A performance intelligent master control method and system based on stage design analysis

Abstract

The application relates to a performance intelligent master control method based on stage design analysis, which comprises the following steps: receiving stage structure data and performing analysis and processing, constructing a stage overall three-dimensional structure model through a three-dimensional modeling algorithm; based on expected performance effect parameters, an initial master control scheme of light, sound, machinery and video multi-system linkage is intelligently generated by adopting a multi-target collaborative optimization algorithm; audience feedback data is collected in real time through a multi-modal sensing array, an audience emotion time sequence curve is generated through emotional semantic analysis, the emotion curve is embedded into a reinforcement learning reward function as a feedback variable, the initial master control scheme is subjected to human-computer emotion coupling optimization, and an optimized master control strategy with enhanced emotional resonance is formed; through machine learning algorithm, multi-physical field coupling simulation deduction and multi-target optimization are performed on a performance process, an executable master control strategy is formed and output to a performance execution system. The application has the effect of improving the control precision of stage performance.

Description

Technical Field

[0001] This invention relates to the technical field of stage intelligent control, and in particular to a method and system for intelligent overall control of performances based on stage design analysis. Background Technology

[0002] With the rapid development of the cultural and performing arts industry and the continuous improvement of audience aesthetic demands, modern stage performances are increasingly showing a trend of multi-system collaboration, high-precision synchronization, and immersive experiences. This poses unprecedented challenges to the overall performance control technology. Currently, the field of overall performance control mainly relies on the traditional stage art design combined with manual programming control mode. Based on the director's artistic conception, technicians independently program and manually coordinate subsystems such as lighting, sound, machinery, and video. However, this work mode, which is centered on human experience, increases the complexity of multi-system collaborative control. The spatiotemporal coupling relationships and signal dependency logic between subsystems need to be repeatedly checked manually. This is prone to resource conflicts, signal collisions, or action misalignments due to oversights. It is not only time-consuming and labor-intensive, but also carries high risks for on-site debugging, reducing the control precision of stage performances. Therefore, improvements are needed. Summary of the Invention

[0003] To effectively improve the control precision of stage performances, this application provides a method and system for intelligent overall control of performances based on stage design analysis.

[0004] Firstly, the above-mentioned inventive objective of this application is achieved through the following technical solution: A method for intelligent overall control of performances based on stage design analysis, the method comprising the following steps: Receive and analyze stage structure data, and construct an overall three-dimensional structure model of the stage that includes stage space geometric features, equipment installation points, and mechanical motion envelopes through three-dimensional modeling algorithms; Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video. The system collects audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data in real time through a multimodal sensor array. The system generates audience emotion time-series curves through emotion semantic analysis. These emotion curves are then embedded as feedback variables into a reinforcement learning reward function. The initial overall control scheme is optimized through human-machine emotion coupling to form an optimized overall control strategy that enhances emotional resonance. Machine learning algorithms are used to perform multi-physics coupling simulation and multi-objective optimization of the performance process, forming an executable overall control strategy and outputting it to the performance execution system.

[0005] By adopting the above technical solutions, compared with the traditional performance control relying on manual programming and experience-based debugging, this solution achieves a leap from "mechanical execution" to "intelligent generation": 3D modeling automatically identifies spatial conflicts, improving efficiency and avoiding safety hazards compared to manual verification; multi-system collaborative optimization automatically generates massive amounts of control instructions, solving the pain points of error-prone manual programming and difficulty in coordinating resource conflicts; real-time feedback and coupling optimization of audience emotions create a new "audience-centric" intelligent performance model, transforming the performance effect from one-way output to two-way interaction, significantly improving audience immersion and satisfaction; multi-physics field deduction and reinforcement learning optimization rehearse extreme conditions in a virtual environment, ensuring the robustness and security of the strategy, avoiding the risks and costs of on-site debugging. Overall, this application transforms performance control from complex manual labor of art + technology into intelligent creation, significantly shortening the production cycle, improving control precision, reducing human error, and effectively improving the overall quality of stage performances.

[0006] In a preferred embodiment, this application can be further configured such that the step of receiving and analyzing stage structure data, and constructing a three-dimensional overall stage structure model including stage space geometric features, equipment installation points, and mechanical motion envelopes using a three-dimensional modeling algorithm, includes the following steps: A multi-source data fusion interface is constructed to receive stage CAD drawings, equipment BIM models, mechanical kinematic parameters, and finite metadata of load-bearing structures. After data cleaning and coordinate system integration, a standardized structural description set is generated. A spatial topology analysis algorithm is used to identify equipment interference areas, line-of-sight obstruction areas, and actor movement conflict areas in the stage structure, and a spatial availability assessment map is generated. Based on the structural description set and availability evaluation map, a three-dimensional stage structure model with a three-dimensional mesh is created using parametric modeling technology. Each device unit in the model is bound to its motion degree of freedom, signal interface and energy consumption attributes. Embedding equipment collision detection constraints and structural safety load constraints in the 3D model, parameter correction feedback is automatically triggered when subsequent schemes violate the constraints.

[0007] In a preferred example, this application can be further configured as follows: the step of intelligently generating an initial overall control scheme for the linkage of multiple systems including lighting, sound, machinery, and video using a multi-objective collaborative optimization algorithm based on expected performance effect parameters includes the following steps: Construct a semantic parser for performance effects, which transforms the director's artistic description text into a set of quantitative parameters for lighting intensity, sound pressure level, mechanical speed, and video frame rate, and establishes a mapping relationship between effect parameters and audience perceived intensity. A multi-system collaborative optimization model is constructed using a graph neural network architecture. The model nodes represent lighting, sound, mechanical, and video subsystems, and the edge weights represent the spatiotemporal coupling strength and signal dependence between the systems. The optimal action timing and intensity curves of each system are solved by embedding device motion smoothness constraints, signal transmission synchronization constraints, and total energy power constraints into the loss function of the optimization model, and the gradient descent algorithm is used to solve the optimal action timing and intensity curves of each system. When generating the master control scheme, the system resource occupancy Gantt chart and signal conflict detection report are output simultaneously. When resource contention or signal collision is detected, the time slice rotation or priority arbitration strategy is automatically enabled.

[0008] In a preferred example, this application can be further configured as follows: In the step of acquiring audience physiological signals, facial expressions, vocal emotions, and on-site atmosphere data in real time through a multimodal sensor array, generating audience emotion time-series curves through emotional semantic analysis, embedding the emotion curves as feedback variables into a reinforcement learning reward function, and optimizing the initial overall control scheme through human-machine emotional coupling to form an optimized overall control strategy with enhanced emotional resonance, the following steps are included: Construct a multimodal audience sensor network, integrating wearable heart rate sensors, infrared thermal imaging facial recognition, microphone array voice pickup, and ambient light intensity sensors to collect data on audience physiological arousal, facial expression valence, voice emotional intensity, and on-site atmosphere activity. Temporal convolutional networks are used to perform emotional semantic fusion on multimodal sensor data to generate audience emotional temporal curves. The peak points of the curves correspond to emotional climaxes, and the trough points correspond to emotional calm intervals. The emotional time-series curve is transformed into emotional weight coefficients of the reinforcement learning reward function. When a downward trend in the emotional curve is detected, the color saturation of the lights is automatically enhanced, the low-frequency gain of the speakers is increased, and the frequency of mechanical movements is accelerated to form an emotional compensation strategy. In a digital twin simulation environment, the effect of emotional compensation strategies on enhancing the audience's emotional curve is simulated. The intensity and timing of compensation are optimized through backpropagation to avoid emotional fatigue caused by overcompensation.

[0009] In a preferred example, this application can be further configured such that the step of performing multi-physics coupling simulation and multi-objective optimization of the performance process using machine learning algorithms to form an executable overall control strategy and output it to the performance execution system includes the following steps: Construct a digital twin simulation environment for the performance process, integrating lighting intensity attenuation models, sound field propagation models, mechanical dynamics models, video rendering delay models, and audience emotional response models to achieve pre-performance calculations that couple multiple physical fields with emotional fields; The agent is trained using a reinforcement learning algorithm. The agent uses an initial overall control scheme as its strategy and the performance realism and audience emotional resonance as its dual-objective reward function. The agent explores the optimal action sequence through Monte Carlo tree search. Random disturbances are injected during the simulation process to assess the robustness margin of the overall control strategy under uncertainties such as actor positioning deviations, equipment signal delays, and sudden changes in audience emotions, and to generate emotional compensation contingency plans for vulnerable links. The optimized strategy is transformed into an executable script, which includes the control command sequence of each system, emergency trigger conditions, manual takeover interface and emotional compensation node, and outputs an interpretability report of the strategy for the director's review.

[0010] In a preferred example, this application can be further configured to include a closed-loop feedback and dynamic strategy adjustment mechanism for the performance process, including the following steps: A three-dimensional spatiotemporal-emotional coupled data acquisition network is constructed. A multimodal sensor array is deployed in the performance area to collect in real time the operating parameters of the equipment, the physical field distribution data of the on-site environmental sensors, the arousal index of the audience's physiological signal sensors, the facial expression valence data of the facial recognition camera, the voice emotion intensity of the microphone array, and the ambient activity of the ambient light sensor. After spatiotemporal registration and feature fusion, a high-dimensional coupled dataset is generated. The coupled dataset is input into the online strategy evaluation module to construct a dual-channel deviation evaluation engine. The first channel calculates the timing deviation between the actual movement trajectory of the device and the preset overall control scheme, and the second channel calculates the morphological deviation between the audience's emotional curve and the target emotional curve. When the deviation of any channel exceeds the dynamic threshold, a strategy fine-tuning request is triggered. An adaptive emotion compensation intensity regulator is constructed. Based on the rate of change and acceleration of the audience's emotion curve, the regulator automatically calculates the compensation gain coefficient. When the rate of decline of emotion exceeds a preset threshold, the compensation intensity is increased, and when the emotion tends to stabilize, the compensation is weakened to avoid oversaturation. The model predictive control algorithm is used to perform rolling optimization of the remaining performance process. In the prediction time domain, a multi-step optimization solution is performed with a dual objective function. The first objective is to minimize the equipment action deviation and energy consumption increment, and the second objective is to maximize the goodness of fit of the audience emotion curve, and output the optimal control increment sequence. During strategy execution, a human-machine collaborative intervention interface is built. When the director or operator manually adjusts the equipment parameters, the system encodes the manual correction as an additional constraint and injects it into the rolling optimizer to achieve adaptive fusion of human-machine decision-making weights. Establish an intelligent matching mechanism for emergency plans. When a fault signal is detected or the audience's emotional curve reaches an extreme low point, the mechanism retrieves the compensation plan that best matches the current performance progress, emotional state, and fault type from the plan database. Based on the output of the emotional compensation intensity regulator, the mechanism dynamically adjusts the action range of the plan to achieve seamless switching of the performance process and a smooth transition of the audience experience. The effect data of each strategy fine-tuning and emergency plan execution is recorded in the strategy replay experience pool. The priority experience replay mechanism is used to train the strategy optimization agent, so that the closed-loop feedback system has the ability to continuously learn from real combat data and improve the strategy itself.

[0011] Secondly, the above-mentioned inventive objective of this application is achieved through the following technical solutions: A performance intelligent control device based on stage design analysis, the device includes: a stage overall three-dimensional structure model construction unit, used to receive stage structure data and perform analysis and processing, and construct an overall three-dimensional structure model of the stage including stage space geometric features, equipment installation points and mechanical motion envelope through a three-dimensional modeling algorithm; The initial master control scheme generation unit is used to intelligently generate an initial master control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video, based on the expected performance effect parameters and using a multi-objective collaborative optimization algorithm. An optimized overall control strategy generation unit is used to collect audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data in real time through a multimodal sensor array. The audience's emotional time-series curve is generated through emotional semantic parsing. The emotional curve is embedded as a feedback variable into the reinforcement learning reward function to optimize the human-machine emotional coupling of the initial overall control scheme and form an optimized overall control strategy with enhanced emotional resonance. The overall control strategy output execution unit is used to perform multi-physics coupling simulation and multi-objective optimization of the performance process through machine learning algorithms, form an executable overall control strategy, and output it to the performance execution system.

[0012] Thirdly, the above-mentioned objectives of this application are achieved through the following technical solutions: An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described intelligent control method for performances based on stage design analysis.

[0013] Fourthly, the above-mentioned objectives of this application are achieved through the following technical solutions: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described intelligent overall control method for performances based on stage design analysis. Attached Figure Description

[0014] Figure 1 This is a flowchart of a performance intelligent control method based on stage design analysis in one embodiment of this application; Figure 2 This is a schematic diagram of a performance intelligent control device based on stage design analysis in one embodiment of this application; Figure 3 This is a schematic diagram of an electronic device according to an embodiment of this application.

[0015] Icon labels: 1. Stage overall 3D structure model construction unit; 2. Initial overall control scheme generation unit; 3. Optimized overall control strategy generation unit; 4. Overall control strategy output execution unit. Detailed Implementation

[0016] The present application will be further described in detail below with reference to the accompanying drawings.

[0017] In one embodiment, such as Figure 1 As shown, this application discloses a performance intelligent control method based on stage design analysis, which specifically includes the following steps: S10: Receives and analyzes stage structure data, and constructs an overall three-dimensional structure model of the stage, including stage space geometric features, equipment installation points, and mechanical motion envelopes, through three-dimensional modeling algorithms. Specifically, during the performance preparation process, the system receives stage CAD drawings (including a large circular rotating stage, multi-layer lifting platforms, and multiple LED ice screen hanging structures) and equipment BIM models (including dozens of computer-controlled lights, multiple line array speakers, and multiple projectors) from the stage designer. Through 3D modeling algorithms, the system identifies spatial interference areas between the lifting platform and the ice screen rigging when the platform is raised to a higher position, automatically marking these as spatial conflict zones. Simultaneously, based on mechanical kinematic parameters, the system calculates the maximum angular velocity limit of the rotating stage; exceeding this value will affect the stability of the performers. In the final constructed 3D stage structure model, each lighting unit is bound to a signal interface and power consumption attribute, and the lifting platform is bound to hydraulic drive degrees of freedom and safety loads. The model is accurate to the millimeter level, providing a non-interference, safe, and reliable virtual physical space for subsequent scheme generation. S20: Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video. The initial overall control scheme includes the spatiotemporal action sequence of each system, signal triggering logic, and resource conflict resolution strategy. Specifically, the director proposed the following effect parameters for a "futuristic and stunning opening": the lighting needed to gradually transition from cool to warm tones with pulses, and the sound system needed to coordinate with the mechanical lifting to create a dynamic sense of pressure. A multi-objective collaborative optimization algorithm was activated: a graph neural network model identified lighting, sound, mechanics, and video as four nodes, with edge weights defining high coupling strength between lighting and mechanics (needing synchronization) and medium coupling strength between sound and video (for rhythm matching). The optimizer determined that during the opening phase, the mechanical platform would rotate with uniform acceleration, while the color and saturation of the lighting would dynamically change, the low-frequency gain of the sound system would gradually increase, and the projected video would synchronously play a starry sky falling animation. An automatic resource conflict resolution strategy was implemented: Recognizing that a large number of lights simultaneously turning on and off would exceed the theater's power supply capacity, the algorithm implemented staggered grouping, dividing the lights into multiple groups and illuminating them sequentially with delays, ensuring visual impact while avoiding power outages.

[0018] S30: Real-time acquisition of audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data through a multimodal sensor array; generation of audience emotion time-series curves through emotional semantic analysis; embedding the emotion curves as feedback variables into the reinforcement learning reward function; optimization of the initial overall control scheme through human-machine emotional coupling; forming an optimized overall control strategy that enhances emotional resonance. Specifically, during the performance, the system uses a multimodal sensor array to perceive audience emotions in real time: infrared thermal imaging cameras installed in the audience area in front of the stage identify facial micro-expressions, wearable heart rate wristbands are provided to some audience members to collect physiological arousal levels, and a microphone array picks up the intensity of applause and cheers from the audience. Midway through the performance, the emotional semantic analysis generated a time-series curve showing that the audience's emotional value declined from its initial peak. Analysis revealed that this was due to slow mechanical movements and monotonous lighting colors in that section, resulting in a lack of emotional engagement. The system embeds the emotional curve into a reinforcement learning reward function, automatically enhancing the overall control strategy: increasing the saturation of lighting colors and adding dynamic flashing effects; increasing the rotation speed of the mechanical platform; and deepening the low-frequency response of the sound system. After implementing the emotional compensation strategy, the audience's emotional value significantly rebounded, and the intensity of applause increased, achieving a closed-loop optimization of human-machine emotional coupling and preventing a lull in the performance. S40: Through machine learning algorithms, the performance process is simulated and optimized using multi-physics coupling, forming an executable overall control strategy and outputting it to the performance execution system; Specifically, the digital twin simulation environment simulates the optimized overall control strategy: integrating a lighting intensity attenuation model (considering theater air dust scattering), a sound field propagation model (considering wall reflection and seat sound absorption), a mechanical dynamics model (considering the hysteresis caused by hydraulic oil temperature rise), a video rendering delay model, and an audience emotional response model (considering the delay of group resonance). The reinforcement learning agent rehearsed multiple times in the virtual environment, discovering that the strategy delayed emotional compensation actions when actors deviated from their positions. Through search optimization, this was changed to trigger compensation actions earlier. The final executable script includes: if the rate of audience emotional decline is detected to be too rapid, emotional compensation is initiated earlier; if the mechanical platform oil temperature is too high, the upper limit of rotation speed is automatically reduced. The script includes an interpretable report: "The advance emotional compensation can offset the delay of group resonance, and the oil temperature limit can ensure the mechanical lifespan." The director reviewed and approved its execution.

[0019] Compared to traditional performance control relying on manual programming and experience-based debugging, this solution achieves a leap from "mechanical execution" to "intelligent generation": S10's 3D modeling automatically identifies spatial conflicts, improving efficiency and avoiding safety hazards compared to manual verification; S20's multi-system collaborative optimization automatically generates massive amounts of control commands with millisecond-level accuracy, solving the pain points of error-prone manual programming and difficulty in coordinating resource conflicts; S30's real-time feedback and coupling optimization of audience emotions creates a new "audience-centric" intelligent performance model, transforming the performance effect from one-way output to two-way interaction, significantly improving audience immersion and satisfaction; S40's multi-physics field deduction and reinforcement learning optimization rehearse extreme conditions in a virtual environment, ensuring the robustness and security of the strategy and avoiding the risks and costs of on-site debugging. Overall, this application transforms performance control from complex manual labor of art + technology into AI-driven intelligent creation, significantly shortening the production cycle, improving control accuracy, and reducing human error, providing a core methodology and feasible technical path for the intelligent upgrading of the performance industry.

[0020] In step S10: Receiving and analyzing stage structure data, and constructing a 3D overall stage structure model including stage space geometric features, equipment installation points, and mechanical motion envelopes using 3D modeling algorithms, the steps include the following: S11: Construct a multi-source data fusion interface to receive stage CAD drawings, equipment BIM models, mechanical kinematic parameters and load-bearing structure finite metadata, and generate a standardized structural description set after data cleaning and coordinate system integration. S12: Use spatial topology analysis algorithms to identify equipment interference areas, sight obstruction areas, and actor movement conflict areas in the stage structure, and generate a spatial availability assessment map; S13: Based on the structural description set and availability assessment map, a three-dimensional stage structure model with a three-dimensional mesh is created using parametric modeling technology. Each device unit in the model is bound to its motion degree of freedom, signal interface and energy consumption attributes. S14: Embed equipment collision detection constraints and structural safety load constraints in the 3D model. When subsequent schemes violate the constraints, parameter correction feedback is automatically triggered.

[0021] In step S20: Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems including lighting, sound, machinery, and video. The steps include the following: S21: Construct a semantic parser for performance effects, which transforms the director's artistic description text into a set of quantitative parameters for lighting intensity, sound pressure level, mechanical speed, and video frame rate, and establishes a mapping relationship between effect parameters and audience perceived intensity. S22: A multi-system collaborative optimization model is constructed using a graph neural network architecture. The model nodes represent the lighting, sound, mechanical, and video subsystems, and the edge weights represent the spatiotemporal coupling strength and signal dependence between the systems. S23: Embed device motion smoothness constraints, signal transmission synchronization constraints, and total energy power constraints into the loss function of the optimization model, and solve the optimal action timing and intensity curves of each system through the gradient descent algorithm; S24: When generating the master control scheme, the system resource occupancy Gantt chart and signal conflict detection report are output simultaneously. When resource contention or signal collision is detected, the time slice rotation or priority arbitration strategy is automatically enabled.

[0022] In this embodiment, the performance requirements include an artistic description: "The opening should create a strong sense of futuristic technological shock, with the lighting gradually transitioning from a deep, cool blue to a blazing white pulse; the mechanical platform should exhibit an oppressive, uniform rise; and the low frequencies of the sound system should gradually increase to produce a chest-shaking effect." The performance effect semantic parser transforms this description into a set of quantifiable parameters: the color temperature range, saturation change curve, and pulse frequency parameters of the lighting subsystem; the low-frequency gain boost and duration of the sound subsystem; the rise speed function and acceleration smoothing coefficient of the mechanical platform; and the animation playback frame rate and color matching degree of the video subsystem. Simultaneously, a mapping relationship is established, such as "oppressive rise" corresponding to the audience's psychological pressure intensity index, and "blazing white pulse" corresponding to the visual impact level, ensuring that the artistic intent is calculable and optimizable.

[0023] When constructing a multi-system collaborative optimization model using a graph neural network, the model nodes include four subsystem nodes: lighting, sound, machinery, and video. Edge weights represent coupling strength: the lighting-machinery edge weight is set to a higher value, reflecting the need for synchronized lighting changes to enhance visual impact when the machinery platform is raised; the sound-video edge weight is set to a medium value, ensuring that the rhythm of the falling stars in the video animation matches the low-frequency enhancement beat of the sound. This architecture enables the optimizer to automatically identify cross-system dependencies, significantly reducing omissions and errors compared to manually identifying logical relationships.

[0024] Triple constraints are embedded in the loss function: a smoothness constraint for equipment motion to prevent sudden jitter during the lifting of the mechanical platform, ensuring the stability of the actors and the visual comfort of the audience; a signal transmission synchronization constraint to force the trigger time difference between the lighting DMX signal and the video SDI signal to be controlled within a very small range, avoiding audio-visual asynchrony; and a total energy power constraint to monitor the total power consumption of all equipment, automatically reducing the brightness of some lights or delaying the video highlight moments when the simultaneous full illumination of all lights and the superposition of bright areas in the video may exceed the limit. After solving the gradient descent algorithm, each subsystem obtains the optimal action timing: the mechanical platform adopts an S-shaped acceleration curve, the color temperature change of the lights is a function of the platform height, and the low-frequency gain of the audio system is precisely triggered when the platform reaches a specific height.

[0025] When generating the master control scheme, the system simultaneously outputs a Gantt chart of system resource usage, displaying the resource utilization rates of lighting, audio, machinery, and video in different time periods. A signal conflict detection report revealed that at the 15-second mark of the opening, the simultaneous startup of 48 moving lights resulted in a high instantaneous current demand, which, combined with the startup current of the video server's hard drive, could potentially trip the main circuit breaker. An automatic priority arbitration strategy was implemented: the moving lights were divided into 6 groups, with each group's startup delayed by a very short time, and the video server's hard drive startup completed several seconds earlier, thus avoiding the current peak. This strategy, while ensuring visual quality, eliminated the risk of resource contention, and the generated scheme could be directly implemented on-site without the need for repeated manual conflict checks.

[0026] The performance effect semantic parser transforms vague directorial descriptions into a computable set of quantifiable parameters, significantly improving the efficiency and accuracy of the conversion from artistic intent to technical implementation, and reducing communication costs and misunderstandings between the director and the technical team. The graph neural network architecture automatically sorts out the complex coupling relationships between multiple systems, significantly reducing the risk of logical omissions and signal collisions compared to traditional manual sorting, and ensuring the rigor of cross-system collaboration. Triple constraint embedding ensures that the optimization process naturally considers equipment safety, signal synchronization, and energy limitations, resulting in highly feasible solutions that avoid common problems in traditional solutions such as equipment jitter, audio-visual asynchrony, and power overload. The automatic resource conflict detection and resolution strategy transforms the tedious work of manually checking for conflicts into automatic algorithmic processing, significantly shortening the solution generation cycle and improving the reliability of the overall control strategy and the safety of on-site implementation. Overall, the S20 steps achieve intelligent transformation and automatic optimization from artistic conception to executable technical solutions, providing core support for high efficiency, high precision, and high reliability in performance overall control.

[0027] In step S30: Real-time acquisition of audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data through a multimodal sensor array; generation of audience emotion time-series curves through emotional semantic analysis; embedding the emotion curves as feedback variables into the reinforcement learning reward function; and optimizing the initial overall control scheme through human-machine emotional coupling to form an optimized overall control strategy that enhances emotional resonance. This step includes the following steps: S31: Construct a multimodal audience sensing network, integrating wearable heart rate sensors, infrared thermal imaging facial recognition, microphone array voice pickup, and ambient light intensity sensors to collect data on audience physiological arousal, facial expression valence, voice emotional intensity, and on-site atmosphere activity. S32: A temporal convolutional network is used to perform emotional semantic fusion on multimodal sensor data to generate a temporal curve of audience emotions. The peak point of the curve corresponds to the emotional climax moment, and the trough point corresponds to the emotional flat interval. S33: Transform the emotional time-series curve into the emotional weight coefficients of the reinforcement learning reward function. When a downward trend in the emotional curve is detected, automatically enhance the color saturation of the lights, increase the low-frequency gain of the speakers, and accelerate the frequency of mechanical actions to form an emotional compensation strategy. S34. In a digital twin simulation environment, preview the effect of emotional compensation strategies on the audience's emotional curve, and optimize the intensity and timing of compensation through backpropagation to avoid emotional fatigue caused by overcompensation.

[0028] In this embodiment, the system constructs a multimodal audience sensing network: infrared thermal imaging cameras are installed in the front, middle, and rear areas of the audience seating area to capture facial expression changes (focus, smiling, surprise, etc.) without the need for wearable devices; wearable heart rate wristbands are provided to some volunteer audience members to monitor changes in physiological arousal in real time; a microphone array is arranged in the theater ceiling to pick up the intensity of applause and the distribution of cheers from the audience; and ambient light sensors are deployed around the stage to sense the activity level of the atmosphere. Before the performance, the system completes the spatiotemporal registration of the sensors, aligning all data streams to the performance timeline to form a three-dimensional sensing network covering the entire venue, providing a multi-dimensional input basis for emotion analysis.

[0029] Midway through the performance, a temporal convolutional network fused and analyzed multimodal sensor data: infrared cameras detected a decrease in the proportion of audience smiles, microphone arrays detected a reduction in the frequency of applause, and heart rate wristband data showed a flattening trend in physiological arousal. The algorithm fused these signals into a temporal curve of audience emotion, clearly revealing an emotional trough. This trough was marked as an "emotional flatness moment," deviating from the emotional climax designed in the script, indicating that the current stage presentation failed to effectively engage the audience's emotions, triggering a subsequent compensation mechanism.

[0030] The system transforms the generated emotional time-series curve into emotional weight coefficients for a reinforcement learning reward function. When a downward trend in the emotional curve is detected, an automatic emotional compensation strategy is triggered: the lighting subsystem enhances the color saturation of the lights, shifting cool tones to more impactful warm tones; the sound subsystem increases low-frequency gain, enhancing the impact and immersion of the sound effects; and the mechanical subsystem accelerates the movement frequency of the stage's rotating platform, strengthening dynamic visual effects. This series of compensation actions works synergistically to create a multi-dimensional stimulation of the audience's audiovisual senses, aiming to raise the emotional curve and bring it back to an emotional climax.

[0031] Before formally implementing emotional compensation, the digital twin simulation environment virtually rehearses the compensation strategy: simulating the impact of enhanced lighting saturation on audience visual comfort to avoid glare from excessive brightness; simulating the impact of increasing low-frequency gain on auditory perception to prevent auditory fatigue from excessive intensity; and simulating the potential safety risks to actors from accelerating mechanical movements. The compensation intensity and timing are optimized through backpropagation algorithms, for example, controlling the color enhancement within a reasonable range, limiting low-frequency boost to a comfortable auditory range, and advancing the starting point of mechanical movement acceleration to match the audience's emotional response delay. The optimized compensation strategy effectively enhances the emotional curve while avoiding emotional fatigue or sensory discomfort caused by overcompensation, achieving precise emotional control.

[0032] Multimodal audience sensor networks enable comprehensive, real-time, and seamless monitoring of audience emotional states. Compared to traditional post-performance questionnaires, the emotional feedback obtained is more authentic, immediate, and representative. The semantic fusion capability of temporal convolutional networks transforms discrete facial expressions, physiological signals, and speech signals into continuous emotional curves, allowing performers to quantitatively grasp the rhythm of audience emotional changes and solving the problem of strong subjectivity in artistic effect evaluation. The human-machine emotional coupling optimization mechanism embeds audience emotion as a core feedback variable into the overall control decision, transforming stage presentation from a one-way output to a two-way interaction, significantly enhancing audience immersion and emotional resonance, and pioneering a new "audience-driven" intelligent performance model. Digital twin pre-performance and compensation optimization strategies avoid the blindness of emotional regulation, ensuring audience comfort while enhancing the performance's appeal and preventing fatigue caused by overstimulation, achieving a refined balance between artistic effect and viewing experience. Overall, this process enables the overall performance control to possess an emotional intelligence closed loop of "perception-understanding-response-optimization," significantly improving the artistic expression of the performance and audience satisfaction.

[0033] In S40: the step of performing multi-physics coupling simulation and multi-objective optimization of the performance process through machine learning algorithms to form an executable overall control strategy and output it to the performance execution system includes the following steps: S41: Construct a digital twin simulation environment for the performance process, integrating lighting intensity attenuation model, sound field propagation model, mechanical dynamics model, video rendering delay model and audience emotional response model, to realize pre-performance calculations coupled with multi-physics field and emotional field; S42: The reinforcement learning algorithm is used to train and optimize the agent. The agent uses the initial overall control scheme as the strategy and the performance realism and audience emotional resonance as the dual objective reward function. The optimal action sequence is explored through Monte Carlo tree search. S43: Inject random disturbances during the simulation process to assess the robustness margin of the overall control strategy under uncertainties such as actor positioning deviations, equipment signal delays, and sudden changes in audience emotions, and generate emotional compensation contingency plans for vulnerable links. S44: Transform the optimized strategy into an executable script. The script contains the control command sequence of each system, emergency trigger conditions, manual takeover interface and emotional compensation node, and outputs an interpretability report of the strategy for the director's review.

[0034] In this embodiment of the application, a closed-loop feedback and dynamic strategy adjustment mechanism for the performance process is also included, specifically comprising the following steps: S51: Construct a three-dimensional spatiotemporal-emotional coupled data acquisition network, deploy a multimodal sensor array in the performance area, and collect in real time the operating parameters of the equipment, the physical field distribution data of the on-site environmental sensors, the arousal index of the audience's physiological signal sensors, the facial expression valence data of the facial recognition camera, the voice emotion intensity of the microphone array, and the ambient activity of the ambient light sensor. After spatiotemporal registration and feature fusion, a high-dimensional coupled dataset is generated. S52: Input the coupled dataset into the online strategy evaluation module to build a dual-channel deviation evaluation engine. The first channel calculates the timing deviation between the actual movement trajectory of the device and the preset overall control scheme. The second channel calculates the morphological deviation between the audience's emotional curve and the target emotional curve. When the deviation of any channel exceeds the dynamic threshold, a strategy fine-tuning request is triggered. S53: Construct an adaptive emotion compensation intensity regulator. This regulator automatically calculates the compensation gain coefficient based on the rate of change and acceleration of the audience's emotion curve. When the rate of decline of emotion exceeds a preset threshold, the compensation intensity is increased, and when the emotion tends to be stable, the compensation is weakened to avoid oversaturation. S54: The model predictive control algorithm is used to perform rolling optimization of the remaining performance process. In the prediction time domain, a multi-step optimization solution is performed with a dual objective function. The first objective is to minimize the equipment action deviation and energy consumption increment, and the second objective is to maximize the goodness of fit of the audience emotion curve, and output the optimal control increment sequence. S55: During strategy execution, a human-machine collaborative intervention interface is built. When the director or operator manually adjusts the equipment parameters, the system encodes the manual correction amount as an additional constraint and injects it into the rolling optimizer to achieve adaptive fusion of human-machine decision-making weights. S56: Establish an intelligent matching mechanism for emergency plans. When a fault signal is detected or the audience's emotional curve reaches an extreme low point, retrieve the compensation plan that best matches the current performance progress, emotional state, and fault type from the plan library. Based on the output of the emotional compensation intensity regulator, dynamically adjust the action range of the plan to achieve seamless switching of the performance process and a smooth transition of the audience experience. S57: Record the effect data of each strategy fine-tuning and emergency plan execution to the strategy replay experience pool, and use the priority experience replay mechanism to train the strategy optimization agent, so that the closed-loop feedback system has the ability to continuously learn from actual combat data and improve the strategy itself.

[0035] During the performance, the system constructs a three-dimensional spatiotemporal-emotional coupled data acquisition network. A multimodal sensor array is deployed in the stage area: operational parameters transmitted by the equipment include the current of the lighting dimmer, the temperature of the mechanical hoist motor, and the status of the sound amplifier; environmental sensors measure the sound pressure level distribution and light intensity uniformity; audience physiological signal sensors collect heart rate variability indices from some volunteers; facial recognition cameras capture the facial expression valence (pleasure, surprise, focus, etc.) of the audience; a microphone array picks up the intensity of applause and cheers; and ambient light sensors monitor the activity level of the theater atmosphere. After all data streams undergo spatiotemporal registration and feature fusion, a high-dimensional coupled dataset is generated, which comprehensively depicts a three-dimensional dynamic picture of equipment status, physical environment, and audience emotional changes.

[0036] The system's online strategy evaluation module constructs a dual-channel deviation evaluation engine. The first channel calculates the temporal deviation between the actual lighting brightness curve and the preset plan in real time, detecting a lag in the rate of brightness increase in a certain section, resulting in insufficient atmosphere creation. The second channel calculates the morphological deviation between the audience's emotional curve and the target curve, detecting a continuous downward trough in the emotional curve, deviating from the emotional climax shape designed in the script. When the deviation in either channel exceeds a dynamic threshold, a strategy fine-tuning request is immediately triggered, avoiding the limitation of a single evaluation dimension failing to fully reflect the performance effect.

[0037] After the adaptive emotional compensation intensity regulator is activated, it analyzes the rate and acceleration characteristics of the audience's emotional curve. When a rapid decline in emotional intensity is detected, the compensation intensity is automatically increased, significantly enhancing the color saturation of the lights and increasing the low-frequency energy output of the speakers. When the emotional curve tends to stabilize, the compensation intensity is proactively reduced to avoid overstimulation that could lead to audience emotional fatigue. This dynamic adjustment mechanism allows emotional compensation to more accurately match the rhythm of the audience's emotional changes, achieving adaptive optimization of the compensation intensity and improving the subtlety of emotional regulation and audience comfort.

[0038] The model predictive control algorithm performs rolling optimization on the remaining performance flow. The optimizer simultaneously solves for two objectives in the prediction time domain: first, minimizing equipment movement deviations and energy consumption increments to avoid frequent and significant equipment adjustments that lead to mechanical wear and energy waste; second, maximizing the goodness of fit between the audience's emotional curve and the target curve to ensure precise achievement of the performance's emotional rhythm. After multiple optimization calculations, the optimal control increment sequence is output, such as "advancing the timing of lighting color switching in subsequent segments" and "appropriately increasing the amplitude of mechanical platform movements," providing forward-looking and optimized incremental instructions for equipment control.

[0039] During a performance, the director noticed a lack of emotional depth in a certain segment and manually increased the color saturation of the lighting. The system captured this manual correction through a human-machine collaborative intervention interface, encoded it as an additional constraint, and injected it into the rolling optimizer. The optimizer then automatically increased the weighting of the lighting adjustment for that segment in subsequent calculations, achieving an adaptive fusion of human and machine decision-making. This fusion mechanism respects the director's artistic intuition while also empowering the AI ​​system to learn from human experience, avoiding artistic biases that might arise from purely algorithmic decision-making.

[0040] When an abnormally high temperature is detected in the mechanical boom motor, the intelligent emergency matching mechanism is immediately activated. The system retrieves the most suitable compensation plan from the plan database that matches the current performance progress, emotional state, and fault type, such as "activating the backup static lighting scheme to replace the mechanical dynamic effects." Simultaneously, based on the output of the emotional compensation intensity regulator, the lighting color and brightness parameters in the plan are dynamically adjusted to ensure the emotional atmosphere is not diminished. The final correction strategy allows the performance to smoothly transition to emergency mode without the audience noticing, ensuring the continuity of the performance and the audience experience.

[0041] After each strategy fine-tuning and emergency plan execution, the system records the effect data of that operation (including changes in audience emotional curves, equipment response, and energy consumption changes) into the strategy replay experience pool. A priority-based experience replay mechanism assigns higher weights to samples that significantly improve emotional impact or successfully handle emergencies, which are then used to train the strategy optimization agent. Through the accumulation of data from multiple performances, the agent gradually learns to regulate emotions more efficiently and select contingency plans more accurately in similar situations. This enables the closed-loop feedback system to continuously learn from real-world data and improve its strategies, resulting in continuous performance enhancement over time.

[0042] The closed-loop feedback and dynamic strategy adjustment mechanism in steps S51-S57 enables the overall performance control system to leap from open-loop execution to closed-loop intelligent evolution. The three-dimensional spatiotemporal-emotional coupled data acquisition network incorporates audience emotions into the overall control feedback loop for the first time, transforming the performance presentation from a one-way output to a two-way interaction, significantly enhancing audience immersion and satisfaction. The dual-channel deviation evaluation engine comprehensively monitors equipment status and emotional effects, ensuring comprehensive control. The adaptive emotional compensation intensity regulator achieves refined and dynamic emotional regulation, avoiding fatigue caused by overstimulation. Rolling optimization of model predictive control provides forward-looking optimal decision-making capabilities, balancing equipment safety and effect achievement. The human-machine collaborative intervention interface preserves the director's artistic leadership, achieving AI assistance rather than replacement. Intelligent matching and correction of emergency plans ensures performance continuity and experience stability in extreme situations. The continuous learning mechanism enables the system to self-evolve, with performance continuously improving with practical experience. Overall, this mechanism endows the overall performance control system with core capabilities of emotional intelligence, dynamic adaptation, and continuous evolution.

[0043] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0044] In one embodiment, a performance intelligent control device based on stage design analysis is provided, which corresponds one-to-one with the performance intelligent control method based on stage design analysis described in the above embodiments. For example... Figure 2 As shown, the intelligent control device for performance based on stage design analysis includes: a stage overall three-dimensional structure model building unit 1, which is used to receive stage structure data and perform analysis and processing, and to build a stage overall three-dimensional structure model containing stage space geometric features, equipment installation points and mechanical motion envelope through a three-dimensional modeling algorithm; Initial master control scheme generation unit 2 is used to intelligently generate an initial master control scheme for the linkage of multiple systems such as lighting, sound, machinery, and video based on the expected performance effect parameters and using a multi-objective collaborative optimization algorithm. The optimized overall control strategy generation unit 3 is used to collect audience physiological signals, facial expressions, voice emotions and on-site atmosphere data in real time through a multimodal sensor array, generate audience emotion time-series curves through emotion semantic parsing, embed the emotion curves as feedback variables into the reinforcement learning reward function, optimize the human-machine emotion coupling of the initial overall control scheme, and form an optimized overall control strategy with enhanced emotional resonance. The overall control strategy output execution unit 4 is used to perform multi-physics coupling simulation and multi-objective optimization of the performance process through machine learning algorithms, form an executable overall control strategy, and output it to the performance execution system.

[0045] Specific limitations regarding the intelligent performance control device based on stage design analysis can be found in the limitations of the intelligent performance control method based on stage design analysis described above, and will not be repeated here. Each module in the aforementioned intelligent performance control device based on stage design analysis can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the electronic device, or stored in the memory of the electronic device as software, so that the processor can call and execute the corresponding operations of each module.

[0046] In one embodiment, an electronic device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 3As shown, this electronic device includes a processor, memory, network interface, and database 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 the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the database. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a performance intelligent control method based on stage design analysis.

[0047] In one embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Receive and analyze stage structure data, and construct an overall three-dimensional structure model of the stage that includes stage space geometric features, equipment installation points, and mechanical motion envelopes through three-dimensional modeling algorithms; Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video. The system collects audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data in real time through a multimodal sensor array. The system generates audience emotion time-series curves through emotion semantic analysis. These emotion curves are then embedded as feedback variables into a reinforcement learning reward function. The initial overall control scheme is optimized through human-machine emotion coupling to form an optimized overall control strategy that enhances emotional resonance. Machine learning algorithms are used to perform multi-physics coupling simulation and multi-objective optimization of the performance process, forming an executable overall control strategy and outputting it to the performance execution system.

[0048] 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: Receive and analyze stage structure data, and construct an overall three-dimensional structure model of the stage that includes stage space geometric features, equipment installation points, and mechanical motion envelopes through three-dimensional modeling algorithms; Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video. The system collects audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data in real time through a multimodal sensor array. The system generates audience emotion time-series curves through emotion semantic analysis. These emotion curves are then embedded as feedback variables into a reinforcement learning reward function. The initial overall control scheme is optimized through human-machine emotion coupling to form an optimized overall control strategy that enhances emotional resonance. Machine learning algorithms are used to perform multi-physics coupling simulation and multi-objective optimization of the performance process, forming an executable overall control strategy and outputting it to the performance execution system.

[0049] 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. When executed, the computer program 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 may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of 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.

[0050] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0051] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A performance intelligent control method based on stage design analysis, characterized in that, The method includes the following steps: Receive and analyze stage structure data, and construct an overall three-dimensional structure model of the stage that includes stage space geometric features, equipment installation points, and mechanical motion envelopes through three-dimensional modeling algorithms; Based on the expected performance effect parameters, a multi-objective collaborative optimization algorithm is used to intelligently generate an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video. The system collects audience physiological signals, facial expressions, voice emotions, and on-site atmosphere data in real time through a multimodal sensor array. The system generates audience emotion time-series curves through emotion semantic analysis. These emotion curves are then embedded as feedback variables into a reinforcement learning reward function. The initial overall control scheme is optimized through human-machine emotion coupling to form an optimized overall control strategy that enhances emotional resonance. Machine learning algorithms are used to perform multi-physics coupling simulation and multi-objective optimization of the performance process, forming an executable overall control strategy and outputting it to the performance execution system.

2. The intelligent overall control method for performances based on stage design analysis according to claim 1, characterized in that, The step of receiving and analyzing stage structure data, and constructing a 3D overall stage structure model that includes stage space geometry, equipment installation points, and mechanical motion envelopes using 3D modeling algorithms, includes the following steps: A multi-source data fusion interface is constructed to receive stage CAD drawings, equipment BIM models, mechanical kinematic parameters, and finite metadata of load-bearing structures. After data cleaning and coordinate system integration, a standardized structural description set is generated. A spatial topology analysis algorithm is used to identify equipment interference areas, line-of-sight obstruction areas, and actor movement conflict areas in the stage structure, and a spatial availability assessment map is generated. Based on the structural description set and availability evaluation map, a three-dimensional stage structure model with a three-dimensional mesh is created using parametric modeling technology. Each device unit in the model is bound to its motion degree of freedom, signal interface and energy consumption attributes. Embedding equipment collision detection constraints and structural safety load constraints in the 3D model, parameter correction feedback is automatically triggered when subsequent schemes violate the constraints.

3. The intelligent overall control method for performances based on stage design analysis according to claim 1, characterized in that, The step of intelligently generating an initial overall control scheme for the linkage of multiple systems, including lighting, sound, machinery, and video, based on the expected performance effect parameters and using a multi-objective collaborative optimization algorithm includes the following steps: Construct a semantic parser for performance effects, which transforms the director's artistic description text into a set of quantitative parameters for lighting intensity, sound pressure level, mechanical speed, and video frame rate, and establishes a mapping relationship between effect parameters and audience perceived intensity. A multi-system collaborative optimization model is constructed using a graph neural network architecture. The model nodes represent lighting, sound, mechanical, and video subsystems, and the edge weights represent the spatiotemporal coupling strength and signal dependence between the systems. The optimal action timing and intensity curves of each system are solved by embedding device motion smoothness constraints, signal transmission synchronization constraints, and total energy power constraints into the loss function of the optimization model, and the gradient descent algorithm is used to solve the optimal action timing and intensity curves of each system. When generating the master control scheme, the system resource occupancy Gantt chart and signal conflict detection report are output simultaneously. When resource contention or signal collision is detected, the time slice rotation or priority arbitration strategy is automatically enabled.

4. The intelligent overall control method for performances based on stage design analysis according to claim 3, characterized in that, The process of acquiring audience physiological signals, facial expressions, vocal emotions, and on-site atmosphere data in real time through a multimodal sensor array, generating audience emotion time-series curves through emotional semantic analysis, embedding these emotion curves as feedback variables into a reinforcement learning reward function, and optimizing the initial overall control scheme through human-machine emotional coupling to form an optimized overall control strategy that enhances emotional resonance includes the following steps: Construct a multimodal audience sensor network, integrating wearable heart rate sensors, infrared thermal imaging facial recognition, microphone array voice pickup, and ambient light intensity sensors to collect data on audience physiological arousal, facial expression valence, voice emotional intensity, and on-site atmosphere activity. Temporal convolutional networks are used to perform emotional semantic fusion on multimodal sensor data to generate audience emotional temporal curves. The peak points of the curves correspond to emotional climaxes, and the trough points correspond to emotional calm intervals. The emotional time-series curve is transformed into emotional weight coefficients of the reinforcement learning reward function. When a downward trend in the emotional curve is detected, the color saturation of the lights is automatically enhanced, the low-frequency gain of the speakers is increased, and the frequency of mechanical movements is accelerated to form an emotional compensation strategy. In a digital twin simulation environment, the effect of emotional compensation strategies on enhancing the audience's emotional curve is simulated. The intensity and timing of compensation are optimized through backpropagation to avoid emotional fatigue caused by overcompensation.

5. The intelligent overall control method for performances based on stage design analysis according to claim 4, characterized in that, The step of using machine learning algorithms to perform multi-physics coupling simulation and multi-objective optimization of the performance process, forming an executable overall control strategy, and outputting it to the performance execution system includes the following steps: Construct a digital twin simulation environment for the performance process, integrating lighting intensity attenuation models, sound field propagation models, mechanical dynamics models, video rendering delay models, and audience emotional response models to achieve pre-performance calculations that couple multiple physical fields with emotional fields; The agent is trained using a reinforcement learning algorithm. The agent uses an initial overall control scheme as its strategy and the performance realism and audience emotional resonance as its dual-objective reward function. The agent explores the optimal action sequence through Monte Carlo tree search. Random disturbances are injected during the simulation process to assess the robustness margin of the overall control strategy under uncertainties such as actor positioning deviations, equipment signal delays, and sudden changes in audience emotions, and to generate emotional compensation contingency plans for vulnerable links. The optimized strategy is transformed into an executable script, which includes the control command sequence of each system, emergency trigger conditions, manual takeover interface and emotional compensation node, and outputs an interpretability report of the strategy for the director's review.

6. The intelligent overall control method for performances based on stage design analysis according to claim 5, characterized in that, It also includes a closed-loop feedback and dynamic strategy adjustment mechanism for the performance process, including the following steps: A three-dimensional spatiotemporal-emotional coupled data acquisition network is constructed. A multimodal sensor array is deployed in the performance area to collect in real time the operating parameters of the equipment, the physical field distribution data of the on-site environmental sensors, the arousal index of the audience's physiological signal sensors, the facial expression valence data of the facial recognition camera, the voice emotion intensity of the microphone array, and the ambient activity of the ambient light sensor. After spatiotemporal registration and feature fusion, a high-dimensional coupled dataset is generated. The coupled dataset is input into the online strategy evaluation module to construct a dual-channel deviation evaluation engine. The first channel calculates the timing deviation between the actual movement trajectory of the device and the preset overall control scheme, and the second channel calculates the morphological deviation between the audience's emotional curve and the target emotional curve. When the deviation of any channel exceeds the dynamic threshold, a strategy fine-tuning request is triggered. An adaptive emotion compensation intensity regulator is constructed. Based on the rate of change and acceleration of the audience's emotion curve, the regulator automatically calculates the compensation gain coefficient. When the rate of decline of emotion exceeds a preset threshold, the compensation intensity is increased, and when the emotion tends to stabilize, the compensation is weakened to avoid oversaturation. The model predictive control algorithm is used to perform rolling optimization of the remaining performance process. In the prediction time domain, a multi-step optimization solution is performed with a dual objective function. The first objective is to minimize the equipment action deviation and energy consumption increment, and the second objective is to maximize the goodness of fit of the audience emotion curve, and output the optimal control increment sequence. During strategy execution, a human-machine collaborative intervention interface is built. When the director or operator manually adjusts the equipment parameters, the system encodes the manual correction as an additional constraint and injects it into the rolling optimizer to achieve adaptive fusion of human-machine decision-making weights. Establish an intelligent matching mechanism for emergency plans. When a fault signal is detected or the audience's emotional curve reaches an extreme low point, the mechanism retrieves the compensation plan that best matches the current performance progress, emotional state, and fault type from the plan database. Based on the output of the emotional compensation intensity regulator, the mechanism dynamically adjusts the action range of the plan to achieve seamless switching of the performance process and a smooth transition of the audience experience. The effect data of each strategy fine-tuning and emergency plan execution is recorded in the strategy replay experience pool. The priority experience replay mechanism is used to train the strategy optimization agent, so that the closed-loop feedback system has the ability to continuously learn from real combat data and improve the strategy itself.

7. A performance intelligent control device based on stage design analysis, applied to the performance intelligent control method based on stage design analysis as described in any one of claims 1 to 6, characterized in that, The device includes: The stage overall three-dimensional structure model building unit (1) is used to receive stage structure data and perform analysis and processing. It constructs an overall three-dimensional structure model of the stage, including stage space geometric features, equipment installation points and mechanical motion envelope, through three-dimensional modeling algorithms. The initial master control scheme generation unit (2) is used to intelligently generate an initial master control scheme for the linkage of multiple systems such as lighting, sound, machinery, and video based on the expected performance effect parameters and using a multi-objective collaborative optimization algorithm. The optimized overall control strategy generation unit (3) is used to collect audience physiological signals, facial expressions, voice emotions and on-site atmosphere data in real time through a multimodal sensor array, generate audience emotional time-series curves through emotional semantic analysis, embed the emotional curves as feedback variables into the reinforcement learning reward function, optimize the human-machine emotional coupling of the initial overall control scheme, and form an optimized overall control strategy with enhanced emotional resonance. The overall control strategy output execution unit (4) is used to perform multi-physics coupling simulation and multi-objective optimization of the performance process through machine learning algorithms, form an executable overall control strategy and output it to the performance execution system.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent overall control method for performance based on stage design analysis as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent overall control method for performance based on stage design analysis as described in any one of claims 1 to 6.

Images