Equipment fault early warning system and method based on Internet of Things

Abstract

The invention discloses an equipment fault early warning system and method based on the Internet of Things, and relates to the technical field of equipment fault early warning, and the method comprises the steps: collecting mechanical operation parameters and production process parameters of equipment, and setting each piece of equipment as an independent intelligent agent; establishing a three-level topological structure; based on the dual-threshold model, judging the running state of the equipment; when the equipment intelligent body detects that the equipment is abnormal, the equipment intelligent body immediately marks the state of the equipment intelligent body as abnormal and generates a feature code; the equipment intelligent agent receiving the feature code analyzes associated parameters in a process associated parameter atlas, and sends an analysis detection result to an adjacent equipment intelligent agent and a central management node; the central management node summarizes analysis results of all the equipment intelligent agents and judges the influence of faults on the fly ash recycling production process; and according to the severity and the influence range of the fault, performing graded early warning, and sending early warning information.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment fault warning, and specifically to an equipment fault warning system and method based on the Internet of Things. Background Art

[0002] In the field of industrial production, especially in the process of fly ash resource treatment, the timely detection and warning of equipment faults are crucial for ensuring production safety and improving production efficiency. Traditional equipment fault diagnosis methods mostly rely on manual inspections, single-point sensor monitoring, or alarm systems based on fixed thresholds, suffering from problems such as poor real-time performance, inaccurate fault location, and inability to predict potential risks. With the development of industrial Internet of Things technology, although some enterprises have achieved remote collection of equipment data, there is a lack of in-depth fusion analysis of multi-source data, making it difficult to establish the correlation between equipment operating states and production processes. In addition, when dealing with complex faults, existing systems cannot achieve collaborative diagnosis between devices, nor can they evaluate the chain impact of faults on the entire production process, resulting in the inability to take effective measures in a timely manner to reduce losses. Summary of the Invention

[0003] The purpose of the present invention is to provide an equipment fault warning system and method based on the Internet of Things to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: An equipment fault warning method based on the Internet of Things, the method includes: Deploy industrial Internet of Things sensors for each device, collect the mechanical operating parameters of the device and the process parameters of production, and set each device as an independent intelligent agent, enabling each device to have data processing and analysis capabilities; Build a communication network based on edge computing, complete the information interaction between each device intelligent agent through the communication network, and establish a three-level topology structure; Each device intelligent agent performs real-time analysis on the collected data according to a preset dual-threshold model, and based on the dual-threshold model, judges the operating state of the device; When the device intelligent agent detects that the device is abnormal, the device intelligent agent immediately marks its own state as abnormal and generates a feature code; through the communication network, the feature code is sent to adjacent device intelligent agents and the central management node; The device intelligent agent that receives the feature code detects whether there is a homologous abnormality in itself, analyzes the associated parameters in the process association parameter map, and sends the analysis and detection results to adjacent device intelligent agents and the central management node; The central management node aggregates the analysis results of each device agent, uses the swarm intelligence algorithm to evaluate the impact degree of the fault, and judges the impact of the fault on the fly ash resource production process; according to the severity and scope of the fault, it conducts hierarchical early warning and sends early warning information.

[0005] According to the above solution, the mechanical operation parameters include rotation speed, vibration, pressure, flow rate, amplitude and temperature; the process parameters include fly ash chemical composition, material particle size distribution, temperature curve, material balance data, gas emission index and filtrate water quality parameters.

[0006] According to the above solution, the dual-threshold model includes a static threshold and a dynamic threshold; for the static threshold, based on the parameters in the device design manual, industry safety standards and the performance indicators provided by the device manufacturer, a fixed threshold range for each parameter is set to form a static threshold library; for the dynamic threshold, by analyzing the historical operation data of the device, a time series analysis method is used to establish a dynamic change model of the parameter, and a dynamic threshold range that changes with the operation time and working conditions is generated; Judging the operation state of the device includes: receiving the mechanical operation parameters and the process parameters in real time, and performing a first-level comparison of the received real-time data with the corresponding thresholds in the static threshold library; when the real-time data does not exceed the static threshold, then performing a second-level comparison with the current threshold range calculated by the dynamic threshold model; at the same time, predicting the future change trend of the data through a time series prediction algorithm to generate a data prediction value, and calculating the deviation between the real-time data and the prediction value; The conditions for determining that the operation state of the device is abnormal include that the real-time parameter value exceeds the static threshold range, the real-time parameter value exceeds the dynamic threshold range, or the deviation between the actual parameter value and the prediction value exceeds the preset standard.

[0007] According to the above solution, the device agent that receives the feature code detects whether there is a homologous abnormality in itself and analyzes the associated parameters in the process-associated parameter map, including: The feature code includes device identification, abnormality type and severity; The device agent that receives the feature code matches and compares the mechanical operation parameters and process parameters collected by itself with the abnormal parameters in the feature code, and the matching comparison includes parameter type matching and numerical feature matching; Based on the process-associated parameter map, calculate the correlation coefficient between the parameters of the device agent itself and the abnormal parameters of the feature code. When the correlation coefficient exceeds the preset threshold, it is determined that there is an abnormality of the same fault type as described in the feature code; According to the causal relationship weight in the process-associated parameter map, analyze the impact degree of the current operation state of the device agent on the faulty device, and the potential impact of the abnormality of the faulty device on the operation state of the device agent; Generate analysis and detection results in combination with the analysis results, where the analysis and detection results include device identification, analysis conclusions, and relevant parameter data.

[0008] According to the above solution, the process-related parameter map includes: Perform Granger causality tests on the mechanical operation parameters and the process parameters. The Granger causality tests include calculating the test statistic between each pair of parameters, setting a significance level threshold, determining the causal relationship with statistical significance based on the significance level threshold, and recording the parameter pairs and causal directions of the causal relationships with statistical significance. Use the parameters with statistically significant causal relationships as the nodes of the process-related parameter map. The nodes include parameter names, device identification, process section codes, and data types; establish directed edges of the process-related parameter map based on the Granger causality analysis results, where the directed edges are determined by the causal direction; and assign weights to each directed edge, where the weights are proportional to the test statistic and reflect the strength of the causal relationship between the parameters.

[0009] According to the above solution, the central management node collects the data uploaded by each device agent. The data uploaded by each device agent includes feature codes, homologous anomaly detection results of adjacent device agents, and real-time status data of each node in the process-related parameter map. Based on the weights of the directed edges of the process-related parameter map, use a directed graph traversal algorithm to identify potential fault propagation paths and screen out effective propagation paths. Construct an evaluation model. The evaluation model includes the abnormal severity level of the fault source device, the number of effective propagation paths, the process criticality coefficient of the affected process sections, and the safety redundancy parameter of the current production stage; use the particle swarm optimization algorithm to calculate the fault impact index. The early warning includes first-level early warning, second-level early warning, and third-level early warning; the first-level early warning includes that the fault impact index exceeds the second warning threshold and involves core production processes; the second-level early warning includes that the fault impact index is between the first warning threshold and the second warning threshold, or affects non-core processes but involves multiple process units; the third-level early warning includes that only the fault source device is abnormal and no effective propagation path is detected. The determination methods of the first warning threshold and the second warning threshold include, based on the statistical analysis of historical fault data, calculating the distribution of the fault impact index for faults of different severity levels, taking the 95% quantile of the fault impact index distribution as the first warning threshold, and the 85% quantile as the second warning threshold, and combining industry general standards and expert experience to correct the thresholds to ensure the rationality of the threshold settings. The early warning information includes the identification and coordinate position of the fault source device, the list of confirmed affected devices and process section numbers, the priority ranking of recommended disposal measures, and the estimated recovery time interval. For the third-level early warning, prompt information is sent to the operation and maintenance personnel, suggesting further inspection; for the second-level early warning and the first-level early warning, an emergency alarm is immediately issued, and the emergency plan is automatically triggered. The emergency plan includes adjusting the operating parameters of some equipment, starting standby equipment, and suspending relevant production links to reduce the losses caused by failures.

[0010] An equipment fault early warning system based on the Internet of Things, which includes a data acquisition module, an agent module, a communication network module, a process association module, a central decision-making module, and a visualization module; The data acquisition module is used to collect mechanical operating parameters and process parameters in real time; The agent module includes a data interface module, a data processing module, a data storage module, and a communication interface module; The communication network module includes a physical layer module, a protocol layer module, and an application layer module; The process association module is used to construct and maintain a causal relationship map of process parameters; The central decision-making module maintains data interaction with the agent module through the communication network module, receives feature codes and homologous anomaly detection results, and issues decision instructions to the agent module; The visualization module displays the equipment status and early warning information.

[0011] According to the above solution, the data interface module receives the mechanical operating parameters and the process parameters in real time; The data processing module preprocesses the received data. The preprocessing includes data filtering, outlier removal, and data standardization; stores the preprocessed data in the data storage module; based on a preset dual-threshold model, compares and analyzes the preprocessed data and historical data to generate an equipment status report and an anomaly feature code; The communication interface module establishes connections with the equipment agent and the central decision-making module according to the communication protocol of the three-level topology structure, transmits the equipment status report and the anomaly feature code, and receives the status information of adjacent equipment agents.

[0012] According to the above solution, the physical layer module adopts a hybrid networking method of industrial Ethernet and industrial wireless network to ensure that the network covers all equipment nodes; The protocol layer module includes an equipment layer, a unit layer, and a central layer. The equipment layer completes point-to-point communication between process-related equipment, the unit layer completes real-time multicast communication of process unit equipment clusters, and the central layer completes star communication between all production line equipment and the central decision-making module; The application layer module includes device status information transmission, abnormal feature code transmission, and network management. The device status information transmission transmits the structured data of the device operation status code and key parameter values. The abnormal feature code transmission transmits the feature codes including device identification, abnormal type, and severity. The network management maintains the communication link status.

[0013] According to the above solution, the three-level topology structure includes device layer topology, unit layer topology, and central layer topology. The device layer topology enables direct communication between devices. The unit layer topology enables cluster communication of devices within a process unit. The central layer topology enables communication between all production line devices and the central decision-making module.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By configuring an independent agent and edge computing capabilities for each device, the present invention realizes distributed real-time data processing, significantly improving the timeliness of fault detection. 2. Through the three-level early warning mechanism combined with the particle swarm optimization algorithm, the present invention realizes the accurate quantitative evaluation of the impact degree of faults. 3. The process-related parameter map constructed by the present invention accurately identifies the fault propagation path between devices, realizing the prediction of cascading faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flowchart of the steps of a device fault early warning method based on the Internet of Things according to the present invention; Figure 2 is a schematic structural diagram of a device fault early warning system based on the Internet of Things according to the present invention. [[ID=twenty-two]] DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment: As Figure 1 - Figure 2 shown, the present invention provides a technical solution, a device fault early warning method based on the Internet of Things. The method includes the steps: S1. Deploy industrial Internet of Things sensors for each device, collect the mechanical operation parameters of the device and the process parameters of the production, and set each device as an independent agent, so that each device has the ability of data processing and analysis; Specifically, the fly ash resource processing plant has 100 core production equipment units, covering key process steps such as crushing, sorting, high-temperature melting, and heavy metal extraction. These units collect real-time mechanical operating parameters, including speed, vibration, pressure, flow rate, amplitude, and temperature, as well as production process parameters. Mechanical operating parameters include fly ash chemical composition, particle size distribution, temperature profiles, material balance data, gas emission indicators, and filtrate water quality parameters. Each unit is equipped with an edge computing unit as an independent intelligent entity, capable of processing 1,000 sets of data per second. This is for illustrative purposes only and is not intended to be limiting.

[0018] S2. Build a communication network based on edge computing, complete information exchange between each device agent through the communication network, and establish a three-level topology structure; Specifically, a hybrid network of industrial Ethernet and 5G wireless network is adopted to construct a three-level topology structure of equipment layer, unit layer and central layer. In the equipment layer, 10 adjacent crushing equipment communicate directly in a star topology, and the data transmission delay is less than 50ms; in the unit layer, the 20 devices in the sorting process unit communicate through multicast, and can transmit 10MB of data per second; the central layer realizes the star connection between the equipment of the entire production line and the central management node to ensure real-time data aggregation.

[0019] S3. Each device agent performs real-time analysis on the collected data based on a preset dual-threshold model and determines the device operating status based on the dual-threshold model. Specifically, the dual threshold model includes static thresholds and dynamic thresholds. For static thresholds, fixed threshold ranges for each parameter are set based on the parameters in the equipment design manual, industry safety standards, and performance indicators provided by the equipment manufacturer to form a static threshold library. For example, for a high-temperature melting furnace, the static threshold for furnace temperature is set at 800-1200°C and the static threshold for pressure is set at 0.1-0.3MPa according to the equipment manual. For dynamic thresholds, a parameter dynamic change model is established by analyzing the historical operating data of the equipment using a time series analysis method, generating a dynamic threshold range that changes with operating time and operating conditions. Furthermore, for example, to generate a dynamic threshold range for a high-temperature melting furnace, the operating data of the high-temperature melting furnace over the past year is collected to establish a temperature dynamic threshold. The formula is: DT t =μ t +2σ t ; Among them, DT t Expressed as the dynamic threshold at time t, μ t Expressed as the average temperature of the same period in the past seven days, σ t Expressed as standard deviation; for example: at 14:00 on a certain day, μ is calculated t =1050℃,σ t =20℃, the dynamic threshold is 1090℃; Furthermore, determining the device operating state includes: receiving mechanical operating parameters and process parameters in real time, and performing a first-level comparison of the received real-time data with the corresponding thresholds in the static threshold library; when the real-time data does not exceed the static threshold, a second-level comparison is made with the current threshold range calculated by the dynamic threshold model; meanwhile, the future change trend of the data is predicted through a time series prediction algorithm to generate a data prediction value, and the deviation between the real-time data and the prediction value is calculated; for example: when the temperature of the melting furnace is detected in real time to be 1120 °C (exceeding the dynamic threshold of 1090 °C), and it is predicted through the time series prediction algorithm that the temperature will continue to rise in the next 10 minutes, an abnormal alarm is triggered. Furthermore, the conditions for determining that the device operating state is abnormal include that the real-time parameter value exceeds the static threshold range, the real-time parameter value exceeds the dynamic threshold range, or the deviation between the actual parameter value and the prediction value exceeds the preset standard.

[0020] S4. When the device agent detects that the device is abnormal, the device agent immediately marks its own state as abnormal and generates a feature code; through the communication network, the feature code is sent to the adjacent device agents and the central management node. Specifically, for example: Device A detects an abnormality in the whole building and generates a feature code: Device A - Vibration Abnormality - Severe, and sends the feature code to the adjacent device agents and the central management node.

[0021] S5. The device agent that receives the feature code detects whether there is a homologous abnormality in itself, analyzes the associated parameters in the process association parameter map, and sends the analysis and detection results to the adjacent device agents and the central management node. Specifically, the device agent that receives the feature code matches the mechanical operation parameters and process parameters collected by itself with the abnormal parameters in the feature code. The matching includes parameter type matching and numerical feature matching; based on the process-related parameter map, calculate the correlation coefficient between the parameters of the device agent itself and the abnormal parameters of the feature code. When the correlation coefficient exceeds the preset threshold, it is determined that there is an abnormality of the same fault type as described in the feature code; for example: after the adjacent device B receives the feature code, compare its own vibration parameters and find that the standard deviation of the vibration value reaches 0.5 m / s² within the past 1 hour (the normal range is 0.1 - 0.3 m / s²), which matches the abnormal parameters of device A, and it is determined that there is an abnormality of the same fault type as described in the feature code; according to the causal relationship weight in the process-related parameter map, analyze the influence degree of the current operation state of the device agent on the faulty device, and the potential influence of the abnormality of the faulty device on the operation state of the device agent; for example: based on the process-related parameter map, there is a causal relationship between the feed flow of device B and the vibration of device A. By calculating the correlation coefficient, the correlation coefficient is obtained as 0.85, which exceeds the preset threshold of 0.7, and it is confirmed that there is a strong correlation; combined with the analysis results, generate the analysis and detection results, and the analysis and detection results include device identification, analysis conclusion, and relevant parameter data.

[0022] Furthermore, perform Granger causality test on the mechanical operation parameters and process parameters. The Granger causality test includes calculating the test statistic between each pair of parameters, setting the significance level threshold, based on the significance level threshold, determining the causal relationship with statistical significance, and recording the parameter pairs and causal directions of the causal relationship with statistical significance; use the parameters of the causal relationship with statistical significance as the nodes of the process-related parameter map, and the nodes include parameter name, device identification, process section code, and data type; establish the directed edges of the process-related parameter map based on the Granger causality analysis results, and the directed edges are determined by the causal direction; and assign weights to each directed edge, and the weights are proportional to the test statistic, reflecting the strength of the causal relationship between the parameters.

[0023] S6. The central management node aggregates the analysis results of each device agent, uses the swarm intelligence algorithm to evaluate the influence degree of the fault, and judges the influence of the fault on the fly ash resource utilization production process; according to the severity and influence range of the fault, conduct hierarchical early warning and send early warning information.

[0024] Specifically, the central management node collects the data uploaded by each device agent. The data uploaded by each device agent includes feature codes, the homologous anomaly detection results of adjacent device agents, and the real-time status data of each node in the process correlation parameter map. Based on the weights of the directed edges in the process correlation parameter map, a directed graph traversal algorithm is used to identify potential fault propagation paths and filter out effective propagation paths. An evaluation model is constructed. The evaluation model includes the anomaly severity level of the fault source device, the number of effective propagation paths, the process criticality coefficient of the affected process section, and the safety redundancy parameter of the current production stage. The particle swarm optimization algorithm is used to calculate the fault impact index. The early warning includes level-one early warning, level-two early warning, and level-three early warning. The level-one early warning includes that the fault impact index exceeds the level-two warning threshold and involves core production processes. The level-two early warning includes that the fault impact index is between the level-one warning threshold and the level-two warning threshold, or affects non-core processes but involves multiple process units. The level-three early warning includes that only the fault source device is abnormal and no effective propagation path is detected. Further, the fault impact index F = α×S + β×P + γ×C + δ×R; where F represents the fault impact index; α, β, γ, and δ represent weight coefficients, and α + β + γ + δ = 1; S represents the anomaly severity of the fault source device; P represents the number of effective propagation paths; C represents the criticality coefficient of the affected process section; R represents the safety redundancy of the current production stage. Further, the method for determining the level-one warning threshold and the level-two warning threshold includes statistically analyzing historical fault data, calculating the distribution of the fault impact index for faults of different severities, taking the 95th percentile of the fault impact index distribution as the level-one warning threshold, and the 85th percentile as the level-two warning threshold, and correcting the thresholds in combination with industry general standards and expert experience to ensure the rationality of the threshold settings. For example: the level-one warning threshold is 3.0, and the level-two warning threshold is 2.0. Further, the early warning information includes the identification and coordinate position of the fault source device, the list of confirmed affected devices and the process section numbers, the priority ranking of the recommended disposal measures, and the estimated recovery time interval. Further, for the level-three early warning, a prompt message is sent to the operation and maintenance personnel, suggesting further inspection. For the level-two early warning and the level-one early warning, an emergency alarm is immediately issued, and the emergency response plan is automatically triggered. The emergency response plan includes adjusting the operating parameters of some devices, starting standby devices, and suspending relevant production links to reduce the losses caused by the fault. For example: the central management node aggregates the analysis results of all device agents to calculate the fault impact index F as 2.5. Since the fault impact index is between the level-one warning threshold and the level-two warning threshold, it is determined as a level-two early warning. An alarm is immediately issued, the operating parameters of relevant devices are automatically adjusted, and standby devices are started to avoid the expansion of the fault. This is only for illustration and not for limitation.

[0025] The present invention provides another technical solution, a device fault warning method based on the Internet of Things, fly ash resource utilization device process-related parameter mapping and fault propagation analysis; For example: The fly ash resource utilization factory includes crushing equipment, sorting equipment, high-temperature melting furnace equipment, heavy metal extraction equipment, and flue gas treatment equipment. Each equipment is interconnected through material flow and energy flow to form a production network; Mechanically operating parameters and production process parameters of the equipment are collected in real time. The mechanically operating parameters include vibration speed, temperature, pressure, motor current, and temperature; the process parameters include fly ash particle size distribution, CaO, heavy metal concentration, and waste gas emission index.

[0026] Perform a moving average filter on the 12-month historical data, eliminate outliers, and retain valid samples; Perform a Granger causality test on the mechanically operating parameters and process parameters. For example: Test the parameters for vibration speed X and rotational speed Y. The Granger causality test formula is: ; Among them, G represents the comprehensive evaluation index; RSS r represents the sum of squared residuals of the constrained model; RSS ur represents the sum of squared residuals of the unconstrained model, m represents the lag order, n represents the number of data points, and k represents the total number of model parameters; Calculate the test statistic between each pair of parameters, and set the significance level threshold. Based on the significance level threshold, determine the causal relationship with statistical significance, and record the parameter pairs and causal directions of the causal relationship with statistical significance. For example: Edge: E1, Node: N1, Weight: 0.82, Causal relationship: For every 1% increase in vibration, the temperature rises by 0.7 °C; When the crushing equipment detects an abnormal situation and the vibration value of equipment C suddenly increases to 5.8 mm / s (static threshold 4.5 mm / s), use the directed graph traversal algorithm to traverse the edges in the directed graph with weights greater than 0.6, identify potential fault propagation paths, filter out effective propagation paths, identify effective propagation paths, and obtain the effective propagation path as: vibration value - temperature - CaO content. This is only for illustrative purposes and not for limitation.

[0027] The present invention provides another technical solution, a device fault warning system based on the Internet of Things. The system includes a data acquisition module, an agent module, a communication network module, a process correlation module, a central decision-making module, and a visualization module; The data acquisition module is used to collect mechanically operating parameters and process parameters in real time; The agent module includes a data interface module, a data processing module, a data storage module, and a communication interface module; The communication network module includes a physical layer module, a protocol layer module, and an application layer module; A process correlation module for constructing and maintaining a process parameter causal relationship graph; A central decision-making module that maintains data interaction with the agent module through a communication network module, receives feature codes and homologous anomaly detection results, and issues decision instructions to the agent module; A visualization module for displaying device status and warning information.

[0028] Specifically, a data interface module that receives mechanical operation parameters and process parameters in real time; A data processing module that preprocesses the received data. The preprocessing includes data filtering, outlier removal, and data standardization; stores the preprocessed data in a data storage module; and performs comparison and analysis on the preprocessed data and historical data based on a preset dual-threshold model to generate a device status report and an anomaly feature code; A communication interface module that establishes connections with device agents and the central decision-making module according to the communication protocol of a three-level topology structure, transmits device status reports and anomaly feature codes, and receives the status information of adjacent device agents.

[0029] Specifically, the physical layer module uses a hybrid networking method of industrial Ethernet and industrial wireless network to ensure that the network covers all device nodes; The protocol layer module includes a device layer, a unit layer, and a central layer. The device layer completes point-to-point communication between process-related devices, the unit layer completes real-time multicast communication of a device cluster in a process unit, and the central layer completes star communication between all production line devices and the central decision-making module; The application layer module includes device status information transmission, anomaly feature code transmission, and network management; device status information transmission transmits structured data of device operation status codes and key parameter values, anomaly feature code transmission transmits feature codes including device identification, anomaly type, and severity, and network management maintains the status of communication links.

[0030] Specifically, the three-level topology structure includes a device layer topology, a unit layer topology, and a central layer topology; the device layer topology enables direct communication between devices, the unit layer topology enables device cluster communication within a process unit, and the central layer topology enables communication between all production line devices and the central decision-making module.

[0031] It is obvious to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. An equipment fault warning method based on the Internet of Things, characterized in that: The method includes: Deploying industrial Internet of Things sensors for each device to collect the mechanical operation parameters of the device and the process parameters of production, and setting each device as an independent agent, enabling each device to have the ability of data processing and analysis; Building a communication network based on edge computing, completing the information interaction between each device agent through the communication network, and establishing a three-level topology structure; Each device agent performs real-time analysis on the collected data according to a preset dual-threshold model, and judges the device operation state based on the dual-threshold model; When the device agent detects that the device is abnormal, the device agent immediately marks its own state as abnormal and generates a feature code; through the communication network, the feature code is sent to the adjacent device agent and the central management node; The device agent that receives the feature code detects whether there is a homologous abnormality in itself, analyzes the associated parameters in the process associated parameter map, and sends the analysis and detection results to the adjacent device agent and the central management node; The central management node summarizes the analysis results of each device agent, uses the swarm intelligence algorithm to evaluate the influence degree of the fault, and judges the influence of the fault on the fly ash resource production process; according to the severity and influence range of the fault, it conducts hierarchical early warning and sends early warning information.

2. The method for warning of device faults based on the Internet of Things according to claim 1, wherein: The mechanical operation parameters include rotational speed, vibration, pressure, flow rate, amplitude and temperature; the process parameters include fly ash chemical composition, material particle size distribution, temperature curve, material balance data, gas emission index and filtrate water quality parameters.

3. The method for warning of device faults based on the Internet of Things according to claim 1, wherein: The dual-threshold model includes a static threshold and a dynamic threshold; for the static threshold, according to the device design manual parameters, industry safety standards and performance indicators provided by the device manufacturer, a fixed threshold range for each parameter is set to form a static threshold library; For the dynamic threshold, by analyzing the historical operation data of the device, a parameter dynamic change model is established by using the time series analysis method to generate a dynamic threshold range that changes with the operation time and working conditions; The judgment of the device operation state includes: receiving the mechanical operation parameters and the process parameters in real time, and performing a first-level comparison of the received real-time data with the corresponding thresholds in the static threshold library; when the real-time data does not exceed the static threshold, it is then compared with the current threshold range calculated by the dynamic threshold model for a second-level comparison; at the same time, the future change trend of the data is predicted through the time series prediction algorithm to generate a data prediction value, and the deviation between the real-time data and the prediction value is calculated; The conditions for determining that the device operation state is abnormal include that the real-time parameter value exceeds the static threshold range, the real-time parameter value exceeds the dynamic threshold range, or the deviation between the actual parameter value and the prediction value exceeds the preset standard.

4. A device fault warning method based on the Internet of Things according to claim 1, characterized in that: The device agent that receives the feature code detects whether there is a homologous abnormality in itself, and analyzes the associated parameters in the process associated parameter map, including: The feature code includes device identification, abnormality type and severity; The device agent that receives the feature code matches the mechanical operation parameters and process parameters collected by itself with the abnormal parameters in the feature code. The matching comparison includes parameter type matching and numerical feature matching; Based on the process-related parameter map, calculate the correlation coefficient between the parameters of the device agent itself and the abnormal parameters of the feature code. When the correlation coefficient exceeds the preset threshold, it is determined that there is an abnormality of the same fault type described in the feature code; According to the causal relationship weight in the process-related parameter map, analyze the influence degree of the current operation state of the device agent on the faulty device, and the potential influence of the abnormality of the faulty device on the operation state of the device agent; Combined with the analysis results, generate an analysis and detection result, which includes device identification, analysis conclusion, and relevant parameter data.

5. The device fault warning method based on the Internet of Things according to claim 4, characterized in that: The process-related parameter map includes: Perform Granger causality test on the mechanical operation parameters and the process parameters. The Granger causality test includes calculating the test statistic between each pair of parameters, setting the significance level threshold, and based on the significance level threshold, determining the causal relationship with statistical significance and recording the parameter pairs and causal directions of the causal relationship with statistical significance; Use the parameters with statistically significant causal relationships as the nodes of the process-related parameter map. The nodes include parameter name, device identification, process section code, and data type; establish the directed edges of the process-related parameter map based on the Granger causality analysis results. The directed edges are determined by the causal direction; and assign weights to each directed edge. The weights are proportional to the test statistic and reflect the strength of the causal relationship between parameters.

6. The method for warning of device faults based on the Internet of Things according to claim 1, wherein: The central management node collects the data uploaded by each device agent. The data uploaded by each device agent includes feature codes, homologous abnormality detection results of adjacent device agents, and real-time status data of each node in the process-related parameter map; Based on the weights of the directed edges of the process-related parameter map, use the directed graph traversal algorithm to identify potential fault propagation paths and screen out effective propagation paths; Construct an evaluation model. The evaluation model includes the abnormal severity level of the fault source device, the number of effective propagation paths, the process criticality coefficient of the affected process section, and the safety redundancy parameter of the current production stage; use the particle swarm optimization algorithm to calculate the fault impact index; The warning includes first-level warning, second-level warning, and third-level warning; the first-level warning includes that the fault impact index exceeds the second warning threshold and involves core production processes; The second-level warning includes that the fault impact index is between the first warning threshold and the second warning threshold, or affects non-core processes but involves multiple process units; the third-level warning includes that only the fault source device is abnormal and no effective propagation path is detected; The determination method of the first warning threshold and the second warning threshold includes, based on the statistical analysis of historical fault data, calculating the distribution of the fault impact index of faults with different severity levels, and combining industry general standards and expert experience to set the first warning threshold and the second warning threshold; The warning information includes the identification and coordinate position of the fault source device, the list of confirmed affected devices and the process section numbers, the priority ranking of recommended disposal measures, and the estimated recovery time interval.

7. An equipment fault warning system based on the Internet of Things, characterized in that: The system includes a data acquisition module, an agent module, a communication network module, a process association module, a central decision-making module, and a visualization module; The data acquisition module is used to collect mechanical operation parameters and process parameters in real time; The agent module includes a data interface module, a data processing module, a data storage module, and a communication interface module; The communication network module includes a physical layer module, a protocol layer module, and an application layer module; The process association module is used to construct and maintain a causal relationship graph of process parameters; The central decision-making module maintains data interaction with the agent module through the communication network module, receives feature codes and homologous anomaly detection results, and issues decision instructions to the agent module; The visualization module displays device status and warning information.

8. The device fault warning system based on the Internet of Things according to claim 7, characterized in that: The data interface module receives the mechanical operation parameters and the process parameters in real time; The data processing module preprocesses the received data, and the preprocessing includes data filtering, outlier removal, and data standardization; The preprocessed data is stored in the data storage module; Based on a preset dual-threshold model, the preprocessed data and historical data are compared and analyzed to generate a device status report and an anomaly feature code; The communication interface module establishes connections with device agents and the central decision-making module according to the communication protocol of the three-level topology structure, transmits the device status report and the anomaly feature code, and receives the status information of adjacent device agents.

9. The device fault warning system based on the Internet of Things according to claim 8, characterized in that: The physical layer module adopts a hybrid networking method of industrial Ethernet and industrial wireless network to ensure that the network covers all device nodes; The protocol layer module includes a device layer, a unit layer, and a central layer. The device layer completes point-to-point communication between process-associated devices, the unit layer completes real-time multicast communication of process unit device clusters, and the central layer completes star communication between all production line devices and the central decision-making module; The application layer module includes device status information transmission, anomaly feature code transmission, and network management; the device status information transmission transmits structured data of device operation status codes and key parameter values, the anomaly feature code transmission transmits feature codes including device identification, anomaly type, and severity, and the network management maintains the communication link status.

10. The device fault warning system based on the Internet of Things according to claim 9, characterized in that: The three-level topology structure includes a device layer topology, a unit layer topology, and a central layer topology; the device layer topology realizes direct communication between devices, the unit layer topology realizes device cluster communication within a process unit, and the central layer topology realizes communication between all production line devices and the central decision-making module.

Images