Phosphoric acid production fault monitoring and early warning method and system based on multi-source data processing

Abstract

This invention relates to the field of fault monitoring technology, addressing the problem that existing technologies struggle to comprehensively utilize multi-source operational data from the phosphoric acid production process to accurately identify production operation status and effectively warn of fault risks. It provides a method and system for fault monitoring and early warning in phosphoric acid production based on multi-source data processing. The method includes: preprocessing multi-source operational data from the phosphoric acid production process to form production process monitoring data; extracting multi-dimensional process state features based on the production process monitoring data; inputting the multi-dimensional process state features into an abnormal state identification model to obtain the identification result corresponding to the current phosphoric acid production process; when the identification result indicates an abnormal operating state, performing trend analysis on the corresponding multi-dimensional process state features to calculate the corresponding fault risk score; and triggering a fault warning of the corresponding level based on the fault risk score and early warning rules. This invention improves the accuracy of fault monitoring and early warning in the phosphoric acid production process.

Description

Technical Field

[0001] This invention relates to the field of fault monitoring technology, and in particular to a method and system for fault monitoring and early warning in phosphoric acid production based on multi-source data processing. Background Technology

[0002] Phosphoric acid is a crucial raw material in fertilizers, food additives, and fine chemicals. Its production process typically involves multiple continuous steps, including reaction, filtration, and evaporation concentration. In actual production, complex coupling relationships exist between process parameters such as reaction temperature, reaction pressure, material flow rate, filtration pressure difference, evaporation temperature, and the concentration of the finished phosphoric acid. Furthermore, the operating status of key equipment like stirring motors and pumps directly impacts the overall production process. Abnormal fluctuations in process parameters or malfunctions in equipment can easily lead to decreased reaction efficiency, obstructed filtration, or abnormal evaporation concentration. In severe cases, it can even cause unstable product quality or equipment failure and shutdown. Therefore, real-time monitoring of the operating status and timely fault warnings during phosphoric acid production are crucial for ensuring stable production, improving product quality, and reducing production risks.

[0003] In existing technologies, fault monitoring and early warning in phosphoric acid production processes typically rely on threshold alarms for single process parameters or monitoring methods based on empirical rules. This involves setting fixed thresholds for parameters such as reaction temperature, pressure, flow rate, or equipment current, triggering an alarm when the monitored value exceeds the threshold. Additionally, some monitoring systems use simple statistical analysis or trend judgment of historical operating data to assist in identifying abnormal states. However, these technologies mostly focus on independent monitoring of single or a small number of parameters, making it difficult to comprehensively utilize multi-source operating data from the production process to systematically analyze the correlation between various process parameters and the overall operating status. Furthermore, when production conditions change or abnormalities gradually evolve, relying solely on threshold alarms or simple trend judgments often fails to identify potential fault risks in a timely manner, leading to delayed warnings or a high false alarm rate.

[0004] Therefore, how to accurately identify the production operation status based on multi-source operation data in the phosphoric acid production process, and effectively assess and warn of failure risks when abnormal conditions occur, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for monitoring and early warning of phosphoric acid production faults based on multi-source data processing, in order to solve the problem that the existing technology is unable to comprehensively utilize multi-source operating data in the phosphoric acid production process to accurately identify the production operation status and effectively warn of fault risks.

[0006] The technical solution adopted in this invention is:

[0007] In a first aspect, the present invention provides a method for monitoring and early warning of phosphoric acid production failures based on multi-source data processing, the method comprising:

[0008] Collect multi-source operational data during the phosphoric acid production process. The multi-source operational data includes at least process parameter data and equipment operating status data. The process parameter data includes at least one of reaction temperature, reaction pressure, material flow rate, filtration pressure difference, evaporation temperature, and finished phosphoric acid concentration. The equipment operating status data includes at least one of stirring motor current and pump vibration signal.

[0009] The multi-source operational data is subjected to time synchronization and data validity verification to form a unified time series of production process monitoring data;

[0010] Based on the production process monitoring data, multi-dimensional process status features reflecting the operating status of the phosphoric acid production process are extracted. The multi-dimensional process status features include at least process parameter fluctuation features, parameter correlation features, and parameter change trend features.

[0011] The multidimensional process state features are input into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process. The identification result includes normal operating state and abnormal operating state.

[0012] When the identification result is the abnormal operating state, trend analysis is performed on the corresponding multi-dimensional process state characteristics to calculate the corresponding fault risk score.

[0013] Based on the fault risk score and the preset warning rules, the corresponding level of fault warning is triggered.

[0014] Preferably, the collection of multi-source operational data during the phosphoric acid production process includes at least process parameter data and equipment operating status data, including:

[0015] The key monitoring process steps in the phosphoric acid production process are identified, including the reaction section, the filtration section, and the evaporation and concentration section.

[0016] In the key monitoring process steps, process parameter data used to characterize the process operation status of the phosphoric acid production process are collected respectively;

[0017] At the key equipment corresponding to the key monitoring process, equipment operation status data is collected to characterize the equipment operation status.

[0018] The collected process parameter data and equipment operating status data are identified according to their respective process stages and collection time to form the multi-source operating data.

[0019] Preferably, the step of performing time synchronization and data validity verification on the multi-source operational data to form unified time-series production process monitoring data includes:

[0020] Based on the acquisition time information corresponding to the multi-source operation data, time stamps are applied to the operation data from different sources to obtain the target time stamps corresponding to the multi-source operation data;

[0021] Based on the target time stamp, the multi-source running data is time aligned so that running data from different sources are mapped to a unified time axis, resulting in multi-source running data with completed time alignment.

[0022] Perform data validity verification on the multi-source running data that has completed time alignment to obtain multi-source running data that has passed the data validity verification. The data validity verification includes at least data integrity verification and outlier detection.

[0023] The production process monitoring data is obtained by filtering and organizing the multi-source operational data that has passed the data validity verification.

[0024] Preferably, the extraction of multi-dimensional process status features reflecting the operating status of the phosphoric acid production process based on the production process monitoring data includes:

[0025] The production process monitoring data is classified and processed according to the type of process parameter to obtain the corresponding process parameter dataset;

[0026] Based on the process parameter dataset, the changes of each process parameter within a preset time window are calculated, and the fluctuation characteristics of the process parameters used to characterize the stability of process operation are extracted.

[0027] Based on the variation relationship between different process parameters in the process parameter dataset, correlation analysis is performed on multiple process parameters to extract parameter correlation features that characterize the degree of mutual influence of process parameters.

[0028] Based on the time series characteristics of the production process monitoring data, the changes of process parameters over time are analyzed, and parameter change trend features used to characterize the evolution of process operation are extracted.

[0029] The process parameter fluctuation characteristics, parameter correlation characteristics, and parameter change trend characteristics are combined to form a multi-dimensional process status characteristic that reflects the operating status of the phosphoric acid production process.

[0030] Preferably, the step of performing correlation analysis on multiple process parameters based on the variation relationship between different process parameters in the process parameter dataset, and extracting parameter correlation features to characterize the degree of mutual influence of process parameters, includes:

[0031] In the process parameter dataset, identify multiple target process parameters that participate in the correlation analysis, and extract the parameter time series data corresponding to each target process parameter within a preset time window;

[0032] The time series data of each target process parameter are divided into sliding time windows, and a corresponding parameter data subsequence is formed within each time window;

[0033] Within each time window, the correlation of data subsequences for any two target process parameters is calculated to obtain the corresponding parameter correlation coefficient;

[0034] Construct a parameter correlation matrix reflecting the interrelationships among multiple process parameters based on the correlation coefficients of each parameter;

[0035] Based on the parameter correlation matrix, parameter correlation features are extracted to characterize the degree of coupling variation of process parameters.

[0036] Preferably, the step of inputting the multi-dimensional process state features into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process includes:

[0037] Based on the changes in process parameters in the production process monitoring data, the current operating condition of the phosphoric acid production process is identified, and the target operating condition corresponding to the current production process is determined.

[0038] Based on the target operating condition, select a target abnormal state identification model corresponding to the target operating condition from the preset abnormal state identification models;

[0039] The multidimensional process state features are input into the target abnormal state identification model to identify the operating state and obtain the corresponding initial state identification result.

[0040] Based on the multidimensional process state characteristics, process parameter coupling relationship data reflecting the mutual influence between multiple process parameters is constructed, and consistency analysis is performed on the process parameter coupling relationship data to obtain consistency analysis results.

[0041] Based on the consistency analysis results, the initial state identification results are verified and corrected to obtain the identification results corresponding to the current phosphoric acid production process.

[0042] Preferably, the process parameter coupling relationship data, which reflects the mutual influence between multiple process parameters, is constructed based on the multidimensional process state characteristics, and consistency analysis is performed on the process parameter coupling relationship data to obtain the consistency analysis results, including:

[0043] Based on the multidimensional process state characteristics, extract the parameter time series data corresponding to each process parameter within a preset time window;

[0044] Based on the time series data of the parameters, multiple process parameters are combined in pairs within the preset time window to construct a set of process parameter combinations.

[0045] For each process parameter combination in the set of process parameter combinations, calculate the correlation index between the corresponding process parameters to reflect the coupling and change relationship between different process parameters;

[0046] Based on the correlation indexes, a process parameter coupling relationship matrix reflecting the interrelationships of multiple process parameters is constructed to obtain the process parameter coupling relationship data;

[0047] The process parameter coupling relationship matrix is ​​compared and analyzed with a pre-established normal operating condition coupling relationship benchmark model to calculate the degree of deviation of the current process parameter coupling relationship from the benchmark model.

[0048] The consistency of the coupling relationship between the current process parameters is evaluated based on the degree of deviation, and the consistency analysis results are obtained.

[0049] Preferably, when the identification result is the abnormal operating state, the step of performing trend analysis on the corresponding multi-dimensional process state characteristics and calculating the corresponding fault risk score includes:

[0050] Based on the time node where the identification result indicates an abnormal operating state, the corresponding abnormal analysis time interval is determined, and the target process state features within the abnormal analysis time interval are extracted from the multi-dimensional process state features.

[0051] The target process state characteristics are divided according to the process parameter type to obtain parameter time series data corresponding to multiple process parameters;

[0052] Within a preset analysis time window, the time series data of each parameter are divided into sliding windows, and at least one of the following factors—the slope of change, the magnitude of change, and the degree of fluctuation—is calculated within each sliding window to obtain the trend change characteristics of each process parameter.

[0053] Based on the trend change characteristics described above, the deviation index of each process parameter from the reference range of normal operation is calculated to characterize the trend intensity of abnormal development of each process parameter.

[0054] The corresponding parameter risk factor value is calculated based on each deviation index, and the parameter risk factor values ​​are integrated according to the process parameter category to obtain a set of failure risk factors, wherein the parameter risk factor value is used to characterize the degree of abnormal risk of the corresponding process parameter.

[0055] Based on the preset risk factor weights, the set of failure risk factors is weighted and calculated, and the weighted calculation results are normalized to obtain the failure risk score corresponding to the current phosphoric acid production process.

[0056] Preferably, the step of calculating the corresponding parameter risk factor value based on each deviation index and integrating the parameter risk factor values ​​according to the process parameter category to obtain the failure risk factor set includes:

[0057] The deviation indexes are normalized to obtain standardized deviation indexes.

[0058] Based on each of the standardized deviation indicators, the corresponding parameter risk factor values ​​are calculated using a preset risk mapping function, wherein the parameter risk factor values ​​are used to characterize the degree of abnormal risk of the corresponding process parameters;

[0059] Based on the phosphoric acid production process flow, the process parameters are categorized into multiple process parameter categories.

[0060] The risk factor values ​​of each parameter are classified according to the process parameter categories, and the risk factor values ​​of parameters in the same category are integrated to obtain the category risk factor values ​​corresponding to each process parameter category.

[0061] The values ​​of each of the aforementioned risk factors are combined to form the set of failure risk factors.

[0062] Secondly, embodiments of the present invention also provide a phosphoric acid production fault monitoring and early warning system based on multi-source data processing, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, wherein when the computer program instructions are executed by the processor, the method of the first aspect described above is implemented.

[0063] In summary, the beneficial effects of the present invention are as follows:

[0064] The present invention provides a method and system for monitoring and early warning of phosphoric acid production faults based on multi-source data processing. By constructing a fault monitoring and early warning mechanism based on multi-source data fusion and status identification, it realizes comprehensive analysis and risk assessment of the operating status of the phosphoric acid production process, thereby effectively solving the technical problem in the prior art that it is difficult to accurately identify the production operation status and effectively warn of potential fault risks based on multi-source operating data. Specifically, firstly, by synchronizing and validating the source operational data, a unified time series of production process monitoring data is formed, ensuring the consistency and reliability of data from different sources in the time dimension. Based on this, multi-dimensional process state features reflecting the operating status of the phosphoric acid production process are extracted from the production process monitoring data, enabling a comprehensive characterization of the dynamic changes in the production process and the correlation between parameters. Subsequently, these multi-dimensional process state features are input into a preset abnormal state identification model to identify the current operating status of the phosphoric acid production process, thus accurately distinguishing between normal and abnormal operating states based on multi-dimensional feature information. When the identification result is an abnormal operating state, trend analysis is further performed on the corresponding multi-dimensional process state features, and a fault risk score is calculated to quantitatively assess the development degree of the abnormal state and potential fault risks. Finally, different levels of fault warnings are triggered according to the fault risk score and preset warning rules, enabling timely risk alerts before faults occur or worsen. This achieves comprehensive utilization of multi-source operational data in the phosphoric acid production process and accurate identification of the production operating status, and effectively assesses and grades potential fault risks, thereby improving the accuracy and timeliness of fault monitoring and warning in the phosphoric acid production process. Attached Figure Description

[0065] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0066] Figure 1 This is a schematic diagram of the overall working process of the phosphoric acid production fault monitoring and early warning method based on multi-source data processing in Embodiment 1 of the present invention;

[0067] Figure 2 This is a flowchart illustrating the extraction of multidimensional process state characteristics reflecting the operating status of the phosphoric acid production process in Embodiment 1 of the present invention.

[0068] Figure 3 This is a flowchart illustrating the process of performing trend analysis on the corresponding multidimensional process state characteristics and calculating the corresponding fault risk score in Embodiment 1 of the present invention.

[0069] Figure 4 This is a schematic diagram of the phosphoric acid production fault monitoring and early warning system based on multi-source data processing in Embodiment 2 of the present invention. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. In the description of the present invention, it should be understood that the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, the element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Where there is no conflict, embodiments of the present invention and the various features thereof can be combined with each other, all of which are within the scope of protection of the present invention.

[0071] Please see Figure 1 Embodiment 1 of the present invention discloses a method for monitoring and early warning of phosphoric acid production failures based on multi-source data processing, the method comprising:

[0072] Collect multi-source operational data during the phosphoric acid production process. The multi-source operational data includes at least process parameter data and equipment operating status data. The process parameter data includes at least one of reaction temperature, reaction pressure, material flow rate, filtration pressure difference, evaporation temperature, and finished phosphoric acid concentration. The equipment operating status data includes at least one of stirring motor current and pump vibration signal.

[0073] Specifically, in the phosphoric acid production process, the first step is to collect operational data generated at the production site to obtain basic information reflecting the operating status of production equipment and processes. Multi-source operational data refers to a collection of data from different monitoring objects and different sensing devices. This typically includes both process parameter data describing the reaction process and equipment operating status data reflecting the equipment's operational status. Process parameter data primarily characterizes the physical or chemical changes in the phosphoric acid production process. For example, reaction temperature and pressure reflect whether the reaction conditions within the reactor are within a reasonable range; material flow rate reflects the stability of material transport in the production system; filtration pressure difference reflects the workload of the filtration equipment; and evaporation temperature and finished phosphoric acid concentration reflect the evaporation and concentration process and the final product quality. Equipment operating status data mainly comes from monitoring the operating status of key equipment. For example, the current of the stirring motor can be used to characterize the stirring load, and the pump vibration signal reflects the mechanical operating status of the conveying equipment. In practice, the aforementioned data can be acquired in real time using data acquisition devices such as temperature sensors, pressure sensors, flow meters, current detection devices, and vibration sensors distributed in key locations such as reaction vessels, filtration equipment, evaporation units, and transfer pumps. This data is then transmitted to the monitoring platform via the industrial control system or data acquisition module. By uniformly acquiring data from different sources, both process operation information and equipment operation information can be obtained simultaneously. This provides a comprehensive data foundation for subsequent integrated analysis of the production process's operational status and avoids the inaccurate judgments caused by relying solely on single parameter monitoring.

[0074] The multi-source operational data is subjected to time synchronization and data validity verification to form a unified time series of production process monitoring data;

[0075] Specifically, since data from different sensing devices often have different sampling periods, recording times, and data integrity, after acquiring multi-source operational data, it is necessary to perform time synchronization and data validity verification on this data to ensure consistency in the time dimension and reliability of data quality. Time synchronization typically refers to time alignment of various types of data according to a unified system time base, so that data collected by different sensors at the same point in time can correspond in the same time series. For example, timestamp correction, unified sampling periods, or interpolation compensation can be used to establish a correspondence between different data such as reaction temperature, material flow rate, and equipment current on the same time axis. Data validity verification involves checking the rationality and completeness of the collected data to identify abnormal, missing, or obviously erroneous data. For example, abnormal values ​​exceeding the physical range can be detected by setting reasonable value ranges, abrupt changes can be identified by continuity judgment, and missing data can be filled or removed. After the above processing, various process parameter data and equipment operating status data are integrated into production process monitoring data with a unified time sequence, enabling data from different sources to be correlated and analyzed within the same time frame, thus providing an accurate and reliable data foundation for subsequent feature extraction and status identification.

[0076] Based on the production process monitoring data, multi-dimensional process status features reflecting the operating status of the phosphoric acid production process are extracted. The multi-dimensional process status features include at least process parameter fluctuation features, parameter correlation features, and parameter change trend features.

[0077] Specifically, after obtaining unified time-series production process monitoring data, it is necessary to further extract multi-dimensional process state characteristics that reflect the overall operating status of the phosphoric acid production process. These multi-dimensional process state characteristics refer to a set of feature information obtained after processing the original monitoring data through statistical analysis, correlation analysis, and trend analysis. These characteristics are used to describe the operating characteristics of the production process from multiple dimensions. Specifically, process parameter fluctuation characteristics are mainly used to characterize the stability of a certain process parameter within a certain time range. For example, the stability of reaction temperature or material flow rate can be reflected by calculating the fluctuation amplitude, standard deviation, or coefficient of variation of the parameter. Parameter correlation characteristics are used to describe the interrelationships between different process parameters. For example, the coupling relationship between reaction pressure and reaction temperature, or between filtration pressure difference and material flow rate, can be reflected through correlation analysis or synergistic change analysis. Parameter change trend characteristics are used to reflect the direction and rate of change of a certain parameter over time. For example, the continuous rise or fall of evaporation temperature or product concentration can be identified by calculating the slope or trend change rate of the parameter. In practical implementation, the production process monitoring data can be segmented using a sliding time window, and the aforementioned characteristic indicators can be calculated separately within each time window. This results in a multi-dimensional process status feature that comprehensively characterizes the production operation status from multiple aspects, including fluctuation characteristics, parameter relationships, and development trends. By performing such feature extraction processing on the raw monitoring data, the data scale can be effectively compressed and key operational information can be highlighted. This allows complex production operation data to reflect the true operating status of the production process in a more representative form, providing more effective input information for subsequent operational status identification.

[0078] The multidimensional process state features are input into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process. The identification result includes normal operating state and abnormal operating state.

[0079] Specifically, after obtaining multi-dimensional process state characteristics that comprehensively reflect the operating characteristics of the phosphoric acid production process, it is necessary to use a pre-defined abnormal state identification model to determine the current operating state of the production process. The abnormal state identification model is a state discrimination model formed by training or constructing historical production operation data. It is mainly used to classify and identify the current operating state of the system based on the input multi-dimensional feature information. In practical implementation, data samples from historical production data that are in a stable operating state and data samples that have experienced faults or abnormal fluctuations can be collected and multi-dimensional process state characteristics of the same type extracted as training samples. A recognition model capable of distinguishing different operating states can be established through data modeling methods. When running, the model receives the multi-dimensional process state characteristics extracted from the current time period as input, comprehensively analyzes the combined relationships between the features, and outputs the corresponding operating state identification results. The identification results typically include at least two categories: normal operating state and abnormal operating state. The normal operating state indicates that the current process parameters and equipment status are generally within a stable and reasonable range, while the abnormal operating state indicates that the relationship or change pattern between some parameters has deviated from the normal operating pattern. In practical applications, this identification process can be continuously and periodically executed within the monitoring system. For example, features can be continuously updated and input into the model for identification within a fixed time window, thereby achieving continuous monitoring of the operating status of the phosphoric acid production process. This multi-dimensional feature-based status identification method comprehensively considers the coordinated changes between multiple process parameters and equipment statuses. Compared to traditional single-parameter threshold judgment methods, it is easier to identify complex abnormal patterns, thus improving the accuracy of operating status judgment.

[0080] When the identification result is the abnormal operating state, trend analysis is performed on the corresponding multi-dimensional process state characteristics to calculate the corresponding fault risk score.

[0081] Specifically, when the identification results indicate that the current production process is in an abnormal operating state, further trend analysis of relevant multi-dimensional process state characteristics is needed to assess the degree of development of the abnormal state and calculate a fault risk score accordingly. Trend analysis is mainly used to determine the direction, rate, and duration of change of various process parameters over a period of time. By quantifying these change patterns, it is possible to more accurately determine whether the abnormal state is continuously deteriorating. In practice, continuous monitoring of various relevant characteristics can be performed within a continuous time window after the occurrence of the abnormal state, and indicators such as the slope, amplitude, and degree of fluctuation of their changes can be calculated. For example, analyzing whether the reaction temperature continues to rise, whether the filtration pressure difference shows a gradual increasing trend, or whether the equipment vibration signal gradually intensifies, etc. After obtaining this trend change information, a risk assessment calculation method can be constructed to transform the changes of each characteristic into corresponding risk indicators, and then perform comprehensive calculations according to preset weights to obtain a fault risk score that characterizes the severity of the current abnormal state. This score is usually expressed in numerical form; the higher the value, the greater the probability or degree of development of the potential fault. By introducing trend analysis and risk scoring calculation, it is possible not only to identify whether there are any anomalies at present, but also to quantitatively assess the development trend of anomalies. This enables the monitoring system to identify the gradually accumulating risks before a fault actually occurs, thereby significantly improving the fault prediction capability.

[0082] Based on the fault risk score and the preset warning rules, the corresponding level of fault warning is triggered.

[0083] Specifically, after obtaining the fault risk score, the system can trigger corresponding level fault warnings based on pre-set warning rules. Warning rules are typically a set of risk thresholds or grading standards set based on production management experience or historical operational data analysis. For example, the risk score may be divided into multiple intervals such as low risk, medium risk, and high risk, with different levels of warning information corresponding to different intervals. When the calculated fault risk score reaches a certain preset threshold, the monitoring system automatically generates a warning signal of the corresponding level and alerts operators through the production monitoring platform, alarm devices, or information prompt interface, enabling relevant personnel to promptly monitor the production status and take necessary adjustments. This risk-score-based hierarchical warning mechanism avoids the frequent false alarms caused by relying solely on simple alarms, while simultaneously providing progressively higher-level warnings of potential faults based on risk severity, thereby improving the effectiveness and practicality of phosphoric acid production process monitoring and early warning.

[0084] Preferably, the collection of multi-source operational data during the phosphoric acid production process includes at least process parameter data and equipment operating status data, including:

[0085] The key monitoring process steps in the phosphoric acid production process are identified, including the reaction section, the filtration section, and the evaporation and concentration section.

[0086] Specifically, in phosphoric acid production, to ensure that subsequent operational status analysis covers key process nodes that significantly impact the production process, it is necessary to first identify the critical monitoring process steps in the phosphoric acid production flow. Critical monitoring process steps refer to the process stages in the entire production process that have a significant impact on product quality, production efficiency, and equipment operational safety; changes in their operational status often directly reflect the overall stability of the production system. In the wet phosphoric acid production process, the reaction section mainly completes the chemical reaction between phosphate rock powder and the acidic medium, which is the core stage of phosphoric acid generation; the filtration section is used to achieve solid-liquid separation in the reaction slurry, and its operational status affects product purity and equipment load; the evaporation and concentration section increases the phosphoric acid concentration by heating and evaporating water, which is a crucial step in ensuring that the final product concentration meets standards. Therefore, during implementation, based on the production process flow diagram and the on-site equipment layout, these sections can be prioritized for monitoring and identified as critical monitoring process steps. By clearly defining the critical process steps during the system design phase, it can be ensured that data collection focuses on covering the process nodes with the greatest impact on production stability, thereby providing representative monitoring data for subsequent operational status analysis.

[0087] In the key monitoring process steps, process parameter data used to characterize the process operation status of the phosphoric acid production process are collected respectively;

[0088] Specifically, after identifying the key monitoring processes, it is necessary to collect process parameter data reflecting the operational status of the production process at each stage. Process parameter data typically refers to process variables acquired in real time through various industrial sensors, used to describe changes in the physical or chemical states during production. For example, in the reaction section, temperature and pressure sensors can be used to acquire reaction temperature and pressure, respectively, while flow meters can acquire the flow rate of materials participating in the reaction or being transported, thus reflecting whether the reaction conditions are stable. In the filtration section, differential pressure sensors can acquire filtration differential pressure data to determine whether the filtration equipment is clogged or under abnormal load. In the evaporation and concentration section, temperature detection devices and concentration detectors can acquire evaporation temperature and the concentration of the finished phosphoric acid, reflecting the operating efficiency of the evaporation system and product quality. In practice, these parameters can be continuously collected by online monitoring instruments located at key locations in the production unit and transmitted to a data processing platform via an industrial control system or data acquisition module. By collecting corresponding process parameter data at different process stages, the operational status of each stage in phosphoric acid production can be comprehensively reflected, providing fundamental information for subsequent comprehensive analysis of the production status.

[0089] At the key equipment corresponding to the key monitoring process, equipment operation status data is collected to characterize the equipment operation status.

[0090] Specifically, besides process parameters, the operating status of the production equipment itself also significantly impacts the stability of the production process. Therefore, it is necessary to collect equipment operating status data at key equipment corresponding to critical monitoring process stages. Equipment operating status data is typically used to characterize information such as mechanical load, vibration, or electrical status of the equipment during operation. This information often reflects whether the equipment is in normal working condition. For example, in the reaction section, the stirring motor drives the stirring device inside the reactor, and its current changes reflect the stirring load. When the material viscosity increases or the equipment malfunctions, the motor current may change significantly. In the material conveying or circulation system, pump vibration signals can reflect the operating status of the pump bearings, impellers, or piping system. When the equipment experiences wear or imbalance, the vibration amplitude often changes abnormally. In actual implementation, this equipment operating information can be collected in real time using current sensors, vibration sensors, or other equipment status monitoring devices and transmitted to the monitoring platform through a data acquisition system. By introducing equipment operating status data, a new dimension for monitoring equipment health status can be added beyond process parameters, thus providing a more comprehensive reflection of the overall operation of the phosphoric acid production system.

[0091] The collected process parameter data and equipment operating status data are identified according to their respective process stages and collection time to form the multi-source operating data.

[0092] Specifically, after collecting process parameter data and equipment operating status data, these data need to be uniformly identified to facilitate subsequent data analysis and management. Specifically, each data record can be identified based on its process stage and collection time. For example, process stage tags such as reaction stage, filtration stage, or evaporation and concentration stage can be added to the data, and the corresponding collection timestamp can be recorded, thus forming data records with clear source and time information. In this way, data from different sources can be accurately distinguished in subsequent processing and can be organized and associated in chronological order. In a practical system, this identification process can be automatically completed in the data acquisition system or data processing platform, that is, process stage identifiers and time information are automatically attached when data enters the system and stored in a unified data structure. After the above processing, the originally scattered process parameter data and equipment operating status data are integrated into structured multi-source operating data, providing a clear data organization foundation for subsequent data analysis operations such as time synchronization processing, feature extraction, and operating status identification.

[0093] Preferably, the step of performing time synchronization and data validity verification on the multi-source operational data to form unified time-series production process monitoring data includes:

[0094] Based on the acquisition time information corresponding to the multi-source operation data, time stamps are applied to the operation data from different sources to obtain the target time stamps corresponding to the multi-source operation data;

[0095] Specifically, after acquiring multi-source operational data, it is necessary to first perform unified time stamping processing on the data according to the acquisition time information corresponding to each type of data. This ensures that data from different sources can be analyzed subsequently within the same time dimension. Acquisition time information is usually automatically added by field sensors, data acquisition modules, or industrial control systems when the data is generated. Essentially, it is a timestamp used to identify the moment the data was generated. Since the phosphoric acid production process involves many types of monitoring equipment, the sampling cycles and data recording methods of different devices may vary. For example, temperature and pressure parameters are usually acquired at fixed time intervals, while vibration or current signals may be recorded at higher frequencies. Therefore, in practical implementation, it is necessary to first read the time information from the original records of various operational data and standardize it according to the system's unified time format, such as uniformly converting it to a system time identifier in seconds or milliseconds. Through this processing, a standardized target time stamp can be added to each process parameter data or equipment operating status data, giving each type of data a clear time attribute in the data structure, laying the foundation for subsequent time alignment and data integration.

[0096] Based on the target time stamp, the multi-source running data is time aligned so that running data from different sources are mapped to a unified time axis, resulting in multi-source running data with completed time alignment.

[0097] Specifically, after obtaining the target timestamp, further time alignment processing is needed for operational data from different sources to ensure they are uniformly mapped to the same time axis. The core of time alignment processing lies in resolving the inconsistency in data time distribution caused by different sampling frequencies from different data sources, ensuring that multiple monitoring parameters form a corresponding relationship at the same time node. In the specific implementation process, a unified system time axis can be constructed first, for example, by establishing a continuous sequence of time nodes according to a preset time interval. Then, each data point is mapped to the nearest time node position based on its corresponding target timestamp. When a certain type of parameter data is missing at a particular time node, interpolation, neighbor-value filling, or historical value preservation can be used to supplement the data, ensuring a complete set of parameters is obtained at the same time node. When some monitoring data have a high sampling frequency, time window averaging, data aggregation, or downsampling can be used to maintain a consistent time granularity with other parameters. Through these methods, different parameter data from the reaction section, filtration section, and evaporation and concentration section can form synchronous records on a unified time axis, thus obtaining multi-source operational data with complete time alignment. This time alignment process enables different types of operational data to be correlated and analyzed on the same time dimension, thereby providing a more accurate data foundation for subsequent production status analysis.

[0098] Perform data validity verification on the multi-source running data that has completed time alignment to obtain multi-source running data that has passed the data validity verification. The data validity verification includes at least data integrity verification and outlier detection.

[0099] Specifically, after time alignment, data validity verification of multi-source operational data is required to ensure the high reliability of data used in subsequent analyses. Data validity verification is mainly used to identify missing records, erroneous records, or abnormal fluctuations that may occur during data acquisition. The process typically includes two aspects: data integrity verification and outlier detection. Data integrity verification mainly determines the completeness of data records by checking whether the parameters in the time series are continuous and whether there are missing sampling points. For example, it checks whether there are consecutive time nodes lacking corresponding parameter records within a certain time period and fills in or marks the missing data according to system-defined rules. Outlier detection is used to identify data that significantly deviates from the normal operating range. For example, when the reaction temperature or evaporation temperature exceeds the equipment's allowable range, the system can identify it based on a preset reasonable range, or it can use statistical analysis methods to detect sudden changes in data with excessively large fluctuations compared to adjacent time points. In practice, these detection rules can be set based on the statistical range of historical operational data, equipment operating specifications, or process control standards, and the data can be batch-checked through automated programs. For data identified as invalid, correction, replacement, or direct removal can be performed as needed to avoid interference from erroneous data in subsequent operational status analysis. This data validity verification process can effectively improve data quality, ensuring that subsequent analysis is based on reliable data.

[0100] The production process monitoring data is obtained by filtering and organizing the multi-source operational data that has passed the data validity verification.

[0101] Specifically, after obtaining data that has passed validity verification, this data needs to be screened and organized to form production process monitoring data that can be directly used for analysis and processing. This process mainly involves structuring the time-aligned and quality-verified data, enabling the originally scattered data to be stored and managed according to a unified data structure. In practice, all parameter data can be sorted in chronological order, and various parameter data corresponding to the same time point can be integrated. For example, monitoring data such as reaction temperature, reaction pressure, material flow rate, filtration differential pressure, and equipment current can be stored as different fields in the same data record. At the same time, the data can also be classified and organized according to process links or equipment categories, making the data structure clearer. For example, parameters related to the reaction section, filtration section, and evaporation and concentration section can be labeled with fields and formed into data records with a unified format in the database or data platform. After such screening and organization, multi-source operating data is transformed into production process monitoring data with a unified time series structure, clear parameter field structure, and reliable data quality, thus providing a stable and standardized data foundation for subsequent data analysis tasks such as process status feature extraction, abnormal status identification, and fault risk assessment.

[0102] Preferably, please refer to Figure 2 The extraction of multi-dimensional process status features reflecting the operating status of the phosphoric acid production process based on the production process monitoring data includes:

[0103] The production process monitoring data is classified and processed according to the type of process parameter to obtain the corresponding process parameter dataset;

[0104] Specifically, after obtaining production process monitoring data with a unified time series structure, the first step is to classify the data according to the type of process parameters. This allows for feature extraction based on the characteristics of different parameters in subsequent analysis. Process parameter types typically refer to various parameter variables used to describe different process states during production, such as reaction temperature, reaction pressure, material flow rate, filtration differential pressure, and evaporation temperature. These parameters often differ in physical meaning, variation patterns, and sampling methods. Therefore, in practice, production process monitoring data can be classified and organized according to parameter names or monitoring equipment identifiers, grouping parameters of the same type into corresponding datasets. For example, temperature parameter datasets, pressure parameter datasets, or flow rate parameter datasets can be constructed. This classification method allows different types of parameters to be called and processed separately in subsequent analysis, thereby improving data processing efficiency and providing a clear data foundation for extracting process state characteristics from different dimensions.

[0105] Based on the process parameter dataset, the changes of each process parameter within a preset time window are calculated, and the fluctuation characteristics of the process parameters used to characterize the stability of process operation are extracted.

[0106] Specifically, after classifying the parameters, the changes in parameters within a preset time window can be calculated based on the datasets of each process parameter, thereby extracting process parameter fluctuation characteristics to characterize the stability of process operation. A preset time window refers to a continuous data interval in time series data divided into fixed time periods, such as several minutes or several sampling periods. Statistical analysis of parameter changes is performed within each time window. Fluctuation characteristics are mainly used to describe the stability of a process parameter within that time range. For example, when reaction temperature or material flow rate fluctuates frequently within a short period, it often indicates instability in the production state. In practice, the fluctuation amplitude of parameters can be characterized by calculating statistical indicators such as the difference between the maximum and minimum values, standard deviation, variance, or coefficient of variation within the time window. The frequency of parameter changes can also be determined by calculating the rate of change between consecutive time points. By quantitatively analyzing the fluctuations of each process parameter in different time windows, a set of fluctuation characteristic indicators reflecting the stability of process operation can be formed. When a production system is in a stable operating state, the fluctuation range of most key process parameters is usually kept within a small range. However, when the system malfunctions, some parameters tend to fluctuate significantly. Therefore, extracting the fluctuation characteristics of process parameters can effectively reflect changes in the stability of the production process.

[0107] Based on the variation relationship between different process parameters in the process parameter dataset, correlation analysis is performed on multiple process parameters to extract parameter correlation features that characterize the degree of mutual influence of process parameters.

[0108] Specifically, besides the changes in individual parameters, different process parameters often have mutual influence relationships. Therefore, it is necessary to conduct correlation analysis on the changes in multiple process parameters to extract parameter correlation features to characterize the degree of mutual influence. In the phosphoric acid production process, many process parameters do not change independently. For example, there may be a synergistic relationship between reaction temperature and reaction pressure, and changes in material flow rate may also affect parameters such as filtration pressure difference or evaporation temperature. When the production system is operating normally, these parameters usually maintain a relatively stable relationship. However, when a certain piece of equipment or process link malfunctions, the original parameter correlation relationship often changes. Therefore, in the specific implementation process, the relationship between different parameters can be calculated based on the process parameter dataset. For example, by analyzing the synchronous change trend, consistency of change direction, or relationship between change magnitude, the degree of mutual influence between parameters can be assessed. Through this correlation analysis, a set of feature information that reflects the synergistic relationship between multiple process parameters can be extracted, enabling the system to not only focus on the changes in individual parameters but also identify changes in the relationship structure between parameters, thereby more comprehensively reflecting the operating status of the production process.

[0109] Based on the time series characteristics of the production process monitoring data, the changes of process parameters over time are analyzed, and parameter change trend features used to characterize the evolution of process operation are extracted.

[0110] Specifically, based on this, it is also necessary to utilize the time-series characteristics of production process monitoring data to analyze the changing patterns of process parameters over time, thereby extracting parameter change trend features to characterize the evolution of process operation. The so-called time-series characteristic refers to the data sequence formed by continuously recording production process monitoring data in chronological order, which reflects the dynamic changes of parameters over time. In practice, the overall direction and rate of change of parameters can be analyzed by calculating the numerical changes of parameters within a continuous time window. For example, it can be used to determine whether the reaction temperature shows a continuous upward trend, whether the evaporation temperature gradually decreases, or whether the filtration pressure difference shows a gradual increasing trend. To obtain more stable trend information, parameter data can be smoothed or a sliding window can be used to calculate the slope of parameter changes, thereby reducing the impact of short-term random fluctuations on trend judgment. In this way, a set of trend features reflecting the evolution of parameters over time can be extracted to describe the dynamic changes in the production process state. When certain process parameters continue to develop in an abnormal direction, these trend features can reflect potential risks in advance, providing an important basis for subsequent abnormal state identification and risk assessment.

[0111] The process parameter fluctuation characteristics, parameter correlation characteristics, and parameter change trend characteristics are combined to form a multi-dimensional process status characteristic that reflects the operating status of the phosphoric acid production process.

[0112] Specifically, after obtaining the fluctuation characteristics, correlation characteristics, and trend characteristics of process parameters, these characteristics need to be comprehensively combined to form a multi-dimensional process state feature that can reflect the overall operating status of the phosphoric acid production process. Multi-dimensional process state features refer to a comprehensive feature vector formed by integrating multiple different types of feature indicators. The features of different dimensions describe the production process status from different perspectives, such as stability, parameter relationships, and trends. In the implementation process, various feature indicators can be combined according to a unified data structure. For example, fluctuation characteristics, correlation characteristics, and trend characteristics can be arranged in a preset order to form a complete set of feature data records, so that each time window corresponds to a set of multi-dimensional feature information. In this way, the originally complex production process monitoring data is transformed into a set of feature information that can describe the operating status from multiple dimensions. This allows the subsequent abnormal state identification model to more comprehensively analyze the operation of the production system and improve the accuracy of abnormal state identification.

[0113] Preferably, the step of performing correlation analysis on multiple process parameters based on the variation relationship between different process parameters in the process parameter dataset, and extracting parameter correlation features to characterize the degree of mutual influence of process parameters, includes:

[0114] In the process parameter dataset, identify multiple target process parameters that participate in the correlation analysis, and extract the parameter time series data corresponding to each target process parameter within a preset time window;

[0115] Specifically, before analyzing the correlation between process parameters, it is necessary to first identify multiple target process parameters involved in the correlation analysis within the process parameter dataset, and then extract the corresponding time series data of these parameters within a preset time window. Target process parameters refer to key process variables that significantly reflect changes in process status or have potential correlations with other parameters during production, such as reaction temperature, reaction pressure, material flow rate, filtration differential pressure, and evaporation temperature. These parameters often participate in a specific process and mutually influence the system's operating status. Time series data refers to a sequence of parameter data recorded continuously in chronological order, reflecting the dynamic process of parameter changes over time. In practice, target process parameters can be selected based on the parameter types recorded in the production process monitoring data, and continuous data sequences corresponding to each parameter can be extracted within a preset time window. For example, temperature, pressure, and flow rate records can be extracted from a unified time axis within that time period, thus forming multiple parameter time series datasets with the same time range. This method ensures that the parameters involved in the analysis are compared and analyzed within the same time range, laying the foundation for subsequent research on the changing relationships between parameters.

[0116] The time series data of each target process parameter are divided into sliding time windows, and a corresponding parameter data subsequence is formed within each time window;

[0117] Specifically, after obtaining the time series data of each target process parameter, this data needs to be further divided into sliding time windows to analyze the local variation relationships between parameters within a smaller time range. A sliding time window refers to a continuous data interval divided into predetermined lengths on the time series, and the entire time series is segmented by gradually moving the window position. In practice, the window length and window movement step can be set first, for example, establishing several continuous windows on the parameter time series at fixed time intervals. Then, the data within each window range is extracted to form corresponding parameter data subsequences. As the time window moves continuously on the time axis, multiple parameter data subsequences for different time periods can be obtained. This processing method can decompose long-term series data into multiple local data segments, enabling subsequent correlation analysis to more accurately reflect the variation relationships of parameters at different time stages, while avoiding ignoring local variation characteristics by relying solely on overall data analysis.

[0118] Within each time window, the correlation of data subsequences for any two target process parameters is calculated to obtain the corresponding parameter correlation coefficient;

[0119] Specifically, within each sliding time window, correlation calculations are needed to further analyze the data relationships between different target process parameters to obtain parameter correlation coefficients that reflect the degree of correlation between parameter changes. Correlation calculation is a statistical analysis method used to measure the strength of the relationship between two variables. Its core purpose is to determine whether two parameters exhibit synchronous or opposite trends during their changes. In practice, data subsequences corresponding to any two target process parameters can be extracted within the same time window, and the relationship between the two sequences can be analyzed using correlation calculation methods, such as calculating the degree of synchronous change of two sets of data over time. When two parameters exhibit similar trends within the time window, the calculated correlation coefficient is usually higher; conversely, when the change patterns of two parameters differ significantly, their correlation coefficients will be relatively lower. By performing pairwise calculations between multiple parameters within each time window, a set of correlation coefficients reflecting the relationship between different parameters can be obtained. These correlation coefficients can quantitatively describe the degree of mutual influence between various process parameters, providing an important basis for further analysis of parameter coupling relationships in the production process.

[0120] Construct a parameter correlation matrix reflecting the interrelationships among multiple process parameters based on the correlation coefficients of each parameter;

[0121] Specifically, after obtaining the correlation coefficients between each parameter, a parameter correlation matrix reflecting the interrelationships among multiple process parameters can be constructed based on these calculation results. A parameter correlation matrix is ​​a matrix structure used to describe the correlation between multiple variables. Its rows and columns correspond to the target process parameters involved in the analysis, and each element in the matrix represents the correlation coefficient between two corresponding parameters. In practical implementation, the pairwise parameter correlation coefficients calculated in the previous step can be arranged according to parameter number or parameter type and filled into the corresponding positions in the matrix, thus forming a complete correlation matrix. This matrix representation can intuitively reflect the overall correlation structure among all target process parameters, such as which parameters have strong correlations and which parameters have weak correlations. The parameter correlation matrix not only clearly describes the overall relationship structure between parameters but also provides a structured data foundation for further extraction of high-level features.

[0122] Based on the parameter correlation matrix, parameter correlation features are extracted to characterize the degree of coupling variation of process parameters.

[0123] Specifically, after constructing the parameter correlation matrix, parameter correlation features can be extracted from this matrix to characterize the degree of coupling change of process parameters. The degree of parameter coupling change refers to the comprehensive performance of the mutual influence and coordinated changes of multiple process parameters during operation. When the production system is in a stable operating state, the parameters usually maintain a relatively stable correlation structure. However, when a certain process link or equipment malfunctions, this correlation structure often changes. In the specific implementation process, relevant features can be extracted by analyzing the overall structural changes in the parameter correlation matrix. For example, the distribution of highly correlated regions in the matrix can be statistically analyzed, the changes in the degree of correlation between each parameter and other parameters can be analyzed, or the numerical changes in the matrix can be comprehensively calculated to obtain a set of feature indicators that can reflect the changes in parameter coupling relationships. These features can describe the coordinated change state between various parameters in the production system from an overall perspective, enabling the subsequent operation status identification model to not only focus on changes in single parameters but also to identify complex parameter coupling change patterns in the production system, thereby more accurately reflecting the actual operation status of the phosphoric acid production process.

[0124] Preferably, the step of inputting the multi-dimensional process state features into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process includes:

[0125] Based on the changes in process parameters in the production process monitoring data, the current operating condition of the phosphoric acid production process is identified, and the target operating condition corresponding to the current production process is determined.

[0126] Specifically, before identifying the operating status, it is necessary to first identify the current operating condition of the phosphoric acid production process based on the changes in various process parameters in the production process monitoring data. Operating condition typically refers to the overall operating mode of the production system at a certain stage, which is mainly determined by multiple factors such as production load, material flow rate, reaction conditions, and equipment operating status. In actual production, the range of changes and interrelationships of various process parameters often differ significantly under different operating conditions. For example, under high-load operating conditions, material flow rate and reaction temperature are usually at higher levels, while under low-load or adjustment conditions, some parameters may show a slower changing trend. Therefore, in practice, a comprehensive analysis of the value range, rate of change, and stability of key parameters can be conducted based on the production process monitoring data. By comparing with pre-set operating condition classification rules or historical operating modes, the current operating status of the production process can be classified, thereby determining the corresponding target operating condition. Through this process, the current operating environment of the system can be clearly defined before status identification, enabling the subsequent identification process to adopt more suitable analysis models for different operating conditions, thereby improving the accuracy of identification.

[0127] Based on the target operating condition, select a target abnormal state identification model corresponding to the target operating condition from the preset abnormal state identification models;

[0128] Specifically, after determining the target operating condition corresponding to the current production process, it is necessary to select a target abnormal state identification model that matches the operating condition from a pre-built set of abnormal state identification models. An abnormal state identification model is an analytical model used to determine whether the operating state of a production process is abnormal based on input feature information. It is typically constructed based on historical production data during the system design phase. Since the normal variation patterns of process parameters may differ under different operating conditions, using the same identification model for all conditions may lead to insufficient adaptability of the model to certain conditions. Therefore, in actual systems, corresponding abnormal state identification models can be established for different operating conditions, and these models can be stored uniformly in a model library. When the system determines that the current production process is in a certain target operating condition, it can call the identification model corresponding to that condition from the model library as the target abnormal state identification model. In this way, subsequent state identification processes can be performed based on models that match the current operating condition, thereby reducing the impact of operating condition differences on the identification results and improving the targeting of abnormal identification.

[0129] The multidimensional process state features are input into the target abnormal state identification model to identify the operating state and obtain the corresponding initial state identification result.

[0130] Specifically, after obtaining the target abnormal state identification model, the aforementioned extracted multidimensional process state features need to be input into the model to identify the operating status of the phosphoric acid production process and obtain the initial state identification result. Multidimensional process state features are a set of comprehensive feature information formed after feature extraction from production process monitoring data. These features include multidimensional feature indicators reflecting process parameter fluctuations, relationships between parameters, and parameter change trends. In practice, these feature information can be organized into a unified feature vector according to the model input structure and input into the target abnormal state identification model for calculation and analysis. After receiving the input features, the model compares the current feature combination with historical normal and abnormal states based on its internally established state discrimination rules or data patterns, thereby outputting the corresponding operating status judgment result. This result is the initial state identification result, which is typically used to preliminarily determine whether there is an abnormal operating condition in the current production process. By utilizing multidimensional feature information for identification, the model can simultaneously consider multiple parameter changes and their relationships during the analysis process, thereby improving the comprehensiveness of the operating status judgment.

[0131] Based on the multidimensional process state characteristics, process parameter coupling relationship data reflecting the mutual influence between multiple process parameters is constructed, and consistency analysis is performed on the process parameter coupling relationship data to obtain consistency analysis results.

[0132] Specifically, to further improve the reliability of state identification, it is necessary to construct process parameter coupling relationship data based on multi-dimensional process state characteristics, reflecting the mutual influence between multiple process parameters, and to perform consistency analysis on this data. Process parameter coupling relationship refers to the mutual influence relationship formed between different parameters during production due to process mechanisms or equipment operation. For example, changes in reaction temperature may affect reaction pressure, while changes in material flow rate may cause changes in filtration pressure difference. In practical implementation, the parameter correlation information and trend information contained in the multi-dimensional process state characteristics can be used to comprehensively calculate the changing relationships between different parameters, thereby forming a data structure describing the coupling relationships between multiple parameters. Subsequently, consistency analysis can be performed on this coupling relationship data to determine whether the current coupling relationship between parameters is consistent with the coupling pattern under normal operating conditions. For example, when the system is in normal operating condition, the parameters usually maintain a relatively stable correlation structure, but when a certain equipment or process link malfunctions, this correlation often changes. By performing consistency analysis on the coupling relationship data, consistency analysis results reflecting whether the current parameter relationship has undergone abnormal changes can be obtained, thus providing a reference for subsequent identification result verification.

[0133] Based on the consistency analysis results, the initial state identification results are verified and corrected to obtain the identification results corresponding to the current phosphoric acid production process.

[0134] Specifically, after obtaining the consistency analysis results, these results can be further used to verify and correct the initial state identification results to obtain a more accurate final identification result. Since the abnormal state identification model is mainly based on feature data, it may be affected by data fluctuations or changes in local features in some cases, leading to misjudgments or unstable judgments. Therefore, in the specific implementation process, the initial state identification results can be comprehensively compared with the consistency analysis results. When the two judgments are consistent, the initial identification result can be directly used as the final identification result; however, when there are significant differences, the initial result can be corrected according to the changes in the coupling relationship. For example, when the model judges it as a normal state but the parameter coupling relationship has significantly deviated from the normal pattern, the identification result can be corrected to an abnormal operating state. By introducing this consistency verification mechanism based on parameter coupling relationships, the misjudgment problem that may be caused by a single model judgment can be effectively reduced, enabling the system to more accurately identify the actual operating state of the phosphoric acid production process.

[0135] Preferably, the process parameter coupling relationship data, which reflects the mutual influence between multiple process parameters, is constructed based on the multidimensional process state characteristics, and consistency analysis is performed on the process parameter coupling relationship data to obtain the consistency analysis results, including:

[0136] Based on the multidimensional process state characteristics, extract the parameter time series data corresponding to each process parameter within a preset time window;

[0137] Specifically, firstly, based on the multidimensional process state characteristics, the time series data of each process parameter within a preset time window are extracted. The multidimensional process state characteristics are typically a set of comprehensive features obtained from production process monitoring data through feature extraction, containing information on the changes of multiple key process parameters at different time points. To analyze the dynamic coupling relationships between these parameters, these feature data need to be organized chronologically, and the corresponding continuous data sequences need to be extracted within the preset time window. For example, a fixed-length time window can be set, and the values ​​of each process parameter over time can be obtained within this time range, thus forming parameter time series data reflecting the dynamic changes of the parameters. In this way, the original monitoring data can be transformed into a time series form that reflects the trend and fluctuation patterns of parameter changes, providing basic data support for subsequent analysis of the correlation between parameters.

[0138] Based on the time series data of the parameters, multiple process parameters are combined in pairs within the preset time window to construct a set of process parameter combinations.

[0139] Specifically, after obtaining the time series data of each process parameter, it is necessary to combine multiple process parameters in pairs within a preset time window based on the time series data to construct a set of process parameter combinations. Specifically, the phosphoric acid production process typically involves multiple key process parameters, such as temperature, pressure, flow rate, and concentration, which often have varying degrees of mutual influence. To comprehensively analyze the coupling relationships between these parameters, all process parameters involved in the analysis can be paired up, forming a parameter combination for every two parameters. For example, when there are multiple target process parameters in the system, multiple parameter combinations can be constructed sequentially according to combination rules, thus forming a set of process parameter combinations containing all possible parameter pairings. By constructing this set of combinations, a unified analytical object can be provided for subsequent analysis of the correlations between different parameters, enabling the system to systematically analyze the interaction relationships between multiple process parameters from an overall perspective.

[0140] For each process parameter combination in the set of process parameter combinations, calculate the correlation index between the corresponding process parameters to reflect the coupling and change relationship between different process parameters;

[0141] Specifically, for each combination of process parameters in the set of process parameter combinations, it is necessary to calculate the correlation index between the corresponding process parameters to reflect the coupling and change relationships between different process parameters. Since different process parameters may exhibit various relationships during production, such as synchronous changes, mutual constraints, or indirect influences, quantitative indicators are needed to describe the strength of the correlation between these parameters. In practice, time series data of the parameters can be used to statistically analyze the changes of each parameter combination within a preset time window and calculate the corresponding correlation index, such as correlation coefficient, covariance, or other statistical indicators that can reflect the relationship between parameter changes. These indicators can determine whether there is a significant synchronous or inverse trend in the changes of two process parameters over time, and the strength of this relationship. The resulting correlation index can intuitively reflect the coupling and change characteristics between different parameters, providing a quantitative basis for constructing the overall parameter relationship structure.

[0142] Based on the correlation indexes, a process parameter coupling relationship matrix reflecting the interrelationships of multiple process parameters is constructed to obtain the process parameter coupling relationship data;

[0143] Specifically, after obtaining the correlation indexes corresponding to each combination of process parameters, a process parameter coupling relationship matrix reflecting the interrelationships of multiple process parameters can be constructed based on these indexes, thereby obtaining the process parameter coupling relationship data. Specifically, the correlation indexes corresponding to each combination of process parameters can be filled into the matrix structure according to a certain arrangement, where the rows and columns of the matrix correspond to different process parameters, and the elements in the matrix represent the correlation degree between corresponding two process parameters. In this way, the originally scattered parameter relationship information can be unified and integrated into a structured matrix, allowing the coupling relationship between multiple process parameters to be expressed in a holistic form. This process parameter coupling relationship matrix not only intuitively reflects the distribution of correlation strength between parameters, but also serves as an important data foundation for describing the structure of parameter interactions in the current production process, providing data support for subsequent comparative analysis and consistency evaluation with the normal operating condition coupling relationship model.

[0144] The process parameter coupling relationship matrix is ​​compared and analyzed with a pre-established normal operating condition coupling relationship benchmark model to calculate the degree of deviation of the current process parameter coupling relationship from the benchmark model.

[0145] Specifically, after obtaining the process parameter coupling relationship matrix, it is necessary to compare and analyze this matrix with a pre-established benchmark model of coupling relationships under normal operating conditions to calculate the degree of deviation of the current process parameter coupling relationship from the benchmark model. Specifically, during the system design or historical data analysis phase, a large amount of production data under normal operating conditions can be used to statistically model the typical correlations between multiple process parameters, thereby establishing a benchmark model reflecting the characteristics of the parameter coupling structure under normal production conditions. This benchmark model is usually stored in the form of a coupling relationship matrix or feature template, used to describe the correlation pattern that should be presented between each parameter under stable production conditions. When a new process parameter coupling relationship matrix is ​​generated during system operation, it can be compared item by item with the benchmark model, for example, by calculating the differences between corresponding elements in the matrix, or by comprehensively evaluating the two through an overall similarity index. Through this comparative analysis, the degree of change of the coupling relationship of each parameter in the current production process relative to the normal pattern can be obtained, thereby quantifying the deviation between the current coupling structure and normal operating conditions, providing an important basis for judging whether the system operating state is abnormal.

[0146] The consistency of the coupling relationship between the current process parameters is evaluated based on the degree of deviation, and the consistency analysis results are obtained.

[0147] Specifically, after obtaining the degree of deviation of the current process parameter coupling relationship from the baseline model, it is necessary to evaluate the consistency of the coupling relationship between the current process parameters based on this degree of deviation, thereby obtaining the consistency analysis results. Specifically, the core of the consistency assessment lies in determining whether the interaction relationship between each process parameter in the current production process still maintains a structural characteristic similar to that of normal operation. If the calculated deviation is small, it indicates that the current parameter coupling relationship is basically consistent with the relationship structure under normal operating conditions, and the change pattern between each parameter still conforms to normal production patterns. At this time, the system can be considered to be in a relatively stable operating state. However, when the deviation increases significantly, it indicates that the correlation pattern between the current parameters has changed significantly, and there may be some abnormal equipment operation, changes in material state, or imbalances in process conditions, which may lead to the destruction of the original coupling relationship between the parameters. Therefore, the degree of deviation can be classified and judged according to pre-set evaluation rules or thresholds, and the corresponding consistency assessment results can be output, such as being judged as "consistent coupling relationship," "slight deviation," or "obvious abnormality." The consistency analysis results can assist in judging the production status from the perspective of parameter relationship structure, and provide an important basis for the subsequent verification and correction of the initial status identification results, thereby improving the reliability and accuracy of the entire production status identification process.

[0148] Preferably, please refer to Figure 3When the identification result indicates an abnormal operating state, trend analysis is performed on the corresponding multi-dimensional process state characteristics to calculate the corresponding fault risk score, including:

[0149] Based on the time node where the identification result indicates an abnormal operating state, the corresponding abnormal analysis time interval is determined, and the target process state features within the abnormal analysis time interval are extracted from the multi-dimensional process state features.

[0150] Specifically, when the system identification result indicates that the current phosphoric acid production process is in an abnormal operating state, the first step is to determine the abnormal analysis time interval based on the time node corresponding to the abnormal operating state, and then extract the target process state features within this time interval from the multi-dimensional process state features. Specifically, the abnormal operating state is usually determined by the identification model at a certain moment or within a certain continuous time period. Therefore, a preset time range can be extended forward or backward from this time node as the center or starting point to form an abnormal analysis time interval for further analysis. This time interval can cover the key data changes before and after the occurrence of the abnormal state, enabling the system to conduct a more comprehensive analysis of the abnormal development trend. After determining this time interval, relevant feature data within this time range can be selected from the extracted multi-dimensional process state feature data, and these data can be used as target process state features for subsequent processing. Through this process, the original large amount of production data can be narrowed down to the analysis scope directly related to the abnormal state, thereby improving the targeting and efficiency of subsequent trend analysis and risk assessment.

[0151] The target process state characteristics are divided according to the process parameter type to obtain parameter time series data corresponding to multiple process parameters;

[0152] Specifically, after obtaining the characteristics of the target process state, these characteristics need to be categorized according to the type of process parameter to obtain parameter time series data corresponding to multiple process parameters. The phosphoric acid production process involves various types of process parameters, such as temperature, pressure, flow rate, concentration, and equipment operating parameters, which perform different control and monitoring functions during production. To analyze the changing trends of different parameters separately, the target process state characteristics need to be organized and classified according to parameter categories, so that parameters of the same type form continuous parameter time series data in the time dimension. Specifically, the values ​​of each process parameter within the anomaly analysis time interval can be arranged in chronological order to construct a time series reflecting the dynamic changes of that parameter. Through this classification and organization method, complex multidimensional feature data can be transformed into several parameter change curves with clear physical meaning, thus providing a foundation for subsequent independent analysis of the changing trends of each parameter.

[0153] Within a preset analysis time window, the time series data of each parameter are divided into sliding windows, and at least one of the following factors—the slope of change, the magnitude of change, and the degree of fluctuation—is calculated within each sliding window to obtain the trend change characteristics of each process parameter.

[0154] Specifically, after obtaining the time series data of each process parameter, these time series need to be divided into sliding windows within a preset analysis time window, and the trend change characteristics of the process parameters need to be calculated within each sliding window. Specifically, a fixed-length analysis time window can be set within the entire anomaly analysis time interval, and it can be continuously slid forward on the time axis according to a certain step size, thus forming multiple continuous time segments. Within each sliding window, statistical analysis can be performed on the parameter data within the corresponding time period. For example, the slope of parameter change over time can be calculated to reflect the speed of parameter rise or fall; the magnitude of parameter change within the window can be calculated to describe the overall range of parameter value change within that time period; or the degree of parameter fluctuation can be calculated to characterize the stability or intensity of parameter value fluctuation. By repeating the above calculations in multiple sliding windows, a series of trend change characteristic data describing the dynamic change characteristics of the parameters can be obtained, thus reflecting the development and change process of each process parameter in the anomaly stage in greater detail.

[0155] Based on the trend change characteristics described above, the deviation index of each process parameter from the reference range of normal operation is calculated to characterize the trend intensity of abnormal development of each process parameter.

[0156] Specifically, after obtaining the trend change characteristics of each process parameter, it is necessary to calculate the deviation index of each process parameter from the reference range of normal operation based on these characteristics, in order to characterize the trend strength of abnormal development of each process parameter. Specifically, during the system establishment phase, a large amount of historical normal production data can be used to statistically determine the reference range of each process parameter under stable operation, such as the normal range of parameter change slope, fluctuation range, or change amplitude. When the current production process is in an abnormal operating state, the calculated trend change characteristics can be compared and analyzed with the corresponding normal reference range, and the degree of difference between the two can be calculated. For example, when the change slope of a certain parameter is significantly higher than the normal range or the fluctuation level increases significantly, it indicates that the parameter is developing in an abnormal direction, and its deviation index will increase accordingly. In this way, the parameter change trend can be compared with the normal operation pattern, thereby quantitatively describing the strength of the current abnormal development of each process parameter.

[0157] The corresponding parameter risk factor value is calculated based on each deviation index, and the parameter risk factor values ​​are integrated according to the process parameter category to obtain a set of failure risk factors, wherein the parameter risk factor value is used to characterize the degree of abnormal risk of the corresponding process parameter.

[0158] Specifically, after obtaining the deviation indices for each process parameter, it is necessary to further calculate the corresponding parameter risk factor values ​​based on these indices, and then integrate the risk factor values ​​of each parameter according to the process parameter category to obtain a set of failure risk factors. The parameter risk factor values ​​are mainly used to characterize the contribution of a single process parameter to the overall abnormal risk of the production process. In practice, the deviation indices of each process parameter can be converted into risk factor values ​​according to certain calculation rules, such as through normalization or function mapping to convert them into risk indicators with uniform dimensions. Subsequently, based on the category or function of different parameters in the process system, parameter risk factors belonging to the same category can be integrated. For example, parameters related to the reaction process can be grouped into one category, and parameters related to equipment operating status can be grouped into another category, and the corresponding risk factor values ​​can be summarized separately. Through this integration process, a set of failure risk factors containing multiple categories of risk factors can be formed, thereby reflecting the potential abnormal risks of the production system from different process dimensions.

[0159] Based on the preset risk factor weights, the set of failure risk factors is weighted and calculated, and the weighted calculation results are normalized to obtain the failure risk score corresponding to the current phosphoric acid production process.

[0160] Specifically, after obtaining the set of failure risk factors, these factors need to be weighted according to preset risk factor weights, and the calculation results need to be normalized to obtain the failure risk score corresponding to the current phosphoric acid production process. Specifically, in the phosphoric acid production system, different types of process parameters typically have different degrees of impact on production safety and equipment stability. Therefore, in the risk assessment process, corresponding weights need to be set for different risk factors. For example, parameters related to critical reaction conditions may have higher weights, while some auxiliary parameters may have relatively lower weights. In actual calculation, the values ​​of each risk factor can be multiplied by their corresponding weights and summed to obtain a comprehensive risk index. Subsequently, to ensure a unified evaluation scale for the risk score, the comprehensive index can be normalized to convert it into a risk score value within a preset range, such as a value between 0 and 1 or 0 and 100. The final failure risk score can intuitively reflect the overall failure risk level of the current phosphoric acid production process, providing a quantitative basis for production managers to make early warning judgments and operational adjustments.

[0161] Preferably, the step of calculating the corresponding parameter risk factor value based on each deviation index and integrating the parameter risk factor values ​​according to the process parameter category to obtain the failure risk factor set includes:

[0162] The deviation indexes are normalized to obtain standardized deviation indexes.

[0163] Specifically, firstly, the deviation indicators need to be normalized to obtain standardized deviation indicators. Since the data range, dimensions, and magnitude of change corresponding to different process parameters may vary significantly when calculating deviation, directly using the original deviation indicators for risk calculation might lead to some parameters having an unreasonable weight in the risk assessment due to their large numerical range. Therefore, in practice, it is necessary to standardize the scale of each deviation indicator. Typically, this can be done by linearly normalizing or scaling the deviation indicators based on a preset standard range or the maximum and minimum values ​​obtained from historical data, converting them into a uniform numerical range, such as dimensionless indicators between 0 and 1. This standardization process eliminates the influence of differences in dimensions or numerical scales between different parameters, allowing the deviation of each parameter to be compared within the same evaluation system, thus providing a more reasonable basis for subsequent risk factor calculations.

[0164] Based on each of the standardized deviation indicators, the corresponding parameter risk factor values ​​are calculated using a preset risk mapping function, wherein the parameter risk factor values ​​are used to characterize the degree of abnormal risk of the corresponding process parameters;

[0165] Specifically, after obtaining the standardized deviation indicators, the corresponding parameter risk factor values ​​need to be calculated using a preset risk mapping function. These parameter risk factor values ​​characterize the degree of abnormal risk associated with the corresponding process parameters. While standardized deviation indicators reflect the degree of deviation of process parameters from their normal state, this deviation does not necessarily exhibit a simple linear relationship with actual production risk. Therefore, a risk mapping function is needed to convert these deviations into parameter risk factor values ​​that better reflect the risk level. This risk mapping function can be set based on actual process experience or historical data analysis. For example, it can use linear mapping functions, exponential functions, or piecewise functions to reflect the characteristic that the risk level may accelerate as the deviation gradually increases. By substituting each standardized deviation indicator into this risk mapping function for calculation, the corresponding parameter risk factor values ​​can be obtained, thus transforming the degree of abnormal deviation of a single parameter into a quantitative indicator that directly reflects the risk level.

[0166] Based on the phosphoric acid production process flow, the process parameters are categorized into multiple process parameter categories.

[0167] Specifically, after calculating the risk factor values ​​for each process parameter, it is necessary to categorize these parameters according to the phosphoric acid production process flow to obtain multiple process parameter categories. Since the phosphoric acid production process typically involves multiple process stages, such as reaction, filtration, concentration, and related equipment operation stages, the process parameters involved in different stages have significant differences in function and role. Therefore, during risk assessment, all process parameters involved in the analysis can be divided into several categories based on the process flow structure or parameter functional attributes. For example, parameters such as reaction temperature and reaction pressure can be classified as reaction process parameters, flow rate and material concentration as material transfer parameters, and equipment operating status parameters as equipment operation parameters. This categorization method allows parameters to be organized at the process system structure level, making the subsequent risk integration process more consistent with the logic of the actual production process.

[0168] The risk factor values ​​of each parameter are classified according to the process parameter categories, and the risk factor values ​​of parameters in the same category are integrated to obtain the category risk factor values ​​corresponding to each process parameter category.

[0169] Specifically, after classifying the process parameters, it is necessary to categorize the risk factor values ​​of each parameter according to the process parameter category, and then integrate the risk factor values ​​of parameters within the same category to obtain the category risk factor value corresponding to each process parameter category. Specifically, multiple parameter risk factor values ​​belonging to the same category can be comprehensively calculated, for example, by averaging, weighted averaging, or using the maximum value, to obtain a category risk factor value that reflects the overall risk level of that category. This integration process avoids the excessive influence of fluctuations in a single parameter on the overall risk assessment, while also comprehensively reflecting the risk level formed by the combined effect of multiple parameters in the same process step. For example, in a certain process step, if multiple parameters simultaneously exhibit large risk factor values, the category risk factor value corresponding to that category will also increase accordingly, thus more accurately reflecting the potential abnormal risks in that process step.

[0170] The values ​​of each of the aforementioned risk factors are combined to form the set of failure risk factors.

[0171] Specifically, after obtaining the category risk factor values ​​corresponding to each process parameter category, these category risk factor values ​​can be combined to form a set of failure risk factors. This set typically consists of multiple category risk factor values, each corresponding to a process parameter category in the production process, describing the risk level of that category under the current operating conditions. By uniformly aggregating risk factors from different categories into the same set, a data structure capable of reflecting the system's risk status from multiple process dimensions can be formed. This set of failure risk factors not only reflects the contribution of different process stages to the overall production risk but also serves as important input data for subsequent comprehensive risk scoring calculations, providing a foundation for further weighted evaluation and failure risk scoring, thereby achieving a systematic and quantitative analysis of potential failure risks in the phosphoric acid production process.

[0172] In addition, combined Figure 1 The phosphoric acid production fault monitoring and early warning method based on multi-source data processing described in Embodiment 1 of the present invention can be implemented by a phosphoric acid production fault monitoring and early warning system based on multi-source data processing. Figure 4 A schematic diagram of the hardware structure of the phosphoric acid production fault monitoring and early warning system based on multi-source data processing provided in Embodiment 2 of the present invention is shown.

[0173] A phosphoric acid production fault monitoring and early warning system based on multi-source data processing may include a processor and a memory storing computer program instructions.

[0174] Specifically, the processor may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement embodiments of the present invention.

[0175] The memory may include a large-capacity storage device for data or instructions. For example, and not limitingly, the memory may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory may include removable or non-removable (or fixed) media. Where appropriate, the memory may be internal or external to a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0176] The processor reads and executes computer program instructions stored in the memory to implement any of the phosphoric acid production fault monitoring and early warning methods based on multi-source data processing in the above embodiments.

[0177] In one example, a phosphoric acid production fault monitoring and early warning system based on multi-source data processing may also include a communication interface and a bus. For example, Figure 4 As shown, the processor, memory, and communication interface are connected via a bus and communicate with each other.

[0178] The communication interface is mainly used to enable communication between various modules, devices, units and / or equipment in the embodiments of the present invention.

[0179] A bus, including hardware, software, or both, couples components of the device together. For example, and not limitingly, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, a bus may include one or more buses. While specific buses are described and illustrated in embodiments of the invention, the invention contemplates any suitable bus or interconnect.

[0180] In summary, the embodiments of the present invention provide a method and system for monitoring and early warning of phosphoric acid production failures based on multi-source data processing.

[0181] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0182] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0183] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant locality, and corresponding operation entry points shall be provided for the user to choose to authorize or refuse.

[0184] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0185] The above description is merely a specific embodiment of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A method for monitoring and early warning of phosphoric acid production failures based on multi-source data processing, characterized in that, The method includes: Collect multi-source operational data during the phosphoric acid production process. The multi-source operational data includes at least process parameter data and equipment operating status data. The process parameter data includes at least one of reaction temperature, reaction pressure, material flow rate, filtration pressure difference, evaporation temperature, and finished phosphoric acid concentration. The equipment operating status data includes at least one of stirring motor current and pump vibration signal. The multi-source operational data is subjected to time synchronization and data validity verification to form a unified time series of production process monitoring data; Based on the production process monitoring data, multi-dimensional process status features reflecting the operating status of the phosphoric acid production process are extracted. The multi-dimensional process status features include at least process parameter fluctuation features, parameter correlation features, and parameter change trend features. The multidimensional process state features are input into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process. The identification result includes normal operating state and abnormal operating state. When the identification result is the abnormal operating state, trend analysis is performed on the corresponding multi-dimensional process state characteristics to calculate the corresponding fault risk score. Based on the fault risk score and the preset warning rules, trigger the corresponding level of fault warning; The step of inputting the multi-dimensional process state features into a preset abnormal state identification model to identify the operating state of the phosphoric acid production process and obtain the identification result corresponding to the current phosphoric acid production process includes: Based on the changes in process parameters in the production process monitoring data, the current operating condition of the phosphoric acid production process is identified, and the target operating condition corresponding to the current production process is determined. Based on the target operating condition, select a target abnormal state identification model corresponding to the target operating condition from the preset abnormal state identification models; The multidimensional process state features are input into the target abnormal state identification model to identify the operating state and obtain the corresponding initial state identification result. Based on the multidimensional process state characteristics, process parameter coupling relationship data reflecting the mutual influence between multiple process parameters is constructed, and consistency analysis is performed on the process parameter coupling relationship data to obtain consistency analysis results. Based on the consistency analysis results, the initial state identification results are verified and corrected to obtain the identification results corresponding to the current phosphoric acid production process; The process parameter coupling relationship data, which reflects the mutual influence between multiple process parameters, is constructed based on the multidimensional process state characteristics. Consistency analysis is then performed on the process parameter coupling relationship data to obtain the consistency analysis results, including: Based on the multidimensional process state characteristics, extract the parameter time series data corresponding to each process parameter within a preset time window; Based on the time series data of the parameters, multiple process parameters are combined in pairs within the preset time window to construct a set of process parameter combinations. For each process parameter combination in the set of process parameter combinations, calculate the correlation index between the corresponding process parameters to reflect the coupling and change relationship between different process parameters; Based on the correlation indexes, a process parameter coupling relationship matrix reflecting the interrelationships of multiple process parameters is constructed to obtain the process parameter coupling relationship data; The process parameter coupling relationship matrix is ​​compared and analyzed with a pre-established normal operating condition coupling relationship benchmark model to calculate the degree of deviation of the current process parameter coupling relationship from the benchmark model. The consistency of the coupling relationship between the current process parameters is evaluated based on the degree of deviation, and the consistency analysis results are obtained. When the identification result indicates an abnormal operating state, trend analysis is performed on the corresponding multi-dimensional process state characteristics to calculate the corresponding fault risk score, including: Based on the time node where the identification result indicates an abnormal operating state, the corresponding abnormal analysis time interval is determined, and the target process state features within the abnormal analysis time interval are extracted from the multi-dimensional process state features. The target process state characteristics are divided according to the process parameter type to obtain parameter time series data corresponding to multiple process parameters; Within a preset analysis time window, the time series data of each parameter are divided into sliding windows, and at least one of the following factors—the slope of change, the magnitude of change, and the degree of fluctuation—is calculated within each sliding window to obtain the trend change characteristics of each process parameter. Based on the trend change characteristics described above, the deviation index of each process parameter from the reference range of normal operation is calculated to characterize the trend intensity of abnormal development of each process parameter. The corresponding parameter risk factor value is calculated based on each deviation index, and the parameter risk factor values ​​are integrated according to the process parameter category to obtain a set of failure risk factors, wherein the parameter risk factor value is used to characterize the degree of abnormal risk of the corresponding process parameter. Based on the preset risk factor weights, the set of failure risk factors is weighted and calculated, and the weighted calculation results are normalized to obtain the failure risk score corresponding to the current phosphoric acid production process.

2. The phosphoric acid production fault monitoring and early warning method based on multi-source data processing according to claim 1, characterized in that, The collection of multi-source operational data during the phosphoric acid production process includes at least process parameter data and equipment operating status data, including: The key monitoring process steps in the phosphoric acid production process are identified, including the reaction section, the filtration section, and the evaporation and concentration section. In the key monitoring process steps, process parameter data used to characterize the process operation status of the phosphoric acid production process are collected respectively; At the key equipment corresponding to the key monitoring process, equipment operation status data is collected to characterize the equipment operation status. The collected process parameter data and equipment operating status data are identified according to their respective process stages and collection time to form the multi-source operating data.

3. The phosphoric acid production fault monitoring and early warning method based on multi-source data processing according to claim 1, characterized in that, The process of synchronizing and validating the multi-source operational data to form a unified time series of production process monitoring data includes: Based on the acquisition time information corresponding to the multi-source operation data, time stamps are applied to the operation data from different sources to obtain the target time stamps corresponding to the multi-source operation data; Based on the target time stamp, the multi-source running data is time aligned so that running data from different sources are mapped to a unified time axis, resulting in multi-source running data with completed time alignment. Perform data validity verification on the multi-source running data that has completed time alignment to obtain multi-source running data that has passed the data validity verification. The data validity verification includes at least data integrity verification and outlier detection. The production process monitoring data is obtained by filtering and organizing the multi-source operational data that has passed the data validity verification.

4. The phosphoric acid production fault monitoring and early warning method based on multi-source data processing according to claim 1, characterized in that, The extraction of multidimensional process status features reflecting the operating status of the phosphoric acid production process based on the production process monitoring data includes: The production process monitoring data is classified and processed according to the type of process parameter to obtain the corresponding process parameter dataset; Based on the process parameter dataset, the changes of each process parameter within a preset time window are calculated, and the fluctuation characteristics of the process parameters used to characterize the stability of process operation are extracted. Based on the variation relationship between different process parameters in the process parameter dataset, correlation analysis is performed on multiple process parameters to extract parameter correlation features that characterize the degree of mutual influence of process parameters. Based on the time series characteristics of the production process monitoring data, the changes of process parameters over time are analyzed, and parameter change trend features used to characterize the evolution of process operation are extracted. The process parameter fluctuation characteristics, parameter correlation characteristics, and parameter change trend characteristics are combined to form a multi-dimensional process status characteristic that reflects the operating status of the phosphoric acid production process.

5. The phosphoric acid production fault monitoring and early warning method based on multi-source data processing according to claim 4, characterized in that, The method involves performing correlation analysis on multiple process parameters based on the variation relationships between different process parameters in the process parameter dataset, and extracting parameter correlation features to characterize the degree of mutual influence among process parameters, including: In the process parameter dataset, identify multiple target process parameters that participate in the correlation analysis, and extract the parameter time series data corresponding to each target process parameter within a preset time window; The time series data of each target process parameter are divided into sliding time windows, and a corresponding parameter data subsequence is formed within each time window; Within each time window, the correlation of data subsequences for any two target process parameters is calculated to obtain the corresponding parameter correlation coefficient; Construct a parameter correlation matrix reflecting the interrelationships among multiple process parameters based on the correlation coefficients of each parameter; Based on the parameter correlation matrix, parameter correlation features are extracted to characterize the degree of coupling variation of process parameters.

6. The phosphoric acid production fault monitoring and early warning method based on multi-source data processing according to claim 1, characterized in that, The step of calculating the corresponding parameter risk factor value based on each deviation index, and integrating the parameter risk factor values ​​according to the process parameter category, yields a set of failure risk factors, including: The deviation indexes are normalized to obtain standardized deviation indexes. Based on each of the standardized deviation indicators, the corresponding parameter risk factor values ​​are calculated using a preset risk mapping function, wherein the parameter risk factor values ​​are used to characterize the degree of abnormal risk of the corresponding process parameters; Based on the phosphoric acid production process flow, the process parameters are categorized into multiple process parameter categories. The risk factor values ​​of each parameter are classified according to the process parameter categories, and the risk factor values ​​of parameters in the same category are integrated to obtain the category risk factor values ​​corresponding to each process parameter category. The values ​​of each of the aforementioned risk factors are combined to form the set of failure risk factors.

7. A phosphoric acid production fault monitoring and early warning system based on multi-source data processing, characterized in that, include: At least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method as described in any one of claims 1-6.

Images