Rail transit energy efficiency balance dynamic monitoring method and system

Abstract

The invention relates to a rail transit energy efficiency balance dynamic monitoring method and system, relates to the field of rail transit energy conservation and monitoring, and solves the problems that in an energy consumption monitoring link, simple statistical analysis and a rule-based system are difficult to adapt to a complex and changeable operating environment due to fixed threshold values and rules, and misjudgment and missed judgment are easy to occur. The method comprises the following steps: according to a comparison result of actual energy consumption of equipment and a personalized energy efficiency standard threshold value, determining whether the corresponding equipment has energy consumption abnormity or not; if yes, the personalized threshold value and historical energy consumption data of the equipment are combined, the built comprehensive energy loss analysis model based on the fault tree and the Bayesian network is utilized, the fault mode of the equipment is started, the operation history, the maintenance record and the environmental factors of the equipment are combined, and Bayesian network reasoning and dynamic fault tree analysis are carried out; and the source reason of the energy loss is positioned and sent to the terminal held by the person in charge. The method has the following effect that the energy efficiency management level of rail transit is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of rail transit energy conservation and monitoring, and in particular to a rail transit energy efficiency balance dynamic monitoring method and system. Background Art

[0002] At present, a variety of technologies have been developed in the field of rail transit energy efficiency monitoring. Some technologies collect data in real time through sensors installed on the equipment, and use simple statistical analysis methods to determine whether the energy consumption is within the normal range. Others use rule-based systems to identify abnormal energy consumption based on pre-set energy consumption standards and equipment operating parameters. In addition, in the field of fault diagnosis and energy loss analysis, there are technologies that use expert systems to build a knowledge base based on the experience and knowledge of experts in the field to infer and judge possible causes of energy loss. In the formulation of solutions, general improvement suggestions are usually given based on the experience of handling similar cases in the past.

[0003] However, existing technologies have obvious shortcomings. In the energy consumption monitoring process, simple statistical analysis and rule-based systems are difficult to adapt to complex and changing operating environments due to fixed thresholds and rules, and are prone to misjudgment and omissions. When analyzing the root causes of energy loss, expert systems rely on expert experience, are highly subjective, and lack the fusion of multi-source data, making it difficult to accurately locate problems. In terms of solution generation, suggestions based on past cases lack specificity and cannot meet the unique needs of the project, resulting in poor implementation results. Summary of the invention

[0004] In order to effectively improve the level of rail transit energy efficiency management, the present application provides a rail transit energy efficiency balance dynamic monitoring method and system.

[0005] In the first aspect, the present application provides a method for dynamic monitoring of energy efficiency balance of rail transit, which adopts the following technical solution: A rail transit energy efficiency balance dynamic monitoring method, comprising: Get relevant data of various external devices; Input relevant data of various external devices into the pre-built fuzzy clustering classification model, and output the classification results of various external devices; Based on the equipment classification results, the life cycle cost analysis method is used to determine the personalized energy efficiency standard thresholds for each type of equipment; According to the comparison result between the actual energy consumption of the equipment and the personalized energy efficiency standard threshold, determine whether the corresponding equipment has energy consumption abnormality; If yes, then the system combines the personalized threshold of the equipment and the historical energy consumption data, uses the built comprehensive energy loss analysis model based on fault tree and Bayesian network, starts from the equipment failure mode, combines the equipment operation history, maintenance records, and environmental factors, and locates the root cause of energy loss through Bayesian network reasoning and dynamic fault tree analysis, and sends it to the terminal held by the person in charge; If no, no notification will be made.

[0006] Optionally, life cycle cost analysis methods are used to determine individual energy efficiency standard thresholds for various types of equipment, including: Analyze whether there is any change in device status based on real-time relevant data of various external devices; If not, maintain the original setting, which is to use the life cycle cost analysis method to determine the personalized energy efficiency standard thresholds for various types of equipment; If yes, the device whose status has changed is taken as the target device, and various information of the target device is collected, wherein the various information of the target device includes real-time operation status, energy consumption data, and usage frequency; A personalized energy efficiency standard threshold of the target device is determined by using a life cycle cost analysis method as an initial personalized threshold, and the initial personalized threshold of the target device and various information of the target device are input into a pre-set and trained adjustment value analysis model based on a reinforcement learning algorithm to output an adjustment value; Based on the adjustment value and the initial personalized threshold, a personalized energy efficiency standard threshold of the target device is calculated and determined.

[0007] Optionally, based on the real-time related data of various external devices, analyzing whether there is a change in the device status includes: Obtain the scenario category of the external device, where the scenarios include daily operation scenarios, peak operation scenarios, equipment maintenance scenarios, extreme weather scenarios, and equipment upgrade and transformation scenarios; According to the correspondence between the scene category of the external device and the device state analysis algorithm, the device state analysis algorithm adopted by different external devices is matched; Based on the real-time relevant data of external devices, a matching device status analysis algorithm is used to analyze whether there is any device status change in the external device.

[0008] In the second aspect, the present application provides a rail transit energy efficiency balance dynamic monitoring system, which adopts the following technical solutions: A rail transit energy efficiency balance dynamic monitoring system comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein the program can implement the rail transit energy efficiency balance dynamic monitoring method as described in the first aspect when loaded and executed by the processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic diagram of the overall process of a method for dynamic monitoring of rail transit energy efficiency balance according to an embodiment of the present application.

[0010] Figure 2 It is a flowchart of another embodiment of the present application for using a life cycle cost analysis method to determine personalized energy efficiency standard thresholds for various types of equipment.

[0011] Figure 3 This is a flowchart of another embodiment of the present application for analyzing whether there is any change in device status based on real-time related data of various external devices. DETAILED DESCRIPTION

[0012] The present application is further described in detail below in conjunction with the accompanying drawings.

[0013] Reference Figure 1 , is a rail transit energy efficiency balance dynamic monitoring method disclosed in this application, comprising: Step S100, obtaining relevant data of various external devices.

[0014] External equipment: In the rail transit system, various facilities related to dynamic monitoring of energy efficiency balance. For example, the traction motor of the train is the power core of the train operation and has a high energy consumption; the escalator in the station consumes electricity continuously, and its energy consumption affects the overall energy efficiency of the station; and the signal system equipment is large in number and runs all day, and the total energy consumption cannot be ignored.

[0015] Related data: Contains information that can reflect the operating status of external equipment, energy consumption characteristics, and other information that affects energy efficiency. It involves both real-time operating parameters of the equipment, such as current and voltage during operation, and historical data accumulated over the long term operation of the equipment, such as average monthly energy consumption and the number of failures in the past year. It also covers equipment operating environment data, such as humidity and altitude.

[0016] There are several ways to obtain relevant data of various external devices, as follows: 1. Sensor collection: Install different sensors on various types of equipment to collect data. For example, pressure sensors are installed in the train braking system to monitor the braking pressure and determine the system status and energy consumption; flow sensors are used in the station ventilation duct to measure the ventilation volume and help optimize the energy saving of the ventilation system. 2. Equipment with built-in system interface: Many rail transit equipment now have intelligent control systems and data interfaces, which are connected to the monitoring system through communication protocols to obtain data. Small equipment such as intelligent lighting control modules may use Bluetooth interfaces, and monitoring personnel can obtain working parameters at a certain distance; large equipment such as train traction control systems often use industrial Ethernet interfaces to achieve high-speed and stable data transmission. 3. Data storage system reading: The data generated by rail transit operations is stored in a special system, and historical data can be read for analysis. Relational databases store structured data, such as basic equipment information and maintenance records, which can be quickly obtained and analyzed using SQL statements; distributed file systems store massive unstructured data, such as equipment operation logs, from which information such as start and stop times and abnormal alarms can be extracted.

[0017] Step S200: input relevant data of various external devices into a pre-built fuzzy clustering classification model, and output classification results of various external devices.

[0018] Fuzzy clustering classification model: Based on fuzzy mathematics, it uses "membership degree" (between 0 and 1) to measure the degree to which a data point belongs to a certain category, breaking the traditional clear classification boundaries, flexibly classifying rail transit external equipment data, and mining the potential characteristics of the data.

[0019] The information related to model construction and training is as follows: 1. Construction: Based on a large amount of historical data of rail transit external equipment, covering operating conditions, energy consumption, load and other information, the model is built by mining data patterns. 2. Training: Divide historical data into training sets and test sets, use the training set to tune the internal parameters of the model, such as cluster centers and membership functions, and then use the test set to evaluate. If the performance does not meet the requirements, adjust the parameters or increase the data for retraining until it meets the requirements.

[0020] The relevant contents of data input and processing are as follows: 1. Input: The real-time parameters, historical energy consumption, environmental parameters, etc. of the equipment obtained in step S100 are formatted and pre-processed before inputting into the model. 2. Processing: Standardize to eliminate dimensional differences, then calculate the degree of membership between data points and cluster centers using the fuzzy clustering algorithm, and iterate and adjust to make the similarity of the same type of data high and the similarity of different types low, and determine the equipment classification.

[0021] The output and application related contents are as follows: 1. Output: Give the equipment classification results and membership, such as "high efficiency and energy saving", "normal energy consumption", "high energy consumption needs to be optimized" and other categories and corresponding values. 2. Application: Provide a basis for energy efficiency management and equipment optimization, summarize the advantages of "high efficiency and energy saving" equipment, analyze the problems of "high energy consumption needs to be optimized" equipment and transform them, and improve system energy efficiency.

[0022] Step S300, based on the equipment classification result, a life cycle cost analysis method is used to determine the personalized energy efficiency standard thresholds for each type of equipment.

[0023] Life cycle cost analysis method: comprehensively consider the cost of the entire process from the birth to the scrapping of the equipment, such as the cost of procurement, use, maintenance, scrapping and other links, so as to analyze energy consumption and other costs and balance the cost and energy efficiency of the equipment throughout its life cycle.

[0024] Personalized energy efficiency standard threshold: Based on device characteristics, usage frequency, operating environment and life cycle cost analysis results, develop exclusive energy consumption judgment standards for different devices to accurately manage device energy efficiency.

[0025] The process of determining the threshold of personalized energy efficiency standards is as follows: 1. Use the equipment classification results: refer to the equipment classification in step S200, and classify equipment with similar characteristics and usage patterns into one category, such as train traction motors and station ventilation equipment. 2. Conduct life cycle cost analysis: Sort out the costs of each stage of each type of equipment, calculate the purchase and transportation costs when purchasing; calculate energy consumption and maintenance costs when using; and estimate dismantling and recycling costs when scrapped. 3. Determine the threshold: Combined with the total cost of the equipment's entire life cycle, industry energy-saving standards and technological trends, set thresholds for equipment in different working conditions. For example, for equipment that is frequently started and stopped, the startup energy consumption standards are appropriately relaxed, and strict control is applied during stable operation.

[0026] Step S400, based on the comparison result between the actual energy consumption of the device and the personalized energy efficiency standard threshold, determine whether the corresponding device has energy consumption abnormality. If yes, execute step S500; if not, execute step S600.

[0027] Among them, actual energy consumption refers to the amount of energy consumed by the equipment during its actual operation, which is collected and recorded in real time through various metering devices installed on the equipment, such as electricity meters and gas meters.

[0028] The comparison process is as follows: During the operation of rail transit equipment, the actual energy consumption data of the equipment is continuously monitored and collected. According to a certain time period, such as every hour or every day, the collected actual energy consumption is compared with the personalized energy efficiency standard threshold determined in step S300. When comparing, focus on the relationship between the energy consumption values ​​and the degree to which the actual energy consumption deviates from the threshold.

[0029] Step S500, combining the personalized threshold of the equipment and the historical energy consumption data, using the built comprehensive energy loss analysis model based on fault tree and Bayesian network, starting from the failure mode of the equipment, combined with the equipment's operation history, maintenance records, and environmental factors, through Bayesian network reasoning and dynamic fault tree analysis, locate the root cause of the energy loss and send it to the terminal held by the person in charge.

[0030] The analysis process is as follows: 1. Use historical energy consumption data to calculate the mean and standard deviation, determine the energy consumption boundary, analyze historical energy consumption according to multiple dimensions, and establish a normal energy consumption mode. 2. Sort out the failure modes of equipment from systems to components, determine the probability of occurrence of each subdivided mode, and build a dynamic fault tree with energy consumption exceeding the preset proportion of the normal boundary as the top event. Use logic gates to describe causal relationships, find all possible fault paths, and select high-frequency fault bottom events as the direction of troubleshooting. 3. Convert fault tree events into Bayesian network nodes and edges, determine the prior probability based on historical equipment failures and industry statistics, input multi-source data to update the posterior probability, and determine the cause of the failure through probability changes.

[0031] The positioning results and applications are as follows: Integrate the fault tree and Bayesian network analysis results, establish an evaluation matrix to score the fault mode, and select factors whose comprehensive scores exceed the preset scores and whose energy consumption increases exceed the preset ratio as the root causes of energy loss. Send the positioning results to the responsible terminal to provide a basis for the subsequent formulation of solutions.

[0032] Step S600: no notification.

[0033] Among them, no notification means that the equipment's operating energy consumption is within a normal and reasonable range, and there is no need to provide special reminders to relevant personnel or initiate additional analysis and processing procedures.

[0034] Reference Figure 2 , using the life cycle cost analysis method to determine the personalized energy efficiency standard thresholds for various types of equipment include: Step S310, based on the real-time related data of various external devices, analyze whether there is a change in the device status. If not, execute step S320; if yes, execute step S330.

[0035] Among them, real-time relevant data of external equipment refers to the data generated by the current operation of various equipment in the rail transit system, such as operating voltage, current size, equipment operating speed, etc., which can intuitively reflect the current operating status of the equipment.

[0036] Equipment status changes: covers changes in equipment operating modes, such as switching from normal operation to energy-saving mode; load changes, such as different loads of ventilation equipment during peak and off-peak periods; also includes equipment failures, starts and stops, etc.

[0037] The general process is as follows: 1. The monitoring system continuously receives real-time data from sensors and the device's own system. 2. Using the preset data analysis algorithm, the real-time data is compared with the standard data set for normal operation of the device. 3. Based on the comparison results, it is determined whether the device status has changed.

[0038] For example, taking subway ventilation equipment as an example, the current sensor collects the fan current data in real time. During normal operation, the current is stable at 5-8A. If the current is suddenly increased to 12A at a certain moment, far exceeding the normal range, the algorithm analysis can determine that the state of the ventilation equipment has changed, which may be due to the blockage of the air duct, resulting in an increase in load.

[0039] Step S320, maintaining the original setting, the original setting is to use the life cycle cost analysis method to determine the personalized energy efficiency standard thresholds of various types of equipment.

[0040] Step S330, taking the device whose status has changed as the target device, and collecting various types of information of the target device.

[0041] Among them, various types of information of the target device include real-time operating status, energy consumption data, and usage frequency.

[0042] The general process is as follows: 1. Once the target device is determined, the system immediately starts the information collection program for the device. 2. Various sensors transmit the real-time monitored data to the data collection terminal, and the equipment management system also uploads the relevant information to the monitoring platform. 3. The monitoring platform preliminarily organizes and classifies the collected data, removes abnormal data, ensures the accuracy and effectiveness of the data, and prepares for subsequent analysis.

[0043] Step S340, use the life cycle cost analysis method to determine the personalized energy efficiency standard threshold of the target device as the initial personalized threshold, and input the initial personalized threshold of the target device and various types of information of the target device into a pre-set and trained adjustment value analysis model based on the reinforcement learning algorithm, and output the adjustment value.

[0044] Among them, the adjustment value analysis model based on the reinforcement learning algorithm is a trained intelligent model that can continuously optimize and learn based on the input target device information using the reinforcement learning algorithm, and output an adjustment value to correct the initial personalized threshold to make it more in line with the actual operation of the device.

[0045] The general process is as follows: 1. Use the life cycle cost analysis method to process the target device's full life cycle cost data to obtain the initial personalized threshold. 2. Integrate the initial personalized threshold with the various types of target device information collected in step S330 and input them into the trained adjustment value analysis model. 3. The model uses the reinforcement learning algorithm to analyze and calculate based on the input data and output the adjustment value.

[0046] Taking the subway escalator as an example, the initial personalized threshold is calculated by analyzing its purchase cost, electricity cost over the years, regular maintenance costs, etc. Then the current running speed (real-time running status), daily power consumption (energy consumption data), number of starts and stops in a day (frequency of use) and other information of the escalator are input into the model together with the initial personalized threshold. Finally, the model outputs the adjustment value, which is used for the subsequent correction of the personalized energy efficiency standard threshold of the escalator.

[0047] Step S350: Calculate and determine the personalized energy efficiency standard threshold of the target device based on the adjustment value and the initial personalized threshold.

[0048] The general process is as follows: 1. Obtain the initial personalized threshold and adjustment value from the corresponding steps and models. 2. According to the preset calculation rules, calculate the adjustment value and the initial personalized threshold. Usually, the adjustment value is added or subtracted from the initial personalized threshold to obtain the final personalized energy efficiency standard threshold. 3. Store the determined personalized energy efficiency standard threshold in the system database so that it can be used as a comparison basis when monitoring the energy consumption of the equipment later.

[0049] Reference Figure 3 , based on the real-time relevant data of various external devices, analyze whether there is any change in the device status including: Step S311, obtaining the scene category of the external device.

[0050] The scenario categories of external equipment refer to the different environments or working stages in which rail transit external equipment operates. The scenarios include daily operation scenarios, peak operation scenarios, equipment maintenance scenarios, extreme weather scenarios, and equipment upgrade and transformation scenarios.

[0051] The daily operation scenario is the normal operating state of the equipment; the peak operation scenario is the period when the passenger flow is large and the equipment is running at high load; the equipment maintenance scenario is the period when the equipment is inspected and maintained; the extreme weather scenario is the operating environment of the equipment in severe weather such as heavy rain and heavy snow; the equipment upgrade and transformation scenario is the stage when the equipment is undergoing functional upgrades or technical transformation.

[0052] The acquisition method is as follows: Time and passenger flow data: Obtain daily time information and passenger flow data of each station through the rail transit operation management system. For example, during the morning and evening rush hours, the passenger flow increases significantly, and combined with the time, it can be judged that it is in the peak operation scene. Equipment maintenance plan and record: Obtain the equipment maintenance plan and actual maintenance record from the equipment maintenance management platform to clarify when the equipment is in the equipment maintenance scene. Meteorological information connection: Connect with the data interface of the meteorological department to obtain real-time meteorological data and determine whether it is in extreme weather scenes, such as heavy rain and heavy snow. Engineering construction and transformation documents: Consult the relevant documents on engineering construction and equipment transformation of rail transit to understand the time and progress of equipment upgrades and transformations, and determine the equipment upgrade and transformation scenes. The general process is as follows: 1. The system regularly obtains relevant information from various data sources. 2. Integrate and analyze the acquired time, passenger flow, weather, maintenance plan and other data. 3. According to the preset scene judgment rules, correspond the external equipment to the corresponding scene category.

[0053] Step S312, matching the device state analysis algorithms used by different external devices according to the correspondence between the scene categories of the external devices and the device state analysis algorithms.

[0054] Device status analysis algorithm: a method used to process real-time data from external devices and identify whether the device status has changed through specific calculation logic and models. For example, algorithms based on data threshold judgment, machine learning model analysis, etc.

[0055] Correspondence between scenario categories and device status analysis algorithms: Pre-set rules clarify which algorithms should be used to analyze the status changes of different external devices in various operating scenarios. The characteristics of device status changes in different scenarios are different, and different algorithms need to be adapted for accurate judgment.

[0056] Specifically, for daily operation scenarios, the device status analysis algorithm uses an algorithm based on threshold judgment. For example, the normal temperature range in the car is set to 24-26°C. When the real-time temperature collected by the temperature sensor exceeds this range and the air conditioning operation mode is normal, the algorithm determines that the air conditioning cooling or heating effect may be abnormal.

[0057] For peak operation scenarios, the equipment status analysis algorithm uses the decision tree algorithm in machine learning. The current, speed, acceleration, passenger flow, etc. are used as input features, and the decision tree establishes a model by learning from historical data. For example, when the train accelerates, the current and acceleration deviate too much from the normal situation, and the passenger flow is at peak time, the decision tree algorithm determines that there may be an abnormality in the traction system.

[0058] Regarding equipment maintenance scenarios, the equipment status analysis algorithm uses a rule-based expert system algorithm. Rules are set according to equipment maintenance standards and expert experience. For example, if the track wear depth exceeds a certain value or the pantograph contact pressure is lower than the standard range, the equipment is considered abnormal.

[0059] For extreme weather scenarios, the equipment status analysis algorithm uses fuzzy logic algorithms. It takes rainfall, humidity, fan operating parameters, etc. as inputs, and uses fuzzy processing, fuzzy reasoning, and defuzzification to determine the equipment status. For example, when rainfall and humidity exceed a certain range, and the fan operating current is abnormal, the fuzzy logic algorithm determines that the ventilation system may be at risk of water ingress or blockage.

[0060] For equipment upgrade and transformation scenarios, the equipment status analysis algorithm uses a comparative analysis algorithm. The equipment operating parameters after the upgrade are compared with the design standard parameters, and the key performance indicators of the equipment before and after the transformation are compared. For example, the signal strength should reach more than 90% of the design standard. If the actual signal strength is lower than this value, the signal system status is judged to be abnormal.

[0061] The general process is as follows: the system uses the scene category and equipment type as search conditions and performs precise matching in the configuration file. For example, if it is determined that a subway ventilation device is in an extreme weather scene, the system will search for the record "Extreme Weather Scene-Ventilation Equipment" in the configuration file and obtain the corresponding device status analysis algorithm name.

[0062] According to the algorithm name obtained from the configuration file, the system searches the algorithm library. After finding the corresponding algorithm, the algorithm code and related dependencies are loaded into the memory to complete the initialization of the algorithm and prepare for the next step of processing the real-time data of the external device.

[0063] Step S313: Based on the real-time relevant data of the external device, a matching device status analysis algorithm is used to analyze whether there is a device status change of the external device.

[0064] The general process is as follows: 1. Data collection: continuously obtain real-time relevant data of external devices from sensors and the device's own system. 2. Algorithm operation: input the collected data into the device status analysis algorithm matched in step S312, and the algorithm processes and analyzes the data according to the preset calculation logic and model. 3. Result judgment: judge whether the device status has changed based on the algorithm analysis results. If the result output by the algorithm exceeds the set range of the normal state or conforms to a specific state change mode, it is determined that the device status has changed; otherwise, the device status is considered normal.

[0065] A method for dynamic monitoring of energy efficiency balance of rail transit also includes steps after collecting various types of information of target equipment, which are as follows: Step S3A0, analyzing whether there are other devices that have a cooperative relationship with the target device. If not, executing step S3B0; if yes, executing step S3C0.

[0066] Synergistic relationship: In rail transit systems, there is a working relationship between equipment that cooperates and influences each other. For example, the traction system and braking system of a train. The traction system is responsible for providing power to move the train forward, while the braking system controls the train to slow down or stop. The two are closely related and have a synergistic relationship.

[0067] Collaborative equipment: other equipment that has a collaborative relationship with the target equipment. If the target equipment is the air conditioning system of the train, then the ventilation system can be considered as its collaborative equipment because they work together to adjust the air environment in the car.

[0068] The acquisition methods are as follows: 1. Equipment layout and connection drawings: By consulting the equipment layout and connection line diagrams of the rail transit system, the physical connection and layout relationship between the equipment can be clarified to determine whether there is a collaborative relationship between the equipment. 2. System function specification: The function specification of the equipment describes in detail the functions of the equipment and how it interacts with other equipment, from which information about the collaborative work of the equipment can be obtained.

[0069] The general process is as follows: 1. After determining the target device, first sort out the devices that are related to the target device in terms of physical location or line connection from the device layout and connection drawings. 2. Then, combined with the system function specification, analyze the coordination relationship between these related devices and the target device in the function implementation process to determine whether there is a synergistic relationship.

[0070] Assume that the target device is an automatic ticket vending machine at a subway station. From the equipment layout diagram, it is found that the automatic ticket vending machine is connected to the network server of the station. By checking the system function specification, it is known that the automatic ticket vending machine needs to obtain ticket information, fare update and other data from the server in real time, and there is an obvious synergistic relationship between the two. Therefore, in step S3A0, it is determined that there is a device that has a synergistic relationship with the automatic ticket vending machine, namely the station network server.

[0071] Step S3B0, maintain the original settings and continue with subsequent steps.

[0072] Among them, the original settings refer to the various parameters and standards set according to previously existing rules, models and methods after the target equipment is determined, including the initial personalized thresholds determined by the life cycle cost analysis method, and the established equipment operation monitoring and analysis processes in the system.

[0073] The general process is as follows: When step S3A0 determines that there are no other devices that have a cooperative relationship with the target device, the system first confirms the original settings from the system configuration file, and continues to collect relevant data of the target device, such as real-time operating status, energy consumption data, etc., according to the data collection frequency and method specified in the original settings.

[0074] The collected data will be subsequently processed and analyzed according to the established analysis process and standards in the original settings to complete the entire energy efficiency balance dynamic monitoring process.

[0075] Step S3C0, collecting various types of information of other devices that have a collaborative relationship with the target device and the existing collaborative relationship data.

[0076] Collaboration relationship data: Specific data describing the interaction and mutual influence between the target device and the collaborative device. For example, the impact of the start and stop of the collaborative device on the energy consumption of the target device, the amount and frequency of data transmission between the two during collaborative work, etc.

[0077] Various information about collaborative devices: real-time operating status, energy consumption data, usage frequency, and other information of other devices that have a collaborative relationship with the target device. This information helps to fully understand the overall situation when the devices work together.

[0078] The general process is as follows: After determining other devices that have a collaborative relationship with the target device, first establish data connections with these collaborative devices and start collecting collaborative relationship data through the device interface.

[0079] Synchronously start collecting data from various monitoring sensors on collaborative equipment to obtain information such as their operating status and energy consumption.

[0080] Regularly extract historical data of collaborative devices from system log records, supplement and verify the collected real-time data, and ensure the integrity and accuracy of the information.

[0081] Step S3D0, input the initial personalized threshold of the target device, various types of information of the target device, various types of information of other devices that have a collaborative relationship with the target device, and the existing collaborative relationship data into the pre-trained system collaborative reinforcement learning threshold adjustment model, and output the adjustment value.

[0082] System collaborative reinforcement learning threshold adjustment model: An intelligent model built based on reinforcement learning algorithm, which is specially used to deal with the impact of the collaborative relationship between devices on the energy efficiency standard threshold. It continuously learns the operation data and collaborative relationship data of the equipment, optimizes its own parameters and strategies, and outputs adjustment values ​​that are more in line with the actual situation.

[0083] Adjustment value: The value output by the system collaborative reinforcement learning threshold adjustment model is used to correct the initial personalized threshold of the target device so that the final personalized energy efficiency standard threshold can more accurately reflect the energy consumption of the device in a collaborative working environment.

[0084] The general process is as follows: The system organizes and preprocesses the relevant data of the target device and collaborative devices in the format required by the model, removes outliers and noise data, and ensures data quality.

[0085] The system collaborative reinforcement learning threshold adjustment model is called and the pre-processed data is input into the model. Based on the reinforcement learning algorithm, the model conducts in-depth analysis and learning of the input data to explore the potential connection between the collaborative relationship between devices and energy consumption.

[0086] Based on the learning and analysis results, the model calculates and outputs an adjustment value, which reflects the impact of the collaborative work of the equipment on the energy consumption standard of the target equipment.

[0087] Step S3E0: Calculate and determine the personalized energy efficiency standard threshold of the target device based on the adjustment value and the initial personalized threshold.

[0088] The general process is as follows: 1. The system extracts the adjustment value output from step S3D0 from the temporary data storage area, and retrieves the initial personalized threshold of the target device from the database. 2. The two values ​​are calculated according to the pre-set calculation rules. Generally, if the adjustment value is a positive number, the adjustment value is added to the initial personalized threshold; if the adjustment value is a negative number, the adjustment value is subtracted to obtain the final personalized energy efficiency standard threshold. 3. The calculated personalized energy efficiency standard threshold is updated and stored in the system database as a comparison basis for subsequent monitoring of the target device energy consumption, so as to determine in real time whether the device energy consumption is within the normal range.

[0089] Among them, the construction process of the system collaborative reinforcement learning threshold adjustment model is as follows: Step S3D1, collects historical operation data of rail transit equipment, covering the operation status, energy consumption data and corresponding system energy consumption of each equipment under different time periods and working conditions.

[0090] Rail transit equipment historical operation data: refers to a series of data generated by various types of equipment in the rail transit system during past operation, including the equipment's operating status (such as on / off, operating speed, operating mode, etc.), energy consumption data (power consumption, gas consumption, etc.) in different time periods (such as working days, holidays, morning and evening peaks, etc.) and different working conditions (normal operation, equipment maintenance, fault repair, etc.), as well as the energy consumption of the entire rail transit system corresponding to the operation of these equipment. These data are the basis for building a system collaborative reinforcement learning threshold adjustment model, which can reflect the operating characteristics and energy consumption patterns of the equipment under various conditions.

[0091] The general process is as follows: 1. Determine the time range and equipment type for data collection, such as collecting the operating data of all subway trains and station equipment in the past year. 2. Extract the corresponding data from the equipment monitoring system, energy management system, and operation and maintenance management records according to the predetermined time range and equipment list. 3. Preliminary sorting of the extracted data, classified storage according to equipment type, time sequence, etc., to facilitate subsequent data preprocessing and model training.

[0092] Step S3D2, preprocessing the collected historical operation data of rail transit equipment, wherein the data preprocessing includes data cleaning and normalization.

[0093] Data cleaning: refers to the review and correction of the collected historical operation data of rail transit equipment to remove incomplete, erroneous, duplicate or noisy data. For example, abnormal energy consumption data caused by sensor failure or erroneous marks in the equipment operation status record need to be corrected through data cleaning to ensure the accuracy and reliability of the data and provide a high-quality data foundation for subsequent analysis and model training.

[0094] Normalization: The process of converting data with different characteristics to the same scale range. In the rail transit equipment operation data, the numerical range of energy consumption data may be very different from the data range of equipment operation speed. Through normalization, these data of different magnitudes can be made comparable, which makes it easier for the model to better learn and process the relationship between data and improve the training effect and accuracy of the model.

[0095] Step S3D3, initialize the system collaborative reinforcement learning threshold adjustment model, and the initial operation is to randomly assign values ​​to the weight matrix and bias vector of the strategy network and the value network.

[0096] Policy network: In reinforcement learning, the policy network is used to generate a series of possible actions and their corresponding probability distributions based on the current state of the environment, that is, to determine what actions the agent should take in the current state. It is the core part of the model decision-making. For example, in the energy efficiency management scenario of rail transit equipment, the policy network will give the probability of adjusting equipment operating parameters (such as changing the train traction power, adjusting the ventilation system wind speed) and other actions based on the current operating status of the equipment, energy consumption data and other information.

[0097] Value network: used to evaluate the expected cumulative rewards that an agent can obtain in the future after taking an action in a certain state, that is, to measure the value of the current state and action. In rail transit, the value network can evaluate the expected effect of adjusting the operating parameters of the equipment on reducing the overall energy consumption of the system or improving energy efficiency, helping the model to judge the pros and cons of the action.

[0098] Weight matrix and bias vector: Weight matrix and bias vector are key parameters in neural networks (including policy networks and value networks). The weight matrix determines the connection strength between the input data and the neuron, and the bias vector adds a fixed value to the output of the neuron. Together, they affect the output of the network. During training, these parameters are constantly adjusted to optimize model performance.

[0099] The general process is as follows: 1. Determine the network structure of the policy network and the value network, including the number of neuron layers, the number of neurons in each layer, etc. For example, build a simple policy network that includes an input layer, two hidden layers, and an output layer. 2. According to the network structure, use a random number generator to generate random values ​​for the weight matrix and bias vector of each layer of the policy network and the value network. For example, for a neural network with 10 neurons in the input layer and 20 neurons in the hidden layer, it is necessary to generate a 10×20 weight matrix and a bias vector of length 20, and their initial values ​​are filled with random numbers. 3. Apply the generated random weight matrix and bias vector to the policy network and the value network to complete the initialization operation of the model, so that the model has preliminary processing capabilities and is ready to accept input data for training.

[0100] Step S3D4, starting from the initial state, the policy network gives the action probability distribution according to the current state, uses the roulette algorithm to select the action to be applied to the simulation environment, records the state, action, reward and next state information, and collects a preset number of records to form a trajectory set.

[0101] Initial state: At the beginning of the reinforcement learning model training, it represents the state of the environment in which the agent is located. For rail transit equipment energy efficiency management, it is the initial operating state of the equipment when the model starts running, including various parameters of the equipment (such as the initial speed of the train, the initial wind speed of the ventilation system, etc.) and surrounding environmental conditions (such as the initial temperature in the station, passenger flow, etc.).

[0102] Action probability distribution: The set of occurrence probabilities of each possible action generated by the policy network based on the current state. For example, in a subway lighting system, possible actions include turning on / off some lamps, adjusting lamp brightness, etc. The action probability distribution indicates the probability of each action being selected in the current state.

[0103] Roulette algorithm: A probability-based random selection algorithm used to select a specific action from the action probability distribution. It is like spinning a roulette wheel divided into different areas, each area corresponds to an action, and the area size is proportional to the action probability. The action corresponding to the area pointed by the pointer when the roulette wheel stops is the selected action.

[0104] Simulation environment: A virtual environment built to train reinforcement learning models, simulating the actual operation of a rail transit system. In this environment, various actions of the equipment can be simulated and their effects, such as changes in energy consumption and equipment status, can be observed.

[0105] Trajectory collection: It consists of a series of records, each of which contains state, action, reward, and next state information. These records reflect the running trajectory of the model in the simulation environment and are an important data basis for model learning and optimization.

[0106] The general process is as follows: 1. The model starts from the initial state and inputs the current state information into the policy network. 2. The policy network calculates and outputs the action probability distribution based on the input state. 3. Use the roulette algorithm to randomly select an action based on the action probability distribution. 4. Apply the selected action to the simulation environment, observe and record the changes in the environment, including the rewards obtained (such as the degree of energy consumption reduction, the improvement of equipment operation stability, etc.) and the next state entered. 5. Repeat the above steps to collect a preset number of records (such as 1,000), and organize these records into a trajectory set for subsequent model training and optimization.

[0107] Step S3D5, using the generalized advantage estimation method, first calculate the TD error, then calculate the advantage function, and judge the quality of the action.

[0108] TD error: Temporal difference error, which is used to measure the difference between the predicted value and the actual value in the current state. In reinforcement learning, TD error is calculated by comparing the value estimate of the current state with the value estimate of the next state plus the reward. It reflects the model's estimated deviation from the environmental changes.

[0109] Advantage function: Indicates the degree of advantage of taking a certain action compared to the average action in the current state. The larger the value of the advantage function, the better the benefit of the action compared to other actions. The advantage function can be used to judge the pros and cons of actions and provide direction for model optimization.

[0110] The acquisition method is disclosed as follows: 1. Value network output: Get the value estimate of the current state and the next state from the value network. The value network outputs the evaluation of the value in the state based on the input state information. These output values ​​are the basic data for calculating the TD error and advantage function. 2. Reward value: In the simulation environment, the feedback reward obtained after performing an action, such as the reward brought by reduced energy consumption or the reward brought by reduced equipment failures in rail transit equipment optimization. These reward values ​​are directly involved in the calculation of TD error and advantage function.

[0111] The general process is as follows: First, based on the value network's estimate of the value of the current state and the next state, as well as the reward obtained after executing the action, the TD error is calculated using the formula. The formula is generally: TD error = reward + γ × next state value estimate - current state value estimate, where γ is a discount factor used to measure the importance of future rewards.

[0112] Next, based on the TD error, the advantage function is calculated using the generalized advantage estimation method. This method comprehensively considers the TD errors of multiple time steps to more accurately evaluate the advantage of the action. For example, the advantage function value is obtained by accumulating the discounted TD errors of multiple time steps.

[0113] The advantages and disadvantages of each action are judged according to the calculated advantage function value. A positive advantage function value indicates that the action is better than the average action; a negative advantage function value indicates that the action is relatively worse.

[0114] Let's take the model of optimizing the energy consumption of subway trains as an example. Assume that the current train is running at a speed of 40km / h. After performing an acceleration action, the speed of the next state becomes 45km / h. The value network estimates the value of the current state to be 0.5, the value of the next state to be 0.6, the reward for executing the action is 0.1 (because the energy consumption is reduced), and the discount factor γ is set to 0.9. First, calculate the TD error: 0.1+0.9×0.6-0.5=0.14. Then use the generalized advantage estimation method to calculate the advantage function. Assume that after a series of calculations, the advantage function value is 0.2. This shows that the acceleration action is advantageous compared to the average action, and in subsequent model training, it will tend to increase the probability of such actions being selected.

[0115] Step S3D6, using the PPO-Clip objective function in the proximal policy optimization algorithm to calculate the policy loss, and adjusting the parameters through gradient descent to minimize the policy loss.

[0116] Proximal Policy Optimization (PPO): An algorithm for optimizing policy networks that aims to improve the performance of policies in reinforcement learning. It improves the stability and efficiency of training by limiting the magnitude of policy updates and preventing drastic changes in the policy during the update process.

[0117] PPO-Clip objective function: The core part of the PPO algorithm, used to calculate the policy loss. It controls the scope of policy updates by clipping the advantage ratio of the new and old policies, ensuring that the policy can develop in a better direction during the optimization process without causing performance degradation due to excessive updates.

[0118] Policy loss: Measures the gap between the current policy and the optimal policy. By minimizing the policy loss, the policy network parameters are optimized so that the policy network can output a better action probability distribution, thereby improving the performance of the model in practical applications.

[0119] The acquisition method is disclosed as follows: 1. Advantage function value: The advantage function value calculated from step S3D5, which reflects the degree of advantage of each action compared to the average action, is the key data for calculating the PPO-Clip objective function. 2. Action probability of the new and old strategies: The action probability output by the strategy network for the same state before and after the update. By comparing the action probabilities of the new and old strategies, the advantage ratio can be calculated, which is then used to calculate the PPO-Clip objective function.

[0120] The general process is as follows: First, obtain the advantage function value calculated in step S3D5, as well as the action probability output for the current state before and after the policy network is updated.

[0121] According to the action probabilities of the new and old strategies, the advantage ratio is calculated. The formula is: advantage ratio = new strategy action probability / old strategy action probability.

[0122] Substitute the advantage ratio and advantage function value into the PPO-Clip objective function for calculation. The general form of the PPO-Clip objective function is: L = min (advantage ratio × advantage function value, clip (advantage ratio, 1-ε, 1+ε) × advantage function value), where the clip function is used to clip the advantage ratio, and ε is a hyperparameter, usually set to a small value, such as 0.2, to control the amplitude of the policy update.

[0123] Through the gradient descent algorithm, the parameters of the policy network are adjusted to minimize the policy loss. During the gradient descent process, the weight matrix and bias vector of the policy network are continuously updated according to the gradient information of the objective function, so that the policy network gradually develops in a better direction.

[0124] Step S3D7, use the mean square error loss function to calculate the value loss and update the parameters by gradient descent.

[0125] Mean square error loss function: It is often used to measure the difference between two numerical sequences. In model training, it calculates the average of the sum of squares of the errors between the predicted value and the true value. By minimizing the mean square error loss function, the predicted value of the model can be made as close to the true value as possible, thus improving the accuracy of the model.

[0126] Value loss: In reinforcement learning, the value network predicts the future value based on the current state. Value loss is the gap between the predicted value and the actual value. By adjusting the parameters of the value network to reduce value loss, the value network can more accurately evaluate the state value.

[0127] The acquisition methods are disclosed as follows: 1. Value network prediction value: Get the predicted value of the future value under the current state from the value network. The value network uses its internal parameters and algorithms to calculate based on the input state information and output the predicted value. 2. Actual value obtained: After performing an action in a simulated environment, the actual value obtained is calculated based on the actual reward and the value feedback of the subsequent state. For example, in the rail transit scenario, the actual reduction in energy consumption and improvement in operating efficiency after adjusting the operating parameters of the equipment are converted into actual value.

[0128] The general process is as follows: First, obtain the predicted value of the value network for the current state and the actual value obtained after performing actions in the simulation environment.

[0129] Substitute the predicted value and the actual obtained value into the mean square error loss function for calculation.

[0130] The gradient descent algorithm is used to update the parameters of the value network according to the calculated value loss. The gradient descent algorithm adjusts the weight matrix and bias vector of the value network in the opposite direction of the gradient of the loss function, so that the value loss gradually decreases and the prediction of the value network becomes more accurate.

[0131] Step S3D8, run the mixed integer linear programming model, calculate the theoretical optimal solution of system energy consumption based on the physical characteristics of the rail transit system and equipment operation constraints, and calculate the gap between the current system energy consumption and the theoretical optimal solution.

[0132] Mixed integer linear programming model: A mathematical optimization model used to deal with linear programming problems involving integer variables and continuous variables. In the field of rail transit, this model can be used to optimize the energy consumption of the system, where the integer variables may represent the number of devices turned on, the number of times the operating mode is switched, etc., and the continuous variables may be the operating parameters of the equipment (such as power, speed, etc.).

[0133] Theoretical optimal solution for system energy consumption: Based on the mixed integer linear programming model, the physical characteristics of the rail transit system (such as the energy conversion efficiency of the equipment, line loss, etc.) and the equipment operation constraints (such as the maximum and minimum operating power of the equipment, operating time limit, etc.) are comprehensively considered to calculate the theoretical value of the minimum system energy consumption or the highest energy efficiency. It provides an ideal reference standard for evaluating the current system energy consumption.

[0134] The acquisition methods are as follows: 1. Data collection: Collect various types of data on the rail transit system, including technical parameters of equipment (such as motor efficiency, transformer loss, etc.), operating constraints (such as the maximum operating speed of the train, the maximum air volume of the ventilation system, etc.) and the system topology (such as power supply network layout, equipment connection relationship, etc.). These data are obtained from technical documents provided by equipment manufacturers, design drawings of rail transit systems, and actual operation monitoring data. 2. Model call: Use professional mathematical optimization software (such as Gurobi, CPLEX, etc.) to call the mixed integer linear programming model, use the collected data as the input parameters of the model, and calculate the theoretical optimal solution for the system energy consumption.

[0135] The general process is as follows: 1. Collect and organize various data related to the energy consumption of the rail transit system to ensure the accuracy and completeness of the data. 2. Input the organized data into the mixed integer linear programming model, set the model's objective function (such as minimizing system energy consumption) and constraints (such as equipment operation restrictions, physical law constraints, etc.). 3. Use mathematical optimization software to solve the model and obtain the theoretical optimal solution for the system's energy consumption. At the same time, obtain the energy consumption data of the actual operation of the current system, monitor it in real time through sensors or extract it from the energy management system. 4. Calculate the gap between the current system energy consumption and the theoretical optimal solution, which can be measured by the difference or percentage between the two.

[0136] Step S3D9, determining the adjustment strategy according to the correspondence between the gap range between the current system energy consumption and the theoretical optimal solution and the adjustment strategy.

[0137] Gap range: Different intervals are divided according to the difference between the current system energy consumption and the theoretical optimal solution. For example, the difference between 0-5% is divided into one range, and 5%-15% is divided into another range. Different ranges reflect the degree of deviation of the system energy consumption from the optimal state.

[0138] Adjustment strategy: A series of model optimization and strategy adjustment methods developed for different gap ranges. For example, when the gap is small, only fine-tuning of model parameters may be required; when the gap is large, the model structure or optimization strategy needs to be reviewed.

[0139] The acquisition method is as follows: Gap calculation result: The gap value between the current system energy consumption and the theoretical optimal solution is obtained from step S3D8. This value is obtained through mathematical calculation and reflects the difference between the actual system energy consumption and the ideal energy consumption.

[0140] Strategy correspondence table: A pre-defined correspondence table between gap ranges and adjustment strategies, stored in the system configuration file or database. This table clearly specifies the specific adjustment strategy corresponding to each gap range, which facilitates quick query and call of the system.

[0141] The general process is as follows: First, obtain the difference value between the current system energy consumption and the theoretical optimal solution from step S3D8. According to the difference value, find the difference range it belongs to in the pre-established strategy correspondence table. After determining the difference range, obtain the adjustment strategy corresponding to the range from the strategy correspondence table.

[0142] Step S3DA, changing the action probability distribution and reward function output by the strategy network according to the determined adjustment strategy.

[0143] Action probability distribution output by the policy network: The probability set of each possible action calculated by the policy network based on the current environment state. In rail transit scenarios, such as the operation strategy of subway trains, the policy network will output the probabilities of different actions such as acceleration, deceleration, and maintaining a constant speed.

[0144] Reward function: It is used to measure the benefits obtained after taking a certain action in a certain state. It is an important basis for guiding the learning of intelligent agents in reinforcement learning models. In energy consumption optimization, a positive reward will be obtained when energy consumption is reduced, and a negative reward will be obtained when energy consumption is increased.

[0145] The general process is as follows: First, the determined adjustment strategy is received from step S3D9. This adjustment strategy is derived based on the gap analysis between the current system energy consumption and the theoretical optimal solution, and is an important basis for subsequent operations.

[0146] In-depth analysis of the impact of the adjustment strategy on the action probability distribution output by the policy network. For example, if the adjustment strategy emphasizes improving the efficiency of train operation, then it is necessary to increase the probability of actions that can achieve this goal (such as optimizing the acceleration and deceleration mode of the train, reasonably allocating power output, etc.) in the action probability distribution. The specific implementation method is to increase the probability values ​​of these beneficial actions by adjusting the parameters of the policy network.

[0147] Make precise changes to the reward function based on the adjustment strategy. For example, in order to encourage the reduction of the idling time of the train, when the action successfully reduces idling, increase the reward value given by the reward function; when the action causes the idling time to increase, increase the penalty, that is, reduce the reward value or give a negative reward.

[0148] Apply the adjusted action probability distribution and reward function to the policy network and the entire reinforcement learning model. This process involves reloading new parameters and configurations to ensure that the model can learn and optimize according to the new rules in subsequent training and decision-making.

[0149] Taking the subway escalator system as an example, assume that the previously determined adjustment strategy is to reduce the energy consumption of the escalator. In the action probability distribution output by the current policy network, the probability of switching the escalator to the low-speed operation mode during low passenger flow periods is 0.4. According to the adjustment strategy, in order to achieve the energy-saving goal more effectively, this probability is increased to 0.7. For the reward function, when the energy consumption of the escalator is reduced by 15%, the reward value is 6, and now it is increased to 10; when the energy consumption does not change, the reward value is reduced from the original 1 to -3. Through these adjustments, the model will more frequently choose to switch the escalator to the low-speed operation mode during low passenger flow periods in the subsequent training process, thereby achieving the goal of reducing the energy consumption of the escalator.

[0150] In step S3DB, the model is retrained based on the adjusted policy network action probability distribution and reward function.

[0151] Adjusted policy network action probability distribution: In step S3DA, the new distribution is obtained by adjusting the original action probability distribution of the policy network according to the adjustment strategy. It reflects the change in the decision-making tendency of the model and guides the model to choose more actions that are conducive to achieving system goals (such as reducing energy consumption and improving operating efficiency) in subsequent training.

[0152] Adjusted reward function: Based on the adjustment strategy of step S3DA, the new function is formed by modifying the original reward function. The new reward function changes the reward values ​​corresponding to different actions and states to more effectively motivate the model to learn the optimal strategy.

[0153] Model training: Using a large amount of training data and optimization algorithms, the model can continuously learn and improve to improve its performance and accuracy in practical applications. In this scenario, model training is to enable the reinforcement learning model to find a better decision strategy based on the adjusted policy network action probability distribution and reward function.

[0154] The general process is as follows: Read the adjusted policy network action probability distribution and reward function from step S3DA to ensure that the data is accurate.

[0155] Prepare training data and preprocess the data, such as cleaning outliers and normalizing, to improve training results.

[0156] Apply the adjusted action probability distribution and reward function to the reinforcement learning model and set the training parameters, such as the number of training rounds, learning rate, etc.

[0157] Start model training. In each round of training, the model selects actions based on the current action probability distribution and obtains reward feedback based on the reward function. Then, the model parameters are updated through optimization algorithms (such as gradient descent) to gradually improve the model performance.

[0158] Regularly evaluate the model during the training process, such as calculating the model's loss value, accuracy and other indicators on the validation set to determine whether the model has converged or achieved the expected performance.

[0159] Step S3DC, after each training cycle, the model performance is evaluated using the validation set data. When the model reaches the set performance index on the validation set, the system collaborative reinforcement learning threshold adjustment model is determined to complete the training.

[0160] Training cycle: The time period during which a model undergoes a complete training process, including multiple rounds of parameter updates and learning. In each training cycle, the model will traverse all training data or perform multiple trainings in specific batches.

[0161] Validation set data: a part of the data divided from the entire data, which is specifically used to evaluate the performance of the model during the training process. It does not participate in the training of the model, and its purpose is to test the generalization ability of the model to unseen data and avoid overfitting of the model.

[0162] Set performance indicators: Model performance measurement standards predetermined according to actual application requirements. For example, in the dynamic monitoring model of rail transit energy efficiency balance, indicators such as energy consumption prediction error within a certain range and action decision accuracy reaching a certain value may be set. When the performance of the model on the validation set reaches these indicators, the model training is considered successful.

[0163] Validation set data: Usually in the data preprocessing stage, the data is divided according to a certain proportion (such as 20%-30%) from the historical operation data of rail transit equipment collected. The division method can be random sampling or stratified sampling to ensure that the validation set data can represent the characteristics of the overall data.

[0164] Performance indicators: Determined by the model developer based on project goals and actual business needs. For example, if the goal is to reduce the energy consumption of the rail transit system by 10%, the average absolute error between the model's predicted energy consumption and the actual energy consumption can be set within 5% as one of the performance indicators.

[0165] The general process is as follows: After each training cycle, the trained model is applied to the validation set data. The model outputs actions according to the state information in the validation set data and the currently learned policy network, and calculates the corresponding rewards.

[0166] Based on the calculation results on the validation set, the performance indicators of the model are calculated, such as the error between the energy consumption predicted by the model and the actual energy consumption in the validation set, and the consistency between the model decision action and the actual optimal action.

[0167] The calculated performance index is compared with the pre-set performance index. If the model reaches the set performance index on the validation set, it means that the model has been trained to the expected level, and the system collaborative reinforcement learning threshold adjustment model is determined to complete the training; if it is not reached, continue to the next round of training cycle, adjust the model parameters, and repeat the training and evaluation process.

[0168] Locating the root causes of energy loss includes: Step S510, acquiring multi-source data, where the multi-source data includes device personalized thresholds, historical energy consumption, operation history, maintenance records, and environmental factors.

[0169] Equipment personalized threshold: The standard value or range of energy consumption, operating parameters, etc. tailored for each device based on the device's own characteristics, operating environment, historical data, etc. For example, the personalized threshold of traction energy consumption for each subway train will vary depending on the model and line characteristics.

[0170] Historical energy consumption: The energy consumption data of the equipment in the past period of time reflects the energy consumption patterns and trends of the equipment. By analyzing the historical energy consumption, we can understand the energy consumption level and energy consumption fluctuations of the equipment during normal operation.

[0171] Operation history: records information such as when the device started and stopped, the operating time, and the operating mode. These data help determine the operating status and stability of the device.

[0172] Maintenance records: Covering the equipment's maintenance time, maintenance content, replaced parts and other information, it can reflect the health status and maintenance status of the equipment, and is of great reference value for analyzing the possible causes of equipment failures caused by abnormal energy consumption.

[0173] Environmental factors: These include the temperature, humidity, altitude and other external environmental conditions in which the equipment operates. These factors may affect the energy consumption of the equipment. For example, in a high temperature environment, the energy consumption of ventilation equipment may increase.

[0174] The general process is as follows: Determine the device scope and time range for which data needs to be obtained, such as obtaining data for the past year for all trains on a subway line.

[0175] Export historical energy consumption and operation history data from the equipment monitoring system and organize them in chronological order and by equipment number.

[0176] Access the equipment management database and extract personalized thresholds and maintenance records for the corresponding equipment to ensure data integrity and accuracy.

[0177] Connect to the environmental monitoring system to collect data on environmental factors in the equipment operation area during the same time period.

[0178] Integrate the acquired multi-source data and store them on a unified data storage platform to facilitate use in subsequent steps.

[0179] Step S520, using historical energy consumption data to calculate the mean and standard deviation to determine the energy consumption boundary, analyzing the historical energy consumption according to multiple dimensions, and establishing a normal energy consumption mode.

[0180] Mean: In statistics, the mean is the sum of a set of data divided by the number of data, representing the average level of the data. In historical energy consumption data, the mean reflects the average energy consumption of the equipment during normal operation.

[0181] Standard Deviation: A statistic that measures the degree of dispersion of data and is used to describe the fluctuation of data relative to the mean. The larger the standard deviation, the greater the fluctuation of energy consumption data; conversely, the smaller the fluctuation.

[0182] Energy consumption boundary: The energy consumption range determined by the mean and standard deviation, generally defined by the mean plus or minus a certain multiple of the standard deviation. The energy consumption within this boundary can be considered as the energy consumption range of the equipment under normal operating conditions.

[0183] Normal energy consumption mode: By analyzing historical energy consumption data according to multiple dimensions (such as time, operating conditions, etc.), the laws and patterns of equipment energy consumption under normal circumstances are summarized, which is an important basis for judging whether the energy consumption is abnormal.

[0184] The general process is as follows: Extract historical energy consumption data from equipment monitoring systems to ensure data integrity and accuracy.

[0185] Use statistical methods to calculate the mean and standard deviation of historical energy consumption data. For example, use the mean() and std() functions in Python's NumPy library for calculation.

[0186] The energy consumption boundary is determined based on the calculated mean and standard deviation, which can generally be set as the mean ± 2 times the standard deviation (the specific multiple can be adjusted based on actual conditions).

[0187] Analyze historical energy consumption data from the time dimension (such as day, week, month, quarter, etc.) and operating condition dimension (such as full load, no load, different speed ranges, etc.), and draw energy consumption trend charts, energy consumption distribution histograms, etc.

[0188] By summarizing and generalizing the results of multi-dimensional analysis, a normal energy consumption model is established to clarify the normal energy consumption range and change pattern of equipment at different times and working conditions.

[0189] Step S530, sort out the failure modes of the equipment from the system to the components, determine the probability of occurrence of each subdivided mode, build a dynamic fault tree with the energy consumption exceeding the normal boundary preset proportion as the top event, use logic gates to describe the cause and effect relationship, and then find out all possible fault paths through qualitative analysis, and regard the events whose occurrence frequency exceeds the preset proportion of the total number of failures as high-frequency fault bottom events, and use the paths of high-frequency fault bottom events as the troubleshooting direction.

[0190] Failure modes: various possible failure manifestations of equipment from the system level to the component level. For example, the failure modes of the braking system of a subway train may include excessive wear of the brake pads, brake fluid leakage, etc.

[0191] Probability of occurrence: The probability of each failure mode occurring during equipment operation, usually obtained through statistical analysis of the equipment's historical failure data.

[0192] Top event: In the fault tree analysis, the final fault event that is of concern and whose cause needs to be analyzed. Here, the energy consumption exceeding the normal boundary by a preset proportion is regarded as the top event, for example, the energy consumption exceeds the normal boundary by 20%.

[0193] Dynamic fault tree: A logical model used to describe the cause-effect relationship of system faults. Unlike traditional fault trees, it can consider the time sequence and dynamic characteristics of events. The cause-effect relationship between each fault event is represented by logic gates (such as AND gates, OR gates, etc.).

[0194] Fault path: Starting from the top event, trace back through logic gates to a series of bottom events that lead to the top event. Each fault path represents a fault combination that may lead to the top event.

[0195] High-frequency fault bottom events: Among all fault paths, bottom events whose occurrence frequency exceeds a preset percentage (such as 30%) of the total fault frequency. These bottom events are common causes of equipment failure and are the focus of troubleshooting.

[0196] The acquisition methods are as follows: 1. Historical equipment failure data: obtained from equipment maintenance records and fault repair systems, recording various types of equipment failure information that occurred in the past, including failure time, failure phenomenon, failure cause, etc. 2. Industry statistical data: obtained by consulting industry reports, communicating with other equipment operating units of the same type, etc., to supplement and verify the failure probability analysis of the unit's equipment.

[0197] The general process is as follows: Comprehensively sort out all possible failure modes of the equipment from systems to components, and refer to the equipment's technical manual, maintenance experience, etc.

[0198] Based on historical equipment failure data and industry statistics, the probability of occurrence of each failure mode is calculated.

[0199] Determine that the energy consumption that exceeds the preset proportion of the normal boundary is the top event. Take the top event as the starting point, use logic gates (such as "AND gate" to indicate that multiple events occur simultaneously to cause the top event, and "OR gate" to indicate that as long as one event occurs, it will cause the top event) to connect various fault modes and build a dynamic fault tree.

[0200] Perform qualitative analysis on the dynamic fault tree and find out all possible fault paths through logical reasoning.

[0201] Count the number of occurrences of bottom events in each fault path, calculate their proportion to the total number of faults, identify bottom events that exceed the preset proportion as high-frequency fault bottom events, and use the paths of these high-frequency fault bottom events as the main troubleshooting direction.

[0202] Step S540, convert the fault tree events into Bayesian network nodes and edges, determine the prior probability based on historical equipment failures and industry statistics, input multi-source data as evidence, use Bayesian theorem to update the posterior probability, and set the probability change to exceed 0.1 to consider that the possibility of the fault cause has increased significantly, so as to calculate the fault probability.

[0203] Bayesian network: A graphical network model based on probabilistic reasoning, consisting of nodes and edges. Nodes represent random variables, which can be equipment failure events in energy loss analysis; edges represent the dependencies between variables, which are quantified through conditional probability tables. It can effectively handle uncertain information and update the judgment of the probability of event occurrence through known information.

[0204] Prior probability: A preliminary estimate of the probability of an event occurring based on past experience or statistical data without additional evidence. For example, the probability of a certain failure occurring is estimated in advance based on historical equipment failures and industry statistics.

[0205] Posterior probability: The probability of an event occurring after obtaining new evidence (such as the current operation data of the equipment, environmental data, and other multi-source data). It more accurately reflects the possibility of an event occurring under the current circumstances.

[0206] The general process is as follows: The events in the fault tree are converted one by one into nodes of the Bayesian network, and the causal relationships between events in the fault tree are converted into edges between the nodes of the Bayesian network.

[0207] Based on historical equipment failures and industry statistics, the prior probability of each node (failure event) is determined, and a corresponding conditional probability table is established to describe the dependencies between nodes.

[0208] The multi-source data obtained from step S510 is input into the Bayesian network as evidence.

[0209] Bayesian theorem is used for probabilistic reasoning to update the posterior probability of each node.

[0210] A probability change threshold is set, such as 0.1. When the posterior probability of a fault event changes by more than this threshold relative to the prior probability, it is considered that the possibility of the cause of the fault has increased significantly and it will be the focus of attention.

[0211] Step S550, integrating the logical structure of the fault tree and the failure probability of the Bayesian network, establishing an evaluation matrix to score the failure mode from two dimensions, and screening factors whose comprehensive scores exceed the preset scores and whose energy consumption increases exceed the preset ratio as the root causes.

[0212] Evaluation matrix: A tool for comprehensive analysis and evaluation, presented in a table format. In this step, the rows and columns of the evaluation matrix represent different failure modes and evaluation dimensions, respectively. By scoring each failure mode in each dimension, the key failure factors can be intuitively compared and screened out.

[0213] Fault mode scoring: Each fault mode is quantitatively evaluated from two dimensions: the logical structure of the fault tree and the fault probability calculated by the Bayesian network. The scoring of the fault tree logical structure dimension can be based on the complexity of the fault path, the number of key devices involved, etc.; the fault probability dimension directly refers to the posterior probability calculated by the Bayesian network.

[0214] Comprehensive score: The weighted sum of the scores of each failure mode in different dimensions in the evaluation matrix. The weights are set according to actual needs and the importance of each dimension. The comprehensive score can comprehensively measure the impact of each failure mode on energy loss.

[0215] The general process is as follows: An evaluation matrix is ​​constructed, in which the rows are various failure modes and the columns are the two evaluation dimensions of the fault tree logical structure and the Bayesian network failure probability.

[0216] For each fault mode, in the dimension of the fault tree logical structure, the complexity of its fault path is analyzed. For example, complex fault paths involving multiple "AND gates" are given lower scores; simple paths that directly lead to top events are given higher scores. In the dimension of fault probability, scores are given according to the size of the posterior probability calculated by the Bayesian network. The higher the probability, the higher the score.

[0217] Set weights for the two evaluation dimensions, for example, the fault tree logic structure weight is set to 0.4, and the Bayesian network failure probability weight is set to 0.6. Calculate the comprehensive score of each fault mode, the formula is: comprehensive score = fault tree logic structure score × 0.4 + Bayesian network failure probability score × 0.6.

[0218] Set a preset score for the comprehensive score, such as 60 points, and set a preset ratio for energy consumption increase, such as 15%. Filter out the factors corresponding to the failure mode whose comprehensive score exceeds the preset score and causes energy consumption increase exceeding the preset ratio, and identify these factors as the root causes of energy loss.

[0219] Taking the subway ventilation system as an example, an evaluation matrix is ​​constructed, in which the failure modes include fan blade damage, motor burnout, and air duct blockage. In the fault tree logical structure dimension, the fan blade damage fault path is simple, with 8 points; the motor burnout involves multiple components, and the fault path is complex, with 4 points; the air duct blockage fault path is relatively simple, with 7 points. In the Bayesian network fault probability dimension, according to the previous calculation, the posterior probability of fan blade damage is 0.15, corresponding to 3 points; the posterior probability of motor burnout is 0.1, with 2 points; the posterior probability of air duct blockage is 0.38, with 7 points. According to the above weights, the comprehensive score is calculated, and the comprehensive score of fan blade damage = 8×0.4+3×0.6=5; the comprehensive score of motor burnout = 4×0.4+2×0.6=2.8; the comprehensive score of air duct blockage = 7×0.4+7×0.6=7. Assuming the preset score is 5 points and the preset increase ratio of energy consumption is 15%, it is found that the blockage of the air duct causes the energy consumption to increase by 20%, which exceeds the preset ratio. Therefore, factors related to the blockage of the air duct are identified as the root cause of energy loss.

[0220] A method for dynamic monitoring of rail transit energy efficiency balance also includes steps in parallel with sending the information to a terminal held by a responsible person, specifically as follows: Step S700, input the root cause of the positioning, relevant data of the equipment and the industry standard information of the equipment into the solution analysis model, and output the solution, wherein the solution includes problem description, solution overview, implementation steps, expected results and risk assessment.

[0221] Solution analysis model: An intelligent model based on data analysis and algorithms that can generate targeted solutions based on various types of input information. The model has been trained with a large amount of data and has the ability to analyze problems and provide effective solution strategies.

[0222] Problem description: Provide a detailed description of the root cause of the energy loss that has been located, including the equipment where the fault occurred, the fault phenomenon, the scope of impact, etc., so that relevant personnel can quickly understand the nature of the problem.

[0223] Solution Overview: Briefly summarize the overall ideas and main methods for solving energy loss problems, and provide a macro solution framework.

[0224] Implementation steps: Break down the solution into specific operational steps, clarify the execution order, responsible person and time node of each step, and ensure that the solution is feasible.

[0225] Expected effect: predict the possible effects that may be achieved in terms of reducing energy loss and improving equipment operating efficiency after implementing the solution, providing a basis for evaluating the effectiveness of the solution.

[0226] Risk assessment: Identify and analyze the risks that may be encountered in the implementation of the solution, including technical risks, operational risks, cost risks, etc., and propose corresponding response measures.

[0227] The general process is as follows: 1. Extract the root cause of energy loss located in step S550. 2. Collect relevant data of the equipment from the equipment monitoring system and equipment management database, such as operating time, number of failures, energy consumption fluctuations, etc. 3. Obtain industry standard information of the equipment through Internet search, industry conference materials, etc., including energy consumption standards, operating parameter standards, etc. 4. Organize the root cause, equipment-related data and industry standard information into a standardized format and input them into the solution analysis model. 5. Based on the input information, the model uses internal algorithms and logic to generate solutions, including problem description, solution overview, implementation steps, expected results and risk assessment.

[0228] Step S800: Send the solution to the terminal held by the person in charge.

[0229] Terminal held by person in charge: usually refers to electronic devices used by personnel responsible for the management, maintenance or operation of rail transit equipment, such as mobile phones, tablets, laptops, etc., to receive and view important work-related information.

[0230] Considering that before sending the solution to the terminal held by the person in charge, the solution can also be simulated and verified, and timely adjustments can be made if there are any problems, so as to better ensure that the solution received by the person in charge meets the needs of the scenario. Therefore, after outputting the solution, the following steps should be taken: Step SA00: Build a simulation environment similar to the actual one on the simulation platform, use digital twin and 3D modeling technology to create a virtual scene based on the actual rail transit information. Import detailed equipment parameters, historical operation data and environmental data to ensure that the simulation environment truly reflects the actual situation.

[0231] Among them, the simulation platform: professional software such as OMNeT++ and SUMO are selected. OMNeT++ is good at simulating communication networks and can simulate the communication between trains and dispatching centers in rail transit; SUMO focuses on traffic flow simulation and can present changes in train running tracks.

[0232] Digital twin technology: Create a virtual model corresponding to the real device or system to reflect its status in real time. For example, build a digital twin model for each subway train and update the virtual model status based on the real-time data of the train sensors (such as speed and position).

[0233] 3D modeling technology: used to build realistic virtual scenes, covering rail lines, stations and equipment. For example, building a 3D model of a subway station accurately presents the location and appearance of platforms, channel layouts, and escalators and other equipment.

[0234] Detailed equipment parameters: including equipment physical and operating parameters. They can be obtained from equipment technical documents, such as the rated power of train traction motors; they can also be collected by equipment sensors, such as train brake pad wear data. For example, the rated power of a train traction motor is 3000kW, which affects the train power and energy consumption simulation.

[0235] Historical operation data: from the rail transit operation database, including equipment operation status and energy consumption data in different time periods (weekdays, peak hours, etc.) and working conditions (normal, fault, etc.). For example, historical data of a station lighting system shows that the lighting system is turned on for a long time and consumes a lot of energy during peak hours, providing a reference for simulation.

[0236] Environmental data: Contains meteorological and geographic information. Temperature and humidity are obtained through the meteorological data interface; altitude and track slope are obtained from the geographic information system. For example, a section of track has a high altitude and a large slope, which affects the energy consumption of the train and needs to be accurately set during simulation.

[0237] In step SB00, a simulation test scheme based on orthogonal experimental design is adopted, multiple levels are set based on working conditions, environment and fault scenarios, and a discrete event simulation algorithm is used to collect energy consumption and equipment operation parameter data at a frequency of once per second. Each test combination is repeated 10 times and the average value is taken.

[0238] Orthogonal test design: This is a scientific test design method. By rationally arranging multi-factor and multi-level test combinations, comprehensive and representative results can be obtained in a relatively small number of tests. For example, in rail transit simulation, factors such as operating conditions, environment, and fault scenarios are considered, and different levels are set for each factor. The operating conditions can be divided into three levels: peak, flat, and low; the temperature in the environmental factors can be set to three levels: high temperature, normal temperature, and low temperature; the fault scenarios can include different types such as failure of key train components and signal system failures as levels. Through the combination of orthogonal tables, it is ensured that the test covers various possible situations, effectively reducing the test workload, while ensuring the reliability and effectiveness of the results.

[0239] Discrete event simulation algorithm: This algorithm treats various activities in the rail transit system as discrete events, such as the arrival and departure of trains, the start and stop of equipment, the occurrence of faults, etc. It simulates based on the time sequence of events, and accurately simulates the dynamic changes of the system over time by building event queues and corresponding processing logic. For example, when the simulated train is running, after the departure event of the train is triggered, according to the predetermined operation plan and speed curve, in the process of advancing the simulation time, the train acceleration, uniform speed, deceleration and other events, as well as equipment failures or environmental changes that may occur on the way are processed in sequence, so as to truly reproduce the operation process of the rail transit system.

[0240] Data collection: During the simulation process, energy consumption and equipment operation parameter data are collected at a frequency of once per second. Energy consumption data covers train operation energy consumption, station equipment energy consumption, etc., such as train traction energy consumption, station lighting and ventilation equipment power consumption, etc. Equipment operation parameters include train speed, acceleration, compartment temperature, equipment operating current, voltage, speed, etc. These data are mainly obtained through the built-in data acquisition module of the simulation platform, which is automatically collected according to the set frequency and parameter category. Each test combination is repeated 10 times and the average value is taken. The purpose is to reduce the impact of random errors in the simulation process and make the collected data more stable and representative, thereby providing a reliable basis for subsequent analysis and program adjustment.

[0241] Step SC00, use deep neural network (DNN) to build a multi-layer perceptron (MLP) structure model, use mean square error (MSE) as loss function, and train with stochastic gradient descent (SGD) algorithm to compare simulation results with expected results to find out the difference. If there is a difference, use genetic algorithm to optimize and adjust the solution, use implementation steps and technical parameters as genes, iterate through selection, crossover, and mutation operations to optimize, and adjust the solution according to the optimal solution.

[0242] Deep Neural Network (DNN) and Multilayer Perceptron (MLP): DNN is a type of neural network with multiple hidden layers that can automatically learn complex patterns and relationships in data. MLP is a common structure of DNN, which consists of an input layer, multiple hidden layers, and an output layer. In this step, the input layer receives simulated collected energy consumption and equipment operation parameter data, such as the speed and energy consumption values ​​of the train at different times. The hidden layer processes and extracts features from the data through nonlinear activation functions, and the output layer outputs relevant results compared with the expected effects. For example, the speed, acceleration, and energy consumption data of a train over a period of operation are input, and after learning and calculation, the MLP outputs a judgment result on whether the energy consumption of the train meets expectations.

[0243] Mean Squared Error (MSE) and Stochastic Gradient Descent (SGD): MSE is a measure of the error between the model's predicted value and the true value. The accuracy of the model is evaluated by calculating the average of the squares of the difference between the predicted value and the true value. In this scenario, it is used to measure the degree of deviation between the simulation results and the expected results. SGD is an optimization algorithm. When training the MLP model, a small portion of data (a batch) is randomly selected each time to calculate the gradient, and the model parameters are updated accordingly, so that the model is continuously optimized to reduce the MSE value. For example, the MSE value of the initial model's predicted train energy consumption and the actual energy consumption is high. After multiple SGD iterations to update the model parameters, the MSE value gradually decreases and the model prediction accuracy improves.

[0244] Genetic algorithm: This is an optimization algorithm based on natural selection and genetic mechanisms. Acquisition method and quantification: Implemented using Python's genetic algorithm library (such as DEAP). First, determine the encoding method of the gene. For example, for the set value of the train speed, real number encoding can be used, and the value range is set according to the actual situation (such as 0-120km / h). Define the fitness function and calculate the fitness of the individual according to the degree of difference between the simulation result and the expected effect. For example, if the expected energy consumption is reduced by 10%, and the simulation result corresponding to a certain individual is only reduced by 5%, then its fitness is relatively low. Set the parameters of the genetic algorithm, such as population size (such as 50 individuals), crossover probability (such as 0.8), mutation probability (such as 0.01), etc. In the iterative process, each generation will select, cross and mutate individuals according to fitness, and after multiple generations (such as 100 generations) of evolution, a better solution is found.

[0245] For example, in the verification of the energy efficiency solution of a certain subway line, the data such as train energy consumption and equipment operating parameters obtained in different time periods are input into the constructed MLP model, and the model is trained by the SGD algorithm to calculate the MSE. If it is found that in a certain simulation scenario, the predicted value of the train energy consumption during peak hours deviates greatly from the expected value, and the MSE reaches 0.3. At this time, the genetic algorithm is used to optimize the train operation plan and equipment control parameters in the solution. Assume that the train's original planned stop time at a certain station is 30 seconds. After genetic algorithm optimization, the stop time is adjusted to 25 seconds, and the operating power of the equipment is also adjusted accordingly. Simulate and verify again to observe whether the MSE is reduced and whether the simulation results are closer to the expected effect. After multiple iterative adjustments, until the MSE is reduced to an acceptable range (such as below 0.05), and the simulation results meet the expected goals of reducing energy consumption and improving equipment operating efficiency, the optimized solution is finally determined.

[0246] Based on the same inventive concept, an embodiment of the present invention provides a rail transit energy efficiency balance dynamic monitoring system, including a memory and a processor, wherein the memory stores a program that can be executed on the processor to implement the following Figures 1 to 3 Procedure for either method.

[0247] The embodiments of this specific implementation method are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, all equivalent changes made based on the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for dynamic monitoring of rail transit energy efficiency balance, characterized in that: include: Get relevant data of various external devices; Input relevant data of various external devices into the pre-built fuzzy clustering classification model, and output the classification results of various external devices; Based on the equipment classification results and the real-time related data of various external equipment, analyze whether there is any change in equipment status; If not, use the life cycle cost analysis method to determine the personalized energy efficiency standard thresholds for each type of equipment; If yes, the device whose status has changed is taken as the target device, and the personalized energy efficiency standard threshold of the target device is determined by using the life cycle cost analysis method as the initial personalized threshold, and various information of the target device is collected, and various information of other devices that have a collaborative relationship with the target device and the existing collaborative relationship data are collected; Inputting the initial personalized threshold of the target device, various information of the target device, various information of other devices that have a collaborative relationship with the target device, and existing collaborative relationship data into a pre-trained system collaborative reinforcement learning threshold adjustment model, and outputting an adjustment value; Based on the adjustment value and the initial personalized threshold, calculate and determine the personalized energy efficiency standard threshold of the target device; According to the comparison result between the actual energy consumption of the equipment and the personalized energy efficiency standard threshold, determine whether the corresponding equipment has energy consumption abnormality; If yes, then the system combines the personalized threshold of the equipment and the historical energy consumption data, uses the built comprehensive energy loss analysis model based on fault tree and Bayesian network, starts from the equipment failure mode, combines the equipment operation history, maintenance records, and environmental factors, and locates the root cause of energy loss through Bayesian network reasoning and dynamic fault tree analysis, and sends it to the terminal held by the person in charge; If no, no notification will be made.

2. A rail transit energy efficiency balance dynamic monitoring method according to claim 1, characterized in that: Based on the real-time data of various external devices, analyze whether there are any changes in the device status, including: Obtain the scenario category of the external device, where the scenarios include daily operation scenarios, peak operation scenarios, equipment maintenance scenarios, extreme weather scenarios, and equipment upgrade and transformation scenarios; According to the correspondence between the scene category of the external device and the device state analysis algorithm, the device state analysis algorithm adopted by different external devices is matched; Based on the real-time relevant data of external devices, a matching device status analysis algorithm is used to analyze whether there is any device status change in the external device.

3. A rail transit energy efficiency balance dynamic monitoring method according to claim 1, characterized in that: The construction process of the system collaborative reinforcement learning threshold adjustment model is as follows: Collect historical operation data of rail transit equipment, covering the operation status, energy consumption data and corresponding system energy consumption of each equipment under different time periods and working conditions; Perform data preprocessing on the collected historical operation data of rail transit equipment, wherein the data preprocessing includes data cleaning and normalization processing; Initialize the threshold adjustment model of system collaborative reinforcement learning. The initial operation is to randomly assign values ​​to the weight matrix and bias vector of the policy network and value network. Starting from the initial state, the policy network gives the action probability distribution according to the current state, uses the roulette algorithm to select actions to be applied to the simulation environment, records the state, action, reward and next state information, and collects a preset number of records to form a trajectory set; Using the generalized advantage estimation method, first calculate the TD error, then calculate the advantage function to judge the quality of the action; The PPO-Clip objective function in the proximal policy optimization algorithm is used to calculate the policy loss, and the parameters are adjusted by gradient descent to minimize the policy loss. Use the mean square error loss function to calculate the value loss and update the parameters by gradient descent; Run a mixed integer linear programming model to calculate the theoretical optimal solution for system energy consumption based on the physical characteristics of the rail transit system and the equipment operation constraints, and calculate the gap between the current system energy consumption and the theoretical optimal solution; Determine the adjustment strategy according to the corresponding relationship between the gap range between the current system energy consumption and the theoretical optimal solution and the adjustment strategy; Changing the action probability distribution and reward function output by the policy network according to the determined adjustment strategy; Retrain the model based on the adjusted policy network action probability distribution and reward function; After each training cycle, the model performance is evaluated using the validation set data. When the model reaches the set performance indicators on the validation set, the system collaborative reinforcement learning threshold adjustment model is determined to complete the training.

4. A rail transit energy efficiency balance dynamic monitoring method according to claim 1, characterized in that: Locating the root causes of energy loss includes: Obtain multi-source data, including device personalized thresholds, historical energy consumption, operation history, maintenance records, and environmental factors; Use historical energy consumption data to calculate the mean and standard deviation to determine energy consumption boundaries, analyze historical energy consumption in multiple dimensions, and establish a normal energy consumption model; Sort out the failure modes of the equipment from the system to the components, determine the probability of occurrence of each subdivided mode, build a dynamic fault tree with the energy consumption exceeding the preset proportion of the normal boundary as the top event, use logic gates to describe the cause and effect relationship, and then find all possible fault paths through qualitative analysis. Events with a frequency exceeding the preset proportion of the total number of faults are regarded as high-frequency fault bottom events, and the paths of high-frequency fault bottom events are used as the troubleshooting direction; Convert fault tree events into Bayesian network nodes and edges, determine prior probabilities based on historical equipment failures and industry statistics, use multi-source data as evidence input, and use Bayesian theorem to update posterior probabilities. When the probability change exceeds 0.1, it is considered that the possibility of the fault cause has increased significantly, and the fault probability is calculated accordingly. By integrating the logical structure of the fault tree and the failure probability of the Bayesian network, an evaluation matrix is ​​established to score the failure modes from two dimensions, and factors with comprehensive scores exceeding the preset scores and energy consumption increases exceeding the preset proportions are screened as root causes.

5. A method for dynamic monitoring of rail transit energy efficiency balance according to any one of claims 1 to 4, characterized in that: It also includes steps parallel to sending to the terminal held by the responsible person, as follows: Input the root cause of the location, relevant data of the equipment, and industry standard information of the equipment into the solution analysis model, and output the solution, where the solution includes problem description, solution overview, implementation steps, expected results, and risk assessment; The solution is sent to the terminal held by the person in charge.

6. A rail transit energy efficiency balance dynamic monitoring system, characterized in that: The invention comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein the program can realize a method for dynamically monitoring the energy efficiency balance of rail transit as claimed in any one of claims 1 to 5 when loaded and executed by the processor.

Images