Building automatic control system intelligent regulation and control method and system based on AI

Abstract

The invention provides an AI-based intelligent regulation and control method and system for a building automatic control system, and the method comprises the steps: carrying out the feature space conversion of state variable monitoring data of the building automatic control system through an adaptive learning network model, and generating a first state feature space matched with a feature mode extracted by a building state recognition network; according to the method, the building automatic control system regulation and control network can be guided to perform adaptive regulation and control decision on a plurality of different fault disturbance events according to the first state feature space, and a plurality of adaptive regulation and control decision results associated with building operation categories are generated, so that actual adaptive regulation and control can be performed on the building automatic control system, and the service life of the building automatic control system can be prolonged. The method is further used for optimizing the building state recognition network, so that a closed-loop intelligent regulation and control and optimization system is formed. According to the method, the coping capacity and the regulation and control precision of the building automatic control system for the fault disturbance event are improved, the self-adaptability and the intelligent level of the system are enhanced, and the operation efficiency and the energy utilization efficiency of the building are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology. Specifically, it relates to an intelligent regulation method and system for a building automation control system based on AI. Background Art

[0002] With the acceleration of the urbanization process, buildings, as an important part of the city, their energy consumption and operation efficiency have received increasing attention. As the core of building management, the building automation control system is responsible for monitoring and regulating various facilities in the building, such as air conditioners, lighting, elevators, etc., to ensure the normal operation of the building and energy conservation and emission reduction. However, traditional building automation control systems often perform regulation based on fixed control logic and preset parameters, and it is difficult to adapt to the complex and changeable situations in the actual operation process of the building.

[0003] During the operation of the building, various fault disturbance events occur from time to time, such as equipment failures, environmental changes, personnel flow, etc. These events will all affect the state of the building, and thus affect the regulation effect of the building automation control system. Traditional building automation control systems lack the ability of adaptive regulation for these fault disturbance events, and often require manual intervention or adjustment of control parameters after the events occur. This not only increases the management difficulty, but also may lead to waste of energy consumption and decline in building operation efficiency.

[0004] In recent years, with the rapid development of artificial intelligence technology, applying AI technology to building automation control systems has become a new trend. By utilizing the powerful learning ability and adaptive ability of AI technology, intelligent regulation of the building automation control system can be achieved, improving the self - adaptability and intelligent level of the system. However, existing intelligent regulation methods for building automation control systems based on AI often have some problems, such as inaccurate feature extraction, inflexible regulation decisions, etc., which limit their effects in practical applications. Summary of the Invention

[0005] In view of the above - mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide an intelligent regulation method for a building automation control system based on AI. The method includes:

[0006] Obtain the state variable monitoring data of the building automation control system;

[0007] The adaptive learning network model performs feature space transformation on the state variable monitoring data to generate a first state feature space that matches a first feature pattern, where the first feature pattern is the feature pattern that the state feature space extracted by the building state recognition network matches;

[0008] According to the first state feature space, guide the building automation system control network to make adaptive control decisions for different fault disturbance events respectively, generate multiple different adaptive control decision results of the building operation categories associated with the first state feature space, and the multiple different adaptive control decision results of the building operation categories associated with the first state feature space are also used to optimize the building state recognition network;

[0009] Based on the multiple different adaptive control decision results of the building operation categories associated with the first state feature space, make an adaptive control decision for the building automation system.

[0010] On the other hand, an embodiment of the present invention further provides an AI-based intelligent control system for a building automation system, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present application performs feature space conversion on the state variable monitoring data of the building automation system through an adaptive learning network model, and generates a first state feature space that matches the feature pattern extracted by the building state recognition network. This method can guide the building automation system control network to make adaptive control decisions for multiple different fault disturbance events according to the first state feature space, and generate multiple adaptive control decision results associated with the building operation categories. These decision results are not only used for actual adaptive control of the building automation system, but also further used to optimize the building state recognition network, thereby forming a closed-loop intelligent control and optimization system. This method improves the ability of the building automation system to respond to fault disturbance events and the control accuracy, enhances the self-adaptability and intelligent level of the system, and effectively improves the operation efficiency and energy utilization efficiency of the building. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a schematic flowchart of the execution of the AI-based intelligent control method for a building automation system provided by an embodiment of the present invention.

[0013] Figure 2 is a schematic hardware architecture diagram of the AI-based intelligent control system for a building automation system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification, Figure 1It is a schematic flowchart of an intelligent regulation method for a building automation system based on AI provided by an embodiment of the present invention. The intelligent regulation method for the building automation system based on AI will be introduced in detail below.

[0015] Step S110, obtain the status variable monitoring data of the building automation system.

[0016] In this embodiment, in the building automation system of a large commercial office building, the acquisition of status variable monitoring data is the basis for effective control and management. The building automation system covers many subsystems, such as the heating, ventilation, and air conditioning (HVAC) system, lighting system, elevator system, and security system. For the HVAC system, the status variable monitoring data includes the temperature values collected by temperature sensors in each area, and these temperature values reflect the real-time temperature conditions of different floors, different rooms, or different office areas. For example, in the lobby on the first floor of the office building, the temperature sensor collects temperature data every certain period (such as 5 minutes). It is possible that the temperature in the lobby is relatively low before the morning rush hour, and as people enter and equipment is turned on, the temperature gradually rises.

[0017] The humidity sensor in the air conditioning system also provides humidity monitoring data. Humidity varies significantly under different seasons and weather conditions. In the humid weather in summer, the humidity may be relatively high, while in the dry weather in winter, the humidity may be relatively low. In addition, equipment operation parameters such as the fan speed and water valve opening in the air conditioning system are also monitored as status variables. The fan speed affects the air circulation speed, and the water valve opening determines the flow rate of cold or hot water, thereby affecting the cooling or heating effect.

[0018] In terms of the lighting system, illuminance sensors are installed on each floor, in each room, or in each public area to collect illuminance data. During the day, the areas near the windows may have sufficient natural light and high illuminance, while the internal areas far from the windows may require more artificial lighting to meet the needs of work or activities. At the same time, the switch status and dimming ratio of the lamps in the lighting system are also monitored status variables. For example, in some meeting rooms, the brightness and switch combination of the lights may need to be adjusted according to different meeting scenarios (such as projection presentations or group discussions).

[0019] In the elevator system, the running speed of the elevator, the floors it stops at, the passenger capacity in the car, and the opening and closing status of the elevator doors are all important status variable monitoring data. For example, during peak hours, the elevator stops at each floor frequently, with a large passenger capacity, and the running speed may be adjusted appropriately according to the load of the elevator to ensure safe and efficient operation.

[0020] The security system includes devices such as access control systems and surveillance cameras. The access control system records data such as the opening and closing times of each door and the verification results of personnel access permissions. The surveillance cameras provide video image data of each surveillance area. Although the video images themselves are a complex data form, state variables such as the direction of personnel flow and the number of personnel can be extracted through image analysis techniques. These data can help managers understand the personnel activities within the building, detect abnormal behaviors in a timely manner, and take corresponding countermeasures.

[0021] By obtaining various state variable monitoring data from different subsystems, the building automation system can comprehensively understand the operating state of the building and provide data support for subsequent analysis, decision-making, and regulation.

[0022] Step S120: The adaptive learning network model performs feature space transformation on the state variable monitoring data to generate a first state feature space that matches the first feature pattern, where the first feature pattern is the feature pattern that the state feature space extracted by the building state recognition network matches.

[0023] Continuing with the example of the building automation system of the above commercial office building. Assume that the adaptive learning network model has undergone preliminary training and has the ability to perform feature space transformation on different types of state variable monitoring data.

[0024] For the state variable monitoring data of the HVAC system, the adaptive learning network model first analyzes the temperature data. Specifically, it not only focuses on the specific temperature values but also considers factors such as the change trend of temperature over different time periods and the correlation of temperatures in different areas. For example, in an office area, if the temperatures in several adjacent rooms show similar rising or falling trends, this may indicate the existence of common influencing factors, such as the overall operating state of the air conditioning unit or the impact of the external climate. The model will integrate these feature information about temperature and transform it from the original numerical space to a new feature space. This new feature space can better reflect the state characteristics of temperature in the entire building's air conditioning system. For example, the temperature values are mapped to a feature space representing temperature stability, temperature uniformity, and the relationship between temperature and comfort.

[0025] For the humidity data, the adaptive learning network model will perform feature space transformation by combining the temperature data and information such as the building structure and personnel density of the building. For example, in an office area with good airtightness and high personnel density, the change in humidity may be closely related to the water vapor exhaled by personnel and the dehumidification ability of the air conditioning system. The model will relate the humidity data to these relevant factors and construct a feature space that includes the mutual relationship between humidity and other factors. This feature space may reflect characteristics such as the impact degree of humidity on the comfort of the building and the operation of equipment under different environmental conditions.

[0026] When the illuminance data in the lighting system is processed by the adaptive learning network model, it will be combined with factors such as time information (such as day or night), personnel activity information (such as whether there is a meeting arrangement in a certain area), and the orientation of the building. For example, for a south-facing meeting room in the afternoon on a sunny day, with sufficient natural lighting, the high illuminance data collected by the illuminance sensor may be associated with the usage plan of the meeting room (whether there is a need for projection presentation) in the transformed feature space. If there is a need for projection presentation, then the high illuminance may be marked as a status feature that needs to be adjusted, while when there is no need for projection presentation, the high illuminance may be regarded as a normal energy-saving status feature.

[0027] In the process of feature space transformation for the status variable monitoring data of the elevator system, the adaptive learning network model will consider the matching degree between the operating rules of the elevator and the personnel flow rules. For example, during the morning rush hour, the frequent stops and high passenger capacity of the elevator are normal operating status features. The model will transform data such as the operating speed, stop floors, and passenger capacity into a feature space that can reflect the service efficiency, operating stability, and matching degree with personnel needs of the elevator. If a certain elevator has frequent failures or abnormal operating speeds during the peak period, these features will be prominently displayed in the transformed feature space for subsequent analysis and decision-making.

[0028] For the access control data in the security system and the personnel flow data obtained from video image analysis, the adaptive learning network model will combine them with factors such as the functional zoning of the building and the security levels of different areas for feature space transformation. For example, in the core office area of an office building, the strict verification results of the access control system and the relatively small personnel flow may be regarded as normal security status features. While in public areas such as the lobby or restaurant, the relatively frequent personnel flow is a normal activity status feature. Through such feature space transformation, the adaptive learning network model can generate a first state feature space that matches the first feature pattern. This first state feature space can comprehensively and accurately reflect the feature relationships of each subsystem of the building under the overall operating state, and is consistent with the feature pattern matched by the state feature space extracted by the building state recognition network, providing a suitable data basis for subsequent fault analysis and regulation decision-making.

[0029] Step S130, according to the first state feature space, guide the building automation system control network to respectively make adaptive regulation decisions on multiple different fault disturbance events for fault disturbance propagation, generate multiple different adaptive regulation decision results of the building operation categories associated with the first state feature space, and the multiple different adaptive regulation decision results of the building operation categories associated with the first state feature space are also used to optimize the building state recognition network.

[0030] Suppose multiple different fault disturbance events occur in the building automation system of a commercial office building.

[0031] For the HVAC system, one possible fault disturbance event is that the refrigeration efficiency of the chiller suddenly drops. According to the first state feature space, which contains feature information such as the temperature and humidity of each area and the operating parameters of air conditioning equipment. The decision optimization unit in the building automation system's regulation network starts to make adaptive regulation decisions for fault disturbance propagation.

[0032] In the first round of iteration, the decision optimization unit uses the first state feature space as the basis for fault diagnosis and obtains information related to the fault disturbance event of the chiller's refrigeration efficiency drop, including information such as the type of chiller and the air conditioning terminal equipment (such as fan coils) that may be affected by the fault. At the same time, initialize the coefficient for measuring the possibility of fault propagation. For example, according to the connection relationship between the chiller and the air conditioning terminal equipment and past fault experience, set the possibility coefficient of the chiller fault propagating to a certain fan coil to be 0.3; initialize the initial value of the index for evaluating the effect of the regulation decision. For example, set the initial evaluation value of controlling the indoor temperature within the comfortable range to be 0.8. Then, based on the initial state information of the chiller, analyze the propagation path of the fault in the current building automation system environment and find that the drop in the chiller's refrigeration efficiency will cause the temperature of the chilled water transported through the pipeline to rise, which in turn affects the refrigeration effect of the fan coils connected to it. The connection relationship between the components in the air conditioning system shows that the chilled water will sequentially affect the fan coils on each floor along the pipeline. The data flow direction is from the chiller to each fan coil, and the interaction mechanism is that the temperature change of the chilled water directly affects the heat exchange capacity of the fan coil. Determine that the potential direction of the fault disturbance event propagating from the chiller to the fan coil is along the pipeline direction, and the possibility size is related to factors such as the distance and insulation condition of the pipeline. Based on these analysis results, generate preliminary regulation decision data, such as increasing the operating frequency of the chiller's compressor and adjusting the water valve opening of some fan coils to increase the chilled water flow rate, attempting to preliminarily suppress or adjust the propagation of the fault disturbance. Apply the preliminary regulation decision data to the current state information of the fault disturbance event to obtain the fault disturbance state information after the first round of iteration. At this time, the operating frequency of the chiller's compressor has increased, and the water valve openings of some fan coils have changed, reflecting the state change of the fault disturbance event after the implementation of the preliminary regulation decision.

[0033] From the second iteration to the (X - 1)-th iteration (assuming X = 5), based on the fault disturbance state information after the previous iteration, the decision optimization unit re-evaluates the propagation of the fault disturbance event in the building automation system. For example, in the second iteration, it is found that although the operating frequency of the compressor of the chiller is increased, the improvement in refrigeration efficiency is not obvious, which may be due to fouling problems in a certain component of the chiller (such as the condenser). Then, according to the new propagation analysis results, the previous regulation decision data is adjusted, such as increasing the cleaning operation of the condenser and further adjusting the water valve opening of the fan coil unit. After adjusting the regulation decision data, the new regulation decision data is applied to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. In the third iteration, it is found that due to the adjustment of the water valve opening of the fan coil unit, abnormal humidity changes occur in some areas. Therefore, measures to regulate humidity are added to the regulation decision, such as turning on the dehumidification equipment in some areas. In the fourth iteration, continue to optimize the regulation decision according to the changes in the fault disturbance state information. For example, according to the actual temperature and humidity requirements of each area, more precisely adjust the water valve opening of the fan coil unit and the operating time of the dehumidification equipment.

[0034] In the X-th iteration (the 5-th iteration), the decision optimization unit extracts the final fault diagnosis features based on the fault disturbance state information after the (X - 1)-th iteration. Analyze the current fault disturbance state information to determine that the final propagation range of the fault disturbance in the entire building automation system has affected the air conditioning comfort of multiple floors, and the final impact degree on each key component (such as the chiller, fan coil unit, dehumidification equipment, etc.). It is found that although the refrigeration efficiency of the chiller has increased, it has not yet returned to the normal level. The operating parameters of the fan coil unit have tended to be stable after multiple adjustments. The opening of the dehumidification equipment has played a certain role in humidity control, but further optimization is still needed. At the same time, no other chain reactions are triggered, such as no impact on the elevator system or lighting system. Then, according to the analysis results, extract the characteristic information reflecting the adaptive regulation decision result of the fault disturbance propagation in the X-th iteration cycle to obtain the target fault diagnosis features.

[0035] The execution strategy unit receives the target fault diagnosis feature from the decision optimization unit, which contains the fault disturbance propagation information after multiple iteration cycles of optimization, as well as the current configuration information of the building automation system, such as the connection relationship, functional attributes (such as cooling capacity, air volume, etc.), and operating parameter thresholds (such as temperature, humidity comfort range, etc.) of system components such as chillers and fan coils. The execution strategy unit analyzes the fault propagation path, impact degree, and potential chain reaction in the target fault diagnosis feature, identifies the chillers and fan coils directly affected by the fault, and the dehumidification equipment indirectly affected, and prioritizes these identified affected components to generate a fault impact range assessment report and a component priority ranking list. According to the fault impact range assessment report and the component priority ranking list, a matching control strategy is searched in the pre-built-in control strategy library. For example, in the case of a decrease in the cooling efficiency of the chiller, control strategies such as adding auxiliary refrigeration equipment and optimizing the operating parameters of the chiller may be found, and the corresponding control scheme is generated based on the matching results. The control scheme is simulated and tested using the simulation environment or digital twin model of the building automation system to generate a simulation test report. In the simulation test, it was found that although adding auxiliary refrigeration equipment can increase the cooling capacity, it will increase energy consumption. Therefore, the control scheme is optimized, such as adjusting the start-up conditions and operating time of the auxiliary refrigeration equipment to generate the final control decision result. The final control decision result is passed to the execution layer of the building automatic control system to guide the building automatic control system to implement control operations according to the final control decision result, such as starting the auxiliary refrigeration equipment according to the optimized time and conditions, and adjusting the operating parameters of the chiller and fan coil unit.

[0036] Similarly, for possible fault disturbance events in the lighting system, such as the failure of the light intensity sensor in a certain area causing abnormal collected data. Based on the relevant characteristic information of the lighting system in the first state feature space, the decision optimization unit begins to make adaptive control decisions for the propagation of fault disturbances. In the iterative process, the fault propagation path is first analyzed. Due to the failure of the light intensity sensor, the decision of the lighting control system connected to it may be affected, thereby affecting the lighting effect of the area. After initializing the relevant parameters, preliminary control decision data is generated, such as switching to a backup light intensity sensor or making preliminary light brightness adjustments based on the historical light data of the area and the current time, personnel activities and other information. After multiple rounds of iterations, the target fault diagnosis feature is finally obtained. The execution strategy unit generates a control plan based on the target fault diagnosis feature, such as repairing or replacing the faulty light intensity sensor, and optimizing the lighting control strategy of the area to ensure that the lighting effect meets the requirements.

[0037] The adaptive regulation decision results corresponding to these different fault disturbance events can also be used to optimize the building status recognition network. For example, when a chiller fails, the data on the relationship between the state characteristics such as temperature, humidity, and equipment operation parameters and the fault accumulated during the adaptive regulation decision-making process can be fed back to the building status recognition network. The building status recognition network can adjust its recognition mode of fault state characteristics based on this data, improving the accuracy and timeliness of recognizing similar faults.

[0038] Step S140, based on multiple different adaptive regulation decision results of the building operation categories associated with the first state feature space, perform adaptive regulation decision-making on the building automation system.

[0039] Still taking the building automation system of a commercial office building as an example. Multiple different adaptive regulation decision results of the building operation categories associated with the first state feature space cover various aspects such as the HVAC system, lighting system, elevator system, and security system.

[0040] For the HVAC system, based on the adaptive regulation decision results generated for fault disturbance events such as the decline in chiller cooling efficiency before, the building automation system performs overall adaptive regulation decision-making. For example, if it is found in multiple adaptive regulation decision results that the air conditioning terminal devices (fan coils) on several floors have difficulty in adjustment or excessive energy consumption, then in the overall adaptive regulation decision-making, local upgrades and renovations can be considered for the air conditioning systems on these floors, such as replacing more efficient fan coils or optimizing the pipeline layout. At the same time, adjust the operation strategy of the entire HVAC system according to the adaptive regulation decision results in different seasons and different time periods. During high-temperature periods in summer, according to the balance consideration of cooling effect and energy consumption in the adaptive regulation decision results, increase the number of operating chillers or raise the cooling capacity setting value in some areas; during low-temperature periods in winter, according to the analysis results of heating demand and energy conservation, optimize the operation parameters of the hot water supply system, such as adjusting the outlet temperature of the hot water boiler and the operating frequency of the circulating pump.

[0041] In terms of the lighting system, a comprehensive adaptive regulation decision is made based on the adaptive regulation decision results for fault disturbance events such as illuminance sensor failures and uneven brightness in the lighting area. If it is found in multiple adaptive regulation decision results that the lighting in some common areas (such as corridors and stairwells) is frequently switched on and off or the brightness is excessively adjusted, the lighting control strategy for these areas may be re-planned. For example, an intelligent lighting algorithm is adopted to more precisely control the switching on and off of lights and the brightness adjustment according to the personnel activity situation and natural daylighting situation. For office areas, according to the working hours and lighting requirements of different departments, combined with the lighting energy consumption analysis in the adaptive regulation decision results, a personalized lighting plan is formulated, such as setting different dimming schedules or adopting a zoning lighting control strategy to improve the overall energy efficiency of the lighting system.

[0042] For the elevator system, an adaptive regulation decision is made according to the adaptive regulation decision results for fault disturbance events such as abnormal elevator running speed and car overload alarm. If it is found in multiple adaptive regulation decision results that the running efficiency of some elevators is low during peak hours, the elevator dispatching algorithm may be adjusted. For example, a strategy of preferentially responding to calls from the upper or lower floors is added, or the elevator stopping strategy on different floors is adjusted to reduce unnecessary stopping times. At the same time, according to the running status and maintenance requirements of the elevator equipment, combined with the equipment fault analysis in the adaptive regulation decision results, the elevator maintenance plan is optimized, such as performing preventive maintenance on components that frequently fail in advance and adjusting the maintenance cycle.

[0043] In terms of the security system, an adaptive regulation decision is made based on the adaptive regulation decision results for fault disturbance events such as misjudgment of the access control system and blurred images of surveillance cameras. If it is found in multiple adaptive regulation decision results that the access control system at a certain entrance has a high misjudgment rate, the access control equipment may be upgraded or the access control verification algorithm may be adjusted. For surveillance cameras, according to the adaptive regulation decision results for faults such as blurred images, operations such as cleaning the camera, adjusting the focal length or replacing aging components are carried out. At the same time, according to the overall running situation of the security system and the security risk analysis in the adaptive regulation decision results, the layout and strategy of security monitoring are optimized, such as increasing the number of surveillance cameras in key areas or adjusting the monitoring range to improve the security prevention ability of the entire building. Through such comprehensive consideration of multiple different adaptive regulation decision results associated with the first state feature space, the building automation system can achieve more intelligent, efficient and stable operation, improving the comfort, security and energy utilization efficiency of the building.

[0044] Based on the above steps, in the embodiment of the present application, the adaptive learning network model is used to perform feature space transformation on the state variable monitoring data of the building automation system, generating a first state feature space that matches the feature patterns extracted by the building state recognition network. This method can, based on the first state feature space, guide the regulation network of the building automation system to make adaptive regulation decisions for multiple different fault disturbance events, generating multiple adaptive regulation decision results associated with the building operation categories. These decision results are not only used for the actual adaptive regulation of the building automation system but also further used to optimize the building state recognition network, thus forming a closed-loop intelligent regulation and optimization system. This method improves the ability of the building automation system to respond to fault disturbance events and the regulation accuracy, enhances the self-adaptability and intelligence level of the system, and effectively improves the operation efficiency and energy utilization efficiency of the building.

[0045] In a possible implementation manner, the regulation network of the building automation system includes a decision optimization unit and an execution strategy unit.

[0046] Step S130 includes:

[0047] Step S131, for each of the multiple different fault disturbance events, using the decision optimization unit to take the first state feature space as the fault diagnosis basis, making an adaptive regulation decision for the fault disturbance propagation of the fault disturbance event for X iteration cycles, generating the target fault diagnosis feature obtained from the adaptive regulation decision of the fault disturbance propagation in the Xth iteration cycle, and using the fault diagnosis feature obtained from the adaptive regulation decision of the fault disturbance propagation in the ath iteration cycle as the input for the adaptive regulation decision of the fault disturbance propagation in the (a + 1)th iteration cycle, where a and X are positive integers and a < X.

[0048] Step S132, using the execution strategy unit to make an adaptive regulation decision for the fault disturbance propagation of the target fault diagnosis feature, generating the adaptive regulation decision result of the building operation category associated with the first state feature space.

[0049] In this embodiment, continuing with the example of a chiller fault in the heating, ventilation, and air conditioning (HVAC) system, which is a fault disturbance event. The decision optimization unit will start processing this fault with the first state feature space as the fault diagnosis basis. The first state feature space contains state feature information related to the HVAC system in multiple aspects such as the temperature and humidity of the entire building and the operating parameters of each air conditioning device. Assume X is 5 and start entering the iteration cycle.

[0050] In the first iteration cycle, the decision optimization unit comprehensively analyzes the situation of chiller failure. Chiller failure may lead to abnormal cold water supply temperature, which in turn affects the cooling effect of the entire building. The decision optimization unit initializes the probability coefficient of chiller failure propagation based on the temperature data in the first state feature space, such as the current temperature and historical temperature change trends of each floor and each room, as well as the connection relationships of various components of the air conditioning system. For example, it is determined that the probability coefficient of chiller failure propagating to adjacent air conditioning terminal devices (such as fan coils) is 0.3. At the same time, the initial values of the evaluation indicators for the control decision effect are initialized. For example, the initial evaluation value for maintaining the indoor comfort temperature range is set to 0.8. Based on the initial state information of the chiller, such as the current refrigeration efficiency and cold water flow rate of the chiller, the propagation path of the failure in the current building automation system environment is analyzed. Since the chiller is connected to multiple fan coils through pipes, after the chiller fails and the cold water temperature rises, the hot water may sequentially affect the fan coils on each floor along the pipes. This is the determined potential direction of failure propagation, and its probability is related to factors such as the insulation condition and distance of the pipes. According to these analysis results, preliminary control decision data is generated, such as increasing the operating frequency of the chiller compressor, adjusting the water valve openings of some fan coils to attempt to increase the cold water flow rate, and suppressing the propagation of the failure. Applying these preliminary control decision data to the current state information of the failure, the failure disturbance state information after the first iteration cycle is obtained. At this time, the operating frequency of the chiller compressor changes, and the water valve openings of some fan coils are adjusted, which reflects the change in the failure state after the implementation of the preliminary control decision.

[0051] In the second iteration cycle, the decision optimization unit is based on the failure disturbance state information after the previous iteration cycle. At this time, it is found that although the operating frequency of the compressor has been increased, the improvement in the refrigeration efficiency of the chiller is not obvious. Through further analysis, it is found that the fouling of the chiller condenser may affect the heat exchange efficiency. Therefore, the failure propagation situation is re-evaluated, and the control decision data is adjusted according to the new analysis results. The cleaning operation of the condenser is increased, and at the same time, the water valve openings of the fan coils are further finely adjusted. After the adjustment, the new control decision data is applied to the current failure disturbance state information to obtain the failure disturbance state information after this iteration cycle.

[0052] In the third iteration cycle, the decision optimization unit continues to be based on the failure disturbance state information of the previous round. At this time, it is found that due to the adjustment of the water valve openings of the fan coils, abnormal humidity changes occur in some areas. Therefore, the failure propagation situation is re-evaluated again, and humidity adjustment measures are added to the control decision, such as turning on the dehumidification equipment in some areas, and applying the new control decision data to the current failure disturbance state information to obtain the failure disturbance state information after this iteration cycle.

[0053] In the 4th iteration cycle, the decision optimization unit still operates according to the fault disturbance status information of the previous round. At this time, it may be found that although the dehumidification equipment has a certain effect on humidity regulation, the refrigeration efficiency of the chiller has not fully returned to normal, and the temperature on some floors has not reached the comfortable range. Therefore, the control decision is further optimized. For example, according to the actual temperature and humidity requirements of each area, the water valve opening of the fan coil unit and the operation time of the dehumidification equipment are adjusted more precisely. Then, the new control decision data is applied to the current fault disturbance status information to obtain the fault disturbance status information after this round of iteration.

[0054] By the 5th iteration cycle, the decision optimization unit extracts the final fault diagnosis features based on the fault disturbance status information after the 4th iteration cycle. Analyze the current fault disturbance status information to determine that the final propagation range of the fault disturbance in the entire building automation system has affected the air conditioning comfort of multiple floors. Although the refrigeration efficiency of the chiller has increased, it has not fully recovered. The operating parameters of the fan coil unit have stabilized after multiple adjustments. The dehumidification equipment has a certain effect on humidity control but still needs to be optimized, and no chain reactions in other systems such as the elevator system and lighting system have been triggered. According to these analysis results, the characteristic information reflecting the adaptive control decision result of the fault disturbance propagation in the 5th iteration cycle is extracted to obtain the target fault diagnosis features.

[0055] After the execution strategy unit receives the target fault diagnosis features from the decision optimization unit, the target fault diagnosis features include the fault perturbation propagation information optimized after 5 iteration cycles and the current configuration information of the building automation system, such as the connection relationships, functional attributes (such as refrigerating capacity, air volume, etc.), and operation parameter thresholds (such as temperature and humidity comfort ranges, etc.) of system components such as chillers and fan coil units. The execution strategy unit analyzes the fault propagation path, influence degree, and potential chain reaction in the target fault diagnosis features, identifies the chillers and fan coil units directly affected by the fault, as well as the dehumidification equipment indirectly affected, and prioritizes these affected components to generate a fault impact range assessment report and a component priority ranking list. According to the fault impact range assessment report and the component priority ranking list, search for matching control strategies in the pre-built control strategy library. For example, for the situation of the decreasing refrigeration efficiency of the chiller, control strategies such as adding auxiliary refrigeration equipment and optimizing the operation parameters of the chiller may be found, and a corresponding control plan is generated based on the matching result. Use the simulation environment or digital twin model of the building automation system to simulate and test the control plan to generate a simulation test report. It is found in the simulation test that adding auxiliary refrigeration equipment can increase the refrigerating capacity but will increase the energy consumption. Therefore, the control plan is optimized, such as adjusting the start-up conditions and operation time of the auxiliary refrigeration equipment, to generate the final control decision result, and the final control decision result is transmitted to the execution layer of the building automation system to guide the building automation system to implement control operations according to the final control decision result, such as starting the auxiliary refrigeration equipment according to the optimized time and conditions, and adjusting the operation parameters of the chiller and fan coil unit, etc.

[0056] Similarly, for the fault perturbation event of the illuminance sensor failure in a certain area of the lighting system, the decision optimization unit starts iteration based on the lighting-related state features in the first state feature space. In each iteration cycle, analyze the fault propagation path. For example, the illuminance sensor failure affects the decision of the lighting control system connected to it, and further affects the lighting effect in this area. In each iteration cycle, adjust the control decision data according to the fault perturbation state information of the previous cycle, such as first switching to the standby illuminance sensor or initially adjusting the light brightness according to the regional historical illuminance data, current time, personnel activities, etc. After obtaining the target fault diagnosis features through multiple iteration cycles, the execution strategy unit generates a control plan based on it, such as repairing or replacing the faulty illuminance sensor, and optimizing the lighting control strategy in this area to ensure that the lighting effect meets the requirements and other series of operations to generate an adaptive control decision result associated with the first state feature space.

[0057] In a possible implementation manner, before step S130, the method further includes:

[0058] Step A110: Obtain the first template status variable monitoring data, the first template status feature space corresponding to the first template status variable monitoring data, and the set number of iteration cycles b determined for the first template status variable monitoring data. The first template status feature space is generated by the sample building status recognition network through feature space conversion of the first template status variable monitoring data. The sample building status recognition network is the building status recognition network or another neural network model with the same network task as the building status recognition network. b is a positive integer not greater than X.

[0059] In this embodiment, the first template status variable monitoring data comes from each subsystem of the building automation system, such as the HVAC system, lighting system, elevator system, and security system mentioned before. For example, for the HVAC system, the first template status variable monitoring data includes the operating parameters of the chiller, such as the compressor speed of the chiller, the temperatures of the condenser and evaporator, and the inlet and outlet water temperatures; it also includes the status data of the terminal air-conditioning equipment such as fan coils, such as the fan speed and the water valve opening. The first template status variable monitoring data in the lighting system includes the illuminance values collected by illuminance sensors in each area, the on / off status of the lamps, and the dimming ratio, etc. The elevator system is data such as the operating speed of the elevator, the car position, the passenger capacity, and the opening and closing status of the elevator doors. The first template status variable monitoring data of the security system covers the personnel entry and exit records of the access control system, the personnel flow information after processing the video data of the surveillance cameras, etc. At the same time, obtain the first template status feature space corresponding to the first template status variable monitoring data. This feature space is generated by the sample building status recognition network through feature space conversion of the first template status variable monitoring data. The sample building status recognition network has the same network task as the actual building status recognition network. Whether it is to identify the overall operating status of the building or analyze the relationship between each subsystem, etc., the tasks are the same. And determine the set number of iteration cycles b. b is a positive integer not greater than X. Assuming X is 5, b can take the value of 3.

[0060] Step A120: The feature conversion module performs feature conversion on the first template status variable monitoring data to generate a first conversion result.

[0061] The feature conversion module performs specific conversion operations on different types of data. For the data of the HVAC system, the feature conversion module integrates and converts the operating parameters of the chiller. For example, it converts data such as compressor speed, condenser and evaporator temperatures into features that can reflect the overall performance and health status of the chiller. By analyzing various factors such as the relationship between compressor speed and condenser temperature, and the impact of inlet and outlet water temperatures on the overall refrigeration efficiency, a comprehensive feature representation is generated, which is part of the first conversion result. For the lighting system, the feature conversion module converts features related to lighting comfort and energy utilization efficiency based on information such as illuminance values and lamp status. For example, it combines the illuminance value with the dimming ratio of the lamp to generate a feature representing the lighting quality and energy-saving level of the area. After the data of the elevator system is processed by the feature conversion module, features related to elevator service quality and operation stability are generated. For example, based on the elevator running speed and car passenger capacity, a feature reflecting the operating efficiency of the elevator under different loads is converted. The data of the security system is converted into features related to personnel safety management and regional security situation. For example, based on the personnel entry and exit records of the access control system and the personnel flow information of the surveillance cameras, a feature representing the safety of personnel activities in a specific area is converted.

[0062] Step A130, the decision optimization unit adds decision interference to the first conversion result for b iteration cycles to generate decision interference addition data. The decision interference addition data includes the target template fault interference features obtained from the decision interference addition in the b-th iteration cycle.

[0063] In the first iteration cycle, the decision optimization unit adds specific decision interferences to each feature in the first conversion result. Taking the HVAC system as an example, for the features related to the performance of the chiller, a simulated fault interference may be added, such as an interference that reduces the refrigeration efficiency of the chiller. Suppose the refrigeration efficiency is reduced by 10%. For the lighting system, interferences may be added to the features representing lighting quality and energy conservation, such as simulating the deviation of the illuminance sensor readings, causing the lighting system to think that the current illuminance is lower than the actual value, thereby affecting the lighting control decision. In the elevator system, interferences are added to the features reflecting the elevator operation efficiency, such as simulating the error of the elevator car weight sensor, resulting in the elevator control system misjudging the car load, and further affecting the elevator operation speed and stop decision. In terms of the security system, interferences are added to the features representing the safety of personnel activities, such as simulating the misjudgment of the access control system, so that a legitimate person entering is misjudged as illegal. As the iteration cycle progresses, each iteration cycle adjusts the decision interference according to the result of the previous cycle and the current system state. After b iteration cycles, decision interference addition data is generated, and the target template fault interference features obtained from the decision interference addition in the bth iteration cycle contain the state features of each subsystem after multiple decision interferences.

[0064] Step A140, using the first template state feature space as the basis for fault diagnosis through the decision optimization unit, making an adaptive regulation decision on the fault perturbation propagation of the target template fault interference features for b iteration cycles, and generating template fault diagnosis features.

[0065] In the first iteration cycle, the decision optimization unit analyzes the fault conditions in the target template fault interference features based on the various feature relationships in the first template state feature space. For example, in a heating, ventilation, and air conditioning (HVAC) system, due to the previously added interference of reduced chiller cooling efficiency, the decision optimization unit analyzes the possible propagation paths of faults according to the association relationships between the chiller and other components (such as fan coils, cooling towers, etc.) in the first template state feature space. It may be determined that the reduced chiller cooling efficiency will cause a decline in the cooling effect of the fan coil, thereby affecting the indoor temperature. Then, preliminary control decisions are generated, such as appropriately increasing the compressor power of the chiller and adjusting the water valve opening of the fan coil. For the lighting system, according to the fault interference of the deviation in the illuminance sensor readings, combined with the relationships between the lighting system and factors such as human activities and time in the first template state feature space, it may be decided to adjust the dimming ratio of the lamps or turn on more lamps to compensate for the perceived insufficient lighting. In the elevator system, for the fault interference of incorrect judgment of the car load, according to the relationship between elevator operation and the safety protection mechanism in the first template state feature space, the speed limit and stop strategy of the elevator are adjusted. In the security system, for the fault interference of incorrect access control judgment, based on the relationship between the access control system and personnel authority management and the monitoring system in the first template state feature space, the access control verification rules are adjusted or the sensitivity of the monitoring alarm is increased. As the iteration cycle progresses, each iteration cycle is adjusted according to the control decision effect and fault propagation situation of the previous cycle. After b iteration cycles, template fault diagnosis features are generated.

[0066] Step A150, the execution strategy unit performs an adaptive control decision on the fault perturbation propagation of the template fault diagnosis features to generate adaptive control decision data.

[0067] The execution strategy unit analyzes the fault states and relevant relationships of each subsystem in the template fault diagnosis features. For the HVAC system, according to the fault states of equipment such as chillers and fan coils and their mutual influence relationships, a suitable control strategy is searched in the pre-set control strategy library. For example, if the chiller fault causes insufficient cooling and the fan coil is operating abnormally, a strategy of simultaneously adjusting the operating parameters of the chiller and the fan coil may be selected, such as further optimizing the compressor frequency of the chiller and the water valve opening ratio of the fan coil. For the lighting system, according to the lighting problems caused by the illuminance sensor fault, strategies such as adjusting the lamp layout or replacing the illuminance sensor are selected. In the elevator system, for the operation problems caused by incorrect judgment of the car load, the execution strategy unit may select strategies such as recalibrating the car weight sensor or adjusting the elevator dispatching algorithm. In the security system, for the situation of incorrect access control judgment, strategies such as updating the access control permission database or upgrading the access control verification algorithm may be adopted. By implementing these strategies, adaptive control decision data is generated.

[0068] Step A160: Extract the state feature space from the adaptive regulation decision data through the sample building state recognition network to generate the template state feature space corresponding to the template fault diagnosis features.

[0069] The sample building state recognition network analyzes the state information of each subsystem in the adaptive regulation decision data. For the HVAC system, it extracts new operating state features reflecting the new operating state of the entire HVAC system from the operating parameters of the adjusted chillers and fan coil units, such as whether the overall refrigeration efficiency has increased and whether the temperatures in each area tend to be stable. In terms of the lighting system, new features related to lighting comfort and energy utilization efficiency are extracted from information such as the adjusted lighting layout and the status of illuminance sensors. The elevator system extracts new features reflecting the operating stability and service quality of the elevator from information such as the recalibrated car weight sensor and the adjusted scheduling algorithm. The security system extracts new features representing personnel safety management and regional security situation from information such as the updated access control permission database and the upgraded verification algorithm, thereby generating the template state feature space.

[0070] Step A170: Determine the state feature space error based on the first template state feature space and the template state feature space.

[0071] Step A180: Optimize the neuron weight information of the decision optimization unit based on the state feature space error until the first convergence requirement is met.

[0072] For example, for the HVAC system, compare the differences in features such as the chiller refrigeration efficiency feature and the fan coil unit operating state feature between the two. Compare the differences in features such as the illuminance distribution feature and the lighting energy-saving feature in the lighting system. Compare the differences in features such as the elevator operating efficiency feature and the accuracy of car load judgment feature in the elevator system. Compare the differences in features such as the personnel safety assessment feature and the access control misjudgment rate feature in the security system. Calculate the overall state feature space error based on these differences. Based on this state feature space error, optimize the neuron weight information of the decision optimization unit until the first convergence requirement is met. During the optimization process, adjust the neuron weights related to each subsystem in the decision optimization unit according to the magnitude and direction of the state feature space error. For example, if the feature difference in the chiller refrigeration efficiency in the HVAC system is large, the neuron weights related to chiller fault diagnosis and regulation decision in the decision optimization unit will be adjusted emphatically. By continuously adjusting the weights, the decision optimization unit can analyze and make decisions more accurately when dealing with similar fault disturbances until the state feature space error reaches the first convergence requirement, that is, the error is within an acceptable range, so as to ensure that the decision optimization unit can work more precisely in the subsequent adaptive regulation decision of actual fault disturbance propagation.

[0073] In a possible implementation manner, step A170 includes:

[0074] Step A171, determining a regulation factor based on the set number of iteration cycles b, where the regulation factor has an inverse association relationship with the set number of iteration cycles b.

[0075] Step A172, calculating the distance between the first template state feature space and the template state feature space.

[0076] Step A173, regulating the distance according to the regulation factor to generate a state feature space error.

[0077] In this embodiment, in the scenario of the building automation system of a commercial office building, the set number of iteration cycles b plays an important role in this process. Since the regulation factor has an inverse association relationship with the set number of iteration cycles b, assuming the value of b is 3 (this is just an example based on the previous setting, not a logical assumption of the hypothesis statement), when the value of b is smaller, the value of the regulation factor is relatively larger. This is because in fewer iteration cycles, the impact of each iteration on the result is relatively larger, so a larger regulation factor is needed to accurately measure the error.

[0078] When calculating the distance between the first template state feature space and the template state feature space, for the heating, ventilation, and air conditioning (HVAC) system, features such as the chiller refrigeration efficiency feature and the fan coil unit operating state feature in the first template state feature space are compared and calculated with the corresponding features in the template state feature space. For example, the chiller refrigeration efficiency may be represented as a specific numerical range or vector form in the first template state feature space, and in the template state feature space, it becomes another numerical range or vector form after a series of operations (such as fault perturbation, adaptive control decision-making, etc.), and a specific mathematical calculation method (such as the Euclidean distance calculation method if the feature space is in vector form) is used to calculate the distance between the two. Similarly, for the lighting system, features such as the illuminance distribution feature and the lamp energy-saving feature also perform such distance calculations; features such as the elevator operation efficiency feature, the accuracy of car load judgment feature of the elevator system, and the personnel safety assessment feature and access control misjudgment rate feature of the security system also perform distance calculations according to their respective feature representation forms.

[0079] Thus, taking the HVAC system as an example, if the distance of characteristics such as the refrigeration efficiency of the chiller is calculated to be a certain value, according to its inverse correlation with b (assuming the adjustment factor is 0.5, which is an example based on the previous relationship), the distance value is multiplied by the adjustment factor to obtain an adjusted value. This value, combined with the values of other subsystems after the same adjustment, forms the state feature space error. The lighting system, elevator system, security system, etc. also participate in the formation process of the entire state feature space error according to their respective adjustment results.

[0080] In a possible implementation manner, the adaptive regulation and control decision result further includes the disturbance regulation estimated in the adaptive regulation and control decision processing of the fault disturbance propagation in b iteration cycles.

[0081] The method further includes:

[0082] Based on the fault disturbance characteristics increased by adding decision interference to the first conversion result in b iteration cycles and the disturbance regulation characteristics estimated in the adaptive regulation and control decision process of the fault disturbance propagation of the target template fault disturbance characteristics in b iteration cycles, determine the disturbance regulation error.

[0083] Step A180 may include:

[0084] Step A181, fuse and calculate the state feature space error and the disturbance regulation error to generate a first training cost.

[0085] Step A182, based on the first training cost, optimize the neuron weight information of the decision optimization unit until the first convergence requirement is met.

[0086] In this embodiment, during the process of adding decision interference to the first conversion result in b iteration cycles, for the HVAC system, the increased fault disturbance characteristics may be the simulation of the reduction of the refrigeration efficiency of the chiller or the abnormal change of the opening of the fan coil water valve. In the adaptive regulation and control decision process of the fault disturbance propagation of the target template fault disturbance characteristics in b iteration cycles, the estimated disturbance regulation characteristics are, for example, the prediction of the adjustment of the compressor power of the chiller and the prediction of the adjustment of the opening of the fan coil water valve. Based on these, determine the disturbance regulation error. Calculate the difference between the fault disturbance characteristics increased during the addition of decision interference and the disturbance regulation characteristics actually estimated in the adaptive regulation and control decision of the fault disturbance propagation. For example, the difference between the simulated value of the reduction of the refrigeration efficiency of the chiller and the predicted value of the recovery of the refrigeration efficiency in the actual regulation and control decision, the difference between the simulated value of the abnormal change of the opening of the fan coil water valve and the predicted adjustment value, etc. Combine these differences to determine the disturbance regulation error.

[0087] Then, for the HVAC system, the part related to the chiller refrigeration efficiency in the state feature space error and the part related to the chiller in the disturbance regulation error are calculated according to a specific fusion calculation method (such as weighted summation, etc.). Similarly, the two parts of the error related to the fan coil unit are also fused and calculated, and then summarized with the fusion results of other subsystems to form the first training cost. Based on the first training cost, the neuron weight information of the decision optimization unit is optimized until the first convergence requirement is met. During this process, the first training cost will guide the adjustment direction and amplitude of the neuron weights of the decision optimization unit. If the first training cost is large, it indicates that the decision result of the current decision optimization unit deviates greatly from the expected result, then a large adjustment will be made to the neuron weights, for example, the neuron weights related to the chiller fault diagnosis and regulation decision are adjusted. With continuous adjustment, the first training cost gradually decreases. When the first convergence requirement is met (for example, the first training cost is less than a set threshold), it indicates that the decision-making ability of the decision optimization unit has reached an ideal state under the current error standard.

[0088] In a possible implementation manner, step A180 may further include: locking the neuron weight information of the feature conversion module and the execution policy unit, and optimizing the neuron weight information of the decision optimization unit based on the state feature space error until the first convergence requirement is met.

[0089] When optimizing the neuron weight information of the decision optimization unit, lock the neuron weight information of the feature conversion module and the execution policy unit. This means that during the optimization of the decision optimization unit, keep the neuron weights of the feature conversion module and the execution policy unit unchanged, and only focus on adjusting the neuron weights of the decision optimization unit. This can ensure that during the optimization process, the adjustment direction and effect of the neuron weights of the decision optimization unit will not be interfered by the changes of the weights of other modules, making the optimization of the decision optimization unit more independent and accurate until it meets the first convergence requirement.

[0090] In a possible implementation manner, before step A120, the method includes:

[0091] Step A111, obtain the second template state variable monitoring data.

[0092] Step A112, perform feature conversion on the second template state variable monitoring data through the feature conversion module to generate a second conversion result.

[0093] Step A113, perform an adaptive regulation decision on the second conversion result for the fault disturbance propagation through the execution policy unit to generate the adaptive regulation decision template state variable monitoring data for the fault disturbance propagation.

[0094] Step A114: Determine the second training cost based on the second template state variable monitoring data and the adaptive regulation decision template state variable monitoring data of the fault disturbance propagation.

[0095] Step A115: Optimize the neuron weight information of the feature transformation module and the execution policy unit based on the second training cost until the second convergence requirement is met.

[0096] Before the feature transformation module performs feature transformation on the first template state variable monitoring data, there are some pre - operations. Obtain the second template state variable monitoring data, which also comes from each subsystem of the building automation system. For the heating, ventilation, and air - conditioning (HVAC) system, it includes the operating parameters of the chiller, the status data of the fan coil units, etc.; for the lighting system, it includes the illuminance sensor values of the lighting system, the lamp status data; for the elevator system, it includes the operating speed, car position, passenger capacity, etc. data; for the security system, it includes the personnel entry and exit records of the access control system, the personnel flow information of the surveillance cameras, etc.

[0097] Next, taking the HVAC system as an example, the operating parameters of the chiller may be converted into a feature form that is more closely related to the overall performance and faults, and the status data of the fan coil units are converted into a feature form that can reflect their impact on the indoor environment. In the lighting system, the illuminance sensor values and lamp status data are converted into features related to lighting comfort and energy utilization efficiency. The operating data of the elevator system are converted into features related to service quality and operating stability, and the personnel entry and exit and flow data of the security system are converted into features related to regional security management.

[0098] Furthermore, for the HVAC system, based on the features related to the chiller and fan coil units in the second conversion result, when assuming that a certain fault occurs in the chiller (such as condenser fouling affecting the refrigeration efficiency), the execution policy unit makes a regulation decision according to the pre - set regulation strategy library (including regulation strategies for different fault scenarios). For example, operations such as adjusting the compressor operating frequency of the chiller and adjusting the water valve opening of the fan coil unit are performed to generate the adaptive regulation decision template state variable monitoring data of the fault disturbance propagation. In the lighting system, when the illuminance sensor fails or the lamp status is abnormal, the execution policy unit adjusts the dimming ratio of the lamps or turns on / off some lamps to generate corresponding data. In the elevator system, when the elevator operating speed is abnormal or the car load is misjudged, operations such as adjusting the elevator docking strategy or speed limit are performed to generate relevant data. In the security system, when the access control system makes a misjudgment or the surveillance camera data is abnormal, operations such as adjusting the access verification rules or increasing the surveillance alarm sensitivity are performed to generate corresponding data.

[0099] For the HVAC system, compare the original operating parameters of the chiller in the monitoring data of the second template state variables with the adjusted operating parameters of the chiller in the monitoring data of the adaptive regulation decision template for fault disturbance propagation, and calculate the difference between the two. Similarly, the difference between the original state data and the adjusted data of the fan coil unit is also calculated. The lighting system compares the difference between the original value and the adjusted value of the illuminance sensor, the difference between the original state and the adjusted state of the lighting fixture, etc. The elevator system compares the difference between the original data such as the running speed, car position, and passenger capacity and the adjusted data. The security system compares the difference between the original data of the personnel entry and exit records of the access control system and the adjusted data, the difference between the original information and the adjusted information of the personnel flow of the monitoring camera, etc., and combines these differences according to a specific calculation method to determine the second training cost.

[0100] If the second training cost is large, it indicates that the current performance of the feature conversion module and the execution policy unit deviates greatly from the expected results, and then their neuron weights will be adjusted. For example, in the feature conversion module, if it is found that the features after converting the operating parameters of the chiller in the HVAC system cannot well reflect the actual fault situation, the neuron weights related to the chiller will be adjusted. In the execution policy unit, if it is found that the regulation decision for the chiller fault cannot effectively improve the system state, the neuron weights related to the chiller fault regulation decision will be adjusted. With continuous adjustment, the second training cost gradually decreases. When the second convergence requirement is met (for example, the second training cost is less than a certain set threshold), it indicates that the feature conversion module and the execution policy unit reach a relatively ideal state under the current error standard.

[0101] In a possible implementation manner, step S131 includes:

[0102] Step S1311, obtain the first state feature space through the decision optimization unit as the basis for fault diagnosis, and at the same time obtain the information related to the current fault disturbance event to be processed, where the information related to the fault disturbance event includes the type of the fault disturbance event and the information of the building automation system components affected.

[0103] Step S1312, initialize the relevant parameters of the fault disturbance event through the decision optimization unit, where the relevant parameters include a coefficient for measuring the possibility of fault propagation and an initial value of an index for evaluating the effect of the regulation decision.

[0104] Step S1313, initialize the information representing the initial state of the fault disturbance event according to the information related to the fault disturbance event to obtain the initial state information.

[0105] Step S1314: In the first round of iteration, the decision optimization unit analyzes the fault disturbance event based on the initialized relevant parameters and the initial state information of the fault, and generates an analysis result. Specifically, first analyze the propagation path of the fault disturbance event in the current building automation system environment, the connection relationships between various components in the building automation system, the data flow, and the interaction mechanism, and determine the potential direction and likelihood of the fault disturbance event spreading from the initially affected component to other components.

[0106] Step S1315: Generate preliminary regulation decision data according to the analysis result of the fault disturbance propagation. The preliminary regulation decision data is generated based on the overall building operation state reflected by the first state feature space and the preliminary analysis of the current fault disturbance, and is used to preliminarily suppress or adjust the propagation of the fault disturbance.

[0107] Step S1316: After generating the preliminary regulation decision data, apply the preliminary regulation decision data to the current state information of the fault disturbance event to obtain the fault disturbance state information after the first round of iteration. The fault disturbance state information reflects the state change of the fault disturbance event after the implementation of the preliminary regulation decision.

[0108] Step S1317: From the second round of iteration to the (X - 1)-th round of iteration, the decision optimization unit re-evaluates the propagation of the fault disturbance event in the building automation system based on the fault disturbance state information after the previous round of iteration, and adjusts the previous regulation decision data according to the new propagation analysis result. After adjusting the regulation decision data, apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. The fault disturbance state information after the current iteration reflects the latest state information of the fault disturbance event after multiple rounds of iteration and regulation decision adjustment.

[0109] Step S1318: In the X-th round of iteration, the decision optimization unit extracts the final fault diagnosis features based on the fault disturbance state information after the (X - 1)-th round of iteration. Specifically, analyze the current fault disturbance state information to determine the final propagation range of the fault disturbance in the entire building automation system, the final impact degree on each key component, and whether other chain reactions are triggered, and then extract the feature information reflecting the adaptive regulation decision result of the fault disturbance propagation in the X-th iteration cycle according to the analysis result to obtain the target fault diagnosis features.

[0110] In a possible implementation manner, step S132 includes:

[0111] Step S1321: Receive, by the execution policy unit, the target fault diagnosis features from the decision optimization unit. The target fault diagnosis features include the fault perturbation propagation information optimized through multiple iteration cycles and the current configuration information of the building automation system. The current configuration information includes the connection relationship, functional attributes, and operating parameter thresholds of system components.

[0112] Step S1322: Analyze the fault propagation path, influence degree, and potential chain reaction in the target fault diagnosis features, identify the affected components directly and indirectly affected by the fault, and prioritize the identified affected components to generate a fault impact scope assessment report and a component priority ranking list.

[0113] Step S1323: Search for matching regulation strategies in the pre-built regulation strategy library according to the fault impact scope assessment report and the component priority ranking list, and generate a corresponding regulation plan based on the matching result. The regulation strategy library contains a predefined set of regulation strategies for various fault scenarios.

[0114] Step S1324: Use the simulation environment or digital twin model of the building automation system to conduct simulation tests on the regulation plan and generate a simulation test report.

[0115] Step S1325: Optimize the regulation plan based on the simulation test report to generate the final regulation decision result. The optimization process involves adjusting regulation parameters, modifying regulation steps, and adding or deleting local regulation measures.

[0116] Step S1326: Transmit the final regulation decision result to the execution layer of the building automation system to guide the building automation system to implement regulation operations according to the final regulation decision result.

[0117] Take the fault perturbation event of fouling on the evaporator of a chiller in the HVAC system as an example. First, the decision optimization unit obtains the first state feature space as the basis for fault diagnosis and simultaneously obtains the information related to the current fault perturbation event to be processed. The type of the fault perturbation event of fouling on the evaporator of the chiller is clearly defined as the fouling problem of the evaporator, and the information of the building automation system components affected by it is mainly the chiller itself and the components connected to it, such as fan coils and cooling towers connected through pipes.

[0118] Next, the decision optimization unit initializes the relevant parameters of the fault disturbance event. Regarding the coefficient used to measure the possibility of fault propagation, since the fouling of the evaporator will affect the refrigeration efficiency of the chiller, which may in turn affect the temperature regulation function of the entire HVAC system. According to the connection relationship between the chiller and other components and past empirical data, the coefficient of the chiller fault propagating to the fan coil unit is set to 0.4, and the coefficient of propagating to the cooling tower is set to 0.3, etc. Regarding the initial values of the indicators used to evaluate the effect of the control decision, the initial evaluation value of keeping the indoor temperature within the comfortable range (for example, 22 - 26 degrees Celsius) is set to 0.8, and the initial evaluation value of controlling the energy consumption of the entire HVAC system within a reasonable range is set to 0.7, etc.

[0119] According to the relevant information of the fault disturbance event, initialize the information representing the initial state of the fault disturbance event to obtain the initial state information. In the initial state of the fouling of the chiller evaporator, the heat exchange efficiency of the evaporator decreases, the refrigeration capacity of the chiller begins to decline, and the outlet water temperature of the chilled water gradually rises. These information constitute the initial state information of the fault.

[0120] In the first round of iteration, the decision optimization unit analyzes the fault disturbance event based on the initialized relevant parameters and the initial state information of the fault. When analyzing the propagation path of the fault disturbance event in the current building automation system environment, considering the connection relationship between each component in the building automation system, the chiller transports chilled water to each fan coil unit through pipelines, and the cooling tower is responsible for cooling the circulating water. The fouling of the evaporator leads to a decrease in refrigeration efficiency and an increase in chilled water temperature. This change will propagate along the pipeline to the fan coil unit, affecting its refrigeration effect. The data flow direction is from the chiller to the fan coil unit, and the interaction mechanism is that the chiller provides a cold source, and the fan coil unit exchanges heat through chilled water to regulate the indoor temperature. Based on these situations, it is determined that the potential direction of the fault disturbance event propagating from the initial affected component (chiller) to other components (fan coil unit, cooling tower) is along the pipeline direction, and the possibility is as set before.

[0121] According to the analysis results of the fault disturbance propagation, generate preliminary control decision data. Based on the global building operation state reflected by the first state feature space and the preliminary analysis of the current fault disturbance, since the decrease in the refrigeration capacity of the chiller affects the indoor temperature, the preliminary control decision data may include appropriately increasing the operating frequency of the chiller compressor to try to increase the refrigeration capacity, and at the same time adjusting the water valve opening of some fan coil units close to the core area to slightly increase the chilled water flow rate to enhance the refrigeration effect, so as to preliminarily suppress or adjust the propagation of the fault disturbance.

[0122] After generating the preliminary regulation decision data, apply the preliminary regulation decision data to the current state information of the fault disturbance event to obtain the fault disturbance state information after the first round of iteration. At this time, the operating frequency of the chiller compressor increases, the opening degree of the water valves of some fan coil units increases, the cooling capacity of the chiller may increase to a certain extent, the rising trend of the chilled water outlet temperature slows down, and the cooling effect of the fan coil units is partially improved. These reflect the state changes of the fault disturbance event after the implementation of the preliminary regulation decision.

[0123] From the second round of iteration to the (X - 1)-th round of iteration (assuming X = 5), the decision optimization unit re-evaluates the propagation of the fault disturbance event in the building automation system based on the fault disturbance state information after the previous round of iteration. In the second round of iteration, it is found that although the operating frequency of the chiller compressor is increased, due to the unsolved problem of evaporator scaling, the improvement of the refrigeration efficiency is limited, and due to the increase in the compressor operating frequency, the energy consumption increases. According to the new propagation analysis results, adjust the previous regulation decision data. Considering the fundamental problem of evaporator scaling, add a plan for evaporator cleaning to the regulation decision, and at the same time adjust the adjustment strategy of the water valve opening degree of the fan coil unit, and make more refined adjustments according to the actual temperature requirements of different regions. For example, appropriately reduce the water valve opening degree in the area closer to the window. After adjusting the regulation decision data, apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. In the third round of iteration, it is found that due to the adjustment of the water valve opening degree of the fan coil unit, the humidity in some areas has abnormal changes. So re-evaluate the fault propagation situation again, add humidity adjustment measures to the regulation decision, such as turning on the dehumidification equipment in some areas and adjusting the operating time of the dehumidification equipment. At the same time, further optimize the evaporator cleaning plan, consider using chemical cleaning methods and determine the specific time and operation process of the cleaning. Then apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after this round of iteration. In the fourth round of iteration, continue to adjust according to the changes in the fault disturbance state information. It is found that although dehumidification measures are taken, the humidity control is still not ideal, and the refrigeration efficiency of the chiller has not fully recovered to the normal level. So further optimize the regulation decision, adjust the operating parameters of the dehumidification equipment, such as the humidity set value and air volume, etc., and at the same time readjust the operating frequency of the chiller compressor according to the cooling capacity demand and energy consumption situation, and estimate the effect after evaporator cleaning and adjust the operation details of the cleaning, etc. Apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after this round of iteration.

[0124] In the Xth iteration (the 5th iteration), the decision optimization unit extracts the final fault diagnosis features based on the fault disturbance state information after the (X - 1)th iteration. By analyzing the current fault disturbance state information, it is determined that the final propagation range of the fault disturbance in the entire building automation system has affected the air conditioning comfort of multiple floors, especially the floors near the chiller are more significantly affected. Regarding the final impact degree on each key component, although the scaling of the evaporator of the chiller has not fully returned to normal after some regulation measures, the refrigeration efficiency is still lower than the normal level. Due to operations such as the water valve opening and humidity adjustment of the fan coil units, the operating parameters of some fan coil units are in a non-ideal but relatively stable state. The operation of the dehumidification equipment has played a certain role in humidity control but still needs further optimization. And no other chain reactions have been triggered, such as no impact on the elevator system or lighting system. Then, according to the analysis results, the characteristic information reflecting the adaptive regulation decision results of the fault disturbance propagation in the Xth iteration cycle is extracted, such as the specific value of the refrigeration efficiency in the final state of the scaling of the chiller evaporator, the final adjustment value of the water valve opening of each fan coil unit, the final operating parameters of the dehumidification equipment, etc., to obtain the target fault diagnosis features.

[0125] Regarding the process of generating the adaptive regulation decision results of the building operation category associated with the first state feature space through the execution policy unit for the adaptive regulation decision of the target fault diagnosis features for fault disturbance propagation.

[0126] The execution policy unit receives the target fault diagnosis features from the decision optimization unit. The target fault diagnosis features contain the fault disturbance propagation information optimized through multiple iteration cycles, as well as the current configuration information of the building automation system. For example, the connection relationship of system components in the current configuration information is that the chiller is connected to each fan coil unit and the cooling tower through pipelines. The functional attribute of the chiller is a device that provides a certain amount of refrigeration capacity, and its operating parameter thresholds include the normal range of the chiller's refrigeration efficiency, the range of the cold water inlet and outlet temperatures, etc. The functional attribute of the fan coil unit is an end device for adjusting the indoor temperature, and the normal range of its water valve opening, etc.

[0127] Analyze the fault propagation path, impact degree, and potential chain reactions in the target fault diagnosis features, identify the chiller and fan coil units directly affected by the fault, as well as the dehumidification equipment indirectly affected, and prioritize the identified affected components. Since the chiller is the source of the fault and has a fundamental impact on the operation of the entire HVAC system, the chiller is listed as the highest priority; the fan coil units are directly affected by the refrigeration effect of the chiller and are listed as the secondary priority; although the dehumidification equipment is for adjusting the humidity problem caused by regulating the water valve opening of the fan coil units and is indirectly affected, it also has a certain impact on the indoor environment and is listed as the third priority. Generate a fault impact range assessment report and a component priority ranking list.

[0128] According to the fault impact range assessment report and the component priority ranking list, search for matching control strategies in the pre-built control strategy library. The control strategy library contains a set of predefined control strategies for various fault scenarios. For the case of fouling in the evaporator of a chiller, control strategies such as thoroughly cleaning the evaporator, replacing some damaged components, and optimizing the operating parameters of the chiller may be found. Generate corresponding control plans based on the matching results. For example, determine to clean the evaporator first, and adjust the operating parameters of the chiller during the cleaning process, such as reducing the operating frequency of the compressor to reduce energy consumption and protect the equipment. At the same time, adjust the opening degree of the fan coil water valve and the operating parameters of the dehumidification equipment according to the actual needs of different areas.

[0129] Use the simulation environment or digital twin model of the building automation system to conduct simulation tests on the control plan and generate a simulation test report. In the simulation test, simulate the actual situation of fouling in the evaporator of the chiller, operate according to the control plan, and observe the changes in the operating parameters of each component, the control effect of indoor temperature and humidity, and the energy consumption situation, etc. It is found that the refrigeration efficiency has been improved after cleaning the evaporator, but if the operating frequency of the compressor is completely reduced according to the setting during the cleaning process, it will cause the cleaning time to be too long and affect the normal operation of the entire system.

[0130] Based on the simulation test report, optimize the control plan to generate the final control decision result. The optimization process involves adjusting control parameters, such as adjusting the operating frequency of the compressor during the evaporator cleaning process to ensure both the cleaning effect and not overly extend the cleaning time; modifying control steps, such as adding a step of jointly debugging the chiller and the fan coil after cleaning; adding local control measures, such as comprehensively checking the entire HVAC system before cleaning the evaporator to ensure that there are no potential problems with other components.

[0131] Transmit the final control decision result to the execution layer of the building automation system to guide the building automation system to implement control operations according to the final control decision result. For example, the execution layer starts the evaporator cleaning work according to the optimized time and conditions, adjusts the operating frequency of the compressor, conducts joint debugging of the chiller and the fan coil according to the new debugging steps, and controls the opening degree of the fan coil water valve and the operating parameters of the dehumidification equipment according to the adjusted parameters, so as to effectively control the faults of the HVAC system and ensure the normal operation of the building.

[0132] For the fault disturbance event of a light intensity sensor failure in a certain area of the lighting system. The decision optimization unit obtains the first state feature space as the basis for fault diagnosis, and at the same time obtains information related to the fault disturbance event. The fault type is a light intensity sensor failure, and the affected components are the lighting control system associated with the sensor and the lamps in this area.

[0133] Initialize relevant parameters. In terms of the coefficient for measuring the possibility of fault propagation, since a fault in the illuminance sensor may cause misjudgment in the lighting control system, which in turn affects all the lamps in this area, the coefficient for propagation to all the lamps in this area is set to 0.8. In terms of the initial values of the indicators for evaluating the effect of the regulation decision, the initial evaluation value for the lighting brightness in this area to meet the normal working requirements (different illuminance standards according to different area functions) is set to 0.8, and the initial evaluation value for the energy consumption of the entire lighting system within a reasonable range is set to 0.7.

[0134] Initialize the initial state information of the fault. When the illuminance sensor fails, the collected illuminance data is inaccurate, which may cause the lighting control system to think that the illuminance in this area is lower than the actual value.

[0135] In the first round of iteration, analyze the fault propagation path. The fault of the illuminance sensor affects the control signals of the lamps connected to it through the connection relationship with the lighting control system. The data flow is from the illuminance sensor to the lighting control system and then to the lamps. The interaction mechanism is that the illuminance sensor provides illuminance information to the lighting control system, and the lighting control system controls the on / off and brightness of the lamps according to this information. It is determined that the possibility of the fault spreading from the illuminance sensor to the lamps is relatively high. Based on these analysis results, generate preliminary regulation decision data, such as switching to a backup illuminance sensor (if available), or preliminarily adjusting the brightness of the lamps according to the historical illuminance data, current time, personnel activities, etc. in this area. For example, appropriately increase the brightness during office hours.

[0136] Apply the preliminary regulation decision data to the current state information of the fault to obtain the fault disturbance state information after the first round of iteration. At this time, if the backup illuminance sensor is switched, the lighting control system receives new accurate illuminance information, and the brightness of the lamps is adjusted according to the new regulation decision.

[0137] From the second round of iteration to the (X - 1)-th round of iteration, re-evaluate the fault propagation situation based on the fault disturbance state information after the previous round of iteration. In the second round of iteration, it is found that although the backup illuminance sensor is switched or the brightness of the lamps is adjusted, due to changes in personnel activities in the area, the lighting effect in some areas is still not ideal. According to the new propagation analysis results, adjust the regulation decision data, such as dynamically adjusting the brightness of the lamps according to the specific positions of the personnel, adopting a zoning control method, dividing the area into different areas such as the office area and the aisle, and adjusting the brightness separately. Apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. In subsequent iterations, continue to optimize the regulation decision according to factors such as personnel activities and time changes. For example, adjust the lighting in the meeting room area to a brightness mode suitable for meetings during the meeting time.

[0138] In the Xth iteration, final fault diagnosis feature extraction is carried out. It is determined that the propagation range of the fault in the entire lighting system is limited to this area. The degree of influence on key components (lamps and lighting control systems) is that the brightness of the lamps has been adjusted multiple times and basically meets the requirements, but the lighting control system has experienced a complex adjustment process due to the failure of the illuminance sensor. No other chain reactions have been triggered, such as having no impact on the HVAC system or elevator system. Feature information reflecting the adaptive regulation decision results of the fault perturbation propagation in the Xth iteration cycle is extracted, such as the final brightness adjustment value of the lamps, the final operating mode of the lighting control system, etc., to obtain the target fault diagnosis features.

[0139] After receiving the target fault diagnosis features, the execution strategy unit analyzes the fault propagation path, degree of influence, and potential chain reactions therein, identifies the illuminance sensor and lamps directly affected by the fault, lists the illuminance sensor as the highest priority, and the lamps as the secondary priority. A fault impact range assessment report and a component priority ranking list are generated.

[0140] According to the report and the list, search for matching regulation strategies in the regulation strategy library, and strategies such as repairing or replacing the faulty illuminance sensor and optimizing the control algorithm of the lighting control system may be found, and corresponding regulation plans are generated.

[0141] Use the simulation environment or digital twin model to simulate and test the regulation plan, such as simulating the activities of personnel at different time periods and observing the lighting effect and energy consumption. It is found that if the illuminance sensor is replaced immediately, it may cause fluctuations in the lighting system in a short period of time.

[0142] Optimize the regulation plan based on the simulation test report, such as adjusting the replacement time of the illuminance sensor to a time period with less personnel activity, and optimizing the control algorithm of the lighting control system so that it can better adapt to changes in personnel activities during the sensor failure. Generate the final regulation decision result and transmit it to the execution layer. The execution layer replaces the illuminance sensor at the appropriate time according to the result and controls the brightness of the lamps according to the optimized algorithm and other operations.

[0143] Regarding the fault perturbation event of misjudgment of the elevator car overload alarm device in the elevator system. The decision optimization unit obtains the first state feature space and information related to the fault perturbation event. The fault type is misjudgment of the car overload alarm device, and the affected components are the elevator control system and the operation of the elevator car.

[0144] Initialize relevant parameters. Regarding the coefficient for measuring the possibility of fault propagation, since misjudgment by the overload alarm device may lead to incorrect operation of the elevator control system, the coefficient for propagation to the elevator control system is set to 0.9. Regarding the initial values of the indicators for evaluating the effectiveness of regulation and control decisions, the initial evaluation value for the elevator to operate safely and reliably (without incorrect overload alarms and abnormal operations) is set to 0.8, and the initial evaluation value for the elevator operation efficiency within the normal range is set to 0.7.

[0145] Initialize the initial state information of the fault. When the car overload alarm device misjudges, it may send an alarm signal when there is no overload, resulting in the elevator stopping operation or restricting the passenger capacity.

[0146] In the first round of iteration, analyze the fault propagation path. The car overload alarm device is connected to the elevator control system. The misjudgment signal will affect the elevator control system's judgment of the car load, and thus affect the elevator's operation decision. The data flow is from the overload alarm device to the elevator control system, and the interaction mechanism is that the overload alarm device provides the car load information to the elevator control system. It is determined that the possibility of the fault propagating from the overload alarm device to the elevator control system is very high. According to the analysis results, generate preliminary regulation and control decision data, such as performing preliminary calibration on the overload alarm device, checking whether the connection of the sensor is normal, and at the same time adjusting the elevator control system's temporary processing strategy for overload alarms, such as performing a simple load review first when an alarm occurs instead of immediately stopping the operation.

[0147] Apply the preliminary regulation and control decision data to the current state information of the fault to obtain the fault disturbance state information after the first round of iteration. At this time, the misjudgment situation of the overload alarm device may be improved after calibration, and the elevator control system's temporary processing strategy begins to take effect.

[0148] From the second round of iteration to the (X - 1)-th round of iteration, re-evaluate the fault propagation situation. In the second round of iteration, it is found that although preliminary calibration and adjustment of the temporary processing strategy have been carried out, due to factors such as vibration during the elevator operation, the overload alarm device still occasionally misjudges. According to the new propagation analysis results, adjust the regulation and control decision data, such as further deeply detecting the sensors of the overload alarm device, replacing the possibly damaged components, and at the same time optimizing the elevator control system's processing logic for overload alarms, comprehensively judging whether there is a real overload based on information such as the car's running speed and acceleration. Apply the new regulation and control decision data to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. In subsequent iterations, continue to optimize the regulation and control decisions according to the actual operation situation of the elevator, such as adjusting the elevator's tolerance for overload alarms according to the passenger demand on different floors.

[0149] In the Xth iteration, final fault diagnosis feature extraction is carried out. It is determined that the propagation range of the fault in the entire elevator system mainly focuses on the elevator car and the elevator control system. The degree of influence on key components (overload alarm device and elevator control system) is that the misjudgment situation of the overload alarm device is significantly reduced after multiple adjustments, and the operation logic of the elevator control system can better handle the overload alarm situation after optimization. No other chain reactions are triggered, such as no impact on the lighting system or HVAC system. Feature information reflecting the adaptive control decision results of the fault disturbance propagation in the Xth iteration cycle is extracted, such as the final calibration parameters of the overload alarm device, the final operation logic of the elevator control system, etc., to obtain the target fault diagnosis features.

[0150] After receiving the target fault diagnosis features, the execution strategy unit analyzes the fault propagation path, influence degree, and potential chain reactions therein, identifies the overload alarm device and the elevator control system directly affected by the fault, lists the overload alarm device as the highest priority, and the elevator control system as the secondary priority. A fault impact range assessment report and a component priority ranking list are generated.

[0151] According to the report and the list, search for matching control strategies in the control strategy library. Strategies such as a comprehensive overhaul of the overload alarm device and an upgrade of the algorithm of the elevator control system may be found, and corresponding control plans are generated.

[0152] Use the simulation environment or digital twin model to simulate and test the control plan. For example, simulate the operation of the elevator under different passenger loads and operating scenarios, and it is found that if the algorithm of the elevator control system is directly upgraded, there may be compatibility problems with the existing equipment.

[0153] Optimize the control plan based on the simulation test report. For example, first conduct a compatibility test on the elevator control system, then upgrade the algorithm in stages. When comprehensively overhauling the overload alarm device, use advanced detection equipment to ensure the overhaul quality. Generate the final control decision result and transmit it to the execution layer. The execution layer performs operations such as overhauling the overload alarm device and upgrading the elevator control system according to the optimized algorithm according to the result.

[0154] For the fault disturbance event of the access control system card reader failure in the security system. The decision optimization unit obtains the first state feature space and information related to the fault disturbance event. The fault type is the access control system card reader failure, and the affected components are the access control system and the associated monitoring system (used to record the entry and exit of personnel).

[0155] Initialize relevant parameters. In terms of the coefficient for measuring the possibility of fault propagation, since a card reader failure may cause the access control system to be unable to correctly identify the identity of personnel, the coefficient transmitted to the access control system is set to 0.9. In terms of the initial values of the indicators for evaluating the effect of regulation decisions, the initial evaluation value for the access control system to accurately identify the identity of personnel (the false positive rate is lower than a certain value) is set to 0.8, and the initial evaluation value for the overall security of the security system to be maintained at a normal level is set to 0.7.

[0156] Initialize the initial state information of the fault. When the card reader of the access control system fails, it cannot correctly read the access card information of personnel, resulting in personnel being unable to enter or exit normally, or there may be a situation where illegal personnel try to enter using the fault loopholes.

[0157] In the first round of iteration, analyze the fault propagation path. The card reader is connected to the access control system. The fault will cause the access control system to be unable to obtain the correct personnel identity information, thereby affecting the normal opening and closing control of the access. The data flow is from the card reader to the access control system, and the interaction mechanism is that the card reader transmits the read access card information to the access control system for identity verification. It is determined that the possibility of the fault spreading from the card reader to the access control system is extremely high. According to the analysis results, generate preliminary regulation decision data, such as restarting the card reader, checking whether the connection line between the card reader and the access control system is loose, and at the same time adjusting the temporary response strategy of the access control system to the card reader failure. For example, during the fault period, adopt the method of manually verifying the identity (such as checking other valid certificates such as work permits) as a supplementary means to control the entry and exit of personnel.

[0158] Apply the preliminary regulation decision data to the current state information of the fault to obtain the fault disturbance state information after the first round of iteration. At this time, the card reader may partially recover its function after being restarted. If the loose situation is found and repaired after checking the connection line, the temporary response strategy of the access control system begins to take effect, and the entry and exit of personnel are managed to a certain extent.

[0159] From the second round of iteration to the (X - 1)-th round of iteration, re-evaluate the fault propagation situation. In the second round of iteration, it is found that although the restart and line inspection have been carried out, the card reader still frequently fails, which may be due to damage to the internal components of the card reader. According to the new propagation analysis results, adjust the regulation decision data, such as arranging professional personnel to conduct internal detection on the card reader, preparing to replace the possibly damaged components, and at the same time optimizing the personnel identity verification process of the access control system during the maintenance period, increasing the level of identity verification. For example, in addition to checking the certificates, it is also necessary to verify the personnel information with the relevant departments. Apply the new regulation decision data to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration. In subsequent iterations, continue to optimize the regulation decision according to the detection results of the card reader and the actual situation of personnel entry and exit, such as adjusting the strictness of manual identity verification according to the security levels of different regions.

[0160] In the X-th iteration, final fault diagnosis feature extraction is performed. It is determined that the propagation range of the fault in the entire security system mainly focuses on the access control system card reader and the access control system. The degree of influence on the key components (card reader and access control system) is that the function of the card reader gradually returns to stability after multiple detection and repair operations, and the personnel authentication process of the access control system can better handle the card reader failure situation after optimization. No other chain reactions are triggered, such as no impact on the elevator system or lighting system. Feature information reflecting the adaptive control decision result of the fault disturbance propagation in the X-th iteration cycle is extracted, such as the final repair status of the card reader, the final personnel authentication process of the access control system, etc., to obtain the target fault diagnosis feature.

[0161] After receiving the target fault diagnosis feature, the execution strategy unit analyzes the fault propagation path, influence degree, and potential chain reactions therein, identifies the card reader and the access control system directly affected by the fault, lists the card reader as the highest priority, and the access control system as the secondary priority. A fault impact range assessment report and a component priority ranking list are generated.

[0162] According to the report and the list, search for matching control strategies in the control strategy library, and strategies such as completely replacing the card reader and upgrading the authentication algorithm of the access control system may be found, and corresponding control plans are generated.

[0163] Use the simulation environment or digital twin model to simulate and test the control plan. For example, simulate the operation of the access control system under different personnel flow and security threat scenarios, and it is found that if the card reader is directly replaced, it may cause incompatibility with some functions of the existing access control system, such as problems with the linkage with the monitoring system.

[0164] Optimize the control plan based on the simulation test report. For example, first conduct compatibility tests on the new card reader to ensure that all functions of the access control system and the monitoring system can work properly together before replacement, and at the same time gradually upgrade the authentication algorithm of the access control system to ensure that it does not affect the normal entry and exit of personnel during the upgrade process. Generate the final control decision result and transmit it to the execution layer. The execution layer performs operations such as replacing the card reader (if necessary) and upgrading the access control system according to the optimized algorithm, so as to ensure the normal operation of the security system and maintain the safety of the building.

[0165] In a possible implementation manner, before step S120, the method further includes:

[0166] Step S101: Obtain the monitoring data of the third template state variable and the corresponding third template state feature space of the monitoring data of the third template state variable. The third template state feature space is generated by the sample building state recognition network through feature space transformation of the monitoring data of the third template state variable.

[0167] In this embodiment, the monitoring data of the third template state variable comes from each subsystem of the building automation system, including the heating, ventilation and air conditioning (HVAC) system, lighting system, elevator system, security system, etc. Taking the HVAC system as an example, the monitoring data of the third template state variable includes various operating parameters of the chiller, such as the rotational speed of the compressor, the temperatures of the condenser and evaporator, the inlet and outlet water temperatures of the chilled water, the flow rate, etc., and also includes the state data of the terminal equipment of the air conditioner, such as the rotational speed of the fan coil unit and the opening degree of the water valve. The monitoring data of the third template state variable in the lighting system covers the illuminance values collected by the illuminance sensors in each area, the on / off state of the lamps, the dimming ratio, etc. The monitoring data of the third template state variable in the elevator system includes the running speed of the elevator, the position of the car, the passenger capacity, the opening and closing state of the elevator door, etc. The security system includes the personnel entry and exit records of the access control system, the processed personnel flow information collected by the monitoring cameras, etc. The third template state feature space is generated by the sample building state recognition network through feature space transformation of the monitoring data of the third template state variable. This feature space can reflect the characteristic relationships of these state variable data in a specific mode. For example, in the HVAC system, it may reflect the correlation between the operating parameters of the chiller and the indoor environmental temperature and humidity, as well as the mutual influence relationship between the components of the air conditioning system.

[0168] Step S102: The feature pattern editing model performs feature pattern editing on the third template state feature space to the conversion result pattern, and generates estimated feature pattern information. The conversion result pattern is used to represent the feature pattern matched by the conversion result generated by the feature conversion module through feature conversion.

[0169] Taking the HVAC system as an example, the feature pattern editing model edits the feature relationships related to the chiller and fan coil in the third template state feature space. If there is a relatively complex non-linear relationship between the refrigeration efficiency of the chiller and the refrigeration effect of the fan coil in the third template state feature space, the feature pattern editing model may edit it into a linear relationship or a specific functional relationship form that is easier to understand and process according to the requirements of the conversion result pattern, so as to generate the estimated feature pattern information. For the lighting system, if the relationship between the illuminance value and the lamp dimming ratio in the third template state feature space is based on a certain statistical law, the feature pattern editing model will edit it into a relationship form that meets the expectations of the conversion result of the feature conversion module according to the conversion result pattern, for example, converting it into a pattern relationship directly related to lighting comfort.

[0170] Step S103, perform adaptive learning by the adaptive learning network model according to the estimated feature pattern information to generate template adaptive learning features that match the first feature pattern.

[0171] The adaptive learning network model analyzes various feature pattern relationships in the estimated feature pattern information. For the HVAC system, according to the edited relationship pattern between the chiller and the fan coil, the adaptive learning network model learns a new feature pattern. For example, under different indoor and outdoor temperature differences and different personnel densities, how the chiller and the fan coil work together to achieve the best refrigeration or heating effect and energy utilization efficiency, which is part of the generated template adaptive learning features. In terms of the lighting system, the adaptive learning network model learns how to adjust the dimming ratio of the lamps to achieve the best lighting effect and energy-saving goal under different time periods (such as day and night) and different area functions (such as office areas and rest areas) according to the edited relationship pattern between the illuminance and the dimming ratio, which also becomes part of the template adaptive learning features.

[0172] Step S104, determine the third training cost according to the template adaptive learning features and the third template state feature space.

[0173] For the HVAC system, compare the characteristics of the chiller and fan coil unit working together in the adaptive learning features of the comparison template with the relationship of the original features in the third template state feature space. If there are differences between the optimal operating parameters of the chiller and fan coil unit under a certain working condition in the adaptive learning features of the template and the original parameters in the third template state feature space, this difference will be quantified and calculated. Similarly, the differences in illuminance and dimming ratio in the lighting system between the adaptive learning features of the template and the third template state feature space are also calculated. By combining these differences, the third training cost is determined. The elevator system and security system also perform similar difference calculations according to their respective characteristics and participate in the determination process of the third training cost.

[0174] Step S105, based on the third training cost, optimize the neuron weight information of the adaptive learning network model and the feature pattern editing model until the third convergence requirement is met.

[0175] Finally, if the third training cost is large, it indicates that the current states of the adaptive learning network model and the feature pattern editing model deviate significantly from the desired states. For the adaptive learning network model, in the HVAC system, if the feature learning results related to the chiller and fan coil unit deviate significantly, the neuron weights related to the chiller and fan coil unit will be adjusted so that more accurate feature patterns that meet the requirements can be generated in subsequent learning. For the feature pattern editing model, if a large training cost is incurred when editing the relationship between the chiller and fan coil unit, the neuron weights related to this relationship editing will be adjusted. As the adjustment continues, the third training cost gradually decreases. When the third convergence requirement is met (for example, the third training cost is less than a certain set threshold), it indicates that the adaptive learning network model and the feature pattern editing model have reached a relatively ideal state under the current error criterion and can provide an accurate basis for the feature space conversion of the subsequent state variable monitoring data.

[0176] Figure 2 The hardware structure diagram of the AI-based intelligent control system 100 for the building automation system for implementing the above-mentioned AI-based intelligent control method for the building automation system provided by the embodiment of the present invention is shown. As Figure 2 shown, the AI-based intelligent control system 100 for the building automation system may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0177] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions that the AI-based intelligent control system 100 of the building automation system uses to execute or complete the exemplary methods described in the present invention.

[0178] In a specific implementation process, one or more processors 110 execute the computer-executable instructions stored in the machine-readable storage medium 120, so that the processors 110 can execute the AI-based intelligent control method of the building automation system as described in the above method embodiments. The processors 110, the machine-readable storage medium 120, and the communication unit 140 are connected through a bus 130. The processors 110 can be used to control the transceiver actions of the communication unit 140.

[0179] For the specific implementation process of the processors 110, reference can be made to the respective method embodiments executed by the above AI-based intelligent control system 100 of the building automation system. Their implementation principles and technical effects are similar, and will not be elaborated here in this embodiment.

[0180] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When a processor executes the computer-executable instructions, the above AI-based intelligent control method of the building automation system is implemented.

[0181] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. An AI-based intelligent control method for building automation system, characterized in that: The method comprises: Obtain state variable monitoring data of the building automatic control system; The state variable monitoring data is subjected to feature space conversion by an adaptive learning network model to generate a first state feature space matching a first feature pattern, wherein the first feature pattern is a feature pattern matched by the state feature space extracted by the building state recognition network; According to the first state feature space, guiding the building automatic control system control network to respectively make adaptive control decisions of fault disturbance propagation for a plurality of different fault disturbance events, generating a plurality of different adaptive control decision results of the building operation category associated with the first state feature space, wherein the plurality of different adaptive control decision results of the building operation category associated with the first state feature space are also used to optimize the building state recognition network; Based on a plurality of different adaptive control decision results of the building operation category associated with the first state feature space, an adaptive control decision is made for the building automatic control system.

2. The AI-based intelligent control method for building automation system according to claim 1 is characterized in that: The building automation system control network includes a decision optimization unit and an execution strategy unit; The method of guiding the building automatic control system control network to perform adaptive control decisions on fault disturbance propagation for a plurality of different fault disturbance events based on the first state feature space, and generating a plurality of different adaptive control decision results for the building operation category associated with the first state feature space, includes: For each of the fault disturbance events in a plurality of different fault disturbance events, the decision optimization unit uses the first state feature space as a fault diagnosis basis, performs adaptive control decisions of fault disturbance propagation for X iteration cycles on the fault disturbance event, generates a target fault diagnosis feature obtained by the adaptive control decision of fault disturbance propagation for the Xth iteration cycle, and uses the fault diagnosis feature obtained by the adaptive control decision of fault disturbance propagation for the ath iteration cycle as an input of the adaptive control decision of fault disturbance propagation for the a+1th iteration cycle, where a and X are positive integers and a<X; The execution strategy unit performs adaptive control decision of fault disturbance propagation on the target fault diagnosis feature to generate an adaptive control decision result of the building operation category associated with the first state feature space.

3. The AI-based intelligent control method for building automation system according to claim 2 is characterized in that: Before the step of guiding the building automatic control system control network to perform adaptive control decisions on fault disturbance propagation for a plurality of different fault disturbance events respectively based on the first state feature space, and generating a plurality of different adaptive control decision results for the building operation category associated with the first state feature space, the method further includes: Obtaining first template state variable monitoring data, a first template state feature space corresponding to the first template state variable monitoring data, and a set number of iteration cycles b determined for the first template state variable monitoring data, wherein the first template state feature space is generated by a sample building state recognition network performing feature space conversion on the first template state variable monitoring data, and the sample building state recognition network is the building state recognition network or other neural network models with the same network task as the building state recognition network; b is a positive integer not greater than X; The feature conversion module performs feature conversion on the first template state variable monitoring data to generate a first conversion result; Performing b-iteration cycles of decision interference addition on the first conversion result by the decision optimization unit to generate decision interference addition data; the decision interference addition data includes the target template fault interference feature obtained by the decision interference addition of the b-th iteration cycle; The decision optimization unit uses the first template state feature space as a fault diagnosis basis to perform adaptive control decisions on the fault disturbance propagation of b iteration cycles on the target template fault interference feature to generate a template fault diagnosis feature; The execution strategy unit performs adaptive control decision of fault disturbance propagation on the template fault diagnosis feature to generate adaptive control decision data; Extracting the state feature space of the adaptive control decision data through the sample building state recognition network to generate a template state feature space corresponding to the template fault diagnosis feature; Determining a state feature space error based on the first template state feature space and the template state feature space; Based on the state feature space error, the neuron weight information of the decision optimization unit is optimized until a first convergence requirement is met.

4. The AI-based intelligent control method for building automation system according to claim 3 is characterized in that: The determining of a state feature space error based on the first template state feature space and the template state feature space includes: Based on the set number of iteration cycles b, determining an adjustment factor, wherein the adjustment factor is in an inverse correlation relationship with the set number of iteration cycles b; Calculating the distance between the first template state feature space and the template state feature space; The distance is adjusted according to the adjustment factor to generate a state feature space error.

5. The AI-based intelligent control method for building automation system according to claim 3 is characterized in that: The adaptive control decision result also includes the disturbance control estimated in the adaptive control decision process of the fault disturbance propagation in b iteration cycles; The method further comprises: Determining a disturbance control error based on a fault disturbance feature added by adding decision disturbances for b iterative cycles to the first conversion result and a disturbance control feature estimated in an adaptive control decision process of fault disturbance propagation for b iterative cycles to the fault disturbance feature of the target template; The optimizing the neuron weight information of the decision optimization unit based on the state feature space error until the first convergence requirement is met includes: The state feature space error and the disturbance control error are fused and calculated to generate a first training cost; Based on the first training cost, the neuron weight information of the decision optimization unit is optimized until a first convergence requirement is met.

6. The AI-based intelligent control method for building automation system according to claim 3 is characterized in that: The optimizing the neuron weight information of the decision optimization unit based on the state feature space error until the first convergence requirement is met includes: The neuron weight information of the feature conversion module and the execution strategy unit is locked, and based on the state feature space error, the neuron weight information of the decision optimization unit is optimized until a first convergence requirement is met.

7. The AI-based intelligent control method for building automation system according to claim 3 is characterized in that: Before the feature conversion module performs feature conversion on the first template state variable monitoring data to generate a first conversion result, the method includes: Obtaining second template state variable monitoring data; Performing feature conversion on the second template state variable monitoring data by the feature conversion module to generate a second conversion result; Performing an adaptive control decision of fault disturbance propagation on the second conversion result through the execution strategy unit to generate state variable monitoring data of an adaptive control decision template of fault disturbance propagation; Determining a second training cost based on the second template state variable monitoring data and the adaptive control decision template state variable monitoring data of the fault disturbance propagation; The neuron weight information of the feature conversion module and the execution strategy unit is optimized based on the second training cost until a second convergence requirement is met.

8. The AI-based intelligent control method for building automation system according to claim 2 is characterized in that: The step of using the first state feature space as a fault diagnosis basis for each of the multiple different fault disturbance events through the decision optimization unit, performing adaptive control decisions of fault disturbance propagation for X iteration cycles on the fault disturbance event, and generating a target fault diagnosis feature obtained by the adaptive control decision of fault disturbance propagation for the Xth iteration cycle includes: The decision optimization unit is used to obtain the first state feature space as a fault diagnosis basis, and simultaneously obtains information related to the fault disturbance event to be processed, wherein the information related to the fault disturbance event includes the type of the fault disturbance event and information on components of the building automatic control system affected; Initializing relevant parameters of the fault disturbance event through the decision optimization unit, wherein the relevant parameters include a coefficient for measuring the possibility of fault propagation and an initial value of an indicator for evaluating the effect of the control decision; Initializing information indicating an initial state of the fault disturbance event according to the relevant information of the fault disturbance event to obtain initial state information; In the first round of iteration, the decision optimization unit analyzes the fault disturbance event based on the initialized relevant parameters and the initial state information of the fault to generate an analysis result. Specifically, the propagation path of the fault disturbance event in the current building automatic control system environment, the connection relationship between the various components in the building automatic control system, the data flow direction and the interaction mechanism are analyzed to determine the potential direction and possibility of the fault disturbance event propagating from the initially affected component to other components. According to the analysis result of the fault disturbance propagation, preliminary control decision data is generated, wherein the preliminary control decision data is generated based on the global operation state of the building reflected by the first state feature space and the preliminary analysis of the current fault disturbance, and is used to preliminarily suppress or adjust the propagation of the fault disturbance; After generating the preliminary control decision data, applying the preliminary control decision data to the current state information of the fault disturbance event to obtain the fault disturbance state information after the first round of iteration, wherein the fault disturbance state information reflects the state change of the fault disturbance event after the implementation of the preliminary control decision; From the second round of iteration to the X-1th round of iteration, the decision optimization unit re-evaluates the propagation of the fault disturbance event in the building automatic control system based on the fault disturbance state information after the previous round of iteration, and adjusts the previous control decision data according to the new propagation analysis result. After adjusting the control decision data, the new control decision data is applied to the current fault disturbance state information to obtain the fault disturbance state information after the current iteration, and the fault disturbance state information after the current iteration reflects the latest state information of the fault disturbance event after multiple rounds of iterations and control decision adjustment; In the Xth round of iteration, the decision optimization unit performs final fault diagnosis feature extraction based on the fault disturbance state information after the X-1th round of iteration; wherein, the current fault disturbance state information is specifically analyzed to determine the final propagation range of the fault disturbance in the entire building automatic control system, the final impact on each key component, and whether other chain reactions have been triggered, and then extracts feature information reflecting the adaptive control decision result of the fault disturbance propagation in the Xth iteration cycle based on the analysis results to obtain the target fault diagnosis feature; The step of performing an adaptive control decision of fault disturbance propagation on the target fault diagnosis feature through the execution strategy unit to generate an adaptive control decision result of the building operation category associated with the first state feature space includes: Receiving, through the execution strategy unit, a target fault diagnosis feature from the decision optimization unit, the target fault diagnosis feature including fault disturbance propagation information optimized through multiple iteration cycles, and current configuration information of the building automatic control system, the current configuration information including connection relationships, functional attributes, and operating parameter thresholds of system components; Analyze the fault propagation path, impact degree and potential chain reaction in the target fault diagnosis characteristics, identify the affected components directly and indirectly affected by the fault, prioritize the identified affected components, and generate a fault impact range assessment report and a component priority ranking list; According to the fault impact scope assessment report and the component priority ranking list, a matching control strategy is searched in a pre-built-in control strategy library, and a corresponding control scheme is generated based on the matching result. The control strategy library contains a set of predefined control strategies for various fault scenarios; Using the simulation environment or digital twin model of the building automation system, simulate the control scheme and generate a simulation test report; Based on the simulation test report, the control scheme is optimized to generate a final control decision result, wherein the optimization process involves adjusting control parameters, modifying control steps, and adding or deleting local control measures; The final control decision result is transmitted to the execution layer of the building automatic control system, and the building automatic control system is guided to implement the control operation according to the final control decision result.

9. The AI-based intelligent control method for building automatic control system according to any one of claims 3 to 7, characterized in that: Before the adaptive learning network model performs feature space conversion on the state variable monitoring data to generate a first state feature space matching the first feature pattern, the method further includes: Acquire third template state variable monitoring data and a third template state feature space corresponding to the third template state variable monitoring data, wherein the third template state feature space is generated by the sample building state recognition network performing feature space conversion on the third template state variable monitoring data; The feature pattern editing model performs feature pattern editing on the third template state feature space to a conversion result pattern to generate estimated feature pattern information; the conversion result pattern is used to represent the feature pattern matched by the conversion result generated by the feature conversion module; Performing adaptive learning based on the estimated feature pattern information through the adaptive learning network model to generate a template adaptive learning feature matching the first feature pattern; Determining a third training cost according to the template adaptive learning feature and the third template state feature space; Based on the third training cost, the neuron weight information of the adaptive learning network model and the characteristic pattern editing model is optimized until the third convergence requirement is met.

10. An AI-based intelligent control system for building automation system, characterized in that: The AI-based building automatic control system intelligent control system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the AI-based building automatic control system intelligent control method described in any one of claims 1 to 9 above.

Images