A filling system data anomaly diagnosis method for VR display

Abstract

The present application relates to the field of data anomaly diagnosis, and more particularly to a filling system data anomaly diagnosis method for VR display. The content includes: collecting device raw data, and performing time synchronization and normalization processing to construct a normalized data vector; modeling the causal relationship of the normalized data vector to obtain causal weights; based on the normalized data vector, converting multi-dimensional physical quantities into a single abnormality measurement value to construct a multi-dimensional deviation quantity; based on the multi-dimensional deviation quantity, performing abnormal score fusion calculation to obtain an abnormal score result, and correcting the abnormal score result to obtain a corrected abnormal score result; based on the corrected abnormal score result, performing anomaly diagnosis to generate an anomaly diagnosis result and performing visual display. The problems of different sampling times of multi-source heterogeneous data, lack of correlation analysis between parameters, unclear abnormal propagation path, and difficulty of traditional fixed threshold-based diagnosis method to accurately identify complex working condition abnormalities are solved.

Description

Technical Field

[0001] This invention relates to the field of data anomaly diagnosis, and more particularly to a method for diagnosing data anomalies in a filling system for VR display. Background Technology

[0002] With the continuous development of industrial applications such as mine backfilling and tailings treatment, backfilling systems are gradually evolving towards larger scale, continuous operation, and automation. Their processes typically include multiple stages such as thickeners, mixing systems, water pumps, and pipeline distribution, with each stage tightly coupled through slurry flow. In actual operation, the system relies on various parameters such as flow rate, pressure, level, current, and power to maintain stable operation. These parameters are collected in real-time by sensors distributed across different devices and uploaded to the monitoring system. However, due to differences in sampling frequency, data accuracy, and transmission mechanisms among different devices, the collected data exhibits asynchrony over time, making it difficult to establish effective correspondences between parameters and affecting the accuracy of subsequent data analysis.

[0003] The operational status monitoring of most filling systems still relies primarily on single parameter thresholds or simple rule-based judgment methods, such as identifying abnormal states by setting upper limits for flow or lower limits for pressure. While this method is simple to implement, it often only identifies local or significant anomalies because it ignores the coupling relationships between different process parameters and the dynamic transmission characteristics of the process flow. It struggles to detect hidden anomalies that gradually propagate from upstream fluctuations to downstream processes in a timely manner. Furthermore, under complex operating conditions, a single parameter may be affected by multiple factors, and its changes do not necessarily directly correspond to the occurrence of a fault, thus easily leading to misjudgments or missed diagnoses. Summary of the Invention

[0004] This invention provides a data anomaly diagnosis method for filling systems for VR display, which solves the problems of asynchronous sampling of multi-source heterogeneous data, lack of correlation analysis between parameters, unclear anomaly propagation path, and the difficulty of accurately identifying complex working condition anomalies by traditional diagnostic methods based on fixed thresholds.

[0005] The present invention provides a method for diagnosing data anomalies in a filling system for VR display, which specifically includes the following steps:

[0006] S1. Collect raw data from the device, perform time synchronization and normalization processing to obtain normalized data, and construct a normalized data vector; perform causal relationship modeling on the normalized data vector to obtain causal weights;

[0007] S2. Based on the normalized data vector, the multidimensional physical quantity is transformed into a single anomaly metric value to construct a multidimensional deviation quantity; based on the multidimensional deviation quantity, an anomaly scoring function is constructed, and anomaly scoring fusion calculation is performed to obtain anomaly scoring results; combined with causal weights, the anomaly scoring results are corrected to obtain corrected anomaly scoring results; based on the corrected anomaly scoring results, anomaly diagnosis is performed to generate anomaly diagnosis results, and these results are then visualized.

[0008] Preferably, the normalized data specifically includes:

[0009] Normalized thickener feed volumetric flow rate, normalized center feed tank liquid level height, normalized mixer shaft power, normalized motor current, normalized water pump outlet pressure, and normalized pipeline flow velocity.

[0010] Preferably, the calculation process of the causal weight specifically includes:

[0011] A flow-driven term is constructed based on the normalized thickener feed volumetric flow rate, and a pressure suppression term is constructed based on the normalized pump outlet pressure. Based on the flow-driven term and the pressure suppression term, a process flow path impedance factor is introduced, and combined with an exponential decay function, a causal function is constructed to obtain the causal weights.

[0012] Preferably, the specific calculation process of the process flow path impedance factor includes:

[0013] Establish a directed relation chain, form node pairs by nodes in the directed relation chain that have process flow relationships, and calculate the actual flow path length and path complexity coefficient of the slurry between node pairs; generate the path impedance factor of the node pairs based on the actual flow path length and path complexity coefficient of the slurry between node pairs; and obtain the process flow path impedance factor by weighted summation of the path impedance factors of all node pairs.

[0014] Preferably, pipeline topology data is introduced, and the actual flow path length of the slurry between node pairs is obtained based on the actual length of the pipeline in the pipeline topology data; the path complexity coefficient between node pairs is calculated based on the included radian of the turning structure in the pipeline topology data.

[0015] Preferably, based on the normalized data vector, normalized historical data is introduced to calculate the reference average values ​​of the thickener feed volumetric flow rate, the water pump outlet pressure, and the energy consumption coupling amount; based on the reference average values ​​of the thickener feed volumetric flow rate, the water pump outlet pressure, and the energy consumption coupling amount, combined with the normalized data vector, a multidimensional deviation is calculated; the multidimensional deviation specifically includes flow rate deviation, pressure deviation, energy consumption coupling deviation, liquid level modulation deviation, and flow velocity consistency deviation.

[0016] Preferably, the specific method for constructing the anomaly scoring function is as follows:

[0017] Based on flow rate deviation, pressure deviation, and energy coupling deviation, a main anomaly term is constructed; based on flow rate consistency evaluation deviation, a flow rate consistency enhancement term is constructed; based on level modulation deviation and flow rate deviation, a level modulation term is constructed; combining the main anomaly term, the flow rate consistency enhancement term, and the level modulation term, an anomaly scoring function is constructed, and the anomaly scoring results are obtained.

[0018] Preferably, the anomaly scoring results are corrected based on the process flow path impedance factor and combined with causal weights to obtain the corrected anomaly scoring results.

[0019] Preferably, the corrected anomaly score is compared with a preset upper limit and lower limit of the dynamic threshold to obtain an anomaly diagnosis result, and the anomaly diagnosis result is transmitted to the VR display module for visualization.

[0020] The beneficial effects of the technical solution of the present invention are:

[0021] 1. By constructing a causal relationship model based on the filling process, the originally isolated process parameter changes are transformed into directional propagation relationships. In particular, by introducing flow-driven terms, pressure-inhibiting terms, and process flow path impedance factors, a causal function that conforms to the actual material conveying law is established, so that anomalies can not only be detected, but also reflect their propagation intensity in the process chain. Compared with the traditional threshold-based static judgment method, it can identify potential "propagation anomalies" and significantly improve the diagnostic capability under complex working conditions.

[0022] 2. Regarding the construction of anomaly scoring, this invention achieves a multi-dimensional characterization of the filling system's operating status by designing multi-dimensional deviations (flow rate deviation, pressure deviation, energy consumption coupling deviation, liquid level modulation deviation, and flow velocity consistency deviation). Among them, energy consumption coupling deviation can reflect abnormal equipment load, flow velocity consistency deviation can identify local blockage or leakage problems, and liquid level modulation deviation can constrain the impact of material supply fluctuations on anomaly judgment. By combining the sum of squares and logarithmic functions, an anomaly scoring function with nonlinear response characteristics is constructed, enabling the filling system to have good sensitivity and distinguishability for both sudden and gradual anomalies, thereby avoiding the problems of missed or false detection caused by traditional single-indicator judgment. Attached Figure Description

[0023] Figure 1 This is a flowchart of a data anomaly diagnosis method for a filling system oriented towards VR display, as described in this invention. Detailed Implementation

[0024] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0026] The following description, in conjunction with the accompanying drawings, details a specific scheme for a VR-oriented filling system data anomaly diagnosis method provided by the present invention.

[0027] See attached document Figure 1 The diagram illustrates a flowchart of a data anomaly diagnosis method for a filling system oriented towards VR display, provided by an embodiment of the present invention. The method includes the following steps:

[0028] S1. Collect raw data from the device, perform time synchronization and normalization processing to obtain normalized data, and construct a normalized data vector; perform causal relationship modeling on the normalized data vector to obtain causal weights;

[0029] Process parameters are continuously collected by sensor acquisition units, such as flow sensors, deployed by staff on equipment like thickeners, mixers, pumps, and hydrocyclones, to obtain raw equipment data. This raw data includes thickener feed volumetric flow rate, central feed tank liquid level, mixer shaft power, motor current, pump outlet pressure, and pipeline flow velocity. The raw equipment data is then time-synchronized using a timestamp-based synchronization method to obtain time-synchronized raw equipment data. Finally, the time-synchronized raw equipment data is normalized using a maximum-minimum method to obtain normalized data, and a normalized data vector is constructed.

[0030]

[0031] in, At the same time point Below is a data vector composed of normalized values ​​of multiple key process parameters, i.e., the normalized data vector; Indicates a unified point in time Below is the normalized feed volumetric flow rate of the thickener; Indicates a unified point in time Below, the normalized liquid level height of the center feed tank; Indicates a unified point in time Below, the normalized mixer shaft power; Indicates a unified point in time Below is the normalized motor current; Indicates a unified point in time Below, the normalized pump outlet pressure; Indicates a unified point in time Below, normalized pipeline flow velocity; store the original equipment data and normalized data into the filling database;

[0032] After obtaining the normalized data vector, causal relationship modeling is performed based on the filling process flow. The technical objective is to quantify the influence relationships between different process parameters. Specifically, the process involves: first, establishing a logical topology based on the filling process flow, treating "thickener feed → feed hopper → mixer → pipeline → water pump" as a directed relationship chain; then, at each time point... Next, the normalized data is substituted into the causal function to calculate the causal weights; the causal function is constructed based on fluid transport-driven models, time correlation analysis models, etc., and its specific expression is as follows:

[0033]

[0034] in, At the same time point The intensity of anomalies or state changes propagating from upstream process nodes to downstream nodes is called the causal weight. It is a flow nonlinear enhancement coefficient used to control the influence of the normalized thickener feed volume flow rate on the propagation intensity. It is determined based on the flow operation ratio, whereby the flow operation ratio... It is the current time point. The ratio of the thickener feed volumetric flow rate in the original equipment data after time synchronization to the rated value of the thickener feed volumetric flow rate obtained from the system design data or equipment nameplate, specifically: ; This is the process flow path impedance factor, and the specific calculation formula is as follows: ,in, Represents node pairs Path impedance factor Represents node pairs The weights of node pairs are determined by the fact that node pairs are composed of nodes in a directed chain that have a process flow relationship (e.g., thickener → mixer). The formula for calculating the weight is: Path impedance factor of node pairs It is obtained by coupling path length and path complexity, and the specific calculation formula is as follows: ,in, For node pairs The actual flow path length of the slurry between them. for The path complexity coefficient between them; the specific calculation process is as follows: extract pipeline topology data from the existing VR model, including the actual length of each pipeline segment, the connection order of each pipeline segment, and the included radian of each turning structure, and construct node pairs. The set of pipe lengths corresponding to the flow path between them , Represents node pairs Between The actual length of the pipeline section Represents node pairs The number of all pipe segments contained in the path between them is calculated by performing a cumulative calculation on the above set of pipe lengths to obtain the node pairs. The actual flow path length of the slurry between The path complexity coefficient is determined by the included radian angle of the turning structure, and the definition is as follows: , Represents node pairs Between The included angle radian of the segmental turning structure; The distance normalization coefficient is obtained by summing the lengths of each pipeline segment in the filling system process path; it is expressed as a power function of the normalized thickener feed volumetric flow rate. As a flow-driven term, it is used to enhance propagation intensity; the square root of the normalized pump outlet pressure. As a pressure suppression term, it is used to suppress propagation; Introducing process flow path resistance factor The decay characteristics are characterized by an exponential decay function; the aforementioned causal function outputs causal weights. This serves as the input for anomaly scoring calculation.

[0035] S2. Based on the normalized data vector, the multidimensional physical quantity is transformed into a single anomaly metric value to construct a multidimensional deviation quantity; based on the multidimensional deviation quantity, an anomaly scoring function is constructed, and anomaly scoring fusion calculation is performed to obtain anomaly scoring results; combined with causal weights, the anomaly scoring results are corrected to obtain corrected anomaly scoring results; based on the corrected anomaly scoring results, anomaly diagnosis is performed to generate anomaly diagnosis results, and these results are then visualized.

[0036] The anomaly scoring stage then begins, with the technical objective of transforming multidimensional physical quantities into a single anomaly metric while maintaining the physical coupling between the parameters. The specific implementation process is as follows: First, based on a window length of... Historical filling data was used to calculate reference averages for thickener feed volumetric flow rate, pump outlet pressure, and energy coupling. The window length was set based on operational stability requirements; for example, 50-100 was used during continuous filling, and 10-30 during startup or load fluctuation. In a specific example, a value of 50 was used here. The historical filling data was normalized and taken from an existing filling database. The reference average for thickener feed volumetric flow rate was used. Reference average value of water pump outlet pressure and the reference average of energy consumption coupling The specific calculation method is as follows:

[0037]

[0038] in, This is an index variable used to iterate through the data points within the history window; Indicates a point in history Below is the normalized feed volumetric flow rate of the thickener; Indicates a point in history Below, the normalized pump outlet pressure; Indicates a point in history Below is the normalized motor current; Indicates a point in history Below, the normalized mixer shaft power; Indicates a point in history The electrical energy input intensity corresponding to the mechanical load is the energy consumption coupling quantity.

[0039] Then, based on the reference averages of the thickener feed volumetric flow rate, pump outlet pressure, and energy consumption coupling, the multidimensional deviation of the current time point is calculated, including:

[0040] Flow deviation This is used to describe the degree of deviation of the feed volume flow rate of a thickener;

[0041]

[0042] Pressure deviation This is used to describe the relative increase in pump outlet pressure;

[0043]

[0044] Energy consumption coupling deviation Used to reflect abnormal motor load conditions:

[0045]

[0046] in, Indicates the current time point Energy consumption coupling quantity under; Used to normalize energy consumption coupling deviation;

[0047] Then, liquid level modulation deviation is introduced. This is used to correct the impact of the feeding status on anomalies. The feeding status refers to the stability of the slurry supplied by the thickener to the subsequent mixing and conveying system, that is, the ability of the filling system to continuously supply stable slurry downstream at the current point in time, which is determined by both the flow input and the liquid level. The specific calculation formula for the liquid level modulation deviation is:

[0048]

[0049] in, It is the ratio of liquid level to flow rate, used to characterize the retention characteristics of slurry in the central feed tank. Its value reflects the degree of matching between the current storage state and the conveying capacity. The design basis of the liquid level modulation deviation is that the liquid level change has periodic and gradual characteristics. Therefore, a sine function is used for smooth modeling.

[0050] Finally, define the flow rate consistency deviation. It is used to detect whether the feed volume flow rate of the thickener matches the pipeline flow rate, thereby identifying local blockages or leaks.

[0051]

[0052] After calculating all deviations, the anomaly score fusion calculation process begins. The technical objective of this process is to uniformly map multidimensional deviations to a single score value. Its construction logic is as follows: based on multivariate Euclidean distance models and nonlinear response enhancement models, a sum-of-squares structure is used to fuse the main deviations (flow deviation, pressure deviation, and energy coupling deviation), a logarithmic function is used to enhance abrupt change sensitivity, and a level modulation term is combined to construct the following anomaly score function:

[0053]

[0054] in, At a certain point in time The anomaly score results are used to describe the degree to which the current system state deviates from normal operation; It is the main anomaly, used to describe the overall anomaly intensity of multidimensional deviation variables; It is a flow consistency enhancement term, used to enhance sensitivity to flow anomalies; This is a level modulation term, used to describe the restraining effect of the liquid level on anomalies.

[0055] Since anomaly propagation is affected by spatial structure, a process flow path impedance factor is introduced, combined with causality weights, to correct the anomaly scoring results, resulting in a corrected anomaly scoring result. The corrected anomaly scoring result is then stored in the filling database.

[0056]

[0057] in, At a certain point in time The revised anomaly score indicates the final degree of anomaly after considering the effects of process path propagation; It is an exponential term used to describe the nonlinear enhancement effect of anomalous propagation; It is the effective propagation strength, which represents the causal propagation capability under the influence of path impedance.

[0058] Furthermore, data at specific time points can be obtained from the existing filling database. The previous N corrected outlier scores were analyzed, and their mean and standard deviation were calculated. The sum of the mean and three times the standard deviation was used as the upper limit of the dynamic threshold. The difference between the mean and three standard deviations is used as the lower limit of the dynamic threshold. Then, the corrected anomaly score is compared with the upper and lower limits of the dynamic threshold to obtain the anomaly diagnosis result: including when When the filling system is in normal operating condition, it is determined that the filling system is in normal operating condition; when When the condition is determined to be a minor abnormality, it indicates that there is an initial deviation in the filling system but no significant failure has yet occurred; when When the condition is determined to be significantly abnormal, the abnormal diagnosis result is transmitted to the VR display module for visualization, thereby realizing a closed-loop abnormal diagnosis process driven by a single correction score.

[0059] In summary, a method for diagnosing data anomalies in a filling system designed for VR demonstrations has been developed.

[0060] The order of the embodiments is for illustrative purposes only and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0061] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0062] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for diagnosing data anomalies in a filling system for VR display, characterized in that, Specifically, the following steps are included: S1. Collect raw equipment data and perform time synchronization and normalization processing to obtain normalized data, including normalized thickener feed volumetric flow rate, normalized central feed tank liquid level height, normalized mixer shaft power, normalized motor current, normalized water pump outlet pressure, and normalized pipeline flow velocity, constructing a normalized data vector; perform causal relationship modeling on the normalized data vector, construct causal functions, and obtain causal weights: ； in, At the same time point Causal weights; Indicates a unified point in time Normalized thickener feed volumetric flow rate; It is the flow nonlinear enhancement coefficient; Indicates a unified point in time Normalized pump outlet pressure; It is the process flow path impedance factor. ,in, Represents node pairs Path impedance factor Represents node pairs The weights are determined by the fact that node pairs consist of nodes in the established directed relation chain that have process flow relationships. , , For node pairs The actual flow path length of the slurry between them. for The path complexity coefficient between them and The specific calculation process is as follows: Pipeline topology data is introduced, and based on the actual length of the pipeline in the pipeline topology data, the actual flow path length of the slurry between node pairs is obtained. , ,in, Represents node pairs Between The actual length of the pipeline section Represents node pairs The number of all pipe segments included in the path between nodes; the path complexity coefficient between node pairs is calculated based on the included radian angle of the turning structures in the pipe topology data. , , Represents node pairs Between The included angle radian of the segmental turning structure; This is the distance normalization coefficient; Indicates flow-driven items; Indicates a pressure suppression term; S2. Based on the normalized data vector, normalized historical data is introduced to calculate the reference average values ​​of the thickener feed volumetric flow rate, the pump outlet pressure, and the energy consumption coupling. Based on these reference average values, combined with the normalized data vector, the multidimensional physical quantities are transformed into single outlier metrics, constructing multidimensional deviations, including flow rate deviation, pressure deviation, energy consumption coupling deviation, liquid level modulation deviation, and flow velocity consistency deviation. Used to describe the degree of deviation in the feed volumetric flow rate and pressure deviation of a thickener. Used to describe the relative increase in pump outlet pressure, energy consumption coupling deviation Used to reflect abnormal motor load conditions and liquid level modulation deviation. Used to correct the impact of feeding status on anomalies, flow rate consistency deviations. This is used to detect whether the feed volumetric flow rate of the thickener matches the pipeline flow rate. Indicates a unified point in time Normalized motor current Indicates a unified point in time Normalized mixer shaft power This represents the reference mean of energy consumption coupling. Indicates a unified point in time Normalized center feed tank liquid level height This represents the reference average value of the feed volume flow rate of the thickener; Based on the multidimensional deviation, an anomaly scoring function is constructed, and anomaly scoring fusion calculation is performed to obtain the anomaly scoring results: ； in, At a certain point in time Abnormal scoring results; It is the main exception; It is a flow rate consistency enhancement item; It is a liquid level modulation item; Based on the process flow path impedance factor and combined with causality weights, the anomaly scoring results are corrected to obtain the corrected anomaly scoring results. : ； Anomaly diagnosis is performed based on the corrected anomaly scoring results, generating anomaly diagnosis results and displaying them visually.

2. The method for diagnosing data anomalies in a filling system for VR display according to claim 1, characterized in that, The corrected anomaly score is compared with the preset upper and lower dynamic thresholds to obtain the anomaly diagnosis result, which is then transmitted to the VR display module for visualization.

Images