A Power Grid 3D Spatial Layout Optimization Method Based on Ergonomics

Abstract

This invention belongs to the field of three-dimensional spatial layout, specifically disclosing a power grid three-dimensional spatial layout optimization method based on human factors engineering. The method includes: modeling the power grid three-dimensional spatial layout process, constructing a layout optimization model, and optimizing the layout optimization model. This scheme utilizes Markov decision processes to dynamically model a multi-dimensional state space and a composite action set, combined with a multi-objective reward function, achieving a dual improvement in dynamic optimization of spatial resource allocation and human-machine collaboration safety. It employs voxel feature extraction based on a three-dimensional convolutional neural network and integrates linear entropy and Lagrange multinomial interpolation to adjust the objective function of the near-end strategy optimization algorithm, accelerating the global convergence speed and enhancing the diversity of layout strategies, thus identifying the optimal power grid three-dimensional spatial layout method.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional spatial layout, specifically to a method for optimizing the three-dimensional spatial layout of power grids based on human factors engineering. Background Technology

[0002] Artificial intelligence can be used to plan the location distribution of power grid facilities in three-dimensional space, which can optimize spatial resources, reduce material consumption and construction difficulty, reduce the possibility of failures, and improve power supply reliability. However, traditional three-dimensional power grid layout methods rely on static models and cannot dynamically respond to environmental changes and operational needs, resulting in rigid spatial resource allocation and low utilization. At the same time, they lack dynamic integration of ergonomic parameters, posing a risk of collision for operators and insufficient interactive safety. Existing reinforcement learning algorithms have low strategy exploration efficiency and slow convergence speed when laying out in three-dimensional space, and are prone to getting trapped in local optima, resulting in poor global optimization capabilities and layout diversity. Summary of the Invention

[0003] To address the aforementioned issues and overcome the shortcomings of existing technologies, this invention provides a ergonomic-based method for optimizing the three-dimensional spatial layout of power grids. Traditional three-dimensional power grid layout methods rely on static models, failing to dynamically respond to environmental changes and operational needs, resulting in rigid spatial resource allocation, low utilization, and a lack of dynamic integration of ergonomic parameters, leading to collision risks and insufficient interactive safety for operators. This solution utilizes Markov decision processes to dynamically model a multi-dimensional state space and composite action set, combined with a multi-objective reward function, achieving a dual improvement in dynamic optimization of spatial resource allocation and human-machine collaboration safety. Furthermore, addressing the low efficiency, slow convergence, and susceptibility to local optima in existing reinforcement learning algorithms for three-dimensional layout, resulting in poor global optimization capabilities and layout diversity, this solution employs voxel feature extraction based on a three-dimensional convolutional neural network and integrates linear entropy and Lagrange multinomial interpolation to adjust the objective function of the near-end strategy optimization algorithm. This accelerates global convergence and enhances the diversity of layout strategies, enabling the identification of the optimal three-dimensional spatial layout method for the power grid.

[0004] The technical solution adopted by this invention is as follows: This invention provides a method for optimizing the three-dimensional spatial layout of power grids based on human factors engineering, specifically including the following steps:

[0005] Step S1: Modeling the three-dimensional spatial layout of the power grid;

[0006] Step S2: Construct a layout optimization model. Use a 3D convolutional neural network model to construct a layout optimization model and perform layout optimization based on the near-end strategy optimization algorithm.

[0007] Step S3: Optimize the layout optimization model by introducing linear and polynomial interpolation into the near-end strategy optimization algorithm to adjust it;

[0008] Further, in step S1, the three-dimensional spatial layout process of the power grid is modeled, specifically as follows: the three-dimensional spatial layout process of the power grid is modeled as a Markov decision process, and the three-dimensional spatial layout process of the power grid is modeled as (S, A, P, R, γ), where S represents the spatial state, A represents the action space, P represents the state transition probability, R represents the reward function, and γ is the coordination factor between the current action space and the future action space.

[0009] The design of the spatial state, action space, and reward function in the three-dimensional spatial layout process of the power grid includes:

[0010] The spatial state in the three-dimensional spatial layout process of the power grid is defined as a multi-dimensional state vector. The multi-dimensional state vector includes the geometric information of the three-dimensional space of the power grid and the power grid equipment, ergonomic parameters, the occupancy status of the power grid equipment, and environmental constraints.

[0011] The action space in the three-dimensional spatial layout process of the power grid is defined as a composite action set, which is a set of discrete and continuous actions for placing power grid equipment;

[0012] The reward function in the three-dimensional spatial layout process of the power grid is defined as a multi-objective reward function, which includes the three-dimensional spatial utilization rate of the power grid and the collision penalty, and the collision penalty includes human-machine collision penalty and equipment collision penalty;

[0013] Further, in step S2, the construction of the layout optimization model is specifically carried out as follows: a three-dimensional convolutional neural network model is created and initialized as the layout optimization model, and the layout optimization model adjusts the three-dimensional spatial layout of the power grid according to the spatial state, action space and reward function;

[0014] The three-dimensional space of the power grid is discretized into a voxel grid, and each voxel grid contains multi-channel features, including the occupancy rate of power grid equipment, the number of power grid equipment and the material. The layout optimization model extracts the high-level features of the voxel grid.

[0015] The layout optimization model is trained using a proximal policy optimization algorithm, which includes a policy network and a value function network. The high-level features of the voxel grid are input into the policy network to output the action distribution. The action distribution of discrete actions is defined as a Softmax probability distribution, and the action distribution of continuous actions is defined as a Gaussian distribution. The value function network estimates the current state value based on the action distribution.

[0016] A 3D modeling engine is integrated into the layout optimization model, and a digital human body model is embedded to simulate the movement path of workers operating power grid equipment. The accessibility score is calculated based on the range of limb movement of the digital human body model as a human-machine collision penalty.

[0017] Furthermore, in step S3, the optimization of the layout optimization model involves introducing linear and polynomial interpolation into the near-end strategy optimization algorithm for adjustment, specifically including the following steps:

[0018] Step S31: Introduce entropy into the near-end policy optimization algorithm to quantify the randomness of the output actions of the layout optimization model. Define the objective function of the near-end policy optimization algorithm using the following formula:

[0019] ;

[0020] ;

[0021] In the formula, It is the objective function of the near-end policy optimization algorithm. This is the original objective function of the near-end policy optimization algorithm. These are strategy parameters. It is a constant whose value is greater than or equal to 0 and less than or equal to 1, and it represents the weight of the entropy value in the objective function. This represents the three-dimensional spatial layout strategy of the power grid. It is the current time step. It is an action in the action space. It is a state within a spatial state. yes The entropy of the distribution strategy;

[0022] Step S32: Add a linear entropy value to the objective function of the near-end policy optimization algorithm. This enhances the randomness of the policy and optimizes the algorithm. The objective function of the near-end policy optimization algorithm after adding the linear entropy value is denoted as the linear entropy objective function, and the formula used is as follows:

[0023] ;

[0024] In the formula, It is the linear entropy objective function. It is a linear entropy value. and These are the initial and final coefficients of the linear entropy. It is the total number of time steps;

[0025] Step S33: Add a Lagrange polynomial interpolation to the objective function of the near-end policy optimization algorithm. The objective function of the near-end policy optimization algorithm after adding the Lagrange polynomial interpolation is denoted as the nonlinear entropy objective function, and the formula used is as follows:

[0026] ;

[0027] ;

[0028] In the formula, It is a nonlinear entropy objective function. It is Lagrange polynomial interpolation. It is the Lagrange interpolation at the current time step. It is the total number of terms in the Lagrange polynomial. Yes Traversal, It is a Lagrange polynomial It is a circular index variable;

[0029] Step S34: Apply the sum of the linear entropy objective function and the nonlinear entropy objective function as the objective function of the final near-end policy optimization algorithm to the layout optimization model.

[0030] The beneficial effects achieved by the present invention using the above solution are as follows:

[0031] (1) In view of the technical problems that traditional three-dimensional power grid layout methods rely on static models, cannot dynamically respond to environmental changes and operational needs, resulting in rigid spatial resource allocation and low utilization, and lack of dynamic integration of human-machine engineering parameters, there is a risk of collision during operation and insufficient interactive safety, this solution uses Markov decision process to dynamically model multi-dimensional state space and composite action set, combined with multi-objective reward function, to achieve a dual improvement of dynamic optimization of spatial resource allocation and human-machine collaboration safety;

[0032] (2) In view of the technical problems of low strategy exploration efficiency, slow convergence speed, and easy getting trapped in local optima when existing reinforcement learning algorithms are deployed in three-dimensional space, resulting in poor global optimization ability and layout diversity, this scheme adopts voxel feature extraction based on three-dimensional convolutional neural network and integrates linear entropy and Lagrange polynomial interpolation to adjust the objective function of the near-end strategy optimization algorithm, which accelerates the global convergence speed and enhances the diversity of layout strategies, and can find the optimal three-dimensional layout method of power grid. Attached Figure Description

[0033] Figure 1 The flowchart illustrates the steps of the ergonomics-based three-dimensional spatial layout optimization method for power grids provided by this invention.

[0034] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0035] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0037] Example 1: See Figure 1 This embodiment provides a power grid three-dimensional spatial layout optimization method based on ergonomics. The present invention provides a power grid three-dimensional spatial layout optimization method based on ergonomics, specifically including the following steps:

[0038] Step S1: Modeling the three-dimensional spatial layout of the power grid;

[0039] Step S2: Construct a layout optimization model. Use a 3D convolutional neural network model to construct a layout optimization model and perform layout optimization based on the near-end strategy optimization algorithm.

[0040] Step S3: Optimize the layout optimization model by introducing linear and polynomial interpolation into the near-end strategy optimization algorithm to adjust it.

[0041] Example 2: See Figure 1 This embodiment is based on the above embodiment. In step S1, the three-dimensional spatial layout process of the power grid is modeled. The specific operation is as follows: the three-dimensional spatial layout process of the power grid is modeled as a Markov decision process. The three-dimensional spatial layout process of the power grid is modeled and represented as (S, A, P, R, γ), where S represents the spatial state, A represents the action space, P represents the state transition probability, R represents the reward function, and γ is the coordination factor between the current action space and the future action space.

[0042] The design of the spatial state, action space, and reward function in the three-dimensional spatial layout process of the power grid includes:

[0043] The spatial state in the three-dimensional spatial layout process of the power grid is defined as a multi-dimensional state vector. The multi-dimensional state vector includes the geometric information of the three-dimensional space of the power grid and the power grid equipment, ergonomic parameters, the occupancy status of the power grid equipment, and environmental constraints.

[0044] The action space in the three-dimensional spatial layout process of the power grid is defined as a composite action set, which is a set of discrete and continuous actions for placing power grid equipment;

[0045] The reward function in the three-dimensional spatial layout process of the power grid is defined as a multi-objective reward function, which includes the three-dimensional spatial utilization rate of the power grid and the collision penalty, and the collision penalty includes human-machine collision penalty and equipment collision penalty.

[0046] Through the above operations, this solution addresses the technical problems of traditional power grid 3D layout methods relying on static models, which cannot dynamically respond to environmental changes and operational needs, resulting in rigid spatial resource allocation, low utilization, and a lack of dynamic integration of ergonomic parameters, leading to collision risks and insufficient interactive safety for operators. This solution utilizes Markov decision processes to dynamically model multi-dimensional state spaces and composite action sets, combined with multi-objective reward functions, to achieve a dual improvement in dynamic optimization of spatial resource allocation and human-machine collaboration safety.

[0047] Example 3: See Figure 1 This embodiment is based on the above embodiment. In step S2, the construction of the layout optimization model is specifically carried out as follows: a three-dimensional convolutional neural network model is created and initialized as the layout optimization model. The layout optimization model adjusts the three-dimensional spatial layout of the power grid according to the spatial state, action space and reward function.

[0048] The three-dimensional space of the power grid is discretized into a voxel grid, and each voxel grid contains multi-channel features, including the occupancy rate of power grid equipment, the number of power grid equipment and the material. The layout optimization model extracts the high-level features of the voxel grid.

[0049] The layout optimization model is trained using a proximal policy optimization algorithm, which includes a policy network and a value function network. The high-level features of the voxel grid are input into the policy network to output the action distribution. The action distribution of discrete actions is defined as a Softmax probability distribution, and the action distribution of continuous actions is defined as a Gaussian distribution. The value function network estimates the current state value based on the action distribution.

[0050] A 3D modeling engine is integrated into the layout optimization model, and a digital human body model is embedded to simulate the movement path of workers operating power grid equipment. The accessibility score is calculated based on the range of limb movement of the digital human body model as a human-machine collision penalty.

[0051] Example 4: See Figure 1 This embodiment is based on the above embodiment. In step S3, the optimization of the layout optimization model involves introducing linear and polynomial interpolation into the near-end strategy optimization algorithm for adjustment. Specifically, this includes the following steps:

[0052] Step S31: Introduce entropy into the near-end policy optimization algorithm to quantify the randomness of the output actions of the layout optimization model. Define the objective function of the near-end policy optimization algorithm using the following formula:

[0053] ;

[0054] ;

[0055] In the formula, It is the objective function of the near-end policy optimization algorithm. This is the original objective function of the near-end policy optimization algorithm. These are strategy parameters. It is a constant whose value is greater than or equal to 0 and less than or equal to 1, and it represents the weight of the entropy value in the objective function. This represents the three-dimensional spatial layout strategy of the power grid. It is the current time step. It is an action in the action space. It is a state within a spatial state. yes The entropy of the distribution strategy;

[0056] Step S32: Add a linear entropy value to the objective function of the near-end policy optimization algorithm. This enhances the randomness of the policy and optimizes the algorithm. The objective function of the near-end policy optimization algorithm after adding the linear entropy value is denoted as the linear entropy objective function, and the formula used is as follows:

[0057] ;

[0058] In the formula, It is the linear entropy objective function. It is a linear entropy value. and These are the initial and final coefficients of the linear entropy. It is the total number of time steps;

[0059] Step S33: Add a Lagrange polynomial interpolation to the objective function of the near-end policy optimization algorithm. The objective function of the near-end policy optimization algorithm after adding the Lagrange polynomial interpolation is denoted as the nonlinear entropy objective function, and the formula used is as follows:

[0060] ;

[0061] ;

[0062] In the formula, It is a nonlinear entropy objective function. It is Lagrange polynomial interpolation. It is the Lagrange interpolation at the current time step. It is the total number of terms in the Lagrange polynomial. Yes Traversal, It is a Lagrange polynomial It is a circular index variable;

[0063] Step S34: Apply the sum of the linear entropy objective function and the nonlinear entropy objective function as the objective function of the final near-end policy optimization algorithm to the layout optimization model.

[0064] By employing the above methods, this solution addresses the technical problems of existing reinforcement learning algorithms in three-dimensional space layout, such as low policy exploration efficiency, slow convergence speed, and susceptibility to local optima, resulting in poor global optimization capabilities and layout diversity. Instead, it adopts voxel feature extraction based on a three-dimensional convolutional neural network and integrates linear entropy and Lagrange multinomial interpolation to adjust the objective function of the near-end policy optimization algorithm. This accelerates the global convergence speed and enhances the diversity of layout strategies, enabling the identification of the optimal three-dimensional power grid layout method.

[0065] Example 5: See Figure 1 This embodiment is based on the above embodiments. The geometric information in the spatial state of Embodiment 1 includes the three-dimensional space of the power grid and the three-dimensional coordinates, orientation, size, and topological relationship of the power grid equipment.

[0066] The ergonomic parameters are based on a digital human body model to limit the accessible area for operation, the length of the maintenance path, and the safe distance.

[0067] The occupancy status of the power grid equipment includes whether the power grid equipment has been deployed and the size of the space it occupies.

[0068] The environmental constraints include physical obstacles in the three-dimensional space of the power grid, such as walls, vegetation and terrain obstacles, as well as regulatory requirements, such as fire separation distances and equipment safety distances;

[0069] The action space is defined as a set of composite actions, which specifically includes adjustments to power grid equipment, such as translation and rotation operations, dynamic addition and deletion of nodes between power grid equipment, and adjustment of wiring paths.

[0070] The reward function is a multi-objective reward function. In practical applications, functional rewards can also be added, including power grid equipment line losses and voltage stability.

[0071] The collision penalty is implemented by setting a minimum safe distance between humans and machines and a minimum safe distance between devices.

[0072] Example 6 is based on the above examples. The reachability score calculated based on the range of motion of the limbs in the digital human body model, as described in Example 3, is used as a human-machine collision penalty. The reachability score is the ratio of the number of operation points that the human body can cover to the total number of operation points.

[0073] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0074] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0075] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for optimizing the three-dimensional spatial layout of an electrical grid based on ergonomics, characterized in that, Specifically comprising the following steps: Step S1: modeling the three-dimensional spatial layout process of the power grid; Step S2: constructing a layout optimization model, using a three-dimensional convolutional neural network model to construct the layout optimization model, and performing layout optimization based on a proximal policy optimization algorithm; Step S3: optimizing the layout optimization model, introducing linear and polynomial interpolation into the proximal policy optimization algorithm to adjust the proximal policy optimization algorithm; In step S1, the three-dimensional spatial layout process of the power grid is modeled, and the specific operation is as follows: the three-dimensional spatial layout process of the power grid is modeled as a Markov decision process, the three-dimensional spatial layout process of the power grid is modeled, and the three-dimensional spatial layout process of the power grid is represented as (S, A, P, R, γ), wherein S represents the spatial state, A represents the action space, P represents the state transition probability, R represents the reward function, and γ is the coordination factor of the current action space and the future action space; The spatial state, action space and reward function in the three-dimensional spatial layout process of the power grid are designed, specifically including: The spatial state in the three-dimensional spatial layout process of the power grid is defined as a multi-dimensional state vector, which includes the geometric information of the three-dimensional space of the power grid and the power grid equipment, the ergonomics parameters, the occupancy state of the power grid equipment and the environmental constraints; The action space in the three-dimensional spatial layout process of the power grid is defined as a composite action set, which is a set of discrete actions and continuous actions for placing power grid equipment; The reward function in the three-dimensional spatial layout process of the power grid is defined as a multi-objective reward function, which includes the utilization rate of the three-dimensional space of the power grid and the collision penalty, and the collision penalty includes the human-machine collision penalty and the equipment collision penalty; In step S2, the layout optimization model is constructed, and the specific operation is as follows: a three-dimensional convolutional neural network model is created and initialized as a layout optimization model, and the layout optimization model adjusts the three-dimensional spatial layout of the power grid according to the spatial state, action space and reward function; The three-dimensional space of the power grid is discretized into a voxel grid, each voxel grid contains multi-channel features, including power grid equipment occupancy rate, power grid equipment number and material, and the layout optimization model extracts high-level features of the voxel grid; The proximal policy optimization algorithm is used to train the layout optimization model, which includes a policy network and a value function network, the high-level features of the voxel grid are input into the policy network to output the action distribution, the action distribution of discrete action is defined as Softmax probability distribution, the action distribution of continuous action is Gaussian distribution, and the value function network estimates the current state value according to the action distribution; In the layout optimization model, a three-dimensional modeling engine is integrated, a digital human body model is embedded, the action path of the staff operating the power grid equipment is simulated, and the reachability score is calculated as the human-machine collision penalty based on the limb activity range of the digital human body model.

2. The ergonomic-based electrical grid three-dimensional spatial layout optimization method of claim 1, wherein, In step S3, the layout optimization model is optimized, and linear and polynomial interpolation is introduced into the proximal policy optimization algorithm for adjustment, specifically including the following steps: Step S31: introducing entropy value into the proximal policy optimization algorithm; Step S32: add a linear entropy value to the objective function of the proximal policy optimization algorithm, and the objective function of the proximal policy optimization algorithm after adding the linear entropy value is denoted as a linear entropy objective function; Step S33: add a Lagrange polynomial interpolation to the objective function of the proximal policy optimization algorithm, and the objective function of the proximal policy optimization algorithm after adding the Lagrange polynomial interpolation is denoted as a nonlinear entropy objective function; Step S34: apply the sum of the linear entropy objective function and the nonlinear entropy objective function as the objective function of the final proximal policy optimization algorithm to the layout optimization model.

Images