Multi-source heterogeneous system and method based on liquid air energy storage plant-level monitoring

Abstract

The invention relates to the technical field of plant-level monitoring systems, in particular to a multi-source heterogeneous system and method for plant-level monitoring based on liquid air energy storage. According to the technical scheme, the method comprises the following steps of S1, collecting and preprocessing multi-source heterogeneous data in real time; s2, constructing a cross-system equipment graph; s3, formulating a hierarchical response intelligent strategy; s31, making a topological graph; s32, according to a hierarchical response strategy, dynamic adjustment is carried out according to the severity degree and the propagation range of the alarm; s33, calculating a weight; and S34, dividing response levels according to the weights. Multi-source heterogeneous data real-time acquisition and preprocessing can align millisecond-level and second-level data streams and eliminate time sequence errors, so that the system alarm association accuracy is improved, six system data formats are unified through JSON structured conversion, the analysis error rate can be reduced, and a basis is provided for large-scale monitoring; a three-flow topology model is established, a key equipment network is constructed, the root cause positioning time is shortened, and the false alarm rate is reduced through a topology weight optimization mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of plant-level monitoring systems, and particularly relates to a multi-source heterogeneous system and method for plant-level monitoring based on liquid air energy storage. BACKGROUND

[0002] In the field of existing power plant integrated systems, products similar to plant-level monitoring information systems are mainly used in traditional energy industries such as thermal power plants, and are applied to calculating the economic performance of power plants and providing optimal strategies for production operation. Patent No. CN120146275A discloses a coal-electricity-aluminum integrated production scheduling system and method, relating to the technical field of integrated production scheduling. The system includes a scheduling strategy subsystem and production site subsystem, planning subsystem, financial subsystem, etc. connected to the scheduling strategy subsystem. The method uses a preset profit and loss balance model to calculate the adjusted work plan and production data and financial data, determines the production scheduling optimization strategy according to the calculation results output by the preset profit and loss balance model, and optimizes the coal-electricity supply and power generation strategy. The system integrates several subsystems to achieve optimization of coal, electricity and aluminum product supply and demand strategies, and realizes the scientificity, economy and efficiency of coal-electricity-aluminum integrated production scheduling. However, the above system does not have the ability of field migration, and the digital and integrated platform suitable for traditional fields is not suitable for liquid air energy storage fields. Most of the integrated scheduling platforms on the market focus more on the economic benefit optimization of traditional industries such as thermal power plants, and their subsystems aim to link the supply and demand strategies of coal, electricity and aluminum products, and the core module is the production plan adjustment driven by financial data. However, liquid air energy storage involves phase change energy transfer and needs to monitor the chain reaction of material state, such as air purity affecting the efficiency of the expander. The existing system can only handle static production factors such as raw coal inventory and electricity price fluctuations. The fundamental contradiction is that the existing integrated platform cannot solve the physical process fault positioning of "cold energy leakage - sudden drop in power generation". If the integrated system of traditional industries is used forcibly, it will cause business mismatch, poor performance and other problems.

[0003] One of the main functions of the plant-level monitoring information system is data collection and integration. Patent No. CN117573995A discloses a method for displaying equipment state data in the production process, which belongs to the technical field of plant-level monitoring information systems. The system includes: S1 layer collects production process data from the bottom control system, S2 layer collects and analyzes data during equipment production and operation, S3 layer performs two-layer calculation on the main parameters in the unit operation process, S4 layer analyzes the parameter statistical indicators, and uses the web end to display the analysis results, and S5 layer monitors the plant data in the management information system in the form of web. The final web interface displays the unit operation parameters, the efficiency of each unit, the public system, and the main auxiliary machine output and energy consumption technical and economic indicators in the form of graphics and trend user diagrams, which can intuitively display the equipment state, alarm information, and unit performance test information, making the equipment management clear and transparent, and allowing for custom filtering conditions to view different types of equipment state. The plant-level monitoring information system uses a five-layer data pipeline to realize equipment parameter visualization, but the data analysis stops at single-index threshold alarm and simple unit efficiency calculation and statistical level. Its main defect is the failure of cross-system coupling. When the compressor system reports a high compressor vibration fault and the GDS system reports a combustible gas alarm, the existing technology only displays the alarm independently and cannot establish a causal chain of refrigerant leakage + compressor surge. Most existing plant-level monitoring information systems only stop at data collection and do not design a multi-system alarm weight linkage mechanism. Moreover, the secondary calculation in this technology only performs simple integration and statistics on important parameters in the unit operation, does not introduce a fault propagation model, and cannot deeply discover the root cause of the equipment failure.

[0004] In the field of existing multi-source heterogeneous systems, the patent with publication number CN120429618A discloses a power distribution network diagram real consistency checking system for multi-source heterogeneous system data collaboration, which includes a topology feature extraction module, a cross-system topology alignment module, and an abnormality tracing module. The topology feature extraction module extracts multi-dimensional attribute features and topology connection relationships of device nodes from the multi-source heterogeneous system to generate dynamic fingerprint vectors. The cross-system topology alignment module receives the dynamic fingerprint vectors and constructs an intermediate semantic layer to output standard fingerprint vectors. By analyzing the differentiated topology rules of the multi-source heterogeneous system, a unified topology model is generated, and the comparison difference values of the standard fingerprint vectors between different systems and the equivalent connection relationships of electrical connection degree characteristics are calculated. The abnormality tracing module receives the comparison difference values and constructs an abnormality propagation tree to generate an abnormality propagation path. By analyzing the coupling weight of the comparison equivalent connection relationships and the comparison difference values, the root node of the abnormality is located. Although the technology achieves the characteristics of multi-source heterogeneous systems and can perform abnormality tracing across systems (solving the technical problem of the existing technology 2, the patent with publication number CN117573995A), the power distribution network checking system aligns the multi-source topology through dynamic fingerprint vectors to achieve consistency verification of "device nodes-electrical connection". However, the topology rules are based on deterministic physical connections (cables / switches). This multi-source heterogeneous characteristic based on physical connections is not suitable for the propagation of material flow in liquid air energy storage, such as cold energy transmission efficiency fluctuations. Moreover, the system cannot be compatible with the six systems of liquid air energy storage projects: compressor system, expander system, integrated control system, distributed control system, electrical system, and flammable toxic gas alarm system. The "abnormality propagation tree" relies on electrical connection degree calculation, but the energy form conversion from the compressor system (mechanical energy) to the cold storage tank (phase change energy) to the electrical system (electric energy) cannot be represented by a unified fingerprint vector.

[0005] Therefore, the present application proposes a multi-source heterogeneous system and method for liquid air energy storage plant-level monitoring. SUMMARY

[0006] The purpose of the present application is to address the problems in the background art by proposing a multi-source heterogeneous system and method for liquid air energy storage plant-level monitoring.

[0007] The technical solution of the present application is a multi-source heterogeneous method for liquid air energy storage plant-level monitoring, which includes the following steps: S1: Real-time acquisition and preprocessing of multi-source heterogeneous data; S2: Constructing a cross-system device map; S3: Developing a hierarchical response intelligent strategy; S31: Developing a topology map; S32: Hierarchical response strategy, dynamically adjusting according to the severity and propagation range of the alarm; S33: Calculating weights; S34. Classify response levels according to weights; S4: Root cause identification and problem backtracking.

[0008] Preferably, step S1 includes: S11: Process the raw data; S12: Design a protocol parsing template for each type of system and dynamically convert the raw data into a unified JSON format; S13: Data standardization processing: S14: The converted data uses the Dynamic Time Warping (DTW) algorithm to align data with different sampling rates to solve the timing misalignment problem caused by the difference in sampling rates of multiple protocols.

[0009] Preferably, step S2 includes: S21: Perform traffic splitting on the system; Distribute distributed control systems, safety instrumented systems, and toxic and combustible gas alarm systems into the control flow; Distribute the compressor system and expander system into the material flow; Distribute the electrical automation system into the energy flow; S22: Reorganize equipment relationships based on energy conversion stages.

[0010] Preferably, step S33 includes: The total weight consists of the base weight (W_base) and the propagation weight (W_spread); The base weight (W_base) is: the weight corresponding to the highest level alarm among all current alarms; The mapping relationship between alarm levels and weights is defined as follows: Level 1 Alarm (Emergency): Weight = 1.0; Level 2 Alarm (Critical): Weight = 0.7; Level 3 Alarm (Warning): Weight = 0.4; Level 4 Alarm (Notification): Weight = 0.1; The propagation weight (W_spread) is calculated based on the number (n) of alarm devices associated in the topology graph. W_spread = min(0.5, 0.1 n) # The propagation weight limit is 0.5, and increases by 0.1 for each additional associated alarm device; Total weight (W_total) = W_base + W_spread If the number of system couplings (n) changes, the alarm escalation mechanism is calculated as follows: .

[0011] Preferably, step S4 includes: S41: Based on thermodynamics or fluid dynamics, a Bayesian engine is introduced to determine the source of the fault. S42: Using dynamic structure learning to construct causal networks; S43: The dual-engine mechanism generates a weighted fusion algorithm to calculate the probability of possible failures; Configure dynamic weight adjustment when sensor data is missing. Automatically upgraded to 0.8 to enhance physical constraints.

[0012] Preferably, step S4 further includes: S44: Historical Data Replay S441: Use the two-layer configuration screen for root cause localization and fault tracing; S442: Select the fault period; S443: View the second-level configuration screen of the unit that experienced the fault during this period; S434: Locate the time of the first alarm, select the playback speed, and perform frame-by-frame analysis.

[0013] Based on a multi-source heterogeneous system for plant-level monitoring of liquid air energy storage, it includes a cross-system alarm correlation engine, a dynamic weight fusion model, and a visualization interaction module.

[0014] Preferably, real-time acquisition and preprocessing of multi-source heterogeneous data will align millisecond-level and second-level data streams, eliminating timing errors; By using JSON structured conversion, the data format of the six major systems is unified, thus integrating the six systems; Establish a three-flow topology model and construct a network for key equipment; In the plant-level monitoring information system, root cause localization is combined with fault tracing screens, and historical operating condition playback functions are set up to visualize the root cause.

[0015] Compared with the prior art, the present invention has the following beneficial technical effects: First, it fills the gap in the field of liquid air energy storage for plant-level monitoring information systems. Real-time acquisition and preprocessing of multi-source heterogeneous data will align millisecond-level and second-level data streams, eliminating timing errors and improving the accuracy of system alarm correlation. By unifying the data formats of the six major systems through JSON structured conversion, the parsing error rate will be reduced, providing a foundation for large-scale monitoring. Second, a three-flow topology model (material flow / energy flow / control flow) is established to construct a network of key equipment, improve root cause localization time, and reduce false alarm rate through topology weight optimization mechanism; for the first time, process-level integration of six major systems is achieved, and the accuracy of cross-system fault propagation path identification is improved. Third, establish a hierarchical intelligent response strategy to reduce unnecessary downtime. Dynamic weight threshold calculation can improve the accuracy of emergency response. Furthermore, resource optimization is significant, and maintenance costs will be reduced by specializing the response strategy through process flow. Fourth, the Bayesian-physical hybrid model improves the accuracy of root cause localization, and the intelligent playback function based on two-layer configuration shortens the fault analysis time and makes it easier for operators to mark the fault propagation path. Attached Figure Description

[0016] Fig. 1 This is a flowchart of the interaction between the Bayesian network and the physical model in this invention; Fig. 2 This is a flowchart of the root cause localization and fault tracing process using a two-layer configuration screen in this invention; Fig. 3 This invention involves constructing a cross-system device map. Detailed Implementation

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The components of the embodiments of the invention described and shown in the accompanying drawings can typically be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention.

[0019] 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.

[0020] Example like Figs. 1-3 As shown, the multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage proposed in this invention includes the following steps: S1: Real-time acquisition and preprocessing of multi-source heterogeneous data; S11: Process the raw data; S12: Design a protocol parsing template for each type of system and dynamically convert the raw data into a unified JSON format; JSON { "tag": "CMP_VIB_001", / / Equipment label (compressor vibration measuring point) "value": 4.8, / / numerical value (mm / s) "timestamp": 1723445062000, / / Unix timestamp (milliseconds) "source": "Modbus / TCP", "range": [0, 10] / / Measuring range } Python def convert_to_json(raw_data, protocol_type): # Protocol parsing template selection if protocol_type == "Modbus / TCP": template = MODBUS_TEMPLATE # Predefined field mapping elif protocol_type == "IEC61850": template = IEC61850_TEMPLATE # Generate structured JSON json_data = { "tag": template["prefix"] + raw_data["address"], "value": apply_calibration(raw_data["value"]), # Range conversion "timestamp": get_precise_timestamp(), # High-precision clock "metadata": { "unit": template["unit"], "range": template["range"] } } return json.dumps(json_data) # Serializes the data into a JSON string.

[0021] S13: Data standardization processing: S14: The converted data uses a Dynamic Time Warping (DTW) algorithm to align data at different sampling rates, addressing timing misalignment issues caused by differences in sampling rates across multiple protocols. An asymmetric stepping mode is used for the compressor vibration signal (10kHz) and the gas concentration signal (1Hz), and a pre-triggered buffer mechanism is employed to handle HART protocol latency.

[0022] For example, slow protocols in combustible gas alarm systems cannot trigger alarms promptly for minor leaks; therefore, a pre-trigger caching mechanism is used. if gas_system.detect_rising_edge(): # Detects rising edge along concentration spikes activate_high_speed_mode(10Hz)# Temporarily increase the sampling rate.

[0023] The Dynamic Time Warping (DTW) algorithm is used to align data at different sampling rates. def dtw align(signal high, signal low): # Interpolate low-frequency signals to high-frequency timestamps x=np.arange(0,len(signal low)) f=interpld(x,signal low,kind='cubic') aligned_low=f(np.linspace(0,len(signal low)-1,len(signal high)) Return aligned low.

[0024] Through the collaborative innovation of JSON+DTW, the accuracy of data acquisition can be improved, and the time alignment error can be reduced from ±500ms in the traditional solution to ±5ms. The memory usage will be reduced by 60% by utilizing the structured characteristics of JSON. The processing latency will also be reduced for the fast response requirements of the safety instrument system. Furthermore, this method has strong reusability. If a new system is added, only the protocol template needs to be added, without changing the core algorithm.

[0025] S2: Construct a cross-system device map; S21: Perform traffic splitting on the system; Distribute distributed control systems, safety instrumented systems, and toxic and combustible gas alarm systems into the control flow; Distribute the compressor system and expander system into the material flow; Distribute electrical automation systems into the energy flow.

[0026] S22: Reorganize the equipment relationships based on the energy conversion stage, breaking away from the traditional grouping method based on physical location, as shown in the table below: The weighting calculation formula is established based on the principles of fluid mechanics, thermodynamics, and cybernetics, or the weighting calculation formula is designed according to the requirements of liquid air energy storage technology (the focus of this patent is not on the calculation of the weighting process formula, but mainly on the system design): Material flow: efficiency × pressure loss × attenuation Energy Flow: Conversion Rate × Health Level × Temperature Compensation Control flow: Delay response × Signal reliability It supports dynamic weight optimization and real-time weight adjustment strategies, breaking through the limitations of traditional static topologies and realizing online weight optimization driven by process parameters.

[0027] def update_weight(edge): if edge.type == "MaterialFlow": # Dynamic correction based on gas purity purity = DCS.get("Air_Purity") edge.weight = (0.8 + 0.2 purity) elif edge.type == "ControlFlow": # Reduce priority when network is congested if network_load>80%: edge.weight = 0.7.

[0028] S3: Develop a tiered response intelligent strategy; S31. Develop a topological map; Alarm devices within the defined radius of the topology graph can be spatially associated, defined by the number of hops outward from the alarming device node as the center in the topology graph. For example, R=2 represents the alarming device itself and its directly adjacent and indirectly adjacent (separated by one device) devices. Within the alarm propagation radius R, if multiple alarming device nodes exist, these alarms are considered potentially associated (belonging to the same fault propagation chain). The radius definition rules are shown in the table below: S32. Tiered response strategy, dynamically adjusted according to the severity and spread of the alarm; Dimension 1: Severity of alarms (the level of a single alarm is provided by each subsystem) Dimension 2: Alarm propagation range (the number of alarm devices associated with an alarm device node within a radius R in the topology graph). The two dimensions are combined to form a total weight, and then the response level is determined based on the total weight.

[0029] S33, Calculate the weights; The total weight consists of the base weight (W_base) and the propagation weight (W_spread); The base weight (W_base) is: the weight corresponding to the highest level alarm among all current alarms; The mapping relationship between alarm levels and weights is defined as follows: Level 1 Alarm (Emergency): Weight = 1.0; Level 2 Alarm (Critical): Weight = 0.7; Level 3 Alarm (Warning): Weight = 0.4; Level 4 Alarm (Notification): Weight = 0.1; The propagation weight (W_spread) is calculated based on the number (n) of alarm devices associated in the topology graph. W_spread = min(0.5, 0.1 n) # The propagation weight limit is 0.5, and increases by 0.1 for each additional associated alarm device; Total weight (W_total) = W_base + W_spread If the number of system couplings (n) changes, the alarm escalation mechanism is calculated as follows: .

[0030] S34. Response levels are determined based on the total weight, as shown in the table below: S4: Root cause identification and problem backtracking.

[0031] S41: Based on thermodynamics or fluid dynamics, a Bayesian engine is introduced to determine the source of the fault. Building upon traditional reliance on physical models (based on thermodynamics / fluid dynamics) to identify and locate fault sources, a Bayesian engine is introduced. Dynamic structure learning is used to construct a causal network, updating node dependencies. For example, when the Bayesian network outputs the hypothesis: tank leakage P=0.85, it needs to be verified through physical equations to see if the data violates physical laws. At this point, a fault or alarm confidence level is generated based on physical constraints. Simultaneously, the Bayesian network calculates a fault alarm confidence level based on existing relevant parameters. Furthermore, this is a bidirectional feedback learning mechanism; when the physical model identifies abnormal parameters, it triggers local relearning of the Bayesian network, updating the conditional probability table (CPT).

[0032] S42: Using dynamic structure learning to construct causal networks; S43: The dual-engine mechanism generates a weighted fusion algorithm to calculate the probability of possible failures, which is more advantageous than a single engine.

[0033] At this point, it is necessary to set dynamically adjustable weights to suit different situations, such as when sensor data is missing. Automatically upgraded to 0.8, enhancing physical constraints. Root cause hypothesis according to... Sort in descending order and keep only The assumption is greater than 0.7.

[0034] Step S4 also includes: S44: Historical Data Replay S441: Use the two-layer configuration screen for root cause localization and fault tracing; S442: Select the fault period; S443: View the second-level configuration screen of the unit that experienced the fault during this period; S434: Locate the time of the first alarm, select the playback speed, and perform frame-by-frame analysis.

[0035] In the plant-level monitoring information system, root cause localization can be used in conjunction with the fault tracing screen. The historical operating condition playback function can be set up to view the changing trends of various parameters related to the operating condition and the equipment within the time interval of the fault occurrence. The steps of the historical data playback function are: historical database - playback engine - time controller - second-level configuration screen - parameter playback - possible root cause marking - correlation analysis interface.

[0036] The process for root cause localization and fault tracing using a two-layer configuration screen is as follows: The operator can select the fault period and view the two-layer configuration screen of the unit that failed during that period, locate the first alarm time, select the playback speed, and perform frame-by-frame analysis. The operator can then use the playback configuration screen to view the changing trends of relevant parameters, mark parameter abrupt change points, and use these parameters to view the status of other devices and parameters within the topology diagram for root cause analysis.

[0037] This system is based on a multi-source heterogeneous system for plant-level monitoring of liquid air energy storage, including a cross-system alarm correlation engine, a dynamic weighted fusion model, and a visualization interaction module. Real-time acquisition and preprocessing of multi-source heterogeneous data will align millisecond-level and second-level data streams, eliminating timing errors.

[0038] This system is expected to improve the reliability of air liquid storage systems by 30%, accelerate the implementation of clean energy, and reduce the risk of grid fluctuations caused by unplanned outages.

[0039] By using JSON structured conversion, the data format of the six major systems is unified, thus integrating the six systems; Establish a three-flow topology model and construct a network for key equipment; The equipment relationships are divided into three types of process flows: material flow (compressor system, expander system), energy flow (expander system, electrical system), and control flow (distributed control system, safety instrumented system, combustible and toxic gas alarm system).

[0040] In the plant-level monitoring information system, root cause localization is combined with fault tracing screens, and historical operating condition playback functions are set up to visualize the root cause.

[0041] The above specific embodiments are merely preferred embodiments of the present invention. Based on the technical solutions of the present invention and the relevant teachings of the above embodiments, those skilled in the art can make various alternative improvements and combinations to the above specific embodiments. The above specific embodiments are merely explanations of the present invention and are not intended to limit the present invention.

Claims

1. A multi-source heterogeneous method for plant-level monitoring of liquid air energy storage, characterized in that, Includes the following steps: S1: Real-time acquisition and preprocessing of multi-source heterogeneous data; S2: Construct a cross-system device map; S3: Develop a tiered response intelligent strategy; S31. Develop a topological map; S32. Tiered response strategy, dynamically adjusted according to the severity and spread of the alarm; S33, Calculate the weights; S34. Classify response levels according to weights; S4: Root cause identification and problem backtracking.

2. The multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage according to claim 1, characterized in that, Step S1 includes: S11: Process the raw data; S12: Design a protocol parsing template for each type of system and dynamically convert the raw data into a unified JSON format; S13: Data standardization processing: S14: The converted data uses the Dynamic Time Warping (DTW) algorithm to align data with different sampling rates to solve the timing misalignment problem caused by the difference in sampling rates of multiple protocols.

3. The multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage according to claim 2, characterized in that, Step S2 includes: S21: Perform traffic splitting on the system; Distribute distributed control systems, safety instrumented systems, and toxic and combustible gas alarm systems into the control flow; Distribute the compressor system and expander system into the material flow; Distribute the electrical automation system into the energy flow; S22: Reorganize equipment relationships based on energy conversion stages.

4. The multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage according to claim 1 or 3, characterized in that: Step S33 includes: The total weight consists of the base weight (W_base) and the propagation weight (W_spread); The base weight (W_base) is: the weight corresponding to the highest level alarm among all current alarms; The mapping relationship between alarm levels and weights is defined as follows: Level 1 Alarm (Emergency): Weight = 1.0; Level 2 Alarm (Critical): Weight = 0.7; Level 3 Alarm (Warning): Weight = 0.4; Level 4 Alarm (Notification): Weight = 0.1; The propagation weight (W_spread) is calculated based on the number (n) of alarm devices associated in the topology graph. W_spread = min(0.5, 0.1 n) # The propagation weight cap is 0.5, and increases by 0.1 for each additional associated alarm device; Total weight (W_total) = W_base + W_spread If the number of system couplings (n) changes, the alarm escalation mechanism is calculated as follows: 。 5. The multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage according to claim 4, characterized in that, Step S4 includes: S41: Based on thermodynamics or fluid dynamics, a Bayesian engine is introduced to determine the source of the fault. S42: Using dynamic structure learning to construct causal networks; S43: The dual-engine mechanism generates a weighted fusion algorithm to calculate the probability of possible failures; Configure dynamic weight adjustment when sensor data is missing. Automatically upgraded to 0.8 to enhance physical constraints.

6. The multi-source heterogeneous method for plant-level monitoring based on liquid air energy storage according to claim 5, characterized in that, Step S4 also includes: S44: Historical Data Replay S441: Use the two-layer configuration screen for root cause localization and fault tracing; S442: Select the fault period; S443: View the second-level configuration screen of the unit that experienced the fault during this period; S434: Locate the time of the first alarm, select the playback speed, and perform frame-by-frame analysis.

7. The multi-source heterogeneous system for plant-level monitoring based on liquid air energy storage as described in claim 6, characterized in that, It includes a cross-system alarm correlation engine, a dynamic weight fusion model, and a visualization and interaction module.

8. The multi-source heterogeneous system for plant-level monitoring based on liquid air energy storage according to claim 7, characterized in that: Real-time acquisition and preprocessing of multi-source heterogeneous data will align millisecond-level and second-level data streams, eliminating timing errors; By using JSON structured conversion, the data format of the six major systems is unified, thus integrating the six systems; Establish a three-flow topology model and construct a network for key equipment; In the plant-level monitoring information system, root cause localization is combined with fault tracing screens, and historical operating condition playback functions are set up to visualize the root cause.

Images