of a local server. A cloud detection module pre-builds a water meter anomaly detection cloud. A communication connection module connects the cloud and the node. A dynamic detection module performs dynamic detection of water meter anomalies. The system solves the technical problem that the existing water meter monitoring technology relies only on water meter readings, resulting in the inability to detect water meter anomalies in real time and accurately, affecting the efficiency and accuracy of water use management, and achieves the technical effect of ensuring a stable supply and accurate measurement of water, reducing the risk of production interruption and waste of water resources.

Description

Technical Field

[0001] The present application relates to the technical field of industrial production water monitoring, and in particular to a smart water meter anomaly detection system with multi-modal sensor fusion. Background Art

[0002] In the scenario of industrial production water management, the demand for accurate monitoring and efficient management of production line water use is particularly prominent. Optimizing water monitoring methods and accurately locating water meter anomalies have become key links in ensuring the normal supply of industrial production water. Traditional industrial production water monitoring is often relatively simple and limited, relying only on the readings of smart water meters. It does not take into account the relationship between the operating parameters of production equipment and water use, and it is difficult to accurately determine whether the water meter is abnormal in real time. In water management, there are inaccuracies, resulting in waste of water resources or affected production. The water meter anomaly detection method is relatively simple and lagging, and cannot effectively respond to the complex water changes in industrial production.

[0003] At present, there is a technical problem in the relevant technologies that water meter monitoring technology only relies on water meter readings, resulting in the inability to detect water meter anomalies in real time, accurately and comprehensively, affecting the efficiency and accuracy of industrial production water management. Summary of the invention

[0004] The present application provides a multi-modal sensor fusion intelligent water meter anomaly detection system, and adopts an equipment composition acquisition module to interactively acquire the target production line composition containing M production equipment preset with intelligent water meters. The production log acquisition module interactively acquires production logs and identifies M associated indicator sets of water consumption related indicators. The sensor equipment deployment module deploys multi-modal sensor equipment for M production equipment according to the indicator set. The water consumption related data set acquisition module collects data based on the indicator set to obtain M data sets. The water consumption prediction edge node construction module uses the data set to construct M nodes distributed on the local server. The cloud detection module pre-builds a water meter anomaly detection cloud. The communication connection module wirelessly connects the cloud and M nodes. The dynamic detection module allows the cloud and nodes to perform dynamic detection of water meter anomalies based on real-time data sets and measured water consumption, thereby achieving the technical effect of ensuring the stable supply and accurate measurement of water for industrial production, reducing the risk of production interruption and waste of water resources through real-time, accurate and comprehensive detection of intelligent water meter anomalies.

[0005] This application provides a multi-modal sensor fusion smart water meter anomaly detection system, including:

[0006] an equipment composition acquisition module, the equipment composition acquisition module is used to interactively obtain the target equipment composition of the target production line, wherein the target equipment composition includes M production equipment, wherein a smart water meter is pre-deployed on the water inlet main pipeline of each of the M production equipment; a production log acquisition module, the production log acquisition module is used to interactively obtain the production log of the target production line, and identify the water consumption related indicators based on the production log to obtain M related indicator sets of the M production equipment; a sensor equipment deployment module, the sensor equipment deployment module is used to deploy sensor equipment for the M production equipment according to the M related indicator sets to obtain M multimodal sensor equipment; a water consumption related data set acquisition module, the water consumption related data set acquisition module is used to identify the production log based on the M related indicator sets Water-related data collection, obtaining M water-related data sets; a water-use prediction edge node construction module, the water-use prediction edge node construction module is used to use the M water-related data sets to construct M water-use prediction edge nodes, wherein the M water-use prediction edge nodes are deployed on the local servers of the M production equipment; a cloud detection module, the cloud detection module is used to pre-build a water meter anomaly detection cloud; a communication connection module, the communication connection module is used to connect the water meter anomaly detection cloud with the M water-use prediction edge nodes using wireless communication; a dynamic detection module, the dynamic detection module is used for the water meter anomaly detection cloud and the M water-use prediction edge nodes to perform dynamic detection of water meter anomalies based on the M real-time associated data sets uploaded by the M multimodal sensing devices and the M measured water consumption uploaded by the M smart water meters.

[0007] The multimodal sensing fusion intelligent water meter anomaly detection system proposed in this application is firstly, the equipment composition acquisition module interactively acquires the target production line composition containing M production equipments with preset intelligent water meters. The production log acquisition module interactively acquires the production log and identifies M associated indicator sets of water consumption related indicators. The sensor equipment deployment module deploys multimodal sensing equipment for M production equipments according to the indicator set. The water consumption related data set acquisition module collects data based on the indicator set to obtain M data sets. The water consumption prediction edge node construction module uses the data set to construct M nodes distributed on the local server. The cloud detection module pre-builds the water meter anomaly detection cloud. The communication connection module wirelessly connects the cloud and M nodes. The dynamic detection module allows the cloud and nodes to perform dynamic detection of water meter anomalies based on real-time data sets and measured water consumption. Through real-time, accurate and comprehensive detection of intelligent water meter anomalies, the technical effect of ensuring the stable supply and accurate measurement of water for industrial production and reducing the risk of production interruption and waste of water resources is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0009] Figure 1 A schematic diagram of the structure of a multi-modal sensor fusion intelligent water meter anomaly detection system provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the structure of a smart water meter anomaly detection system with multimodal sensor fusion provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: equipment composition acquisition module 10, production log acquisition module 20, sensor equipment deployment module 30, water use related data set acquisition module 40, water use prediction edge node construction module 50, cloud detection module 60, communication connection module 70, dynamic detection module 80. DETAILED DESCRIPTION

[0012] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0013] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0014] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0015] The present application embodiment provides a multi-modal sensor fusion intelligent water meter anomaly detection system, such as Figure 1 As shown, the system comprises:

[0016] The equipment composition acquisition module 10 is used to interactively obtain the target equipment composition of the target production line, wherein the target equipment composition includes M production equipment, wherein a smart water meter is pre-arranged on the water inlet main pipe of each of the M production equipment. Specifically, the equipment composition acquisition module 10 obtains the detailed equipment composition information of the target production line by interacting with relevant systems or personnel, screens and analyzes the acquired information, and determines the M production equipment that requires industrial water. For each determined production equipment, special attention will be paid to the smart water meter pre-arranged on its water inlet main pipe. The smart water meter has a unique function that can directly calculate the actual water consumption during the interval between two water consumption data collections based on the measurement results of two water consumption data. Although the target production line may contain more equipment, the equipment composition acquisition module 10 focuses on the M production equipment that requires industrial water, because subsequent water consumption monitoring, analysis, and abnormality detection operations are mainly carried out around these production equipment with water demand. For example, in a large manufacturing plant, the target production line may include multiple processes and multiple equipment. The equipment composition acquisition module 10 will identify M production equipment such as stamping equipment and cleaning equipment that need to consume industrial water during operation, and clarify the layout of smart water meters on their water inlet main pipes, so as to prepare for subsequent water use monitoring and management work.

[0017] The production log acquisition module 20 is used to interactively obtain the production log of the target production line, and identify water consumption related indicators based on the production log to obtain M related indicator sets of the M production equipment. Specifically, the production log acquisition module 20 obtains a comprehensive and detailed production log by interacting with the relevant system or data source of the target production line. The production log contains various information of the target production line during operation, such as the operation time, operation status, output, energy consumption and other data of the equipment. After obtaining the production log, the production log acquisition module 20 will conduct in-depth analysis and mining of the massive data, focusing on the information related to water consumption, and using specific algorithms and data analysis methods to identify key indicators that can reflect the water consumption situation. For M production equipment, the production log acquisition module 20 performs targeted analysis and processing respectively, and extracts indicators closely related to the water consumption of the equipment from the production log according to the characteristics and operation rules of each production equipment, forming M independent related indicator sets. For example, for a specific production equipment, related indicators such as water consumption per unit time and water consumption per unit output may be determined based on factors such as its operation time, output and the raw materials used, and these indicators are integrated into a related indicator set. The production log acquisition module 20 can generate an exclusive related indicator set for each production equipment, which provides an important data basis and basis for subsequent operations such as accurate prediction of water consumption and detection of water meter anomalies.

[0018] In a possible implementation, the production log acquisition module 20 further includes: a first operation log calling unit, which is used to call and obtain a first operation log of a first production device from the production log, wherein the first production device is any one of the M production devices, and the first operation log includes multiple time series change data of multiple operation indicators. Specifically, the first operation log calling unit clearly defines the target to be processed, that is, to accurately find any one of the M production equipment from the entire production log covering a large amount of information, and define it as the first production equipment. The production log contains detailed records of various aspects of the production line. The first operation log calling unit searches and filters in the data according to pre-set identification, coding or other specific identification methods, and analyzes the structure and format of the data to ensure that the required information is accurately extracted. The multiple time-series change data of multiple operation indicators contained in the first operation log are a dynamic description of the operating status of the first production equipment. The operation indicators cover key parameters such as the speed, temperature, pressure, and current of the equipment, and the multiple time-series change data reflect the specific values ​​and change trends of the parameters at different time points. By obtaining comprehensive and detailed operation log data, basic material is provided for the subsequent in-depth analysis of the relationship between equipment operation and water use.

[0019] The first water use log acquisition unit is used to obtain the first water use log of the first production equipment from the production log. Specifically, when the first water use log acquisition unit acquires the water use log of the first production equipment, it establishes a connection with the data source of the production log and obtains access rights. According to the identification of the first production equipment, it uses a specially designed search algorithm and query statement to accurately find records related to the water use of the equipment from the production log. The water use record will record in detail the key information such as the start time, end time, and water consumption of each water use to form the first water use log. The acquired water use log will also be stored in a predetermined format to ensure data consistency and readability.

[0020] A correlation analysis unit, the correlation analysis unit is used to use the multiple time-series change data and the multiple water consumption change data as analysis benchmarks, use the Pearson correlation coefficient to analyze the correlation between the multiple operating indicators and the water consumption, and filter the multiple operating indicators to obtain a first correlation indicator set based on the analysis results. Specifically, after obtaining the multiple time-series change data in the first operation log and the multiple water consumption change data in the first water consumption log, the correlation analysis unit first pre-processes the data, including checking the integrity of the data, removing possible abnormal values ​​or erroneous data, and aligning the time-series change data of each operating indicator, such as the operating speed of the equipment, with the water consumption change data according to the same time series, and using the Pearson correlation coefficient calculation formula to calculate each pair of operating indicator data and water consumption data. During the calculation process, the data at each time point will be multiplied and summed, squared and summed, etc., and finally a correlation coefficient value reflecting the strength of the linear relationship between the two is obtained. According to a pre-set threshold, those operating indicators that are significantly correlated with water consumption are screened out to form a first correlation indicator set.

[0021] An associated indicator set acquisition unit, the associated indicator set acquisition unit is used to identify water consumption associated indicators based on the production log and obtain the M associated indicator sets of the M production equipment. Specifically, the associated indicator set acquisition unit uses the processing flow of the first production equipment as a template, and for each of the M production equipment, the following steps are repeated in sequence: calling the operation log and water consumption log corresponding to the equipment, performing data preprocessing, alignment and calculation of the Pearson correlation coefficient, filtering out associated indicators according to the threshold, and forming a unique associated indicator set for each device. The entire process needs to be performed through an automated script or program to ensure that each step is executed accurately and without error, and the processing results are recorded and monitored in a timely manner so that backtracking and adjustments can be made when problems arise.

[0022] In a possible implementation, the association relationship analysis unit further includes: a first time series change data retrieval subunit, the first time series change data retrieval subunit is used to retrieve first time series change data of a first operation indicator from the first operation log, wherein the first operation indicator is any one of the multiple operation indicators. Specifically, the first time series change data retrieval subunit searches in the first operation log obtained, and the search is based on pre-set rules and identifiers to accurately locate any one of the multiple operation indicators, called the first operation indicator, and extracts the first time series change data corresponding to the first operation indicator from a large number of data fields. The data is usually stored in the form of a time series, reflecting the specific numerical changes of the indicator at different time points.

[0023] The data alignment subunit is used to align the first time series change data and multiple water consumption change data, and to segment the alignment results based on the preset time window data to obtain H local change data and H local water consumption of H acquisition cycles. Specifically, after receiving the first time series change data and multiple water consumption change data, the data alignment subunit will perform a time alignment operation on the two sets of data. In order to ensure that the data at the corresponding time points can be accurately matched in the subsequent analysis, the aligned results are segmented according to the preset time window data. Through segmentation, H local change data (corresponding to the first operating indicator) and H local water consumption of H acquisition cycles are obtained. The data in each acquisition cycle has relatively independent and complete characteristics, which is convenient for more detailed analysis.

[0024] The Pearson correlation coefficient calculation subunit is used to calculate and obtain the H Pearson correlation coefficients of the H local change data and the H local water consumption. Specifically, the Pearson correlation coefficient calculation subunit uses the calculation formula of the Pearson correlation coefficient for each collection period to calculate the local change data and the local water consumption within the period, and finally obtains the H Pearson correlation coefficients corresponding to each of the H collection periods by calculating the covariance, standard deviation, etc. The coefficients quantitatively describe the linear correlation degree between the first operating indicator and the water consumption in each collection period.

[0025] A weight coefficient configuration subunit, the weight coefficient configuration subunit is used to configure H weight coefficients based on H collection cycles, and use the H weight coefficients to perform weighted calculation on the H Pearson correlation coefficients to obtain a first correlation coefficient of the first operating indicator. Specifically, the weight coefficient configuration subunit will configure a corresponding weight coefficient for each collection cycle according to the characteristics, importance or other preset standards of the H collection cycles. The weight coefficient reflects the relative importance of different collection cycles in the overall analysis. The H weight coefficients are used to perform weighted calculation on the H Pearson correlation coefficients calculated previously. Through weighting, the importance of each collection cycle is comprehensively considered to obtain a first correlation coefficient that can more accurately reflect the overall correlation between the first operating indicator and the water consumption.

[0026] The correlation coefficient calculation subunit is used to calculate and obtain multiple correlation coefficients of the multiple operating indicators by analogy. Specifically, the correlation coefficient calculation subunit takes the method of calculating the correlation coefficient of the first operating indicator as an example, repeats the above process for multiple operating indicators in sequence, thereby calculating the correlation coefficient corresponding to each operating indicator, and finally obtaining multiple correlation coefficients.

[0027] A numerical relationship comparison subunit, the numerical relationship comparison subunit is used to preset a correlation threshold, and by comparing the numerical relationship between the correlation threshold and the multiple correlation coefficients, a plurality of first correlation indicators are obtained from the multiple operating indicators, and the multiple first correlation indicators constitute the first correlation indicator set. Specifically, the numerical relationship comparison subunit first presets a correlation threshold, and compares the threshold with the multiple calculated correlation coefficients one by one. If a certain correlation coefficient is greater than or equal to the correlation threshold, the corresponding operating indicator is screened out and defined as the first correlation indicator. Through the screening process, multiple first correlation indicators are obtained, and these indicators together constitute the first correlation indicator set. The indicator set reflects the operating indicators that are strongly correlated with water consumption, and provides an important basis for subsequent analysis and decision-making.

[0028] The sensor equipment deployment module 30 is used to deploy sensor equipment for the M production equipment according to the M associated indicator sets to obtain M multimodal sensor equipment. Specifically, the sensor equipment deployment module 30 first obtains M associated indicator sets, which are determined for the M production equipment respectively and reflect the specific operation and water usage characteristics of each production equipment. For the first production equipment, the sensor equipment deployment module 30 will analyze its associated indicator set and select the appropriate sensor type according to the data type and measurement requirements involved in the indicator set. For example, if the associated indicator set indicates that parameters such as temperature, pressure and flow need to be measured, then sensors that can measure these parameters will be selected, and the selected multiple types of sensors will be integrated into one device to form a multimodal sensor device for the first production equipment. The device can collect and measure multiple different types of data or signals at the same time. To meet the monitoring needs of the production equipment, follow the same steps to process the second, third, and up to the Mth production equipment in turn. During the entire deployment process, it is necessary to consider practical issues such as the installation position and method of the sensor, as well as the connection and communication method with the production equipment to ensure that the sensor can accurately and stably acquire data, and that the data can be transmitted and processed in a timely and effective manner. For example, for certain high temperature, high humidity or strong vibration environments, it is necessary to select sensors with corresponding protection and adaptability, and adopt appropriate installation and fixing methods. Finally, complete the deployment of sensor equipment for M production equipment, and obtain M multimodal sensor devices that can comprehensively and accurately monitor the operation and water usage of each production equipment.

[0029] The water use-related data set acquisition module 40 is used to collect water use-related data from the production log based on the M related indicator sets to obtain M water use-related data sets. Specifically, the water use-related data set acquisition module 40 acquires the M related indicator sets determined previously. For the first related indicator set, the water use-related data set acquisition module 40 will search and filter in the data ocean of the production log according to the various indicators and conditions set therein. For example, if a certain related indicator is the relationship between the operating temperature and water consumption of the production equipment within a specific time period, the water use-related data set acquisition module 40 will find the operating temperature data and the corresponding water consumption data of the equipment within this period from the production log. Through precise extraction and sorting, a series of water use-related data related to the first related indicator set is acquired to form the first water use-related data set. The second related indicator set is processed according to the same method and steps. The corresponding water use-related data is collected from the log to construct the second water use-related data set, and this cycle is repeated to process the third, fourth, and Mth related indicator sets in turn, thereby obtaining M water use-related data sets corresponding to each related indicator set. In the process of collecting data, it is necessary to ensure the accuracy and completeness of the data. Possible data missing or abnormal situations need to be properly processed and marked for subsequent analysis and use. For example, if the production log data for a certain period of time is missing, a reasonable interpolation method may be used or it may be estimated and supplemented according to the rules of historical data. Finally, M comprehensive, accurate and targeted water use-related data sets are successfully obtained, providing a solid data foundation for subsequent water use analysis and prediction.

[0030] The water use prediction edge node construction module 50 is used to construct M water use prediction edge nodes using the M water use related data sets, wherein the M water use prediction edge nodes are deployed on the local servers of the M production equipment. Specifically, the water use prediction edge node construction module 50 obtains M water use related data sets. For the first water use related data set, the water use prediction edge node construction module 50 uses data analysis and modeling algorithms to deeply mine and analyze the various parameters and relationships in the data set. For example, a time series prediction model, a machine learning algorithm, etc. are used to establish a model that can predict future water use conditions based on historical water use related data. Based on the model, the first water use prediction edge node is constructed. The node includes functions such as data processing, prediction calculation, and result output. The constructed first water use prediction edge node is deployed to the local server of the first production equipment. During the deployment process, it is necessary to ensure that the node is compatible with the hardware and software environment of the server and can run stably and efficiently. According to the same steps, the second, third, and Mth water use related data sets are processed in sequence. For each data set, a corresponding water consumption prediction edge node was built and deployed on the local server of the corresponding production equipment. During the construction and deployment process, network communication, data security, and permission management issues also need to be considered. For example, ensure the security of data transmission between nodes and servers, and restrict illegal access and operation to prediction nodes. Finally, M water consumption prediction edge nodes were successfully deployed on the local servers of M production equipment, realizing the localization, real-time, and accurate prediction capability of the water consumption of each production equipment.

[0031] In a possible implementation, the water use prediction edge node construction module 50 further includes: a first water use associated data set acquisition unit, the first water use associated data set acquisition unit is used to collect water use associated data for the production log based on the first associated indicator set to obtain a first water use associated data set, wherein the first water use associated data set includes multiple first associated time series data of the multiple first associated indicators. Specifically, the first water use associated data set acquisition unit first acquires the first associated indicator set, and performs detailed data collection on the production log according to the requirements and rules of the indicator set, and accurately screens out water use data related to the first associated indicator set from a large amount of data in the production log, wherein the data includes multiple first associated time series data of multiple first associated indicators, reflecting the corresponding relationship between these associated indicators and water use conditions at different time points.

[0032] A data integration unit, the data integration unit is used to traverse the multiple first associated time series data to locate multiple groups of change nodes, perform data integration on the multiple groups of change nodes, and perform time series sorting on the integration results to obtain Q stable operation intervals. Specifically, the data integration unit traverses the multiple first associated time series data obtained, locates multiple groups of change nodes through specific algorithms and judgment conditions, and the change nodes represent changes in water use patterns, equipment operating status or other related factors, integrates the multiple groups of change nodes, merges and organizes related data, and sorts the integration results in chronological order, thereby determining Q stable operation intervals, and the stable operation intervals reflect the relatively stable state of equipment operation and water use under specific conditions.

[0033] An interval division unit, the interval division unit is used to divide the multiple first associated time series data and the first water use log using the Q stable operation intervals to obtain multiple sample water use associated data, wherein each sample water use associated data includes a sample stable operation time, a sample associated data group, and a sample water consumption. Specifically, the interval division unit uses the Q stable operation intervals obtained above to divide the multiple first associated time series data and the first water use log, and through the division, a plurality of sample water use associated data are obtained, each sample water use associated data includes detailed information such as the sample stable operation time (i.e., the length of time within the stable operation interval), the sample associated data group (the associated indicator data corresponding to the interval), and the sample water consumption.

[0034] A prediction model construction unit, the prediction model construction unit is used to construct a standard water consumption prediction model based on a back propagation neural network, and use the multiple sample water consumption associated data as training data to perform localized update of the standard water consumption prediction model to obtain a first water consumption prediction model. Specifically, the prediction model construction unit constructs a standard water consumption prediction model based on a back propagation neural network, and the model can predict water consumption based on input equipment-related indicators and operating time information, and use the multiple sample water consumption associated data obtained previously as training data and input them into this standard water consumption prediction model, and locally update and optimize the model, and by continuously adjusting the parameters of the model, enable it to better adapt to the water use characteristics of specific production equipment, and finally obtain the first water consumption prediction model.

[0035] A node deployment unit, the node deployment unit is used to construct a first water consumption prediction edge node based on the first water consumption prediction model, and deploy the first water consumption prediction edge node on the local server of the first production equipment. Specifically, the node deployment unit constructs a first water consumption prediction edge node according to the first water consumption prediction model, and the node has the function of predicting water consumption according to input data. The first water consumption prediction edge node is deployed on the local server of the first production equipment, so that it can process and predict the water consumption of the production equipment in real time locally.

[0036] The edge node construction unit is used to construct the M water use prediction edge nodes by analogy using the M water use related data sets. Specifically, the prediction edge node construction unit takes the processing method of the first production equipment as an example, performs the same operation on the other M-1 production equipment in turn, and uses the corresponding water use related data sets to repeat the process of data acquisition, integration, interval division, model construction and node layout, and finally constructs M water use prediction edge nodes to achieve accurate prediction and real-time monitoring of the water use conditions of all production equipment in the entire production line.

[0037] In a possible implementation, the water use prediction edge node construction module 50 further includes: an update window unit, which is used to extract the minimum value from the Q stable operation intervals as the water use data update window. Specifically, the update window unit will perform a detailed analysis and comparison on the Q stable operation intervals determined previously, extract the minimum value from the interval, and the minimum value is selected as the water use data update window. The determination of the update window is based on a comprehensive consideration of the system operation stability and data timeliness, aiming to ensure that the data update frequency can reflect the changes in the system without being too frequent to cause resource waste.

[0038] An update window constraint unit is used for the M multimodal sensing devices to periodically map and upload M real-time associated data sets to the M water use prediction edge nodes with the water use data update window as a constraint. Specifically, the update window constraint unit uses the extracted water use data update window as a key constraint condition. The M multimodal sensing devices will upload the M real-time associated data sets monitored in real time to the corresponding M water use prediction edge nodes according to the constraint and at a certain period. The periodic upload ensures that the water use prediction edge nodes can obtain the latest data in a timely manner for real-time analysis and prediction.

[0039] The measured water consumption uploading unit is used for the M smart water meters to upload the water consumption data update window as a constraint, and periodically maps and uploads the M measured water consumption to the M water consumption prediction edge nodes. Specifically, the operation of the measured water consumption uploading unit is similar to that of the update window constraint unit. The M smart water meters also use the water consumption data update window as a constraint. According to the same cycle, the measured M measured water consumption is uploaded to the M water consumption prediction edge nodes. The water consumption prediction edge nodes can simultaneously have real-time associated data and measured water consumption, providing comprehensive data support for water meter anomaly detection.

[0040] The unit for acquiring the result of a single test is used for the M water consumption prediction edge nodes to perform water meter anomaly detection according to the M real-time associated data sets and the M measured water consumption, and obtain M single test results. Specifically, after the M water consumption prediction edge nodes receive the M real-time associated data sets and the M measured water consumption, they will use specific algorithms and logic to perform water meter anomaly detection. By comparing, analyzing and calculating these two sets of data, each water consumption prediction edge node will obtain a single test result, and finally obtain M single test results, which respectively reflect whether the water meter status of each production equipment is normal.

[0041] The real-time faulty water meter positioning unit is used for synchronizing the M measured water consumption to the water meter abnormality detection cloud for water meter abnormality verification and locating the real-time faulty water meter if any of the M single-unit detection results indicates that the water meter is abnormal. Specifically, if any of the M single-unit detection results indicates that the water meter is abnormal, the M water consumption prediction edge nodes will synchronously transmit the M measured water consumption to the water meter abnormality detection cloud, and the cloud will conduct a deeper and more comprehensive analysis and verification of these data, and compare them with preset standards, models or other reference data, so as to accurately locate the real-time faulty water meter and achieve accurate diagnosis and positioning of water meter abnormalities.

[0042] In a possible implementation, the monomer detection result acquisition unit further includes: a predicted water consumption acquisition subunit, which is used to input the M real-time associated data sets into M water consumption prediction models in the M water consumption prediction edge nodes to obtain M predicted water consumption. Specifically, the predicted water consumption acquisition subunit inputs the acquired M real-time associated data sets into the corresponding M water consumption prediction models respectively. The water consumption prediction model is trained and optimized in advance and can predict water consumption based on the input real-time associated data. After the data is input, the model will output the predicted water consumption for each production equipment through internal calculation and analysis, and finally obtain M predicted water consumption.

[0043] The water meter anomaly detection subunit is used to detect water meter anomalies by comparing the data consistency of the M predicted water consumption and the M measured water consumption, and obtain the M individual detection results. Specifically, the water meter anomaly detection subunit compares the M predicted water consumption and the M measured water consumption obtained at the same time one by one. The focus of the comparison is the data consistency of the two, including the size of the values, the trend of changes, etc. Through detailed comparison and analysis, if there is a large deviation between the predicted water consumption and the measured water consumption or it does not conform to the normal water use rules, it will be judged as a water meter anomaly. After one-by-one detection and judgment of the M groups of data, M individual detection results are finally obtained to clarify whether there is an abnormality in the water meter of each production equipment.

[0044] In a possible implementation, the real-time fault water meter positioning unit further includes: an analysis taboo setting subunit, the analysis taboo setting subunit is used to set the first measured water consumption corresponding to the first production equipment as an analysis taboo if the single-unit detection result of the first production equipment in the M single-unit detection results is a water meter abnormality. Specifically, the analysis taboo setting subunit receives the M single-unit detection results, checks the results one by one, and if it is found that the single-unit detection result of the first production equipment shows that the water meter is abnormal, then the first measured water consumption corresponding to the first production equipment will be marked as an analysis taboo, which means that in some subsequent analysis steps, the first measured water consumption will not be directly used or considered temporarily to avoid interference with the analysis results.

[0045] The synchronization subunit is used for the M water consumption prediction edge nodes to synchronize the M measured water consumption to the water meter anomaly detection cloud. Specifically, the task of the synchronization subunit is to allow the M water consumption prediction edge nodes to simultaneously transmit the M measured water consumption obtained by each node to the water meter anomaly detection cloud. The synchronization operation ensures that the cloud can have comprehensive measured water consumption data, providing a complete information basis for subsequent verification and analysis.

[0046] The first water use efficiency verification branch activation subunit is used to activate the first water use efficiency verification branch from the M water use verification branches based on the analysis taboo. Specifically, the first water use efficiency verification branch activation subunit will operate according to the previously set analysis taboo, and based on the taboo condition, accurately activate the first water use efficiency verification branch from the M water use verification branches, and the branch will be specifically used to further verify and analyze the water use situation of the first production equipment.

[0047] The first validation water consumption acquisition subunit is used to remove the first measured water consumption from the M measured water consumptions, and then synchronize to the first water efficiency branch for analysis to obtain the first validation water consumption. Specifically, the first validation water consumption acquisition subunit will remove the first measured water consumption marked as analysis taboo from the M measured water consumptions, and synchronize the remaining measured water consumption to the first water efficiency branch for analysis, thereby obtaining the first validation water consumption.

[0048] The real-time faulty water meter determination subunit is used to determine the smart water meter of the first production equipment as the real-time faulty water meter if the first validation water consumption is inconsistent with the first measured water consumption. Specifically, the real-time faulty water meter determination subunit compares the first validation water consumption with the first measured water consumption. If the two are inconsistent, the smart water meter of the first production equipment is determined to be a real-time faulty water meter. Through comparison and judgment, the water meter with abnormality can be accurately determined, so that maintenance or replacement measures can be taken in time.

[0049] The cloud detection module 60 is used to pre-build the water meter anomaly detection cloud. Specifically, the cloud detection module 60 determines the required hardware resources, such as the configuration of the server, storage capacity, etc., to meet the needs of large-scale data processing and analysis, builds the cloud infrastructure, including establishing network connections, setting up security protection mechanisms, ensuring the security and stability of data during transmission and storage, installing and configuring related software and services, involving database management systems for storing and managing large amounts of water use data; data analysis tools and algorithms for detecting water meter anomalies; and monitoring and management tools to track the operating status of the cloud in real time. After the software configuration is completed, the system is initialized, the input and output formats of the data are defined, the rules and processes for data processing are set, and communication protocols with other modules (such as water use prediction edge nodes) are established. According to the pre-set detection standards and algorithms, the program code for water meter anomaly detection is written. The code will be used to analyze and judge the uploaded data in real time, conduct comprehensive testing and optimization, simulate different water use conditions and water meter data, verify the accuracy and reliability of cloud detection, repair and optimize the problems found, ensure that the water meter anomaly detection cloud can run efficiently and accurately, and provide powerful cloud detection support for the entire system.

[0050] In a possible implementation, the cloud detection module 60 further includes: a water use log calling unit, which is used to call and obtain the M water use logs of the M production equipment from the production log. Specifically, the water use log calling unit establishes a connection with a database or data system storing production logs, and accurately extracts the M water use logs corresponding to the M production equipment from the production log through specific query statements and screening conditions. The water use logs record in detail the water use of each production equipment at different times.

[0051] The mean calculation unit is used to perform mean calculation on the Q stable operation intervals to obtain the water use data collection period. Specifically, the mean calculation unit obtains the data of the Q stable operation intervals, calculates and processes the relevant data in each stable operation interval, obtains a value that can represent the stable operation characteristics by averaging, and finally determines the water use data collection period by combining the mean calculation results of all stable operation intervals. The period reflects a more reasonable data collection interval under normal production conditions.

[0052] A multiple-group associated water consumption acquisition unit, wherein the multiple-group associated water consumption acquisition unit is used to divide the M water consumption logs by the water consumption data collection period after data alignment of the M water consumption logs, and obtain multiple groups of associated water consumption, wherein each group of associated water consumption includes the water consumption of M stages of the M production equipment. Specifically, the multiple-group associated water consumption acquisition unit first performs data alignment on the M water consumption logs to ensure that the data of each production equipment at the same time point can correspond, and divides the M water consumption logs by using the previously determined water consumption data collection period. Through the division, multiple groups of associated water consumption are obtained, and each group of associated water consumption covers the water consumption of the M production equipment at a specific stage, reflecting the relationship between water consumption between production equipment.

[0053] A linear regression analysis unit, the linear regression analysis unit is used to perform linear regression analysis on the multiple groups of associated water consumption to obtain M water consumption correlation functions of the M production equipment. Specifically, the linear regression analysis unit receives multiple groups of associated water consumption data, uses a linear regression analysis method to model and calculate each group of data, and through analysis, obtains a water consumption correlation function for each production equipment that can describe the relationship between its water consumption and other factors, and finally obtains M water consumption correlation functions corresponding to each of the M production equipment.

[0054] A water meter anomaly detection cloud construction unit, the water meter anomaly detection cloud construction unit is used to construct M water use verification branches based on the M water use correlation functions, and connect the M water use verification branches in parallel to complete the construction of the water meter anomaly detection cloud. Specifically, the water meter anomaly detection cloud construction unit constructs an independent water use verification branch for each function based on the M water use correlation functions. The branch can detect and judge according to the input data and the corresponding water use correlation function. The M water use verification branches are connected in parallel to form a complete detection architecture, completing the construction of the water meter anomaly detection cloud, so that it can comprehensively consider the water use relationship between multiple production equipment and conduct comprehensive and accurate detection of water meter anomalies.

[0055] The communication connection module 70 is used to connect the water meter anomaly detection cloud and the M water use prediction edge nodes by wireless communication. Specifically, the communication connection module 70 performs system initialization operation, sets the communication parameters of the water meter anomaly detection cloud and the M water use prediction edge nodes, including but not limited to network frequency band, encryption method, transmission protocol, etc., searches for available wireless communication networks, including Wi-Fi, Bluetooth, mobile data network, etc. After determining the available network, it will select the most suitable wireless communication method according to the preset priority and network quality evaluation criteria. The communication connection module 70 will establish a communication link between the water meter anomaly detection cloud and the first water use prediction edge node, involving steps such as sending a connection request, verifying identity information, and negotiating communication parameters. Once the connection is successfully established, the module will monitor and evaluate the communication quality in real time to ensure the stability and reliability of data transmission. In the same way, the second, third, and Mth water use prediction edge nodes will establish a wireless communication connection with the water meter anomaly detection cloud. During the entire connection process, the communication connection module 70 will also handle possible connection interruptions, signal interference, and other problems. When an abnormal situation occurs, it will automatically try to reconnect or switch to an alternative communication method to ensure the continuity of data transmission. After completing all connections, the communication connection module 70 will continue to monitor the status of the entire communication network and be ready to make adjustments and optimizations at any time to ensure efficient and stable data interaction between the water meter anomaly detection cloud and the M water consumption prediction edge nodes.

[0056] Dynamic detection module 80, the dynamic detection module 80 is used for the water meter anomaly detection cloud and the M water consumption prediction edge nodes to dynamically detect water meter anomalies based on the M real-time associated data sets uploaded by the M multimodal sensing devices and the M measured water consumption uploaded by the M smart water meters. Specifically, the dynamic detection module 80 first prepares the water meter anomaly detection cloud and the M water consumption prediction edge nodes to receive data. The M multimodal sensing devices upload the M real-time associated data sets collected in real time. At the same time, the M smart water meters will also upload the corresponding M measured water consumption to the corresponding locations. After receiving these data, the water meter anomaly detection cloud will immediately process and analyze them, and use the preset algorithms and models to compare and calculate the various indicators in the real-time associated data set with the measured water consumption. By analyzing the relationship between the two, the change trend and whether they conform to the normal water use rules, it is determined whether the water meter has an abnormality. For the M water consumption prediction edge nodes, each will perform a real-time associated data set and measured water consumption. Localized processing and detection: Each node will quickly analyze the data based on its own specific rules and models, and preliminarily determine the water meter status of the production equipment to which it belongs. During the detection process, the water meter anomaly detection cloud and M water consumption prediction edge nodes will continuously update the data in real time, and dynamically adjust and re-judge based on the new data. If the data at a certain moment is found to fluctuate abnormally or deviate from the normal range, further in-depth analysis will be conducted to confirm whether there is indeed a water meter anomaly. The entire dynamic detection process is continuous to ensure that water meter anomalies can be discovered in a timely manner. Once an anomaly is detected, the system will immediately issue an alarm or take appropriate measures to facilitate timely maintenance or adjustments to ensure the accuracy of water metering and the normal operation of production.

[0057] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0058] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. The multi-modal sensor fusion intelligent water meter anomaly detection system is characterized by: The system comprises: An equipment configuration acquisition module, the equipment configuration acquisition module is used to interactively obtain a target equipment configuration of a target production line, wherein the target equipment configuration includes M production equipment, wherein a smart water meter is pre-arranged on a water inlet main pipeline of each of the M production equipment; A production log acquisition module, the production log acquisition module is used to interactively obtain the production log of the target production line, and identify water consumption related indicators based on the production log to obtain M related indicator sets of the M production equipment; A sensor equipment deployment module, the sensor equipment deployment module is used to deploy sensor equipment for the M production equipment according to the M associated indicator sets, and obtain M multimodal sensor equipment for monitoring the operation and water usage of each production equipment; A water use related data set acquisition module, the water use related data set acquisition module is used to collect water use related data from the production log based on the M related indicator sets to obtain M water use related data sets; A water use prediction edge node construction module, wherein the water use prediction edge node construction module is used to construct M water use prediction edge nodes using the M water use association data sets, wherein the M water use prediction edge nodes are deployed on local servers of the M production devices; A cloud detection module, which is used to build a water meter anomaly detection cloud; A communication connection module, the communication connection module is used to connect the water meter anomaly detection cloud with the M water consumption prediction edge nodes by wireless communication; A dynamic detection module is used in the water meter anomaly detection cloud and the M water consumption prediction edge nodes to dynamically detect water meter anomalies based on the M real-time associated data sets uploaded by the M multimodal sensing devices and the M measured water consumption uploaded by the M smart water meters.

2. The system according to claim 1, characterized in that The production log acquisition module also includes: A first operation log calling unit, the first operation log calling unit is used to call and obtain a first operation log of a first production device from the production log, wherein the first production device is any one of the M production devices, and the first operation log includes a plurality of time series change data of a plurality of operation indicators; a first water use log obtaining unit, the first water use log obtaining unit being used to obtain a first water use log of a first production device from the production log; A correlation analysis unit, the correlation analysis unit is used to use the multiple time series change data and the multiple water consumption change data as analysis benchmarks, use the Pearson correlation coefficient to analyze the correlation between the multiple operating indicators and the water consumption, and filter the multiple operating indicators to obtain a first correlation indicator set based on the analysis result; The associated indicator set acquisition unit is used to identify water consumption associated indicators based on the production log and obtain the M associated indicator sets of the M production equipment.

3. The system according to claim 2, characterized in that The association relationship analysis unit also includes: A first time series change data retrieving subunit, the first time series change data retrieving subunit is used to retrieve first time series change data of a first operation indicator from the first operation log, wherein the first operation indicator is any one of the multiple operation indicators; A data alignment subunit, the data alignment subunit is used to align the first time series change data and the multiple water consumption change data, and segment the alignment result based on the preset time window data to obtain H local change data and H local water consumption of H collection cycles; A Pearson correlation coefficient calculation subunit, the Pearson correlation coefficient calculation subunit is used to calculate and obtain H Pearson correlation coefficients of the H local change data and the H local water consumption; A weight coefficient configuration subunit, the weight coefficient configuration subunit is used to configure H weight coefficients based on the H acquisition cycles, and use the H weight coefficients to perform weighted calculation on the H Pearson correlation coefficients to obtain a first correlation coefficient of the first operation indicator; A correlation coefficient calculation subunit, the correlation coefficient calculation subunit is used to calculate and obtain multiple correlation coefficients of the multiple operating indicators by analogy; A numerical relationship comparison subunit, the numerical relationship comparison subunit is used to preset a correlation threshold, and by comparing the numerical relationship between the correlation threshold and the multiple correlation coefficients, a plurality of first correlation indicators are obtained from the multiple operating indicators, and the plurality of first correlation indicators constitute the first correlation indicator set.

4. The system according to claim 3, characterized in that The water use prediction edge node building module also includes: a first water use associated data set acquisition unit, the first water use associated data set acquisition unit being used to collect water use associated data for the production log based on the first associated indicator set to obtain a first water use associated data set, wherein the first water use associated data set includes a plurality of first associated time series data of the plurality of first associated indicators; A data integration unit, the data integration unit is used to traverse the plurality of first associated time series data to locate a plurality of groups of change nodes, perform data integration on the plurality of groups of change nodes, and perform time series sorting on the integration results to obtain Q stable operation intervals; An interval division unit, the interval division unit is used to divide the plurality of first associated time series data and the first water use log by using the Q stable operation intervals to obtain a plurality of sample water use associated data, wherein each sample water use associated data includes a sample stable operation time, a sample associated data group, and a sample water use amount; A prediction model construction unit, the prediction model construction unit is used to construct a standard water consumption prediction model based on a back propagation neural network, and use the multiple sample water consumption associated data as training data to perform localized update of the standard water consumption prediction model to obtain a first water consumption prediction model; A node deployment unit, the node deployment unit is used to obtain a first water consumption prediction edge node based on the first water consumption prediction model, and deploy the first water consumption prediction edge node on a local server of the first production equipment; The edge node construction unit is used to construct the M water use prediction edge nodes by analogy using the M water use associated data sets.

5. The system according to claim 4, characterized in that The cloud detection module also includes: A water use log calling unit, the water use log calling unit is used to call and obtain M water use logs of the M production equipment from the production log; A mean value calculation unit, the mean value calculation unit is used to perform mean value calculation on the Q stable operation intervals to obtain a water consumption data collection period; A multiple-group associated water consumption acquisition unit, wherein the multiple-group associated water consumption acquisition unit is used to divide the M water consumption logs by the water consumption data collection period after aligning the data of the M water consumption logs, so as to obtain multiple groups of associated water consumption, wherein each group of associated water consumption includes the water consumption of M stages of the M production equipment; A linear regression analysis unit, the linear regression analysis unit is used to perform a linear regression analysis on the multiple groups of associated water consumption to obtain M water consumption correlation functions of the M production equipment; A water meter anomaly detection cloud-based construction unit is used to construct M water use verification branches based on the M water use association functions, and connect the M water use verification branches in parallel to complete the construction of the water meter anomaly detection cloud.

6. The system according to claim 5, characterized in that The water use prediction edge node building module also includes: An updating window unit, wherein the updating window unit is used to extract a minimum value from the Q stable operation intervals as a water use data updating window; An update window constraint unit, wherein the update window constraint unit is used for the M multimodal sensing devices to periodically map and upload M real-time associated data sets to the M water use prediction edge nodes with the water use data update window as a constraint; A measured water consumption uploading unit, wherein the measured water consumption uploading unit is used for the M smart water meters to upload the M measured water consumptions with the water consumption data update window as a constraint, and periodically map and upload the M measured water consumptions to the M water consumption prediction edge nodes; A single detection result acquisition unit, wherein the single detection result acquisition unit is used for the M water consumption prediction edge nodes to perform water meter anomaly detection according to the M real-time associated data sets and the M measured water consumption to obtain M single detection results; A real-time faulty water meter locating unit is used to synchronize the M measured water consumption to the water meter anomaly detection cloud to perform water meter anomaly verification and locate the real-time faulty water meter if any of the M single-unit detection results is a water meter abnormality.

7. The system according to claim 6, characterized in that The monomer detection result acquisition unit also includes: A predicted water consumption acquisition subunit, wherein the predicted water consumption acquisition subunit is used to input the M real-time associated data sets into M water consumption prediction models in the M water consumption prediction edge nodes to obtain M predicted water consumption; The water meter anomaly detection subunit is used to perform water meter anomaly detection by comparing the data consistency of the M predicted water consumption and the M measured water consumption to obtain the M single-unit detection results.

8. The system according to claim 7, characterized in that The real-time fault water meter positioning unit also includes: an analysis taboo setting subunit, the analysis taboo setting subunit being used for setting a first measured water consumption corresponding to the first production equipment as an analysis taboo if the single unit detection result of the first production equipment in the M single unit detection results is that the water meter is abnormal; A synchronization subunit, wherein the synchronization subunit is used for the M water consumption prediction edge nodes to synchronize the M measured water consumption to the water meter anomaly detection cloud; a first water use effectiveness verification branch activation subunit, the first water use effectiveness verification branch activation subunit being used to activate a first water use effectiveness verification branch from the M water use verification branches based on the analysis taboo; A first validation water consumption acquisition subunit, the first validation water consumption acquisition subunit is used to remove the first measured water consumption from the M measured water consumptions, and then synchronize to the first water efficiency branch for analysis to obtain the first validation water consumption; A real-time faulty water meter determination subunit is used to use the smart water meter of the first production equipment as the real-time faulty water meter if the first verification water consumption is inconsistent with the first measured water consumption.

Images