Method and system for mechanical and electrical equipment failure prediction based on BAS system

Abstract

The embodiment of the application provides a kind of based on BAS system's electromechanical equipment failure prediction method and system, first based on the historical operation log of BAS system constructs equipment state benchmark library, then through the real-time monitoring network of BAS system obtains equipment multi-source monitoring data set, covers component sensor data, system control instruction data and environmental feedback data.Then the state alignment verification result indicating matching degree is generated to the state alignment verification processing of multi-source monitoring data set and equipment state benchmark library.Based on the result, failure feature level derivation processing is carried out, and the equipment failure precursor feature chain containing abnormal transmission path is generated.Finally, according to the evolution trend of failure precursor feature chain, equipment failure dynamic risk assessment report is generated, and is synchronized to the operation and maintenance decision module of BAS system, effectively reduces the probability of equipment failure, improves the operation reliability and stability of electromechanical equipment, reduces the economic loss and safety risk caused by equipment failure.

Description

Technical Field

[0001] This application relates to the fields of rail transit environment and equipment monitoring systems and electromechanical equipment maintenance technology, specifically, to a method and system for predicting electromechanical equipment faults based on a BAS system. Background Technology

[0002] In urban rail transit, the Building Automation System (BAS) plays a crucial role, enabling centralized monitoring, management, and control of various electromechanical equipment to ensure the efficient and stable operation of the facilities. As the core component of the BAS, the operational status of the electromechanical equipment directly impacts the performance and safety of the entire system.

[0003] Currently, troubleshooting for electromechanical equipment primarily involves reactive repair and scheduled maintenance. Reactive repair involves fixing equipment after a malfunction occurs, which leads to prolonged downtime, impacting normal facility operation, increasing maintenance costs, and indirect losses due to downtime. Scheduled maintenance involves inspecting and maintaining equipment at predetermined intervals; however, these methods lack specificity and cannot be adjusted based on the actual operating condition of the equipment, potentially resulting in over-maintenance or under-maintenance. Over-maintenance increases unnecessary costs and wastes resources, while under-maintenance fails to detect potential problems in a timely manner, potentially leading to sudden malfunctions.

[0004] While existing fault prediction technologies can provide early warnings of equipment failures to some extent, most of them rely on a single data source or simple parameter threshold judgments. They do not fully consider the complex relationships between equipment under different operating modes and the synergistic effects between multiple data sources, making it difficult to accurately and comprehensively predict equipment failures and failing to meet the high requirements of modern facilities for the reliable operation of electromechanical equipment. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a method and system for predicting mechanical and electrical equipment faults based on a BAS system.

[0006] In conjunction with the first aspect of this application, a method for predicting faults in electromechanical equipment based on a BAS system is provided, applied to a BAS-based electromechanical equipment fault prediction system, the method comprising:

[0007] A device status benchmark library is constructed based on the historical operation logs of the BAS system. The device status benchmark library contains parameter association rules for the normal state of the device under different operating modes and component collaborative behavior patterns.

[0008] The BAS system acquires a multi-source monitoring data set of the equipment through its real-time monitoring network. The multi-source monitoring data set includes component sensor data, system control command data, and environmental feedback data.

[0009] The multi-source monitoring data set and the equipment status benchmark library are subjected to status alignment verification processing to generate a status alignment verification result. The status alignment verification result is used to indicate the degree of matching between the monitoring data and the benchmark library in terms of parameter dimensions, time scale and correlation.

[0010] Based on the state alignment verification results, fault feature hierarchy derivation processing is performed to generate a device fault precursor feature chain. The fault precursor feature chain contains a progressive fault propagation path from parameter anomaly to component anomaly and then to system anomaly.

[0011] A dynamic risk assessment report for equipment failure is generated based on the evolution trend of the fault precursor feature chain, and the dynamic risk assessment report for equipment failure is synchronized to the operation and maintenance decision module of the BAS system.

[0012] In conjunction with the second aspect of this application, a fault prediction system for electromechanical equipment based on a BAS system is provided. The fault prediction system for electromechanical equipment based on a BAS system includes a machine-readable storage medium and a processor. The machine-readable storage medium stores machine-executable instructions. When the processor executes the machine-executable instructions, the fault prediction system for electromechanical equipment based on a BAS system implements the aforementioned fault prediction method for electromechanical equipment based on a BAS system.

[0013] In conjunction with a third aspect of this application, a computer-readable storage medium is provided, wherein computer-executable instructions are stored therein, and when the computer-executable instructions are executed, the aforementioned method for predicting electromechanical equipment faults based on a BAS system is implemented.

[0014] Combining any of the above aspects, a device status benchmark library is constructed based on the historical operation logs of the BAS system. This benchmark library covers parameter association rules and component collaborative behavior patterns for the normal state of the device under different operating modes, fully taking into account the complexity and diversity of device operation. A multi-source monitoring data set of the device is acquired through the BAS system's real-time monitoring network, including component sensor data, system control command data, and environmental feedback data, comprehensively reflecting the device's operating status from multiple dimensions and improving data integrity and reliability.

[0015] A state alignment verification process is performed on the multi-source monitoring data set and the equipment state benchmark library. The generated state alignment verification results accurately indicate the degree of matching between the monitoring data and the benchmark library in terms of parameter dimensions, time scale, and correlation. Based on the state alignment verification results, fault feature hierarchy derivation is performed, and the generated equipment fault precursor feature chain clearly presents the progressive anomaly propagation path from parameter anomaly to component anomaly and then to system anomaly, which helps to deeply understand the generation and development process of faults. According to the evolution trend of the fault precursor feature chain, a dynamic risk assessment report of equipment faults is generated and synchronized to the operation and maintenance decision module of the BAS system. This realizes the organic combination of fault prediction and operation and maintenance decision-making, which can provide operation and maintenance personnel with equipment fault risk information in a timely and accurate manner, enabling them to take targeted maintenance measures in advance, effectively reduce the probability of equipment faults, improve the operational reliability and stability of electromechanical equipment, and reduce economic losses and safety risks caused by equipment faults. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained in conjunction with these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the fault prediction method for electromechanical equipment based on a BAS system provided in this application embodiment. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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.

[0019] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0020] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] Figure 1 This illustration shows a flowchart of a BAS-based electromechanical equipment fault prediction method according to an embodiment of this application. It should be understood that in other embodiments, the order of some steps in the BAS-based electromechanical equipment fault prediction method of this embodiment can be shared based on actual needs, or some steps can be omitted or maintained. The detailed components of this BAS-based electromechanical equipment fault prediction method are as follows:

[0022] This embodiment provides a method for predicting electromechanical equipment faults based on a BAS (Building Automation System). This method is mainly applied to electromechanical equipment in urban rail transit, such as chiller units in central air conditioning systems. The BAS system is used for real-time monitoring and fault prediction to ensure the stable operation of the equipment. The steps of this method will be described in detail below.

[0023] Step S110: Construct a device status benchmark library based on the historical operation logs of the BAS system. The device status benchmark library contains parameter association rules and component collaborative behavior patterns for the normal state of the device under different operating modes.

[0024] In this embodiment, the chiller unit of a central air conditioning system is taken as the object. The BAS system records the historical operating logs of the chiller unit under various operating conditions. First, it is necessary to filter the operating data of the chiller unit during the fault-free period from these historical operating logs. This data will serve as the basis for building the equipment status baseline library. Different operating modes include the chiller unit's start-up mode, stable operation mode, and shutdown mode. For the start-up mode, the operating data may include the compressor's start-up current, voltage, start-up time, and the initial temperatures of components such as the condenser and evaporator; the data for the stable operation mode includes the compressor's operating frequency, condensing temperature, evaporating temperature, cooling water flow rate, and chilled water flow rate; the data for the shutdown mode may involve the compressor's shutdown time and the temperature drop rate of each component.

[0025] Step S111: Filter the operating data subset of the equipment during the fault-free period from the historical operating log of the BAS system. The operating data subset includes complete operating cycle data of the equipment in startup mode, stable operation mode and shutdown mode.

[0026] During the screening process, it is necessary to first determine the criteria for judging whether the chiller unit is fault-free. For example, within a certain period of time, the BAS system should not record any fault alarm information about the chiller unit, and all operating parameters should be within the normal range specified by the equipment manufacturer. Then, operating data within the time period that meets these criteria is extracted from the historical operating logs to form a subset of operating data. For start-up mode, the complete operating cycle refers to the time period from when the compressor starts up to when it reaches a stable operating state; the complete cycle of stable operating mode refers to the time period during which the equipment operates continuously and stably; and the complete cycle of shutdown mode is the time period from when the equipment starts to when it stops operating completely. For example, the complete operating cycle of start-up mode may be 5 minutes, and the current, voltage, temperature, and other data within these 5 minutes will be included in the subset of operating data.

[0027] Step S112: Perform parameter association mining processing on the subset of running data, extract the dependency relationship between different device parameters under the same running mode, and generate parameter association rules. The parameter association rules include the positive association threshold, the negative association threshold, and the independent association range between parameters.

[0028] For the selected subset of operating data, taking stable operation mode as an example, it is necessary to analyze the dependencies between different parameters. For instance, in stable operation, the parameters condensing temperature and cooling water flow rate typically show a positive correlation: an increase in cooling water flow rate usually leads to a decrease in condensing temperature. Analysis of a large amount of historical data can determine the threshold for this positive correlation, which is the range of temperature change when the cooling water flow rate varies within a certain range. For example, if analysis shows that when the cooling water flow rate changes between Q1 and Q2, the condensing temperature changes between T1 and T2, then the positive correlation threshold can be expressed as the temperature range corresponding to that flow rate range. A negative correlation may exist between the compressor's operating frequency and the evaporating temperature; when the compressor's operating frequency increases, the evaporating temperature decreases. Data mining can determine the threshold for this negative correlation. For parameters with weaker correlations, such as condenser pressure and evaporator pressure, they may be in an independent correlation range in stable operation mode, meaning a change in one parameter will not significantly affect the other. Analysis can determine their respective independent variation ranges as the independent correlation ranges.

[0029] Step S113: Analyze the timing relationship of actions between components in the runtime data subset and extract the component collaborative behavior pattern. The component collaborative behavior pattern includes the time delay range of component actions, action sequence constraints, and upper limit of collaborative action frequency.

[0030] The components of a chiller unit include a compressor, condenser fan, expansion valve, and water pump. In startup mode, the actions of these components follow a specific time sequence. For example, during startup, the water pump typically starts first, then the expansion valve opens, followed by the compressor, and finally the condenser fan. By analyzing the action time records in a subset of the operating data, the time delay range between the actions of each component can be determined. For instance, after the water pump starts, the expansion valve opens within a time range of t1 to t2. Action sequence constraints explicitly define the order in which components act, preventing any reversal of the order; for example, the water pump cannot start only after the compressor has started. The upper limit for the frequency of coordinated actions refers to the maximum number of times related components can coordinate their actions per unit time. For example, in stable operation mode, the frequency of coordinated adjustment between the condenser fan and water pump cannot exceed f times per hour to avoid damage caused by excessively frequent component actions.

[0031] Step S114: Standardize and encode the parameter association rules and component collaborative behavior patterns, and store them in the device status benchmark library according to the operation mode. The storage structure of the device status benchmark library includes a mode identifier field, a parameter association field, and a collaborative behavior field.

[0032] Standardized coding converts parameter association rules and component collaborative behavior patterns into a unified data format to facilitate storage and retrieval of the equipment status baseline library. For example, for positive association thresholds in parameter association rules, a specific character combination represents the parameter name, association type, and threshold range, such as "condensing temperature - cooling water flow rate - positive association - T1 - T2 - Q1 - Q2," where T1 and T2 are the ranges of condensing temperatures, and Q1 and Q2 are the ranges of cooling water flow rates. For component collaborative behavior patterns, the coding might be "startup mode - water pump - expansion valve - compressor - condenser fan - t1 - t2 - sequence constraint - upper limit f." The equipment status baseline library is categorized and stored according to startup mode, stable operation mode, shutdown mode, etc. Each entry for a mode includes a mode identifier field, such as "startup mode 001"; a parameter association field, storing the coding of all parameter association rules under that mode; and a collaborative behavior field, storing the coding of component collaborative behavior patterns under that mode.

[0033] Step S120: Obtain a multi-source monitoring data set of the device through the real-time monitoring network of the BAS system. The multi-source monitoring data set includes component sensor data, system control command data, and environmental feedback data.

[0034] The BAS system's real-time monitoring network covers all key components of the chiller unit, enabling the real-time collection of various data related to equipment operation. This data will be aggregated into a multi-source monitoring dataset.

[0035] Step S121: Call the sensor management module of the BAS system to collect real-time operating data of key components of the equipment as component sensor data. The component sensor data includes temperature sensing data, vibration sensing data and pressure sensing data.

[0036] The sensor management module manages the sensors distributed across the key components of the chiller unit. Temperature sensing data includes the compressor's suction and discharge temperatures, the condenser's inlet and outlet water temperatures, and the evaporator's inlet and outlet water temperatures. This data is collected in real time by temperature sensors installed at their respective locations; for example, the temperature sensor at the compressor's suction port collects data every 10 seconds. Vibration sensing data comes from vibration sensors installed on rotating parts such as the compressor and water pump, used to monitor the vibration of these components during operation. The data includes information such as vibration amplitude and frequency. Pressure sensing data includes condenser pressure, evaporator pressure, and compressor suction and discharge pressures, which are obtained in real time by pressure sensors.

[0037] Step S122: Access the control command log module of the BAS system, extract the control command sequence currently received by the device as system control command data, the system control command data includes start command, speed adjustment command and stop command.

[0038] The control command log module records all control commands sent by the BAS system to the chiller unit. Start-up commands are those that control the chiller unit to begin operation, including information such as start-up time and target parameters at startup, for example, "2024-05-20 08:00:00 Start-up target condensing temperature 35℃". Speed ​​control commands are used to adjust the compressor's operating frequency, water pump, and fan speeds, such as "2024-05-20 09:30:00 Speed ​​control compressor frequency increased by 5Hz". Shutdown commands are those that control the chiller unit to stop operation, including shutdown time and shutdown method, for example, "2024-05-20 18:00:00 Shutdown normal shutdown". When extracting the currently received control command sequence, a time range needs to be determined, such as control commands from the past hour, to form system control command data.

[0039] Step S123: Obtain real-time feedback data from the BAS system's environmental monitoring subsystem as environmental feedback data. The environmental feedback data includes temperature and humidity data, airflow velocity data, and dust concentration data of the area where the equipment is located.

[0040] The environmental monitoring subsystem is deployed in the chiller room. It collects temperature and relative humidity data (e.g., temperature 25°C, relative humidity 60%) using temperature and humidity sensors. Airflow velocity data is monitored by airflow sensors, reflecting air movement within the room (e.g., airflow velocity 0.5 m / s). Dust concentration data is acquired by dust sensors to assess air quality within the room (e.g., dust concentration 0.1 mg / m³). This environmental feedback data is transmitted in real-time to the BAS system as part of the overall environmental feedback data.

[0041] Step S124: Perform timestamp synchronization processing on the component sensor data, system control command data, and environmental feedback data to generate a multi-source monitoring data set with time series consistency.

[0042] Because the acquisition and generation times of component sensor data, system control command data, and environmental feedback data may differ, timestamp synchronization processing is required to ensure the accuracy of subsequent analysis. First, a precise timestamp is added to each data item, with millisecond-level accuracy. Then, using a uniform time interval as a benchmark, such as 1 second, data collected within that interval are grouped into data from the same time point. For multiple data points of the same type within the same time interval, averaging or taking the last data point can be used. For example, if a temperature sensor collects three data points within a 1-second time interval: 25℃, 25.1℃, and 25.2℃, the average value of 25.1℃ is taken as the temperature data for that time point. After this processing, the generated multi-source monitoring dataset maintains consistency in the time series, facilitating subsequent state alignment verification and other processing.

[0043] Step S130: Perform state alignment verification processing on the multi-source monitoring data set and the equipment state benchmark library to generate state alignment verification results. The state alignment verification results are used to indicate the degree of matching between the monitoring data and the benchmark library in terms of parameter dimensions, time scale and correlation.

[0044] The status alignment verification process compares the real-time acquired multi-source monitoring data set with the data in the corresponding operating mode in the equipment status benchmark library to determine the matching status of the two in multiple aspects.

[0045] Step S131: Extract the current operating mode identifier from the multi-source monitoring data set, and retrieve the corresponding parameter association rules and component collaborative behavior patterns from the equipment status benchmark library based on the operating mode identifier as the target benchmark subset.

[0046] The multi-source monitoring dataset contains information on the current operating mode of the chiller unit. For example, by analyzing the start command in the system control command data or combining the changing trends of component sensor data, it can be determined that the current operating mode is stable, and the corresponding operating mode identifier, such as "Stable Operating Mode 002," can be extracted. Then, based on this identifier, a search is performed in the equipment status benchmark database to find the corresponding parameter association rules and component collaborative behavior patterns under the stable operating mode. These data are then combined into a target benchmark subset.

[0047] Step S132: Verify the parameter dimensions of the multi-source monitoring data set. Verify that the monitoring data includes all parameter types involved in the parameter association rules in the target benchmark subset, and generate parameter dimension matching tags.

[0048] The parameter association rules in the target baseline subset involve multiple parameter types. For example, in stable operation mode, these might include parameters such as compressor operating frequency, condensing temperature, evaporating temperature, cooling water flow rate, and chilled water flow rate. Validating the parameter dimensions of the multi-source monitoring dataset involves checking whether the dataset contains data for all these parameter types. If all parameter types are present, the generated parameter dimension match is marked as "match"; if any parameter type is missing, it is marked as "mismatch," and the missing parameter type is recorded. For example, if the multi-source monitoring dataset lacks the parameter type of cooling water flow rate, the parameter dimension match is marked as "mismatch, cooling water flow rate missing."

[0049] Step S133: Calculate the time window length of the multi-source monitoring data set and the time deviation between the time window length and the typical operating cycle in the target reference subset, and generate a time scale matching degree value. The time scale matching degree value is calculated by the absolute value of the difference between the time window length and the typical operating cycle.

[0050] First, determine the time window length of the multi-source monitoring data set, which is the time span of the data covered by the set. For example, from 10:00:00 to 10:05:00, the time window length is 300 seconds. The typical operating cycle in the target reference subset refers to the standard time length for the equipment to complete a complete operating process under the corresponding operating mode. For example, the typical operating cycle in stable operating mode is 1800 seconds. Then, calculate the absolute value of the difference between the time window length and the typical operating cycle, i.e., |300-1800| = 1500 seconds. This absolute value of the difference is the time scale matching degree value, which reflects the difference between the time span of the multi-source monitoring data set and the standard cycle.

[0051] Step S134: Analyze the degree of conformity between the actual correlation between parameters in the multi-source monitoring data set and the correlation rules of parameters in the target benchmark subset, and generate the correlation matching rate. The correlation matching rate is calculated by the overlap ratio between the actual correlation and the rule correlation.

[0052] For each parameter association rule in the target benchmark subset, analyze whether the actual association relationship between the corresponding parameters in the multi-source monitoring dataset conforms to the rule. For example, a parameter association rule stipulates that under stable operation mode, when the cooling water flow rate is between Q1 and Q2, the condensing temperature should be between T1 and T2. In the multi-source monitoring dataset, count the number of times the condensing temperature is within the range of T1 to T2 when the cooling water flow rate is within the range of Q1 to Q2, and the total number of times the cooling water flow rate is within this range. The association matching rate is the ratio of the former to the latter. Assuming that in the multi-source monitoring dataset, the total number of times the cooling water flow rate is within the range of Q1 to Q2 is 100, and the number of times the condensing temperature is within the range of T1 to T2 is 80, then the association matching rate of this rule is 0.8. For all parameter association rules in the target benchmark subset, calculate the matching rate in the above manner, and then take the average of these rates as the total association matching rate.

[0053] Step S135: Integrate the parameter dimension matching marker, time scale matching degree value and association relationship matching rate to generate state alignment verification result.

[0054] The parameter dimension matching marker, time scale matching value, and correlation matching rate are integrated according to a certain format to form the state alignment verification result. For example, the state alignment verification result can be expressed as "Parameter dimension matching marker: mismatch, missing cooling water flow; Time scale matching value: 1500 seconds; Correlation matching rate: 0.75". The above results can clearly reflect the matching status between the multi-source monitoring data set and the target baseline subset in terms of parameter dimension, time scale, and correlation.

[0055] Step S140: Based on the state alignment verification results, perform fault feature hierarchy derivation processing to generate a device fault precursor feature chain. The fault precursor feature chain contains a progressive fault propagation path from parameter anomaly to component anomaly and then to system anomaly.

[0056] Based on the various mismatches reflected in the state alignment verification results, the possible fault characteristics are deduced step by step, and a fault precursor feature chain is constructed according to the hierarchy of parameter anomalies, component anomalies, and system anomalies to clarify the transmission path and development trend of anomalies.

[0057] Step S141: When the parameter dimension matching is marked as mismatch in the state alignment verification result, locate the missing parameter type and mark it as a parameter-level outlier.

[0058] If the parameter dimension matching is marked as mismatch, it means that the multi-source monitoring data set is missing certain parameter types from the target baseline subset, and these missing parameter types need to be located and marked.

[0059] Step S1411: Extract the parameter dimension matching markers from the state alignment verification results and determine whether there are any missing parameter types.

[0060] The specific content of the parameter dimension matching mark can be obtained from the state alignment verification result, such as "mismatch, missing cooling water flow". This can be used to determine that there is a missing parameter type, namely cooling water flow.

[0061] Step S1412: If there is a missing parameter type, trace the sensor status of the missing parameter from the original acquisition records of the multi-source monitoring data set, and determine the sensor failure or data transmission interruption as the cause of the missing parameter.

[0062] Review the raw acquisition records of the multi-source monitoring dataset to understand the acquisition status of the cooling water flow rate parameter. If the raw records show that the sensor for this parameter has never returned data, or the returned data is consistently invalid, it may be due to a sensor malfunction. If normal data was previously returned, but suddenly stopped after a certain point in time, and the acquisition of other parameters is normal, it may be due to a data transmission interruption, such as a faulty transmission line or a problem with the data transmission module.

[0063] Step S1413: Generate a parameter missing type identifier based on the missing reason. The parameter missing type identifier includes sensor fault type and data transmission interruption type.

[0064] If the cause of the missing parameter is determined to be a sensor malfunction, the parameter missing type identifier can be "Sensor malfunction - Cooling water flow sensor"; if it is a data transmission interruption, the identifier is "Data transmission interruption - Cooling water flow data transmission line".

[0065] Step S1414: Associate the missing parameter type, missing reason, and missing parameter type identifier to generate a parameter-level anomaly. The parameter-level anomaly includes a parameter name field, a missing reason field, and a missing type field.

[0066] Associate the cooling water flow rate (parameter name field), sensor fault (missing reason field), and "sensor fault - cooling water flow sensor" (missing type field) to form parameter-level anomalies, such as "parameter name: cooling water flow rate; missing reason: sensor fault; missing type: sensor fault - cooling water flow sensor".

[0067] Step S142: When the time scale matching degree value in the state alignment verification result exceeds the preset deviation threshold, analyze the non-periodic fluctuation characteristics of the parameters within the monitoring data time window and mark them as component-level anomalies.

[0068] When the time scale matching value exceeds the preset deviation threshold, it indicates that there is a significant difference between the time window of the multi-source monitoring data set and the typical operating cycle of the target benchmark subset. It is necessary to analyze the fluctuation of parameters within this time window to determine whether there are component-level anomalies.

[0069] Step S1421: Extract the time scale matching degree value from the state alignment verification result and determine whether it exceeds the preset deviation threshold.

[0070] The preset deviation threshold is set based on the characteristics of the equipment and actual operating experience. For example, in stable operating mode, the preset deviation threshold is 200 seconds. If the time scale matching value is 300 seconds, exceeding the preset deviation threshold, further analysis is required.

[0071] Step S1422: If the preset deviation threshold is exceeded, perform fluctuation frequency analysis on the parameters within the time window of the multi-source monitoring data set and calculate the number of fluctuations of the parameters per unit time.

[0072] In this embodiment, the time window is 300 seconds, and parameters such as compressor operating frequency, condensing temperature, and evaporating temperature are selected for fluctuation frequency analysis. For the compressor operating frequency, the number of times its value changes within these 300 seconds is recorded. Assuming a total of 15 changes, the fluctuation frequency per unit time (1 second) is 15 divided by 300, which gives 0.05 fluctuations per second. Similarly, similar calculations are performed for the condensing temperature and evaporating temperature to obtain their respective fluctuation frequencies. For example, the fluctuation frequency of the condensing temperature is 0.03 fluctuations per second, and the fluctuation frequency of the evaporating temperature is 0.04 fluctuations per second.

[0073] Step S1423: Identify the periodic characteristics of parameter fluctuations and generate periodic fluctuation identifiers and non-periodic fluctuation identifiers.

[0074] By analyzing the fluctuations of various parameters within a time window, it can be determined whether they exhibit periodicity. For example, under normal and stable operation, the compressor's operating frequency may fluctuate regularly according to a certain period, such as adjusting the frequency every 30 seconds. In this case, a periodic fluctuation label is generated, such as "Compressor Operating Frequency - Periodic Fluctuation". However, if the fluctuation of the condensing temperature does not show a clear pattern, sometimes rising rapidly and sometimes falling slowly, without a fixed period, a non-periodic fluctuation label is generated, such as "Condensing Temperature - Non-periodic Fluctuation".

[0075] Step S1424: Perform amplitude analysis on the parameters of non-periodic fluctuations, and calculate the difference between the fluctuation amplitude and the reference amplitude as the abnormal amplitude.

[0076] For parameters marked as non-periodic fluctuations, such as condensing temperature, it is necessary to determine their fluctuation amplitude. First, find the maximum and minimum values ​​of this parameter within a time window; the difference between the two is the fluctuation amplitude. Assuming the maximum value of the condensing temperature within the time window is 40℃ and the minimum is 30℃, then the fluctuation amplitude is 10℃. The baseline amplitude is the fluctuation amplitude of this parameter under normal stable operating conditions, obtained from the target baseline subset of the equipment status baseline library; for example, the baseline amplitude is 5℃. Then, the abnormal amplitude is 10℃ minus 5℃, resulting in 5℃.

[0077] Step S1425: Associate and record the parameter name, number of fluctuations and abnormal amplitude of non-periodic fluctuations to generate component-level anomaly points. The component-level anomaly points include parameter name field, number of fluctuations field and abnormal amplitude field.

[0078] The condensation temperature (parameter name field), fluctuation frequency (0.03 times per second field), and abnormal amplitude (5℃ field) are linked and recorded to form a component-level anomaly point, for example, "Parameter name: condensation temperature; fluctuation frequency: 0.03 times per second; abnormal amplitude: 5℃". This indicates that the component containing the condensation temperature parameter may have an anomaly because its fluctuations are non-periodic and the abnormal amplitude is large.

[0079] Step S143: When the correlation matching rate in the state alignment verification result is lower than the preset matching threshold, identify the transmission direction and influence range of abnormal correlation between parameters and mark it as a system-level anomaly.

[0080] If the correlation matching rate is lower than the preset matching threshold, it indicates that there is a significant difference between the actual correlation between parameters in the multi-source monitoring data set and the parameter correlation rules in the target benchmark subset, and further analysis is needed to determine system-level anomalies.

[0081] Step S1431: Extract the correlation matching rate from the state alignment verification result and determine whether it is lower than the preset matching threshold.

[0082] The preset matching threshold is set based on the correlation relationship during normal device operation. For example, in stable operation mode, the preset matching threshold is 0.8. If the correlation matching rate in the state alignment verification result is 0.75, which is lower than the preset threshold, further analysis and processing are required.

[0083] Step S1432: If the value is lower than the preset matching threshold, perform graph model construction processing on the actual correlation between parameters in the multi-source monitoring data set to generate a parameter correlation graph. The nodes of the parameter correlation graph represent parameter types, and the edges represent the correlation strength between parameters.

[0084] Parameters such as compressor operating frequency, condensing temperature, cooling water flow rate, evaporating temperature, and chilled water flow rate are selected as nodes to analyze their actual correlations. For example, changes in compressor operating frequency affect both condensing and evaporating temperatures, changes in cooling water flow rate affect condensing temperature, and changes in chilled water flow rate affect evaporating temperature. The strength of the correlation is represented by the thickness of the edges; the stronger the correlation, the thicker the edge. Assuming the correlation strength between compressor operating frequency and condensing temperature is 0.8, between compressor operating frequency and evaporating temperature is 0.7, between cooling water flow rate and condensing temperature is 0.6, and between chilled water flow rate and evaporating temperature is 0.5, these relationships are represented using a graphical model to generate a parameter correlation graph.

[0085] Step S1433: Extract a standard parameter correlation graph from the target reference subset of the device status reference library, wherein the standard parameter correlation graph contains the standard correlation strength between parameters.

[0086] A standard parameter correlation graph under stable operating conditions is obtained from the target reference subset. The nodes in this graph are parameters such as compressor operating frequency, condensing temperature, cooling water flow rate, evaporating temperature, and chilled water flow rate. The thickness of the edges represents the standard correlation strength. For example, the standard correlation strength between compressor operating frequency and condensing temperature is 0.9, between compressor operating frequency and evaporating temperature is 0.8, between cooling water flow rate and condensing temperature is 0.7, and between chilled water flow rate and evaporating temperature is 0.6.

[0087] Step S1434: Calculate the edge difference set between the parameter association graph and the standard parameter association graph, wherein the edge difference set includes edges with enhanced association strength and edges with weakened association strength.

[0088] The correlation strength of corresponding edges in the parameter correlation graph and the standard parameter correlation graph is compared one by one. For the edge between compressor operating frequency and condensing temperature, the correlation strength in the parameter correlation graph is 0.8, while it is 0.9 in the standard correlation graph, a difference of -0.1, indicating a weakening correlation strength. Similarly, the edge between compressor operating frequency and evaporating temperature has a correlation strength of 0.7 in the parameter correlation graph and 0.8 in the standard correlation graph, also a difference of -0.1, indicating a weakening correlation strength. Likewise, the edge between cooling water flow rate and condensing temperature has a correlation strength of 0.6 in the parameter correlation graph and 0.7 in the standard correlation graph, a difference of -0.1, indicating a weakening correlation strength. Finally, the edge between chilled water flow rate and evaporating temperature has a correlation strength of 0.5 in the parameter correlation graph and 0.6 in the standard correlation graph, a difference of -0.1, indicating a weakening correlation strength. These edges with varying correlation strengths are collected to form an edge difference set, for example, "edges with weakening correlation strength: compressor operating frequency - condensing temperature, compressor operating frequency - evaporating temperature, cooling water flow rate - condensing temperature, chilled water flow rate - evaporating temperature".

[0089] Step S1435: Analyze the propagation path of the change in association strength in the edge difference set, and determine the starting parameters and the range of affected parameters of abnormal association.

[0090] The edge difference set reveals a weakening of the correlation strength across multiple edges. Further analysis of these edge relationships shows that the compressor operating frequency is the central parameter correlated with multiple parameters, and its weakened correlation with condensing and evaporating temperatures may be the starting point of the anomalous correlation. The affected parameters include condensing temperature, evaporating temperature, cooling water flow rate, and chilled water flow rate, as these parameters are correlated with the compressor operating frequency or other affected parameters. For example, the weakened correlation between the compressor operating frequency and condensing temperature may affect the regulating effect of cooling water flow rate on condensing temperature, thus impacting the parameter correlations of the entire system.

[0091] Step S1436: Associate and record the starting parameters, transmission paths and affected parameter ranges associated with the anomaly to generate a system-level anomaly point. The system-level anomaly point includes a starting parameter field, a transmission path field and an affected range field.

[0092] The system associates and records the compressor operating frequency (starting parameter field), "compressor operating frequency → condensing temperature; compressor operating frequency → evaporating temperature; cooling water flow rate → condensing temperature; chilled water flow rate → evaporating temperature" (transmission path field), and "condensing temperature, evaporating temperature, cooling water flow rate, chilled water flow rate" (affected range field) to generate system-level anomalies. For example, "starting parameter: compressor operating frequency; transmission path: compressor operating frequency → condensing temperature; compressor operating frequency → evaporating temperature; cooling water flow rate → condensing temperature; chilled water flow rate → evaporating temperature; affected range: condensing temperature, evaporating temperature, cooling water flow rate, chilled water flow rate".

[0093] Step S144: Perform time-series correlation analysis on the parameter-level anomalies, component-level anomalies, and system-level anomalies to determine the time sequence and impact intensity of the anomalies propagating from the parameter level to the component level and then to the system level.

[0094] The occurrence times of parameter-level, component-level, and system-level anomalies were collected, and time-series correlation analysis was performed. For example, a parameter-level anomaly (missing cooling water flow) appeared at 10:00:00, a component-level anomaly (non-periodic fluctuation in condensing temperature) appeared at 10:02:00, and a system-level anomaly (compressor operating frequency correlation anomaly) appeared at 10:05:00. This determined the time sequence of anomaly propagation: parameter-level anomalies appeared first, followed by component-level anomalies, and finally system-level anomalies. Simultaneously, the influence strength between each anomaly was analyzed. The influence strength of parameter-level anomalies on component-level anomalies can be measured by the correlation between them, for example, set to 0.6; the influence strength of component-level anomalies on system-level anomalies is set to 0.7. These influence strengths were derived from the analysis of similar anomaly propagation patterns in historical data.

[0095] Step S145: Construct a fault precursor feature chain based on the time sequence and influence intensity. The fault precursor feature chain includes anomaly point type, anomaly propagation time interval, and anomaly influence intensity parameters.

[0096] Based on the time sequence and impact intensity, a fault precursor characteristic chain is constructed. The anomaly types are categorized as parameter-level anomalies, component-level anomalies, and system-level anomalies, in that order. The anomaly propagation time interval refers to the time difference between the occurrence of one anomaly and the occurrence of the next; the time interval from a parameter-level anomaly to a component-level anomaly is 2 minutes, and the time interval from a component-level anomaly to a system-level anomaly is 3 minutes. The anomaly impact intensity parameters are 0.6 and 0.7, as determined in step S144. Therefore, the fault precursor characteristic chain can be represented as: "parameter-level anomaly → (time interval 2 minutes, impact intensity 0.6) → component-level anomaly → (time interval 3 minutes, impact intensity 0.7) → system-level anomaly".

[0097] Step S150: Generate a dynamic risk assessment report for equipment failure based on the evolution trend of the fault precursor feature chain, and synchronize the dynamic risk assessment report for equipment failure to the operation and maintenance decision module of the BAS system.

[0098] By analyzing the evolution trend of the fault precursor characteristic chain, the risk of equipment failure is assessed, and corresponding reports are generated and promptly transmitted to the operation and maintenance decision-making module so that appropriate measures can be taken.

[0099] Step S151: Extract the abnormal propagation time interval and abnormal influence intensity parameters from the fault precursor feature chain.

[0100] The anomaly propagation time interval is extracted from the fault precursor feature chain, namely 2 minutes from the parameter-level anomaly point to the component-level anomaly point and 3 minutes from the component-level anomaly point to the system-level anomaly point; the anomaly impact intensity parameters are 0.6 and 0.7.

[0101] Step S152: Based on the anomaly propagation time interval, predict the time required for the anomaly to propagate from the current level to the next level, and generate an anomaly propagation time prediction value.

[0102] Based on the evolution of similar fault precursor characteristic chains in historical data, and combined with the current anomaly propagation time interval, the time required for an anomaly to propagate from the system-level anomaly point to the next level (i.e., equipment failure) is predicted. Assume that the analysis shows the predicted anomaly propagation time is 5 minutes under the current circumstances.

[0103] Step S153: Based on the anomaly impact intensity parameter, calculate the degree of impact of the anomaly on the equipment function and generate a function impact degree value.

[0104] The abnormal impact intensity parameter is calculated comprehensively. For example, by using a weighted average method, 0.6 and 0.7 are assigned the same weight, and the result is (0.6+0.7)÷2=0.65. This value is the functional impact value, which reflects the degree of impact of the abnormality on the function of the chiller unit.

[0105] Step S154: Combine the predicted value of the anomaly transmission time and the value of the degree of functional impact to construct a fault risk assessment model. The fault risk assessment model includes a risk level calculation function and a risk evolution curve.

[0106] By processing the predicted values ​​of anomaly propagation time and functional impact, a model capable of assessing equipment failure risk is constructed.

[0107] For example, step S1541: normalize the predicted value of the abnormal transmission time to generate a time risk factor in the range of 0 to 1.

[0108] The predicted abnormal transmission time is 5 minutes, with the longest possible transmission time set at 10 minutes and the shortest at 0 minutes. The normalization method is to use (maximum time - predicted abnormal transmission time) ÷ (maximum time - shortest time), i.e., (10-5) ÷ (10-0) = 0.5, resulting in a time risk factor of 0.5.

[0109] Step S1542: Normalize the functional impact value to generate a functional risk factor in the range of 0 to 1.

[0110] The functional impact value is 0.65, which is already in the range of 0 to 1, so it can be directly used as the functional risk factor, i.e., 0.65.

[0111] Step S1543: Perform a weighted summation of the time risk factor and the functional risk factor to generate a comprehensive risk index. The weight coefficients of the weighted summation are dynamically adjusted according to the equipment type.

[0112] For the aforementioned critical equipment in the chiller unit, the weighting coefficient for the time risk factor is set to 0.6, and the weighting coefficient for the functional risk factor is set to 0.4. The comprehensive risk index = time risk factor × 0.6 + functional risk factor × 0.4 = 0.5 × 0.6 + 0.65 × 0.4 = 0.3 + 0.26 = 0.56.

[0113] Step S1544: Divide the risk level according to the numerical range of the comprehensive risk index, and the risk level includes low risk level, medium risk level and high risk level.

[0114] The overall risk index is defined as follows: 0 to 0.3 is low risk, 0.3 to 0.7 is medium risk, and 0.7 to 1 is high risk. Since the current overall risk index is 0.56, it falls within the medium risk category.

[0115] Step S1545: Based on the historical evolution data of the fault precursor feature chain, fit the trend curve of risk changing over time to generate a risk evolution curve, wherein the risk evolution curve includes the rate of change parameter of the risk index over time.

[0116] Historical evolution data of precursory characteristic chains of similar past failures were collected, including comprehensive risk indices at different time points. For example, in a similar past event, the comprehensive risk index was 0.3 at time point t1, 0.4 at t2, 0.5 at t3, and 0.6 at t4. Based on this data, a trend curve of risk changing over time was fitted, showing an upward trend. By calculating the ratio of the difference in risk index between adjacent time points to the time difference, the rate of change parameter of the risk index over time was obtained; for example, the calculated rate of change parameter was 0.02 per minute.

[0117] Step S1546: Combine the risk level classification rules and risk evolution curves to generate a fault risk assessment model that includes a risk level calculation function and a risk evolution curve.

[0118] By combining the risk level classification rules (i.e., the correspondence between the comprehensive risk index and the risk level) and the risk evolution curve (including a change rate parameter of 0.02 per minute), a fault risk assessment model is formed. This model can calculate the comprehensive risk index, determine the risk level, and predict future risk changes based on the input anomaly transmission time prediction value and functional impact degree value, and predict future risk changes through the risk evolution curve.

[0119] Step S155: Calculate the current fault risk level of the equipment and the risk evolution trend in the future time period using the fault risk assessment model, and generate a dynamic fault risk assessment report for the equipment.

[0120] The predicted anomaly propagation time (5 minutes) and functional impact value (0.65) are input into the fault risk assessment model, calculating the current fault risk level as medium risk. Based on the risk evolution curve and the rate of change parameter (0.02 per minute), the risk evolution trend over the next 10 minutes is predicted. For example, the comprehensive risk index is expected to reach 0.66 in 5 minutes and 0.76 in 10 minutes, entering the high risk level. This information is then compiled into a dynamic equipment fault risk assessment report, which includes the current risk level, predicted anomaly propagation time, functional impact value, and predicted risk evolution trend.

[0121] Step S156: Call the interface protocol of the BAS system to transmit the dynamic risk assessment report of equipment failure to the storage unit and display unit of the operation and maintenance decision module.

[0122] The dynamic risk assessment report for equipment failure is sent to the operation and maintenance decision module via the BAS system's interface protocol, such as the BACnet protocol. The storage unit of the operation and maintenance decision module stores the report for later query and analysis; the display unit shows the main contents of the report, such as the current risk level and future risk trends, to the operation and maintenance personnel, enabling them to understand the equipment failure risk situation in a timely manner and take corresponding maintenance measures.

[0123] In the above embodiments, the electromechanical equipment fault prediction system based on the BAS system for performing the above method embodiments has at least one processor, a control module (chipset) coupled to at least one of the processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load to / output device coupled to the control module, and a network interface coupled to the control module.

[0124] The processor may include at least one single-core or multi-core processor, and may include any combination of general-purpose processors or special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). For some alternative implementations, the BAS-based electromechanical equipment fault prediction system can serve as the gateway or other electronic device described in the embodiments of this application.

[0125] In some alternative implementations, a BAS-based electromechanical equipment fault prediction system may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor fused with the at least one computer-readable medium and configured to execute the instructions to implement the module thereby performing the actions described in this disclosure.

[0126] In one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processors and / or any suitable device or component communicating with the control module.

[0127] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0128] The memory can be used, for example, to load and store data and / or instructions for a BAS-based electromechanical equipment fault prediction system. In one embodiment, the memory may include any suitable volatile memory, such as suitable DRAM.

[0129] In one embodiment, the control module may include at least one load-to-output controller to provide an interface to the NVM / storage device and (at least one) load-to-output device.

[0130] For example, an NVM / storage device can be used to store data and / or instructions. An NVM / storage device may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one optical disc (CD) drive, and / or at least one digital universal optical disc (DVD) drive).

[0131] NVM / storage devices may include storage resources that are physically part of a device on which an electromechanical fault prediction system based on a BAS system is installed, or that can be accessed by the device without being part of the device. For example, an NVM / storage device may be accessed over a network via at least one load-to-output device.

[0132] At least one load-to-output device can provide an interface for the BAS-based electromechanical equipment fault prediction system to communicate with any other suitable device. The load-to-output device may include communication components, pinyin components, sensor components, etc. A network interface can provide an interface for the BAS-based electromechanical equipment fault prediction system to communicate over at least one network. The BAS-based electromechanical equipment fault prediction system can wirelessly communicate with at least one component of a wireless network based on at least one wireless network prior and / or protocol, such as accessing a wireless network based on communication priors.

[0133] In one embodiment, at least one of the processors may be integrated with the logic of at least one controller of the control module (e.g., a memory controller module). In one embodiment, at least one of the processors may be integrated with the logic of at least one controller of the control module to form a system-level integration. In one embodiment, at least one of the processors may be fused with the logic of at least one controller of the control module on the same die. In one embodiment, at least one of the processors may be fused with the logic of at least one controller of the control module on the same die to form a system-on-a-chip (SoC).

[0134] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0135] This invention discloses a computer read storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to execute the steps in the BAS-based electromechanical equipment fault prediction method described in the foregoing embodiments.

[0136] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the BAS-based electromechanical equipment fault prediction method described in the foregoing embodiments.

[0137] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0138] Finally, it should be noted that the above-disclosed embodiments are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit them. 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.

Claims

1. A method for predicting faults in electromechanical equipment based on a BAS system, characterized in that, The method includes: A device status benchmark library is constructed based on the historical operation logs of the BAS system. The device status benchmark library contains parameter association rules for the normal state of the device under different operating modes and component collaborative behavior patterns. The BAS system acquires a multi-source monitoring data set of the equipment through its real-time monitoring network. The multi-source monitoring data set includes component sensor data, system control command data, and environmental feedback data. The multi-source monitoring data set and the equipment status benchmark library are subjected to status alignment verification processing to generate a status alignment verification result. The status alignment verification result is used to indicate the degree of matching between the monitoring data and the benchmark library in terms of parameter dimensions, time scale and correlation. Based on the state alignment verification results, fault feature hierarchy derivation processing is performed to generate a device fault precursor feature chain. The fault precursor feature chain contains a progressive fault propagation path from parameter anomaly to component anomaly and then to system anomaly. A dynamic risk assessment report for equipment failure is generated based on the evolution trend of the failure precursor feature chain, and the dynamic risk assessment report for equipment failure is synchronized to the operation and maintenance decision module of the BAS system. The step of performing state alignment verification processing on the multi-source monitoring data set and the equipment state benchmark library to generate state alignment verification results includes: Extract the current operating mode identifier from the multi-source monitoring data set, and retrieve the corresponding parameter association rules and component collaborative behavior patterns from the equipment status benchmark library based on the operating mode identifier as the target benchmark subset; The parameter dimensions of the multi-source monitoring data set are verified to include all parameter types involved in the parameter association rules in the target benchmark subset, and parameter dimension matching tags are generated. Calculate the time window length of the multi-source monitoring data set and the time deviation between the time window length and the typical operating cycle in the target reference subset, and generate a time scale matching degree value. The time scale matching degree value is calculated by the absolute value of the difference between the time window length and the typical operating cycle. The degree of conformity between the actual correlation between parameters in the multi-source monitoring data set and the correlation rules of parameters in the target benchmark subset is analyzed to generate a correlation matching rate. The correlation matching rate is calculated by the overlap ratio between the actual correlation and the rule correlation. The parameter dimension matching markers, time scale matching scores, and association relationship matching rates are integrated and processed to generate state alignment verification results; The step of performing fault feature hierarchy derivation processing based on the state alignment verification results to generate a device fault precursor feature chain includes: When the parameter dimension matching is marked as mismatch in the state alignment verification result, the missing parameter type is located and marked as a parameter-level outlier. When the time scale matching degree value in the state alignment verification result exceeds the preset deviation threshold, the non-periodic fluctuation characteristics of the parameters within the monitoring data time window are analyzed and marked as component-level anomalies. When the correlation matching rate in the state alignment verification result is lower than the preset matching threshold, the transmission direction and influence range of abnormal correlation between parameters are identified and marked as system-level anomalies. Time-series correlation analysis is performed on the parameter-level anomalies, component-level anomalies, and system-level anomalies to determine the time sequence and impact intensity of anomalies propagating from the parameter level to the component level and then to the system level. Based on the time sequence and the intensity of influence, a fault precursor feature chain is constructed. The fault precursor feature chain includes parameters such as the anomaly type, the anomaly propagation time interval, and the anomaly influence intensity.

2. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, The device status baseline library built based on the historical operation logs of the BAS system includes: Filter the operation data subset of the equipment during the fault-free period from the historical operation log of the BAS system. The operation data subset includes complete operation cycle data of the equipment in start-up mode, stable operation mode and shutdown mode. The subset of operational data is subjected to parameter association mining processing to extract the dependency relationships between parameters of different devices under the same operational mode and generate parameter association rules. The parameter association rules include the positive association threshold, the negative association threshold, and the independent association range between parameters. Analyze the timing relationship of actions between components in the runtime data subset, and extract the component collaborative behavior pattern. The component collaborative behavior pattern includes the time delay range of component actions, action sequence constraints, and upper limit of collaborative action frequency. The parameter association rules and component collaborative behavior patterns are standardized and encoded, and stored in the device status benchmark library according to the operation mode. The storage structure of the device status benchmark library includes a mode identifier field, a parameter association field, and a collaborative behavior field.

3. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, The acquisition of multi-source monitoring data sets of equipment through the real-time monitoring network of the BAS system includes: The sensor management module of the BAS system is invoked to collect real-time operating data of key components of the equipment as component sensor data. The component sensor data includes temperature sensing data, vibration sensing data and pressure sensing data. Access the control command log module of the BAS system, extract the sequence of control commands currently received by the device as system control command data, which includes start command, speed adjustment command and stop command; The real-time feedback data of the BAS system's environmental monitoring subsystem is obtained as environmental feedback data, which includes temperature and humidity data, airflow velocity data, and dust concentration data of the area where the equipment is located. The sensor data, system control command data, and environmental feedback data of the aforementioned components are time-stamped and synchronized to generate a multi-source monitoring data set with time series consistency.

4. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, When the parameter dimension matching in the state alignment verification result is marked as mismatched, the missing parameter type is located and marked as a parameter-level outlier, including: Extract parameter dimension matching markers from the state alignment verification results to determine if there are any missing parameter types; If there is a missing parameter type, trace the sensor status of the missing parameter from the original acquisition records of the multi-source monitoring data set to determine the sensor failure or data transmission interruption as the cause of the missing parameter. A parameter missing type identifier is generated based on the cause of the missing information. The parameter missing type identifier includes sensor fault type and data transmission interruption type. The missing parameter type, missing reason, and missing parameter type identifier are associated and recorded to generate a parameter-level anomaly. The parameter-level anomaly includes a parameter name field, a missing reason field, and a missing type field.

5. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, When the time-scale matching degree value in the state alignment verification result exceeds a preset deviation threshold, the non-periodic fluctuation characteristics of the parameters within the monitoring data time window are analyzed and marked as component-level anomalies, including: Extract the time-scale matching degree value from the state alignment verification result and determine whether it exceeds a preset deviation threshold. If the preset deviation threshold is exceeded, the parameters within the time window of the multi-source monitoring data set are subjected to fluctuation frequency analysis processing to calculate the number of fluctuations of the parameters per unit time. Identify the periodic characteristics of parameter fluctuations and generate periodic fluctuation identifiers and non-periodic fluctuation identifiers; Amplitude analysis is performed on the parameters of non-periodic fluctuations, and the difference between the fluctuation amplitude and the benchmark amplitude is calculated as the abnormal amplitude. The parameter name, number of fluctuations, and abnormal amplitude of non-periodic fluctuations are associated and recorded to generate component-level anomaly points. The component-level anomaly points include parameter name field, number of fluctuations field, and abnormal amplitude field.

6. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, When the correlation matching rate in the state alignment verification result is lower than a preset matching threshold, the transmission direction and impact range of abnormal correlations between parameters are identified and marked as system-level anomalies, including: Extract the correlation matching rate from the state alignment verification results and determine whether it is lower than a preset matching threshold; If the value is below a preset matching threshold, a graph model is constructed to analyze the actual correlation between parameters in the multi-source monitoring data set, generating a parameter correlation graph. The nodes in the parameter correlation graph represent parameter types, and the edges represent the correlation strength between parameters. A standard parameter correlation graph is extracted from the target reference subset of the equipment status reference library, and the standard parameter correlation graph contains the standard correlation strength between parameters. Calculate the set of edge differences between the parameter association graph and the standard parameter association graph, wherein the set of edge differences includes edges with increased association strength and edges with decreased association strength; Analyze the propagation path of the change in association strength in the edge difference set to determine the starting parameters and the range of affected parameters of abnormal associations; The system-level anomaly point is generated by associating and recording the starting parameters, transmission path, and affected parameter range associated with the anomaly. The system-level anomaly point includes the starting parameter field, transmission path field, and affected range field.

7. The method for predicting electromechanical equipment faults based on a BAS system according to claim 1, characterized in that, The step of generating a dynamic risk assessment report for equipment failure based on the evolution trend of the failure precursor feature chain, and synchronizing the dynamic risk assessment report for equipment failure to the operation and maintenance decision module of the BAS system, includes: Extract the anomaly propagation time interval and anomaly influence intensity parameters from the fault precursor feature chain; Based on the anomaly propagation time interval, predict the time required for an anomaly to propagate from the current level to the next level, and generate an anomaly propagation time prediction value. Based on the aforementioned anomaly impact intensity parameter, the degree of impact of the anomaly on the equipment function is calculated, and a functional impact degree value is generated; By combining the predicted value of the anomaly transmission time and the value of the functional impact, a fault risk assessment model is constructed. The fault risk assessment model includes a risk level calculation function and a risk evolution curve. The fault risk assessment model is used to calculate the current fault risk level of the equipment and the risk evolution trend in the future time period, and to generate a dynamic fault risk assessment report of the equipment. The BAS system's interface protocol is invoked to transmit the device fault dynamic risk assessment report to the storage and display units of the operation and maintenance decision module.

8. A fault prediction system for electromechanical equipment based on a BAS system, characterized in that, The system includes a processor and a computer-readable storage medium storing machine-executable instructions, which, when executed by a computer, implement the electromechanical equipment fault prediction method based on any one of claims 1-7.

Images