Artificial intelligence-based heating and ventilation design evaluation method, device and computer equipment

Abstract

The application relates to an artificial intelligence-based heating ventilation design evaluation method, device and computer equipment. The method comprises the following steps: acquiring a heating ventilation design image according to a heating ventilation model, the heating ventilation design image comprising at least one heating ventilation object, inputting the heating ventilation design image into a preset neural network, and performing design relationship identification and evaluation on the heating ventilation object to obtain an identification evaluation result, so that the heating ventilation model can be adjusted according to the identification evaluation result, and the heating ventilation model does not need to be adjusted by relying on artificial experience, so that the adjustment efficiency is high and accurate.

Description

Technical Field

[0001] This application relates to the field of architectural design technology, and in particular to an artificial intelligence-based HVAC design evaluation method, apparatus, and computer equipment. Background Technology

[0002] In the traditional HVAC design process, the design results (i.e., HVAC design models) are obtained through manual design. After the design is completed, the results are optimized manually based on years of industry experience.

[0003] However, in the traditional design optimization process, relying solely on human experience to adjust the HVAC model may result in low adjustment efficiency and incomplete optimization, leading to low quality of the design model. Summary of the Invention

[0004] Therefore, it is necessary to provide an AI-based HVAC design evaluation method, device, and equipment to address the aforementioned technical problems.

[0005] In a first aspect, embodiments of this application provide an artificial intelligence-based HVAC design evaluation method, the method comprising:

[0006] Based on a 3D building model, HVAC design images that meet the requirements are obtained. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0007] The HVAC design image is input into a preset evaluation neural network to identify HVAC objects and evaluate design relationships in the HVAC design image. Based on the evaluation results, the HVAC design image is classified, and design annotations are given for HVAC design images that do not meet the design standards.

[0008] In one embodiment, obtaining HVAC design images that meet the requirements based on a three-dimensional building model includes:

[0009] Obtain the refrigerant system, chilled water system, and ventilation system from the 3D building model;

[0010] The HVAC objects in the refrigerant system, chilled water system, and ventilation system are located.

[0011] Based on the HVAC object, HVAC design images that meet the requirements are extracted according to preset rules.

[0012] In one embodiment, the step of extracting HVAC design images that meet the requirements based on the HVAC object according to preset rules includes:

[0013] Control the virtual camera's viewpoint to capture HVAC design images that meet the requirements at a fixed angle;

[0014] The fixed angle is determined based on the floor level, indoor or outdoor location of the HVAC object.

[0015] The acquired images possess clear features, significantly reducing the number of training samples required for neural networks, thus solving the technical problem of limited training samples in the construction field. Furthermore, BIM models can achieve realistic rendering, addressing the shortage of physical images.

[0016] In one embodiment, providing design annotations for HVAC design images that do not conform to design standards includes:

[0017] Establish a HVAC design standard library in advance and identify HVAC objects in HVAC design images that do not meet the design standards;

[0018] Based on the HVAC object recognition results, retrieve the corresponding design standards from the HVAC design standard library and annotate the design standards in the HVAC design image;

[0019] The floor where the HVAC design object that does not meet the design standards is located is marked, and the mark is visible in the three-dimensional building model.

[0020] Secondly, embodiments of this application provide an artificial intelligence-based HVAC design evaluation method, the method comprising:

[0021] Based on the HVAC design model, an HVAC design image is obtained; the HVAC design image includes at least one HVAC object.

[0022] The HVAC design image is input into a neural network that includes several design relationship constraints to identify the design relationships between the HVAC objects and obtain the design relationships between each HVAC object.

[0023] The design relationships between various HVAC objects are matched with the design relationship constraints to obtain the identification and evaluation results; the design relationships between HVAC objects are used to characterize at least one of the refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships.

[0024] In one embodiment, the training process of the neural network includes:

[0025] The sample model of the HVAC system with pre-annotated HVAC objects is converted from 3D graphics to 2D images to obtain multiple initial HVAC design image samples.

[0026] The initial HVAC design image sample is preprocessed to obtain an HVAC design image sample;

[0027] The HVAC design image samples are input into a preset initial neural network. Design relationships are extracted based on the classification method of decision tree, and frequent itemsets are obtained. The neural network is then obtained through constraints.

[0028] Thirdly, embodiments of this application provide an artificial intelligence-based HVAC model evaluation device, the device comprising:

[0029] The acquisition module is used to acquire HVAC design images that meet the requirements based on a three-dimensional building model. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0030] The identification and evaluation module is used to input the HVAC design image into a preset evaluation neural network, identify the HVAC objects in the HVAC design image and evaluate the design relationships, classify the HVAC design image according to the design relationship evaluation results, and provide design annotations for HVAC design images that do not meet the design standards.

[0031] In one embodiment, the apparatus further includes:

[0032] The annotation module is used to pre-establish a HVAC design standard library, identify HVAC design images that do not conform to the design standards, retrieve the corresponding design standards from the HVAC design standard library based on the HVAC object identification results, and annotate the design standards in the HVAC design images; and mark the floors where the HVAC design objects that do not conform to the design standards are located, and the marks are visible in the three-dimensional building model.

[0033] Fourthly, embodiments of this application provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0034] Based on a 3D building model, HVAC design images that meet the requirements are obtained. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0035] The HVAC design image is input into a preset evaluation neural network to identify HVAC objects and evaluate design relationships in the HVAC design image. Based on the evaluation results, the HVAC design image is classified, and design annotations are given for HVAC design images that do not meet the design standards.

[0036] Fifthly, embodiments of this application provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0037] Based on the HVAC design model, an HVAC design image is obtained; the HVAC design image includes at least one HVAC object.

[0038] The HVAC design image is input into a neural network that includes several design relationship constraints to identify the design relationships between the HVAC objects and obtain the design relationships between each HVAC object.

[0039] The design relationships between various HVAC objects are matched with the design relationship constraints to obtain the identification and evaluation results; the design relationships between HVAC objects are used to characterize at least one of the refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships.

[0040] Sixthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the following steps:

[0041] Based on a 3D building model, HVAC design images that meet the requirements are obtained. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0042] The HVAC design image is input into a preset evaluation neural network to identify HVAC objects and evaluate design relationships in the HVAC design image. Based on the evaluation results, the HVAC design image is classified, and design annotations are given for HVAC design images that do not meet the design standards.

[0043] In a seventh aspect, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the following steps:

[0044] Based on the HVAC design model, an HVAC design image is obtained; the HVAC design image includes at least one HVAC object.

[0045] The HVAC design image is input into a neural network that includes several design relationship constraints to identify the design relationships between the HVAC objects and obtain the design relationships between each HVAC object.

[0046] The design relationships between various HVAC objects are matched with the design relationship constraints to obtain the identification and evaluation results; the design relationships between HVAC objects are used to characterize at least one of the refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships.

[0047] The aforementioned AI-based HVAC design evaluation method, apparatus, and computer equipment acquire HVAC design images, input these images into a pre-set neural network, and perform design relationship identification and evaluation on HVAC objects to obtain evaluation results. It can also annotate non-compliance. By using a neural network to identify and evaluate design relationships of HVAC objects, the evaluation results can be automatically obtained. The HVAC model can be adjusted based on these results, eliminating the need for manual adjustments and resulting in high efficiency and accuracy. Attached Figure Description

[0048] Figure 1 This is an internal structural diagram of a computer device in one embodiment;

[0049] Figure 2 This is a flowchart illustrating an AI-based HVAC design evaluation method in one embodiment.

[0050] Figure 3 This is a flowchart illustrating the detailed steps of step S201 in one embodiment;

[0051] Figure 4 This is a flowchart illustrating the detailed steps of step S202 in one embodiment;

[0052] Figure 5 This is a flowchart illustrating an AI-based HVAC design evaluation method in one embodiment.

[0053] Figure 6 This is a schematic diagram of the structure of a smart HVAC design evaluation device in one embodiment. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0055] The AI-based HVAC design evaluation method provided in this application can be applied to... Figure 1The computer device shown includes a processor, memory, network interface, database, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The database stores the neural network described in the following embodiments; a detailed description of the neural network is provided in the following embodiments. The network interface of the computer device can be used to communicate with other external devices via a network connection. Optionally, the computer device can be a server, a desktop computer, a personal digital assistant, or other terminal devices such as tablets, mobile phones, etc., or it can be a cloud or remote server. This application does not limit the specific form of the computer device. The display screen of the computer device can be a liquid crystal display or an e-ink display. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device casing, or an external keyboard, touchpad, or mouse. Of course, input devices and displays may not be part of the computer equipment; they can be external devices to the computer equipment.

[0056] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0057] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0058] It should be noted that the executing entity of the following method embodiments can be a HVAC intelligent design evaluation device, which can be implemented as part or all of the aforementioned computer equipment through software, hardware, or a combination of software and hardware. The following method embodiments are described using a computer equipment as an example of the executing entity.

[0059] Figure 2 This is a flowchart illustrating an AI-based HVAC design evaluation method as provided in one embodiment. This embodiment relates to the process of using a computer device to perform assisted design on an HVAC model using a neural network.

[0060] like Figure 2 As shown, it includes:

[0061] Step S201: Obtain a HVAC design image that meets the requirements based on the three-dimensional building model. The HVAC design image includes at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0062] In this embodiment of the invention, the three-dimensional building model is a three-dimensional graphics software product, which may contain all or part of the data information of the building and related data within the building. To evaluate the HVAC design within the three-dimensional building model, it is necessary to first obtain the HVAC model within the real-time three-dimensional building model.

[0063] The computer device of this invention can obtain HVAC design images by recognizing HVAC models. Since HVAC design images are two-dimensional images and cannot be directly derived from three-dimensional building models, they require cropping processing; that is, cropping the two-dimensional image at an arbitrary angle with defined direction and coordinates. Furthermore, the cropping of the two-dimensional image should be performed systematically to improve recognition speed and accuracy.

[0064] To evaluate the design of an HVAC model, it is necessary to assess the relationships, locations, parameters, and other aspects of the HVAC objects involved in the model. Optionally, HVAC objects may include indoor units, outdoor units, condensate pipes, fans, etc.

[0065] Step S202: Input the HVAC design image into a preset evaluation neural network to identify and evaluate the HVAC objects in the HVAC design image and evaluate their design relationships. Classify the HVAC design image according to the evaluation results and provide design annotations for HVAC design images that do not meet the design standards.

[0066] In this embodiment of the invention, the preset evaluation neural network is an artificial intelligence neural network trained on images and / or data. It is trained to identify HVAC objects in HVAC design images and to evaluate and classify the identification results.

[0067] In this embodiment, the relationship between HVAC objects is evaluated. For example, the design relationship refers to the relationship between different HVAC objects (the arrangement between indoor and outdoor units), or it can be the relationship between the attributes of different HVAC objects (e.g., valve material). This embodiment does not limit this.

[0068] Specifically, each HVAC object and related HVAC objects have certain design standards, which can be derived through neural network training constraints, thereby enabling identification and classification.

[0069] In this embodiment, HVAC design images that meet the evaluation criteria are categorized, while those that do not meet the criteria are classified. Based on this, HVAC models that do not meet the criteria and have design problems can be derived. Further, through manual identification, design annotations can be provided for the HVAC design images that do not meet the design criteria.

[0070] The aforementioned AI-based HVAC design evaluation method acquires HVAC design images based on an HVAC model. These images include at least one HVAC object. The images are then input into a pre-set neural network, which identifies and evaluates the design relationships between the HVAC objects, yielding the evaluation results. Using images extracted from the model results in high efficiency and accuracy. Furthermore, utilizing a neural network to identify and evaluate the design relationships between HVAC objects automatically generates evaluation results, allowing for adjustments to the HVAC model without relying on human experience, thus ensuring efficient and accurate adjustments.

[0071] In one embodiment, such as Figure 3 The diagram shown is a detailed flowchart of step S201, including:

[0072] Step S2011: Obtain the refrigerant system, chilled water system, and ventilation system from the 3D building model;

[0073] Step S2012: Locate the HVAC objects in each of the refrigerant system, chilled water system, and ventilation system;

[0074] Step S2013: Based on the HVAC object, extract the HVAC design image that meets the requirements according to preset rules.

[0075] A 3D building model, such as a 3D model of a building, includes HVAC systems, water supply and drainage systems, electromechanical systems, etc. The HVAC model is constructed by acquiring the HVAC objects (components, pipes, equipment) involved in the refrigerant system, chilled water system, and ventilation system, etc. This HVAC model serves as the evaluation object, based on image recognition and calculation.

[0076] Refrigerant systems, chilled water systems, and ventilation systems encompass various HVAC components, including equipment, piping, and their relative design relationships. Therefore, HVAC design images can be obtained from these systems. However, training neural networks requires a large amount of data, and the availability of training data in the construction field is limited, making it difficult to meet training requirements. It is advisable to consider acquiring high-quality, easily identifiable HVAC design images to improve the accuracy and efficiency of the neural network's recognition.

[0077] In this embodiment of the invention, the computer device acquires a heating and ventilation model, and converts the three-dimensional graphics of the heating and ventilation model into a two-dimensional image according to preset rules to obtain multiple heating and ventilation design images. This enables the automatic acquisition of heating and ventilation design images from a complete design model, thus achieving a higher degree of automation and further improving recognition efficiency and accuracy.

[0078] In this embodiment of the invention, the computer device can traverse a three-dimensional building model to obtain a HVAC model, and then convert the HVAC model from a three-dimensional to a two-dimensional image. Further, optionally, the computer device does not need to filter the obtained two-dimensional HVAC design images, because the obtained HVAC design images are all images with good clarity. The acquired images have clear features, which greatly reduces the number of training samples for the neural network, thus solving the technical problem of very few training samples in the construction field. In addition, the BIM model can achieve real-scene rendering, solving the problem of the shortage of physical images.

[0079] Because 3D building models serve as a design integration carrier, they contain corresponding component names, attributes, and so on. For example, the outdoor unit model is an independent virtual component in a 3D building model. The information of this virtual component includes the component's name, attributes, 3D data, and so on. Depending on the requirements, the brand and model of the outdoor unit can also be labeled.

[0080] Furthermore, to meet the requirement of extracting HVAC design images that meet the needs based on the HVAC object according to preset rules, the method further includes the following refined steps:

[0081] The virtual camera viewpoint is controlled to capture HVAC design images that meet the requirements at a fixed angle; the fixed angle is determined based on the floor, indoor or outdoor location of the HVAC object.

[0082] In one embodiment, the 3D building model can adjust the angle of the virtual camera within the software, allowing browsing of any position within the 3D building model. Based on this, HVAC models can be displayed from different virtual camera perspectives to meet different needs, thereby capturing HVAC design images.

[0083] For example, indoor units located indoors can be photographed from an upward angle, similar to the angle and position of a person looking up at an indoor unit from inside a building; a fixed angle is assumed to be 65° upward, without limitation, or it can be a fixed range, such as 45° to 70°.

[0084] Specifically, if the HVAC system involves outdoor units located near exterior walls, a uniform overhead view can be used for photography. Using this method of capturing HVAC design images at a fixed angle ensures similarity in angle between images and facilitates neural network training. The clear distinction between features in the images makes feature extraction and calculation easier, reducing training time and the number of training samples. Traditional neural network training requires hundreds of thousands of image samples, each with varying angles and backgrounds, making processing complex and resulting in low training accuracy.

[0085] Using a uniform image with a wall as the background can reduce the workload of background processing; in addition, since the training and recognition images are similar, less image feature extraction is required, the training speed is faster, and fewer influencing factors can improve recognition accuracy.

[0086] In one embodiment, the HVAC design image that meets the requirements includes requirements such as placing HVAC objects in the center of the image as much as possible, covering as many HVAC objects as possible, and being able to identify the relationships between HVAC objects; minimizing the proportion of the background wall in the image, etc.

[0087] In addition, the floor to which the HVAC design image is captured will be recorded. This floor is used to mark HVAC objects that do not conform to the design relationship, so as to facilitate the location of floors that do not conform to the design, making it easier for designers to identify and locate them, and thus making it easier to modify the design.

[0088] Furthermore, such as Figure 4 The diagram shown is a detailed flowchart of step S202. Step S202 specifically includes:

[0089] Step S2021: Establish a HVAC design standard library in advance, and identify HVAC objects in HVAC design images that do not conform to the design standards.

[0090] For example, designs that comply with national standards such as the "Code for Fire Protection Design of Buildings GB 50016-2014 (2018 Edition)," the "Technical Standard for Smoke Control and Exhaust Systems in Buildings GB 51251-2017," and the "Code for Design of Heating, Ventilation and Air Conditioning of Civil Buildings GB 50736-2012" are required for architectural design. Creating a design comparison database from these standards facilitates understanding which parts of the design process fail to meet these standards.

[0091] After the design relationship is evaluated by the neural network, non-compliant HVAC design images are identified, and the HVAC objects in the corresponding HVAC design images can also be identified. Standards can be queried in the standard library based on the attributes of the HVAC objects.

[0092] Step S2022: Based on the HVAC object recognition result, retrieve the corresponding design standard from the HVAC design standard library and annotate the design standard in the HVAC design image.

[0093] For example, the installation method of the outdoor unit—ground installation, wall installation, or roof installation—should be marked in the HVAC design drawings as outputs for designers to review.

[0094] Step S2023, and mark the floor where the HVAC design object that does not meet the design standards is located, the mark being visible in the three-dimensional building model.

[0095] In this step, the floor information carried in the output is input into the 3D building model. The corresponding floor is located in the 3D building model and marked, indicating that the HVAC design of that floor does not meet the design requirements. The HVAC design images can be viewed to see the relevant design standards. Then, based on the HVAC design objects, the HVAC objects can be found in the floor, thus facilitating the modification of the 3D building model.

[0096] The aforementioned AI-based HVAC design evaluation method utilizes a neural network to identify design relationships in HVAC design images, revealing the relationships between various HVAC objects. This allows for the determination of whether the current HVAC model meets design requirements, thus avoiding the problems of low efficiency, low accuracy, incomplete adjustments, and high human learning costs associated with traditional manual adjustments and optimizations based on experience. Furthermore, this method automatically identifies design relationships in the HVAC model using neural networks and extracts images from the model, significantly improving recognition efficiency and enabling more accurate and comprehensive identification of design relationships. This also greatly increases accuracy while reducing human learning costs, resulting in substantial savings in time and manpower. After evaluation and identification, the relationships can be marked on the 3D building model for easy review and modification by designers.

[0097] Figure 5 This is a flowchart of an artificial intelligence-based HVAC design evaluation method provided by the present invention, the method comprising:

[0098] Step S301: Based on the HVAC design model, obtain an HVAC design image; the HVAC design image includes at least one HVAC object;

[0099] Step S302: Input the HVAC design image into a neural network that includes several design relationship constraints, identify the design relationships between the HVAC objects, and obtain the design relationships between each HVAC object;

[0100] Step S303: Match the design relationships between each HVAC object with the design relationship constraints to obtain the identification and evaluation results; the design relationships between the HVAC objects are used to characterize at least one of the refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships.

[0101] In one embodiment, the HVAC design model can be independently designed using design software, or it can be selected from a 3D building model; then, it is adopted... Figure 1 The evaluation method shown obtains HVAC design images and inputs them into a neural network for evaluation.

[0102] The neural network includes several design relationship constraints. The HVAC design image is input into the neural network, and the design relationships of the HVAC objects are identified to obtain the design relationships between each HVAC object. The design relationships between each HVAC object are matched with the design relationship constraints to obtain the identification evaluation result (i.e., the degree of matching between each design relationship and the corresponding design relationship constraint).

[0103] Furthermore, the design relationships include at least one of the following: refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships. Specifically, the refrigerant system design relationships include the design relationships for the location and installation method of the outdoor unit, the location and installation method of the indoor unit, and the design relationships for the refrigerant piping path and connection settings. The chilled water system design relationships include the design relationships for the location of the equipment room and the installation and positioning design relationships for the chilled water system terminals. The ventilation system design relationships include the design relationships for the fan model, installation location, and installation method, the design relationships for the specifications and installation methods of the terminal air outlets, and the design relationships for the pipe type, duct wall thickness, reinforcement method, connection method, and duct type.

[0104] The design relationship constraints are pre-trained design relationships. In this embodiment of the invention, for example, after design relationship identification, it is found that a 10-square-meter machine room has 10 outdoor units. Matching this design relationship with the design relationship constraints (the learned constraint is that a 10-square-meter machine room has 6 outdoor units) will result in a completely mismatched identification evaluation result. This identification evaluation result characterizes the degree of matching between the design relationships and design relationship constraints between various HVAC objects.

[0105] The aforementioned AI-based HVAC design evaluation method involves inputting HVAC design images into a pre-set neural network. This neural network is trained using images labeled with HVAC objects to uncover patterns in the various design relationships between different HVAC objects. These patterns are then used as design relationship constraints. Therefore, this neural network can identify design relationships in the HVAC design images, obtaining the relationships between different HVAC objects. These relationships are then matched with the design relationship constraints to determine the degree of matching between the design relationships and the discovered constraints, resulting in an evaluation result. This result determines whether the current HVAC model meets the design requirements. If the degree of matching between the design relationships and the discovered constraints is high, the current HVAC model is considered to meet the design requirements, indicating high design quality. Conversely, if the degree of matching is low, the current HVAC model is considered to not meet the design requirements, indicating low design quality.

[0106] Furthermore, the identification and evaluation results are displayed for easy viewing.

[0107] Furthermore, the training process of the neural network includes:

[0108] The HVAC system sample model with pre-labeled HVAC objects is converted from 3D graphics to 2D images to obtain multiple initial HVAC design image samples; the initial HVAC design image samples are preprocessed to obtain HVAC design image samples; the HVAC design image samples are input into a preset initial neural network, and design relationships are extracted based on the classification method of decision tree, and frequent itemsets are obtained. The neural network is obtained through constraints.

[0109] Optionally, based on the above embodiments, the training process of the neural network is a learning process of design relationship constraints in the HVAC system, including:

[0110] In this embodiment of the invention, a computer device can acquire multiple HVAC design images as learning samples. These sample models are labeled with tags representing HVAC objects, such as indoor units, outdoor units, and condenser pipes. The computer device converts the pre-labeled HVAC system sample models from three-dimensional to two-dimensional to obtain multiple initial HVAC design image samples. Optionally, labeling tools such as LabelMe or VIA can be used to label indoor units, outdoor units, and condenser pipes. Optionally, indoor units, outdoor units, and condenser pipes can also be classified.

[0111] It should be noted that the above design relationship constraints have three metrics: support, confidence, and lift. Support: The support of X→Y represents the probability that {X,Y} appears in the total itemset. Confidence: The confidence of X→Y represents the probability that Y is derived from the rule X→Y given the occurrence of the precondition X; that is, the probability that Y may also exist if X exists. Generally, a confidence level of 95% or higher is considered a rule that must be satisfied, while a confidence level of 60-70% is considered a rule that should be satisfied as much as possible. Lift: The lift of X→Y represents the ratio of the probability that Y is also present given X to the overall probability of Y occurring, which can be expressed as P(Y|X) / P(Y). Optionally, the above design relationships can broadly include the following four categories: Boolean association rules, quantification rules, one-dimensional and multi-dimensional rules, and single-level and multi-level association rules. For example, a Boolean association rule could be that Y must exist if X exists; that is, if Y does not exist, the requirement is not met. Quantification rules can be rules for the quantification value of a certain indicator of HVAC objects, such as the diameter of the pipe being above 3cm and below 25cm; single-dimensional and multi-dimensional rules can be two-dimensional design relationships between indoor units and drainage locations and ceiling heights; single-layer and multi-layer association rules can be single-layer association rules between the main air outlet and the next level air outlet, and multi-layer association rules between the main air outlet and the terminal air outlet.

[0112] In this embodiment of the invention, the computer device inputs the above-mentioned HVAC design image samples into an initial neural network. The neural network can hierarchically classify the HVAC objects and other entity objects in the above-mentioned HVAC design image samples based on a decision tree, and extract the design relationships of different HVAC objects based on the hierarchical structure. Then, based on these design relationships, frequent itemsets are obtained to obtain a neural network including design relationship constraints.

[0113] Classification using decision trees can involve a tree structure built according to a series of rules for classification and prediction. The top node of the decision tree is the root node, and each node forms a new node downwards. Nodes without branches are leaf nodes, and each leaf node corresponds to a decision, i.e., a possible classification result. During computation, the tree is traversed from the root node downwards. Each node corresponds to an attribute, and different branches are selected for different attribute values, finally reaching a leaf node to complete the classification. Decision tree algorithms have a simple structure, high classification accuracy, and good robustness to noisy data.

[0114] Alternatively, object detection and segmentation techniques from Mast R-CNN and TensorFlow can be used to perform deep learning on HVAC objects. Based on the labeled images of HVAC objects, the HVAC objects can be identified to obtain a trained model.

[0115] It is understandable that the neural network training method described in the above evaluation method is at least partially related to... Figure 2 The embodiments shown are the same.

[0116] In this embodiment, a computer device can convert a pre-annotated HVAC system sample model from 3D to 2D to obtain multiple initial HVAC design image samples. These initial HVAC design image samples are then preprocessed to obtain HVAC design image samples. These samples are then input into an initial neural network, where design relationships are extracted using a decision tree classification method, and frequent itemsets are obtained, resulting in a neural network that includes design relationship constraints. This method, by mining and learning the design relationships in the HVAC system sample model, can uncover more hidden patterns and obtain more comprehensive design relationship constraints. Therefore, it can comprehensively and effectively identify HVAC models, better supplementing existing manual design methods and further improving the design quality of HVAC systems.

[0117] In one embodiment, such as Figure 6 As shown, an artificial intelligence-based HVAC model evaluation device is provided, comprising:

[0118] The acquisition module 701 is used to acquire a HVAC design image that meets the requirements based on a three-dimensional building model. The HVAC design image includes at least one HVAC object among a refrigerant system, a chilled water system, and a ventilation system.

[0119] The identification and evaluation module 702 is used to input the HVAC design image into a preset evaluation neural network, identify the HVAC objects in the HVAC design image and evaluate the design relationships, classify the HVAC design image according to the design relationship evaluation results, and provide design annotations for HVAC design images that do not meet the design standards.

[0120] Furthermore, in one embodiment, the HVAC intelligent design evaluation device further includes a labeling module 703, which is used to pre-establish an HVAC design standard library, identify HVAC design images that do not conform to the design standards, retrieve the corresponding design standards from the HVAC design standard library based on the HVAC object identification results, label the design standards in the HVAC design images, and mark the floors where the HVAC design objects that do not conform to the design standards are located, and the marks are visible in the three-dimensional building model.

[0121] The acquisition of HVAC design images that meet the requirements based on the above-mentioned 3D building model is achieved through the following methods:

[0122] Obtain the refrigerant system, chilled water system, and ventilation system from the 3D building model;

[0123] The HVAC objects in the refrigerant system, chilled water system, and ventilation system are located.

[0124] Based on the HVAC object, HVAC design images that meet the requirements are extracted according to preset rules.

[0125] The step of extracting HVAC design images that meet the requirements based on the HVAC object according to preset rules is achieved in the following way:

[0126] Control the virtual camera's viewpoint to capture HVAC design images that meet the requirements at a fixed angle;

[0127] The fixed angle is determined based on the floor level, indoor or outdoor location of the HVAC object.

[0128] The provision of design annotations for HVAC design images that do not conform to design standards is achieved through the following methods:

[0129] Establish a HVAC design standard library in advance and identify HVAC objects in HVAC design images that do not meet the design standards;

[0130] Based on the HVAC object recognition results, retrieve the corresponding design standards from the HVAC design standard library and annotate the design standards in the HVAC design image;

[0131] The floor where the HVAC design object that does not meet the design standards is located is marked, and the mark is visible in the three-dimensional building model.

[0132] The device provided by this invention can automatically identify the design relationships of HVAC models using neural networks, greatly improving identification efficiency and enabling more accurate and comprehensive identification of the current HVAC model's design relationships. This significantly increases accuracy while reducing human learning costs, thus greatly saving time and manpower. After evaluation and identification, the relationships can be marked on the 3D building model for easy review and modification by designers.

[0133] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0134] Based on a 3D building model, HVAC design images that meet the requirements are obtained. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0135] The HVAC design image is input into a preset evaluation neural network to identify HVAC objects and evaluate design relationships in the HVAC design image. Based on the evaluation results, the HVAC design image is classified, and design annotations are given for HVAC design images that do not meet the design standards.

[0136] In one embodiment, the computer program performs the following steps when it is also executed by the processor:

[0137] Establish a HVAC design standard library in advance and identify HVAC objects in HVAC design images that do not meet the design standards;

[0138] Based on the HVAC object recognition results, retrieve the corresponding design standards from the HVAC design standard library and annotate the design standards in the HVAC design image;

[0139] The floor where the HVAC design object that does not meet the design standards is located is marked, and the mark is visible in the three-dimensional building model.

[0140] In one embodiment, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0141] Based on the HVAC design model, an HVAC design image is obtained; the HVAC design image includes at least one HVAC object.

[0142] The HVAC design image is input into a neural network that includes several design relationship constraints to identify the design relationships between the HVAC objects and obtain the design relationships between each HVAC object.

[0143] The design relationships between various HVAC objects are matched with the design relationship constraints to obtain the identification and evaluation results; the design relationships between HVAC objects are used to characterize at least one of the refrigerant system design relationships, chilled water system design relationships, and ventilation system design relationships.

[0144] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0145] Based on a 3D building model, HVAC design images that meet the requirements are obtained. The HVAC design images include at least one HVAC object among the refrigerant system, chilled water system and ventilation system.

[0146] The HVAC design image is input into a preset evaluation neural network to identify HVAC objects and evaluate design relationships in the HVAC design image. Based on the evaluation results, the HVAC design image is classified, and design annotations are given for HVAC design images that do not meet the design standards.

[0147] It should be clear that the process of the processor executing the computer program in the above embodiments is consistent with the execution process of each step in the above method, as can be seen in the description above.

[0148] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0149] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An artificial intelligence-based heating and ventilation design evaluation method, characterized by, The method comprises: obtaining a heating and ventilation design image meeting the requirements based on a three-dimensional building model, wherein the heating and ventilation design image comprises at least one heating and ventilation object in a refrigerant system, a chilled water system and a ventilation system; inputting the heating and ventilation design image into a preset evaluation neural network, identifying the heating and ventilation object in the heating and ventilation design image and evaluating a design relationship, classifying the heating and ventilation design image according to the evaluation result of the design relationship, and giving a design annotation for the heating and ventilation design image not meeting the design standard, wherein the training process of the evaluation neural network comprises: converting a heating and ventilation system sample model pre-labeled with the heating and ventilation object from a three-dimensional graph to a two-dimensional image to obtain a plurality of initial heating and ventilation design image samples; pre-processing the initial heating and ventilation design image samples to obtain heating and ventilation design image samples; inputting the heating and ventilation design image samples into a preset initial neural network, extracting a design relationship based on a decision tree classification method, and obtaining a frequent item set, and obtaining the evaluation neural network through constraint.

2. The method of claim 1, wherein, The method comprises: obtaining a heating and ventilation design image meeting the requirements based on a three-dimensional building model, wherein the heating and ventilation design image comprises at least one heating and ventilation object in a refrigerant system, a chilled water system and a ventilation system; positioning the heating and ventilation object in each system in the refrigerant system, the chilled water system and the ventilation system; intercepting a heating and ventilation design image meeting the requirements based on the heating and ventilation object according to a preset rule.

3. The method of claim 2, wherein, The method comprises: controlling a virtual camera perspective to intercept a heating and ventilation design image meeting the requirements at a fixed angle; the fixed angle is determined according to a floor, an indoor space or an outdoor space where the heating and ventilation object is located.

4. The method of claim 1, wherein, The method comprises: pre-establishing a heating and ventilation design standard library, identifying the heating and ventilation object in the heating and ventilation design image not meeting the design standard; according to the identification result of the heating and ventilation object, calling a corresponding design standard in the heating and ventilation design standard library, and labeling the design standard in the heating and ventilation design image; and marking a floor where the heating and ventilation object not meeting the design standard is located, wherein the marking is visible in the three-dimensional building model.

5. An artificial intelligence-based heating and ventilation design evaluation method, characterized by, The method comprises: obtaining a heating and ventilation design image based on a heating and ventilation design model, wherein the heating and ventilation design image comprises at least one heating and ventilation object; inputting the heating and ventilation design image into a neural network comprising a plurality of design relationship constraint conditions, identifying a design relationship between the heating and ventilation objects to obtain the design relationship between each heating and ventilation object; wherein the training process of the neural network comprises: converting a heating and ventilation system sample model pre-labeled with the heating and ventilation object from a three-dimensional graph to a two-dimensional image to obtain a plurality of initial heating and ventilation design image samples; pre-processing the initial heating and ventilation design image samples to obtain heating and ventilation design image samples; inputting the heating and ventilation design image samples into a preset initial neural network, extracting a design relationship based on a decision tree classification method, and obtaining a frequent item set, and obtaining the neural network through constraint. The design relationship between each heating and ventilation object is matched with the design relationship constraint condition to obtain an identification evaluation result; the design relationship between the heating and ventilation objects is used to represent at least one of a refrigerant system design relationship, a cold water system design relationship, and a ventilation system design relationship.

6. The method of claim 5, wherein, The metric indicators of the design relationship constraint condition include support, confidence, and promotion.

7. An artificial intelligence-based heating and ventilation model evaluation device, characterized by, The apparatus comprises: The acquisition module is configured to acquire a heating and ventilation design image that meets a requirement based on a three-dimensional building model, wherein the heating and ventilation design image includes at least one heating and ventilation object in a refrigerant system, a cold water system, and a ventilation system. The identification evaluation module is configured to input the heating and ventilation design image into a preset evaluation neural network, identify and evaluate the design relationship of the heating and ventilation object in the heating and ventilation design image, classify the heating and ventilation design image according to the design relationship evaluation result, and give a design annotation for the heating and ventilation design image that does not meet the design standard, wherein the training process of the evaluation neural network includes converting a heating and ventilation system sample model, in which the heating and ventilation object is pre-labeled, from three-dimensional graphics to a two-dimensional image to obtain a plurality of initial heating and ventilation design image samples, pre-processing the initial heating and ventilation design image samples to obtain heating and ventilation design image samples, inputting the heating and ventilation design image samples into a preset initial neural network, extracting the design relationship based on a decision tree classification method, and obtaining a frequent item set to obtain the evaluation neural network through constraint.

8. The apparatus of claim 7, wherein, The apparatus further comprises: The labeling module is configured to pre-establish a heating and ventilation design standard library, identify the heating and ventilation object in the heating and ventilation design image that does not meet the design standard, retrieve the corresponding design standard in the heating and ventilation design standard library according to the identification result of the heating and ventilation object, label the design standard in the heating and ventilation design image, and mark the floor on which the heating and ventilation object that does not meet the design standard is located, wherein the mark is visible in the three-dimensional building model. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the method of any one of claims 1 to 6.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed to implement the steps of the method of any one of claims 1 to 6.

Images