A method for automatically checking before inputting into intelligent regulation and control of building heating and ventilation system

Abstract

This invention discloses an automatic pre-implementation check method for intelligent control of building HVAC systems. The method includes: standardizing the definition of connections and associated equipment within the building HVAC system to obtain a standardized process description language for the system and storing it in a database; selecting algorithm availability based on the building HVAC system to determine the corresponding control algorithm and input / output parameters; verifying the dependency relationship of input / output parameters based on a dynamic parameter dependency verification mechanism for the physical links of the building HVAC system; and conducting a baseline assessment of the control adaptability data for input / output parameters with successful dependency verification, and performing control logic simulation and verification by calling APIs. This invention can efficiently help AI algorithm users or developers conduct thorough self-checks before implementing algorithms, ensuring their reliability before deployment; it also helps AI algorithm practitioners optimize algorithm engineering.

Description

Technical Field

[0001] This application relates to the field of HVAC control technology, and more specifically, to an automatic inspection method applicable to the intelligent control of building HVAC systems before implementation. Background Technology

[0002] With the development of green and intelligent buildings, HVAC systems, as a major component of building energy consumption, are increasingly adopting intelligent control technologies. Intelligent control typically relies on big data analytics, artificial intelligence algorithms (such as predictive control and adaptive learning), and IoT platforms to achieve dynamic optimization control of parameters such as temperature, humidity, fresh air volume, and energy consumption.

[0003] However, although numerous algorithms for intelligent control of building HVAC systems have emerged, most are still in the demo or experimental stage and have not been widely used. Furthermore, due to significant differences in the design concepts of these algorithms, such as many algorithms only considering the optimization of single process nodes or single links, they have not formed a closed-loop and long-term operation system dominated by algorithm control.

[0004] Before integrating or switching the intelligent control system into the actual operating environment, due to project differences and system initialization, the lack of a systematic pre-implementation inspection mechanism may lead to the following problems:

[0005] 1. Incorrect HVAC process identification leads to mismatch in AI algorithm selection: Building HVAC systems are divided into heating and cooling systems based on their supply methods. The process characteristics of these heating and cooling systems differ significantly, such as the form of energy station, heat exchange station, and terminal equipment. If the self-inspection mechanism cannot fully consider these influencing factors, it will lead to incorrect process identification and automatically select the wrong algorithm for input.

[0006] 2. Incomplete identification of dependent parameters leads to the inability of AI algorithm to start: It is currently difficult to develop a general algorithm in building HVAC systems. Due to different algorithm design principles, the dependent algorithm parameter conditions are inconsistent, which leads to the inability to start the algorithm when it is directly applied.

[0007] 3. Insufficient data affects AI algorithm training: AI algorithm adjustment is divided into two situations based on the data situation: no data or a small amount of data (generally referring to less than 1 month of running data) and a large amount of data (generally referring to more than 1 month of running data). Different strategies will be adopted before the algorithm is put into training for different data. For example, when the amount of data is less than 1 month, the algorithm is in a waiting training state and is not enough to be put into training.

[0008] 4. Disordered control parameters prevent AI algorithms from selecting appropriate control parameters: Building HVAC systems involve numerous control parameters, but AI algorithms generally control a limited number of parameters. If all parameters are included in the algorithm, it will lead to disordered control of similar process components. For example, the main pipe has a branch pipe with parallel bypass pipes and two parallel electric valves, namely electric valve #1 and electric valve #2. If it cannot be effectively identified during self-test, the algorithm will control randomly, which is not the optimal choice for the system.

[0009] Therefore, there is an urgent need for a systematic, quantifiable, and intelligent inspection method specifically designed for the pre-implementation of intelligent control of building HVAC systems, in order to improve the reliability and success rate of system deployment. Summary of the Invention

[0010] In view of the above problems, the purpose of this invention is to provide an automatic inspection method for intelligent control of building HVAC systems before implementation. Through a self-inspection mechanism encompassing process identification, algorithm selection, parameter verification, and control simulation, it can efficiently help AI algorithm users or developers conduct thorough self-inspections before algorithm implementation, ensuring the reliability of the algorithm. Simultaneously, it also helps AI algorithm practitioners optimize algorithm engineering. By implementing process identification, algorithm usability verification, and parameter verification, it effectively avoids scenarios of algorithm mismatch and parameter disorder control, preventing potential failures in advance. It provides quantifiable and traceable inspection processes and evaluation indicators, facilitating engineering implementation and quality control. Due to the inclusion of the process identification step, this self-inspection mechanism is highly adaptable to various building HVAC systems; it also supports the verification of multiple algorithms, meeting the self-inspection needs of different AI algorithms before implementation.

[0011] The first aspect of this invention provides an automatic pre-implementation check method for intelligent control of building HVAC systems, comprising:

[0012] The connection relationships and related equipment involved in building HVAC systems are standardized and defined to obtain a standardized process description language for building HVAC systems;

[0013] The standardized construction process description language for building HVAC systems is stored in the database;

[0014] Based on the building's HVAC system, algorithm availability is selected to determine the corresponding control algorithm;

[0015] The input and output parameters are determined according to the control algorithm.

[0016] Based on the dynamic parameter dependency verification mechanism for physical links of building HVAC systems, the input and output parameters are verified for dependencies; the dependency verification includes belonging dependency verification, multiple path dependency verification, and exclusion dependency verification.

[0017] For input and output parameters that have successfully verified dependencies, an adaptive data baseline assessment is performed. Based on a two-stage dynamic baseline data availability verification mechanism, during the training preparation period, a data continuity coverage verification based on the minimum effective period is performed; during the operation preparation period, an empirical-statistical hybrid anomaly identification based on HVAC operation modes is performed.

[0018] After completing the baseline assessment of the control adaptability data, the control logic simulation and verification are performed by calling the API;

[0019] Once the control logic simulation and verification are completed, the automatic check before the intelligent control of the building HVAC system is put into operation is confirmed.

[0020] In this solution, the standardized definition of the connection relationships and related equipment involved in the building HVAC system results in a standardized construction process description language for the building HVAC system, including:

[0021] Based on the dynamic topology modeling mechanism, the connection relationships and associated equipment involved in the building HVAC system are sequentially modeled in both structural and functional dimensions, undergoing event-driven model evolution, constructing an extensible semantic text library, and generating an API-based self-describing output interface. This transforms the connection relationships and associated equipment involved in the building HVAC system from physical entities into a standardized process description language for building HVAC systems.

[0022] This plan also includes:

[0023] The standardized construction process description language for building HVAC systems is packaged into an API and stored in a database.

[0024] In this solution, the step of selecting the appropriate control algorithm based on the building's HVAC system includes:

[0025] Based on the building HVAC system, one or more key features are selected from the corresponding process characteristics;

[0026] By analyzing one or more of the key features, the corresponding control algorithm is determined.

[0027] In this solution, the dependency verification specifically involves:

[0028] Generate strong binding logic between input / output parameters and their respective functional units;

[0029] Based on the functional units identified by the process model, determine whether the key auxiliary parameters of the functional units are configured in the building HVAC system.

[0030] If the aforementioned key auxiliary parameters are not configured, the verification is deemed to have failed, and a prompt message is generated based on the unconfigured key auxiliary parameters.

[0031] In this scheme, the verification of the multiple-choice path dependency relationship specifically includes:

[0032] When multiple optional control paths exist, each optional control path is evaluated based on the equipment operating characteristics to determine the optimal control path; the equipment operating characteristics include at least the equipment operating time, historical regulation sensitivity curve, and current energy efficiency ratio;

[0033] Activate the input and output parameters associated with the optimal control path, and control the input and output parameters associated with other optional paths to enter the standby locked state.

[0034] In this solution, the exclusionary dependency verification specifically involves:

[0035] Construct an exclusion relationship table;

[0036] If the input / output parameters exist in the exclusion relationship table, the verification is deemed to have failed, the input / output parameters are marked as exclusion parameters, and a prompt message is generated based on the exclusion parameters.

[0037] In this scheme, the data continuity coverage verification based on the minimum effective period specifically refers to:

[0038] Based on the minimum effective period, the frequency compliance of input and output data collection is verified to determine whether the data point collection time interval meets the preset minimum sampling requirement and whether the effective data coverage is greater than or equal to the preset minimum coverage.

[0039] When the data point collection time interval meets the preset minimum sampling requirement and the effective data coverage is greater than or equal to the preset minimum coverage, the collection frequency compliance verification is considered successful; otherwise, the collection frequency compliance verification fails.

[0040] After the compliance verification of the collection frequency is successful, the integrity verification of the operating mode is carried out. Based on the identified type of building HVAC system process, it is determined whether the input and output data cover typical operating conditions.

[0041] If yes, the modal integrity verification will succeed; otherwise, the modal integrity verification will fail.

[0042] In this solution, the empirical-statistical hybrid anomaly identification based on HVAC operation modes specifically refers to:

[0043] Construct a lightweight outlier self-detection algorithm that includes an empirical rule layer and a dynamic statistics layer;

[0044] A knowledge base of HVAC experts is established in the experience rule layer, and typical anomaly patterns are preset based on the knowledge base of HVAC experts.

[0045] In the dynamic statistics layer, a combination of moving window IQR and sliding Z-score is used for dynamic statistics, outliers are marked and their distribution is recorded to generate an anomaly heatmap.

[0046] In this solution, the step of performing control logic simulation and verification by calling APIs includes:

[0047] The API is used to query parameters, and the obtained parameters are controlled. The API is then verified based on a feedback mechanism that provides a given response.

[0048] Once all APIs have been verified, the control logic simulation and verification are complete.

[0049] This invention discloses an automatic pre-implementation check method for intelligent control of building HVAC systems. The method includes: standardizing the definition of connections and associated equipment within the building HVAC system to obtain a standardized process description language for the system and storing it in a database; selecting algorithm availability based on the building HVAC system to determine the corresponding control algorithm and input / output parameters; verifying the dependency relationship of input / output parameters based on a dynamic parameter dependency verification mechanism for the physical links of the building HVAC system; and conducting a baseline assessment of the control adaptability data for input / output parameters with successful dependency verification, and performing control logic simulation and verification by calling APIs. This invention can efficiently help AI algorithm users or developers conduct thorough self-checks before implementing algorithms, ensuring their reliability before deployment; it also helps AI algorithm practitioners optimize algorithm engineering.

[0050] Compared with the prior art, the present invention has the following advantages:

[0051] 1. Introducing a self-inspection mechanism for dynamic process identification: Traditional building HVAC algorithms generally adopt a static configuration method for process identification. For example, a set of static configuration files defines the basic process characteristics of the project, such as the project being a refrigeration system, the energy system being a chiller unit, including a manifold, and five branch pipes. This invention designs a general process description language that can be applied to the self-inspection business scenario before the AI ​​algorithm is deployed, effectively solving the problem of dynamic process identification.

[0052] 2. Introduction of a dynamic self-checking mechanism for algorithm selection: Companies in the building HVAC field generally invest less in research and development. Most of their research is based on a single basic algorithm and does not fully incorporate process characteristics to design multiple algorithms. At the same time, the algorithm selection mechanism is usually determined manually. This invention provides a mechanism based on process identification and matching algorithms, which solves the problem of multiple algorithms being put into joint operation and the algorithm being dynamically selected based on process identification.

[0053] 3. Three new dependency model verification mechanisms were designed to solve the problem of self-checking for complex parameter dependencies: HVAC system is a closed-loop link formed by main link + multiple branches. The overall process description is relatively complex, and the parameters are interdependent and mutually influential. This invention proposes three types of parameter dependency solutions: membership relationship, exclusion relationship, and multiple selection relationship, which fully match various process parameter dependency problems.

[0054] 4. Introducing a data quality self-checking mechanism and supporting algorithm self-checking from the start of training: Most self-checks are based on the model after the algorithm is trained, which has weak adaptability. This invention introduces a self-check before AI algorithm training, that is, it provides a basis for self-checking for projects that need to be retrained when there is insufficient data, which enhances the universality of self-checking and indirectly improves the adaptability of the algorithm. Attached Figure Description

[0055] Figure 1 This invention provides a flowchart of an automatic pre-implementation inspection method for intelligent control of building HVAC systems.

[0056] Figure 2 A flowchart of the dependency verification method provided by this invention is shown;

[0057] Figure 3 A flowchart of the multiple-selection path dependency verification method provided by the present invention is shown. Detailed Implementation

[0058] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0059] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0060] Figure 1 The flowchart illustrates an automatic pre-implementation inspection method for intelligent control of building HVAC systems provided by the present invention.

[0061] like Figure 1 As shown, this invention discloses an automatic inspection method for intelligent control of building HVAC systems before implementation, comprising:

[0062] S101 standardizes the connection relationships and related equipment involved in building HVAC systems, resulting in a standardized process description language for building HVAC systems.

[0063] S102 stores the standardized construction process description language for building HVAC systems into the database;

[0064] S103, Select the algorithm availability based on the building HVAC system and determine the corresponding control algorithm;

[0065] S104, determine the input and output parameters according to the control algorithm;

[0066] S105, based on the dynamic parameter dependency verification mechanism for physical links of building HVAC systems, performs dependency verification on input and output parameters; dependency verification includes ownership dependency verification, multiple path dependency verification, and exclusion dependency verification.

[0067] S106, Adjust and adapt data baseline evaluation for input and output parameters that have successfully verified the dependency relationship. Based on the data availability verification mechanism of the two-stage dynamic baseline, data continuity coverage verification is performed based on the minimum effective period during the training preparation period; and empirical-statistical hybrid anomaly identification is performed based on the HVAC operation mode during the operation preparation period.

[0068] S107 After completing the baseline assessment of the control adaptability data, control logic simulation and verification are performed by calling the API;

[0069] S108, after the control logic simulation and verification are completed, confirm that the automatic check before the intelligent control of the building HVAC system is put into operation is completed.

[0070] According to an embodiment of the present invention, before the automatic check begins before the intelligent control of the building HVAC system is put into operation, the software implementer initializes the project, including initializing the basic building information, HVAC system equipment, signal points associated with the equipment, and the physical and logical locations (process locations) of the equipment. Relatively complete input is required to automatically generate process relationships and provide basic data for process identification.

[0071] The process description language for building HVAC systems is constructed by those skilled in the art based on actual needs. It can be built upon the knowledge of HVAC experts. Building HVAC systems vary depending on the project type, and process identification is a complex process. If the problem of dynamic process identification cannot be solved, the algorithm cannot effectively identify and implement the system, and it is difficult to achieve universality and large-scale use. To address these issues, a new and efficient process description language for building HVAC systems (HVAC-DSL) is constructed, specifically serving the process identification stage. It standardizes the definitions of connection relationships (such as energy type, transmission and distribution methods, and terminal layout) and related equipment (such as units, heat exchange stations, and electric valves) involved in the building HVAC system. The resulting standardized process description language is packaged into an API and stored in a database, allowing the system to update it in real-time or event-driven, ensuring the model remains up-to-date.

[0072] Depending on the design logic of different control algorithms, the input and output parameters are usually different. For the field of building HVAC systems, the input parameters include, but are not limited to: outdoor meteorological parameters, weather forecast results, chilled water supply and return temperatures, chilled water flow rate, cooling water supply and return temperatures, cooling water flow rate, and all energy consumption points; the output parameters include, but are not limited to: target supply water temperature, number of air source units, chilled water pump frequency, circulating pump frequency, cooling pump frequency, and start-up and shutdown settings.

[0073] To ensure that the input and output parameter sets of AI control algorithms maintain semantic consistency and control safety with the actual system's process structure and operational logic before being integrated into the building's HVAC system, a dynamic parameter dependency verification mechanism for the physical links of the HVAC system is proposed. This mechanism breaks through the limitations of traditional parameter integrity verification, which only focuses on "existence" or "whether it exceeds the limit," and instead identifies the functional coupling, control mutual exclusion, and path selectivity between parameters from the system engineering logic level. A novel parameter dependency configuration design scheme is proposed, which realizes the definition of parameters and the maintenance of dependencies. The definition of parameters is relatively straightforward, importing the required parameters according to the algorithm requirements or importing all parameters according to the process system for maintenance. However, the maintenance of dependencies is more complex, mainly due to the complexity of process relationships. Several types of dependencies are defined: belonging dependencies, multiple-choice path dependencies, and exclusive dependencies.

[0074] Meanwhile, to address the issue of AI algorithm training failure and control inaccuracies caused by inconsistent data acquisition quality, missing historical data, or abnormal interference in building HVAC systems, this invention proposes a data availability verification mechanism based on a two-stage dynamic baseline. Its core lies in treating data quality not as a static attribute, but as a "system adaptability criterion" for the safe deployment of AI algorithms, thus achieving a technological leap from "passive quality inspection" to "active access." This mechanism verifies two core data requirements of the AI ​​control algorithm: during the training preparation phase, it performs data continuity coverage verification based on the minimum effective period; and during the operation preparation phase, it performs empirical-statistical hybrid anomaly identification based on HVAC operation modes.

[0075] After all the above steps have been verified, the system calls the API control algorithm for verification testing. When all APIs have successfully completed verification, the entire closed-loop evaluation of the self-test input is completed, and the control logic simulation and verification are completed.

[0076] Finally, based on the test results of the above steps, once all self-checks are passed, the algorithm is ready to be put into operation, thereby realizing automatic pre-operation checks for intelligent control of building HVAC systems.

[0077] According to embodiments of the present invention, the connection relationships and associated equipment involved in a building HVAC system are standardized and defined to obtain a standardized construction process description language for the building HVAC system, including:

[0078] Based on the dynamic topology modeling mechanism, the connection relationships and related equipment involved in the building HVAC system are sequentially modeled in both structural and functional dimensions, undergoing event-driven model evolution, constructing an extensible semantic text library, and generating an API-based self-describing output interface. This transforms the connection relationships and related equipment involved in the building HVAC system from physical entities into a standardized process description language for building HVAC systems.

[0079] It should be noted that a building HVAC system process description language (HVAC-DSL) is constructed based on the system's pre-set dynamic topology modeling mechanism. This enables the system to autonomously understand the process and context before AI control is implemented, transforming the HVAC system from a "physical entity" into a "computable logical model." This endows the AI ​​algorithm with the ability to autonomously understand the process structure in unknown systems, solving the fundamental problems of existing technologies where AI relies on manual configuration and is difficult to reuse across projects. This overcomes the bottlenecks of traditional AI control algorithms in building HVAC systems, such as "poor scenario adaptability, high deployment costs, and weak generalization ability" caused by heterogeneous system structures and variable equipment configurations.

[0080] The HVAC system process description language not only covers traditional equipment parameters and connection relationships (such as energy type, transmission and distribution form, terminal layout, unit configuration and valve logic), but also introduces mechanisms such as hierarchical semantic tags, functional unit abstraction and dynamic topology expression, mapping the physical system into a logical model with engineering semantic interpretability. It is a domain-specific HVAC system process description language.

[0081] This design supports the following features:

[0082] (1) Structure-function dual-dimensional modeling: While describing the physical connection, label the functional roles of each component (such as "cold source main supply", "heat recovery bypass" and "variable flow regulation unit"), so that the AI ​​algorithm can understand the system operation logic and not just the data structure;

[0083] (2) Event-driven model evolution: When equipment is added or removed, pipeline is reconfigured or control logic is changed in the system, the model is dynamically updated through the edge gateway or BMS interface to ensure that the process model is synchronized with the actual system state in real time;

[0084] (3) Construct an extensible semantic text library: The built-in ontology knowledge base in the HVAC field supports automatic classification and semantic derivation of new equipment types and new system forms, and improves the compatibility with unknown topologies;

[0085] (4) Generate API-based self-describing output interface: The constructed semantic process model is encapsulated as a standardized RESTful API service. The AI ​​control algorithm actively calls the corresponding interface of the service during the initialization phase, automatically parses the system topology, identifies key control loops, extracts adjustable parameter boundary conditions, and completes the semantic alignment of "algorithm-system".

[0086] The core innovation of this step lies in standardizing traditional static data and upgrading it into a system self-description capability that is "semantically driven, dynamically evolving, and functionally understandable." This enables AI algorithms to "understand" the system's process logic before they are integrated into the control system, significantly reducing engineering configuration costs, improving algorithm deployment efficiency and cross-project migration capabilities, and forming the prerequisite foundation for the safe and reliable deployment of AI regulation.

[0087] According to an embodiment of the present invention, it further includes:

[0088] The standardized process description language for building HVAC systems is packaged into an API and stored in a database.

[0089] It should be noted that after determining the standardized process description language for building HVAC systems, it is packaged into an API for AI algorithms to call. Before the AI ​​algorithm is deployed, it must first load the API through the corresponding API-based self-description output interface to obtain the relevant status.

[0090] According to an embodiment of the present invention, algorithm availability selection is performed based on the building HVAC system to determine the corresponding control algorithm, including:

[0091] Based on the building HVAC system, select one or more key features from the corresponding process characteristics;

[0092] By analyzing one or more key features, the corresponding control algorithm can be determined.

[0093] It should be noted that the algorithms for building HVAC systems are primarily designed based on process characteristics. Different process characteristics employ different algorithms or operators. Therefore, after identifying the process characteristics through the steps described above, one or more key characteristics are selected from the corresponding process characteristics based on the building HVAC system. These key characteristics include whether the energy station is controllable, whether it includes a heat exchange station, whether the transmission and distribution system is controllable, and the form of the monitoring terminals.

[0094] The analysis process for key features is as follows:

[0095] (1) Whether the energy station is controllable: Taking the heating scenario as an example, if the energy station is municipal heating and charged by area, there is no need for regulation, and the algorithm without energy station regulation is selected; if the energy station has the conditions for regulation, the algorithm including energy station regulation is selected.

[0096] (2) Whether it includes a heat exchange station: Generally, energy forms are divided into direct supply from a single network and distributed regulation with a heat exchange station. This has a large difference in the construction of the algorithm. Based on the data of process identification, we determine whether to select an algorithm that includes heat exchange station regulation.

[0097] (3) Whether the transmission and distribution system is controllable: The transmission and distribution system generally includes controllable components such as water pumps and regulating valves. Depending on the actual situation, the quality regulation algorithm of the energy station can be selected, or the quality and quantity regulation algorithm of the transmission and distribution system can be selected.

[0098] (4) Monitoring terminal form: The energy terminals of buildings have various forms such as radiators, underfloor heating, and sub-panels, depending on the form of heating and cooling. Based on this consideration, these differentiated parameters are transmitted to the algorithm for early training to assist the algorithm in selection.

[0099] Based on the selection of the above key features, it can not only assist in algorithm design, but also provide support for self-checking and identification before the algorithm is deployed, and can determine the key algorithms and operators used before the algorithm is deployed.

[0100] Figure 2 A flowchart of the dependency verification method provided by this invention is shown.

[0101] like Figure 2As shown in the embodiment of the present invention, the dependency verification specifically includes:

[0102] S201 generates strong binding logic between input / output parameters and their respective functional units;

[0103] S202, based on the functional units identified by the process model, determine whether the building HVAC system is equipped with key auxiliary parameters for functional units;

[0104] S203 If the key auxiliary parameters are not configured, the verification is deemed to have failed, and a prompt message is generated based on the unconfigured key auxiliary parameters.

[0105] It should be noted that the key auxiliary parameters of each functional unit are determined through the strong binding logic between the input / output parameters and the functional unit to which they belong. There can be one or more key auxiliary parameters. If not all key auxiliary parameters are configured in the building HVAC system, the self-testing system will directly determine it as "structural mismatch", trigger the verification failure, and generate a prompt message indicating the missing item based on the unconfigured key auxiliary parameters.

[0106] For example, when the process model identifies a "plate heat exchange station" as a functional unit in the system, input or output parameters such as "primary side inlet water temperature," "secondary side target supply water temperature," and "heat exchanger bypass valve opening" are marked as key auxiliary parameters for that unit. If the AI ​​control strategy declares support for heat exchange station optimization but does not configure any of the above key parameters, the verification fails, and a prompt message indicating the missing item is generated based on the unconfigured key auxiliary parameters.

[0107] Figure 3 A flowchart of the multiple-selection path dependency verification method provided by the present invention is shown.

[0108] like Figure 3 As shown in the embodiment of the present invention, the multiple path dependency verification specifically involves:

[0109] S301 When multiple optional control paths exist, each optional control path is evaluated based on the equipment operating characteristics to determine the optimal control path; the equipment operating characteristics include at least the equipment operating time, historical regulation sensitivity curve and current energy efficiency ratio;

[0110] S302, activate the input and output parameters associated with the optimal control path, and control the input and output parameters associated with other optional paths to enter the standby locked state.

[0111] It should be noted that, typically, parallel equipment links or multi-path distribution systems (such as multiple air source heat pumps in parallel or dual primary pump frequency converter redundancy configurations) have multiple selectable control paths. This paper proposes a parameter path optimization mechanism based on operational efficiency priority for parallel equipment links or multi-path distribution systems. After identifying multiple selectable control paths, this mechanism does not perform random or default selection. Instead, it automatically evaluates and activates the parameter set associated with the optimal control path by combining operational characteristics such as equipment runtime, historical adjustment sensitivity curves, and current energy efficiency ratio. The optimal control path can be determined through weighted calculation of various operational characteristics or by analysis using a pre-built preset control path analysis model based on historical operational characteristics. Simultaneously, it controls the input and output parameters associated with other selectable paths to enter a standby locked state.

[0112] According to an embodiment of the present invention, the exclusionary dependency verification specifically involves:

[0113] Construct an exclusion relationship table;

[0114] If the input and output parameters exist in the exclusion relationship table, the verification is deemed to have failed, the input and output parameters are marked as exclusion parameters, and a prompt message is generated based on the exclusion parameters.

[0115] It should be noted that the exclusion relationship table is set up by those skilled in the art based on actual needs and the historical operation data of the building's HVAC system. To avoid control disruption, in addition to identifying the process flow, it is necessary to effectively verify the exclusion parameters. When an input or output parameter in the exclusion table is automatically identified and marked as an exclusion parameter, the verification is deemed to have failed. The input or output parameter is then marked as an exclusion parameter, and a prompt message is generated based on the exclusion parameter to remind the user to correct the exclusion relationship.

[0116] For example, if the unit's supply water temperature setting is set, and then the return water temperature setting parameter is initialized, these are mutually exclusive operations for the control business. They cannot be adjusted simultaneously, meaning there is an exclusionary relationship. Accurate identification and timely correction are required to avoid algorithmic disorder in adjustment.

[0117] According to an embodiment of the present invention, the data continuity coverage verification based on the minimum effective period specifically includes:

[0118] Based on the minimum effective period, the frequency compliance of input and output data collection is verified to determine whether the data point collection time interval meets the preset minimum sampling requirement and whether the effective data coverage is greater than or equal to the preset minimum coverage.

[0119] When the data point collection time interval meets the preset minimum sampling requirement and the effective data coverage is greater than or equal to the preset minimum coverage, the collection frequency compliance verification is considered successful; otherwise, the collection frequency compliance verification fails.

[0120] After the compliance verification of the collection frequency is successful, the integrity verification of the operating mode is carried out. Based on the identified type of building HVAC system process, it is determined whether the input and output data cover typical operating conditions.

[0121] If yes, the modal integrity verification will succeed; otherwise, the modal integrity verification will fail.

[0122] It should be noted that, given the significant differences in data foundations among different projects, a "minimum effective training period" (MVTW) is first defined, preferably 30 consecutive days, which can be dynamically adjusted according to the seasonality of system operation (e.g., for winter heating systems only, it can be shortened to the continuous heating period).

[0123] The system automatically scans all input and output variables that have passed dependency validation and performs the following checks within the MVTW window:

[0124] Data collection frequency compliance: Verify whether the data point collection time interval within the minimum effective period meets the preset minimum sampling requirement (e.g., ≤15 minutes), and calculate whether the effective data coverage is greater than or equal to the preset minimum coverage (e.g., ≥85%).

[0125] Operational modal integrity: Based on the identified building HVAC system process type (such as "air source heat pump + fan coil unit"), determine whether the data covers typical operating conditions (such as start-up, full load, variable load, and diurnal temperature difference response), and avoid model bias caused by only collecting steady-state data.

[0126] Unlike simply counting "whether there is data", this invention jointly determines the "sufficiency of time" and "representativeness of working conditions" of data coverage to form an admission threshold for the effectiveness of AI training, ensuring that the algorithm has basic learning materials before it is deployed.

[0127] According to an embodiment of the present invention, the empirical-statistical hybrid anomaly identification based on HVAC operation modes specifically includes:

[0128] Construct a lightweight outlier self-detection algorithm that includes an empirical rule layer and a dynamic statistics layer;

[0129] Establish an expert knowledge base for the HVAC field at the experience rule layer, and pre-set typical anomaly patterns based on the expert knowledge base for the HVAC field.

[0130] In the dynamic statistics layer, a combination of moving window IQR and sliding Z-score is used for dynamic statistics, outliers are marked and their distribution is recorded, and an anomaly heatmap is generated.

[0131] It should be noted that, in order to balance universality and robustness, a lightweight outlier self-detection algorithm (LAS) was designed, which includes an empirical rule layer and a dynamic statistical layer, thereby integrating universality and robustness as the basis for judgment.

[0132] The system includes an embedded HVAC expert knowledge base (built based on HVAC expert knowledge) within the experience rule layer. Based on this knowledge base, typical anomaly modes are preset, such as: chilled water supply and return water temperature difference > 7℃ and normal flow rate → possible sensor drift; outdoor temperature > 28℃ but cooling tower outlet water temperature < 22℃ → data logic contradiction; unit operating status is "off" but power > 5% of rated power → equipment status false alarm, etc.

[0133] In the dynamic statistical layer, a combination of moving window IQR (interquartile range) and sliding Z-score is used. Sliding Z-score is suitable for normally distributed data; it calculates the number of standard deviations a data point differs from the mean, and values ​​exceeding a preset threshold are considered outliers. Moving window IQR is suitable for non-normally distributed data; it calculates the interquartile range and sets upper and lower bounds (e.g., q1 - 1.5 × iqr and q3 + 1.5 × iqr), and values ​​outside these ranges are considered outliers. The choice of method should be based on the data distribution. Sliding Z-score is more intuitive but sensitive to distribution, while moving window IQR is more robust and universal, and can be combined with visualization techniques to improve accuracy. The preset threshold is designed to be "wide but evolving." For example, an initial preset threshold of leniency (e.g., Z < 3.0) allows for some noise, ensuring that low-quality systems can pass the initial screening. Simultaneously, during the dynamic statistical process, outliers are marked and their distribution is recorded, generating an anomaly heatmap for subsequent weighted training or feature masking in the algorithm.

[0134] During the baseline assessment of adaptive control data, a closed-loop linkage is established with the process model to implement differentiated verification strategies: for systems identified as "high-precision intelligent buildings," the data quality threshold is automatically tightened (e.g., coverage ≥95%, Z-score threshold = 2.5); for systems identified as "old renovation projects," a wide threshold mode is activated, and the system is marked as "data-restricted operation," limiting the scope of AI control and prioritizing safety. Therefore, data verification is no longer an isolated process, but a dynamically adapted intelligent gating mechanism based on the system's process attributes, truly achieving adaptive access with "one system, one strategy."

[0135] According to an embodiment of the present invention, control logic simulation and verification are performed by calling an API, including:

[0136] By calling the API to query parameters, the obtained parameters are controlled, and the API is verified based on the feedback mechanism of the given response.

[0137] Once all APIs have been verified, the control logic simulation and verification are complete.

[0138] It should be noted that the system uses a preset API control algorithm to call the API for control logic simulation and verification. This API control algorithm is similar to a simulation environment, which allows users to freely call the API to query and control relevant parameters, while also providing a feedback mechanism for the response.

[0139] All information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals (including but not limited to signals transmitted between user terminals and other devices) involved in this application have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the "process characteristics" and "input / output parameters" involved in this disclosure were obtained under full authorization.

[0140] This invention discloses an automatic pre-implementation check method for intelligent control of building HVAC systems. The method includes: standardizing the definition of connections and associated equipment within the building HVAC system to obtain a standardized process description language for the system and storing it in a database; selecting algorithm availability based on the building HVAC system to determine the corresponding control algorithm and input / output parameters; verifying the dependency relationship of input / output parameters based on a dynamic parameter dependency verification mechanism for the physical links of the building HVAC system; and conducting a baseline assessment of the control adaptability data for input / output parameters with successful dependency verification, and performing control logic simulation and verification by calling APIs. This invention can efficiently help AI algorithm users or developers conduct thorough self-checks before implementing algorithms, ensuring their reliability before deployment; it also helps AI algorithm practitioners optimize algorithm engineering.

[0141] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0142] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0143] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0144] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for automatic pre-implementation inspection of intelligent control of building HVAC systems, characterized in that, include: The connection relationships and related equipment involved in building HVAC systems are standardized and defined to obtain a standardized process description language for building HVAC systems; The standardized construction process description language for building HVAC systems is stored in the database; Based on the building's HVAC system, algorithm availability is selected to determine the corresponding control algorithm; The input and output parameters are determined according to the control algorithm. Based on the dynamic verification mechanism of parameter dependency for physical links of building HVAC systems, the input and output parameters are verified for dependency. The dependency verification includes dependent relationship verification, multiple path dependency verification, and exclusion dependency verification; For input and output parameters that have successfully verified dependencies, an adaptive data baseline assessment is performed. Based on a two-stage dynamic baseline data availability verification mechanism, during the training preparation period, a data continuity coverage verification based on the minimum effective period is performed; during the operation preparation period, an empirical-statistical hybrid anomaly identification based on HVAC operation modes is performed. After completing the baseline assessment of the control adaptability data, the control logic simulation and verification are performed by calling the API; Once the control logic simulation and verification are completed, the automatic check before the intelligent control of the building HVAC system is put into operation is confirmed to be complete. The standardized definition of the connection relationships and related equipment involved in the building HVAC system yields a standardized construction process description language for the building HVAC system, including: Based on the dynamic topology modeling mechanism, the connection relationships and associated equipment involved in the building HVAC system are sequentially modeled in both structural and functional dimensions, undergoing event-driven model evolution, constructing an extensible semantic text library, and generating an API-based self-describing output interface. This transforms the connection relationships and associated equipment involved in the building HVAC system from physical entities into a standardized process description language for building HVAC systems.

2. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, Also includes: The standardized construction process description language for building HVAC systems is packaged into an API and stored in a database.

3. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The step of selecting an algorithm based on the building's HVAC system and determining the corresponding control algorithm includes: Based on the building HVAC system, one or more key features are selected from the corresponding process characteristics; By analyzing one or more of the key features, the corresponding control algorithm is determined.

4. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The dependency relationship verification specifically involves: Generate strong binding logic between input / output parameters and their respective functional units; Based on the functional units identified by the process model, determine whether the key auxiliary parameters of the functional units are configured in the building HVAC system. If the aforementioned key auxiliary parameters are not configured, the verification is deemed to have failed, and a prompt message is generated based on the unconfigured key auxiliary parameters.

5. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The specific steps of the multiple-selection path dependency verification are as follows: When multiple optional control paths exist, each optional control path is evaluated based on the equipment operating characteristics to determine the optimal control path; the equipment operating characteristics include at least the equipment operating time, historical regulation sensitivity curve, and current energy efficiency ratio; Activate the input and output parameters associated with the optimal control path, and control the input and output parameters associated with other optional paths to enter the standby locked state.

6. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The exclusionary dependency verification specifically involves: Construct an exclusion relationship table; If the input / output parameters exist in the exclusion relationship table, the verification is deemed to have failed, the input / output parameters are marked as exclusion parameters, and a prompt message is generated based on the exclusion parameters.

7. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The data continuity coverage verification based on the minimum effective period is specifically as follows: Based on the minimum effective period, the frequency compliance of input and output data collection is verified to determine whether the data point collection time interval meets the preset minimum sampling requirement and whether the effective data coverage is greater than or equal to the preset minimum coverage. When the data point collection time interval meets the preset minimum sampling requirement and the effective data coverage is greater than or equal to the preset minimum coverage, the compliance verification of the collection frequency is confirmed to be successful. Conversely, the compliance verification of the collection frequency fails; After the compliance verification of the collection frequency is successful, the integrity verification of the operating mode is carried out. Based on the identified type of building HVAC system process, it is determined whether the input and output data cover typical operating conditions. If so, the modal integrity verification is successful; Conversely, the operational modal integrity verification fails.

8. The automatic inspection method for intelligent control of building HVAC systems according to claim 1, characterized in that, The empirical-statistical hybrid anomaly identification based on HVAC operation modes specifically refers to: Construct a lightweight outlier self-detection algorithm that includes an empirical rule layer and a dynamic statistics layer; A knowledge base of HVAC experts is established in the experience rule layer, and typical anomaly patterns are preset based on the knowledge base of HVAC experts. In the dynamic statistics layer, a combination of moving window IQR and sliding Z-score is used for dynamic statistics, outliers are marked and their distribution is recorded to generate an anomaly heatmap.

9. The automatic inspection method for intelligent control of building HVAC systems according to claim 2, characterized in that, The control logic simulation and verification via API calls includes: The API is used to query parameters, and the obtained parameters are controlled. The API is then verified based on a feedback mechanism that provides a given response. Once all APIs have been verified, the control logic simulation and verification are complete.

Images