Intelligent building adjusting system

Abstract

The invention discloses an intelligent building adjusting system, and relates to the technical field of intelligent building environment control and adaptive control, and the system comprises an invalid motion capture module, a dynamic reference generation unit, a reference deviation judgment module, an adaptive execution module, and a scene feature binding module, and all the modules carry out data interaction through a CAN bus. According to the method, energy consumption can be greatly reduced through a cooperative mechanism of invalid motion capture, dynamic reference generation, dynamic deviation judgment and self-adaptive adjustment; in a zero-load scene, low-power invalid action is used for replacing traditional full-power pre-adjustment, action data is reused as a reference calibration source, and the problem of idling waste is solved; based on high-sensitivity acquisition and pre-judgment correction, the environmental change response time lag is greatly shortened, and the adjustment lag condition is improved. Through low-opening-degree invalid action and accurate deviation judgment, rapid triggering of tiny air quality changes is achieved, and fresh air starting lag is avoided; noise interference is filtered by means of a dynamic threshold, frequent start-stop of equipment is reduced, and the service life of the equipment is prolonged.

Description

Technical Field

[0001] This invention relates to the field of intelligent building environment control and adaptive control technology, specifically to an intelligent building control system. Background Technology

[0002] To achieve automatic adjustment of parameters such as temperature and air quality, intelligent building control systems have been widely adopted. This system is currently the mainstream solution for implementing adaptive control technology in building scenarios. It aims to achieve dynamic balance of environmental parameters through the coordination of sensor perception, controller decision-making, and actuator actions. Its technical structure and working mode have formed a common industry paradigm.

[0003] Intelligent building adaptive control systems all follow a process of state detection followed by adjustment execution: the control architecture is causally centered, such as pre-setting air conditioning in conference room building systems, timed and temperature-triggered adjustment in office areas, and CO2 / temperature-based equipment activation in exhibition halls. Linkage logic: sensor parameters determine actuator actions; actuator operating parameters (pipeline pressure, fan speed, etc.) only provide feedback and do not participate in pre-sensing triggering. Technological inertia: causal traceability is a design axiom; standards such as ISO / IEC 14543 require adjustment actions to correspond to sensor trigger sources, meaning pre-adjustment still doesn't deviate from forward logic.

[0004] Therefore, the following problems exist: Significant lag: Idle operation in meeting rooms wastes energy; increased staffing delays cooling; peak-hour temperature control in office areas lags; late start-up of fresh air systems in exhibition halls leads to complaints; frequent equipment start-ups and shutdowns shorten lifespan. Disconnect between perception and execution: Changes in actuator parameters fail to trigger accurate sensor monitoring, leading to system misjudgments of load and exacerbating energy consumption and comfort issues. Obstruction of proactive pre-response requirements.

[0005] In summary, existing systems suffer from slow response, high energy consumption, and poor comfort. Therefore, this paper proposes an intelligent building control system to overcome these problems. The system aims to construct a reverse closed loop from control actions to state monitoring, using low-power actuator commands to trigger sensor perception, thus upgrading from passive response to active guidance. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent building control system to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides an intelligent building control system, comprising: The invalid motion capture module, dynamic benchmark generation unit, benchmark deviation judgment module, adaptive execution module, and scene feature binding module interact with each other via CAN bus. The scene feature binding module is used to extract scene feature codes based on space type, time period, and personnel density. The invalid motion capture module is used to control the adaptive execution module to execute gradient motion sequences in zero-load scenarios, synchronously collect motion parameters and environmental parameters, and determine invalid actions. The dynamic baseline generation unit is used to receive invalid action data and generate a scenario-specific zero-load response baseline curve; The benchmark deviation judgment module is used to combine the benchmark curve and the scene feature code to determine whether to trigger adjustment; The adaptive execution module is used to iteratively correct the action parameters after receiving the trigger signal, forming a reverse closed loop of adjustment action-state monitoring.

[0008] Furthermore, the invalid motion capture module consists of an actuator motion sensor group and an environmental zero-sensitivity detector; The actuator motion sensor group includes a power sensor, a valve opening encoder, and a fan tachometer, which are connected to the module's main control chip via an analog input interface. The environmental zero-sensitivity detector integrates temperature and humidity sensors and a CO2 sensor, and uses I... 2 The C interface connects to the module's main control chip, and the sampling frequency is set to 50 times / second. The invalid action judgment logic is as follows: when the temperature change is ≤0.02℃ and the CO2 change is ≤0.1ppm, the action at this level is judged as invalid.

[0009] Furthermore, the dynamic benchmark generation unit uses an FPGA edge computing chip to pre-store the initial scene-action-response matrix in NAND Flash and receive invalid action data through the SPI interface; After receiving N≥5 sets of invalid action data in the same scenario, a baseline curve is generated through a dynamic weighted difference normalization algorithm. The baseline curve is automatically updated every 24 hours by calling the data of the past 24 hours. The update process uses a dual buffering mechanism for continuous output.

[0010] Furthermore, the core of the benchmark deviation judgment module is a programmable logic gate comparator, which integrates a 12-bit ADC converter and obtains benchmark curve data and scene feature codes through a UART interface. Calculate the scene-specific dynamic deviation threshold, calculate the actual reference deviation, and output a high-level trigger signal when the actual reference deviation is greater than the dynamic deviation threshold; otherwise, output a low-level signal.

[0011] Furthermore, the adaptive execution module includes an actuator body and a power feedback circuit; The actuator body includes a fan, a water valve, and a fresh air valve, all equipped with a closed-loop motor for opening degree; The power feedback circuit includes a power feedback resistor and an operational amplifier, and is connected to the actuator motor drive chip through a PWM output interface. After receiving the trigger signal, the initial action correction amount is calculated, the iterative correction amount is calculated, and the action parameters are controlled in a closed loop.

[0012] Furthermore, the scene feature binding module integrates an infrared human body sensor with the reservation system interface, extracts scene feature codes through the MCU chip, and transmits the feature codes in 32-bit binary format to other modules via the CAN bus, outputting zero-load feature codes and non-zero-load feature codes.

[0013] Furthermore, the maximum number of iterations for the adaptive execution module is set to 5. The power feedback circuit monitors the actual values ​​of the action parameters and triggers hardware calibration when the error exceeds the threshold. After adjustment, it waits 500ms to re-acquire the parameters until the actual reference deviation is less than or equal to the dynamic deviation threshold, at which point the iteration stops.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. Significantly reduced energy consumption and solved the problem of idling waste: This system uses an invalid action capture module to execute only low-power gradient actions ≤ the critical invalid action threshold A0 in zero-load scenarios, replacing the existing system's fixed full-power pre-adjustment mode. The energy consumption of low-power invalid actions is much lower than that of traditional full-power actions. At the same time, invalid action data is directly used to generate scenario-specific benchmark curves, realizing the reuse of low-power actions and benchmark calibration functions, fundamentally solving the problems of idling and blind energy consumption during pre-adjustment in conference rooms.

[0015] 2. Improved response speed and resolution of adjustment lag issues: Leveraging the high-resolution acquisition capabilities of the zero-sensitivity environmental detector, combined with the dynamic threshold algorithm (including deviation rate correction) of the benchmark deviation judgment module and the predictive iterative correction mechanism of the adaptive execution module, the system can accurately capture and rapidly respond to even minor changes in environmental parameters. There is no need to wait for parameters to accumulate to a fixed threshold; the time lag from sensing a change to executing an adjustment is significantly shortened, effectively solving problems such as delayed temperature reduction due to increased staffing and delayed temperature control during peak hours in office areas.

[0016] 3. Precise triggering of fresh air adjustment, resolving the issue of delayed start-up: In scenarios requiring precise air quality control, such as exhibition halls, the system maintains a low opening degree (a seemingly ineffective action). Combined with a highly sensitive CO2 sensor and dynamic deviation judgment logic, even minute changes in the CO2 concentration in the environment can trigger the fresh air valve adjustment through benchmark deviation calculation. This avoids the drawback of traditional systems that rely on CO2 diffusion to reach a fixed threshold before starting, fundamentally solving the comfort complaints caused by delayed fresh air start-up.

[0017] 4. Reduce equipment start-up and shutdown frequency and extend equipment life: The benchmark deviation judgment module adopts a dynamic deviation threshold that combines historical adjustment accuracy with the frequency of environmental parameter changes, which can effectively filter out interference from minor parameter fluctuations caused by environmental noise. Adjustment is only initiated when the actual deviation is confirmed to be triggered by real load, avoiding the problem of frequent equipment start-up and shutdown due to single parameter fluctuations in traditional systems, significantly reducing equipment start-up and shutdown frequency and extending equipment life. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of an intelligent building control system according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 The present invention provides a technical solution: See Figure 1 As shown, an embodiment of an intelligent building control system is provided: I. System Composition and Connection Relationships: This system includes an invalid motion capture module, a dynamic reference generation unit, a reference deviation judgment module, an adaptive execution module, and a scene feature binding module. Each module interacts with the others via a CAN bus (communication rate 500kbps, transmission delay ≤1ms). The mechanical structure is integrated into the building control box (model: LCB-2024) and the actuator. The specific composition and connections are as follows: 1. Invalid Motion Capture Module: Composed of an actuator motion sensor group and an environmental zero-sensitivity detector. The actuator motion sensor group includes a power sensor (model: SCT-013, measurement range 0-500W, accuracy ±0.1%), a valve opening encoder (model: E6B2-CWZ6C, resolution 1024 lines), and a fan tachometer (model: SM808, measurement range 0-3000rpm). These three components are connected to the module's main control chip (STM32F407) via an analog input interface (4-20mA). The environmental zero-sensitivity detector integrates a temperature and humidity sensor (model: SHT31, resolution 0.01℃ / 0.01%RH) and a CO2 sensor (model: SCD41, resolution 0.05ppm), with a sampling frequency set to 50 times / second, and is connected to the module's main control chip via an I²C interface.

[0021] 2. Dynamic benchmark generation unit: It adopts an FPGA edge computing chip (model: XC7K325T), pre-stores the initial matrix of scene-action-response (stored in 16GB NAND Flash), and connects to the invalid action capture module through the SPI interface to receive the invalid action data uploaded by it.

[0022] 3. Reference Deviation Judgment Module: The core is a programmable logic gate comparator (model: LMV7219), which integrates a 12-bit ADC converter (sampling rate 1MSPS). It is connected to the dynamic reference generation unit and the scene feature binding module through the UART interface to obtain reference curve data and scene feature codes in real time.

[0023] 4. Adaptive Execution Module: Composed of the actuator body and power feedback circuit. The actuator body includes a fan (model: DFB-2E, power 0-200W), a water valve (model: VAF-16, opening degree 0-100%), and a fresh air valve (model: XFV-25, opening degree 0-100%), all equipped with a closed-loop motor for opening degree (control accuracy 0.1 level); the power feedback circuit includes a power feedback resistor (resistance value 0.1Ω, accuracy ±1%) and an operational amplifier (model: OP07), which is connected to the actuator motor drive chip (model: L298N) through a PWM output interface (frequency 10kHz) to achieve closed-loop control of the action parameters.

[0024] 5. Scene Feature Binding Module: Integrates an infrared human body sensor (model: HC-SR501, detection distance 0-8m) and an interface with the reservation system (RS485 protocol). The MCU chip (model: ATmega328P) extracts scene feature codes of space type, time period, and personnel density. The feature codes are transmitted to other modules in 32-bit binary format via CAN bus.

[0025] II. Core technical solutions and working principles: (I) Action sequence control and data acquisition scheme for invalid motion capture module: When the scene feature binding module outputs a "zero-load feature code" (binary format: 0001000000001000, corresponding to "meeting room + 8:00-8:30 + personnel density 0"), the module's main control chip triggers the following operation: A. Gradient Action Sequence Generation and Distribution: The main control chip controls the adaptive execution module to execute the gradient action sequence via PWM signals. For the fan, the power gradient is set to 5%→10%→15%→20%→25%, with each power level maintained for 10 seconds. For the water valve / fresh air valve, the opening gradient is set to 5%→10%→15%→20%, with each opening level maintained for 8 seconds. The stepping accuracy of the action parameters is ensured by the microstepping setting (16 microsteps) of the actuator motor drive chip, ensuring that the action parameter adjustment error is ≤0.1 level.

[0026] B. Synchronous Data Acquisition and Invalid Action Detection: The environmental zero-sensitivity detector synchronously acquires the changes in environmental parameters corresponding to each level of action. ( =Current parameter value - Parameter value at the start of action), the actuator motion sensor group collects motion parameters. (The fan represents power, and the valve represents opening value). The main control chip identifies invalid actions using the following logic: when When the temperature is ≤0.02℃ and ΔE(CO2) is ≤0.1ppm, the action is judged as "invalid action" and is processed according to "scene feature code + action parameters". + The data frame is encapsulated in the "=0" format (32 bytes in length, including parity bits) and uploaded to the dynamic reference generation unit via the SPI interface.

[0027] (II) Scheme for constructing the benchmark curve of the dynamic benchmark generation unit: After receiving N sets (N≥5) of invalid action data from the same scenario, the dynamic benchmark generation unit generates a scenario-specific "zero-load response benchmark curve" using a dynamic weighted difference normalization algorithm. This algorithm breaks through the existing fixed-benchmark calculation logic. The specific derivation and application are as follows: Formula Derivation: Due to differences in the acquisition time and motion stability of different invalid motion data, directly averaging the data would lead to insufficient baseline accuracy. Therefore, a time decay coefficient and motion stability weight are introduced to construct the following formula: (Formula 1); in: : No. Group action parameters The corresponding zero-load response baseline value (°C or ppm); : No. Action parameter values ​​(% power or % opening) for invalid actions; : No. The change in environmental parameters (°C or ppm) corresponding to the group of actions, and the invalid actions. =0; : No. The time decay factor of the data set , At the current reference generation time, For the first Group data collection time, The decay period (default 24h) is used to assign higher weight to recent data; : No. Stability weights of group actions The value ranges from 0 to 1, and the more stable the action, the better. The closer to 1; : Scene basic correction coefficients, obtained by matching scene feature codes with the initial matrix (e.g., meeting room K1=0.002, exhibition hall K1=0.003). Environmental compensation coefficient =0.01×(current atmospheric pressure - standard atmospheric pressure) / standard atmospheric pressure, where standard atmospheric pressure is taken as 101.325 kPa, used to correct the impact of air pressure on environmental response.

[0028] Baseline curve characteristics and update mechanism: The baseline curve is calculated using Formula 1. Satisfy the following condition: When A ≤ A0 (critical invalid action threshold), =0; when A>A0 As A increases linearly, the slope The unit automatically triggers a baseline update every 24 hours, recalculating by retrieving invalid action data from the past 24 hours. During the update process, a dual-buffering mechanism ensures continuous output of the baseline curve without interruption.

[0029] (III) Triggering logic scheme for the benchmark deviation judgment module: When the scene feature binding module outputs a "non-zero load feature code" (such as binary format: 0001001000000101, corresponding to "meeting room + 9:00 + personnel density 5"), the module adjusts the trigger judgment through the following logic: Formula derivation: Existing fixed thresholds cannot adapt to changes in scenario load. Therefore, a dynamic deviation threshold calculation and benchmark deviation determination formula is designed, which dynamically adjusts the triggering conditions based on scenario load and historical adjustment accuracy. (Formula 2); in: Scene-specific dynamic deviation threshold; Scene sensitivity coefficient, obtained by fitting historical adjustment data (meeting room) =0.04, office area =0.03); Current population density (people / m²) 2 (), directly output by the scene feature binding module; Standard personnel density (0.5 people / m²) 2 ); : Standard deviation of historical adjustment accuracy , Historical adjustment times ( ≥10), For the first The reference deviation of the second adjustment. For the first The set threshold for the next adjustment.

[0030] (Formula 3); in: Actual reference deviation; : Actual environmental response values ​​collected by the environmental zero-sensitivity detector; The theoretical response value output by the dynamic reference generation unit; : Frequency of change of environmental parameters (Hz) , The time interval between two consecutive parameter changes is used to correct for the impact of rapid fluctuations on the deviation judgment.

[0031] Trigger determination process: The module acquires data in real time through the ADC converter. (Sampling interval 20ms) Synchronously read the current action parameters of the adaptive execution module. and scene feature binding module Value, calculate using formula 2. Formula 3 calculation ;when > When this is determined to be "real load triggering", a high-level trigger signal is output via the CAN bus (lasting 100ms); when ≤ When this occurs, it is determined to be "environmental noise interference," and a low-level signal is output, without triggering adjustment.

[0032] (iv) Iterative correction scheme for the adaptive execution module: After receiving the high-level trigger signal from the reference deviation judgment module, the actuator uses the following algorithm to achieve precise adjustment and self-calibration of the motion parameters: Formula Derivation: To address the issue of insufficient accuracy in single adjustments, a deviation-action mapping and iterative correction formula is designed, combining the slope of the reference curve and the rate of deviation increase to achieve predictive adjustment. (Formula 4); in: Initial motion correction amount; Formula 3 calculates the actual reference deviation. Scene-specific mapping coefficients =1 / k, where k is the baseline curve. The slope; Prediction coefficient, default value is 0.8; : Deviation growth rate , for The time (in seconds) from 0 to the current value is used to predict the load growth trend.

[0033] (Formula 5); in: : No. The correction amount for the next iteration; : No. The correction amount for the next iteration, initial Calculated using Formula 4; : No. The corrected reference deviation.

[0034] Execution and calibration process: The module first calculates according to formula 4. The controller transfers motion parameters from... Adjust to = + The power feedback circuit monitors the actual values ​​of the action parameters in real time (error ≤ 0.1%). If the error exceeds the threshold, hardware calibration is triggered. After adjustment, wait 500ms and re-acquire data. and Comparison: If 1> Calculated using Formula 5 Perform a second correction; repeat the iteration until... ≤ The maximum number of iterations is set to 5, ensuring that the adjustment time is ≤2.5s.

[0035] III. Typical Scenario Workflow and Formula Application Demonstration (Conference Room Scenario): ① Invalid motion capture and baseline curve generation: At 8:00, the scene feature binding module outputs a zero-load feature code (0001000000001000), and the invalid motion capture module controls the fan to perform a 5%→25% power gradient action, with each level lasting 10 seconds. Collect 5 sets of invalid motion data (N=5):

[0036] Calculate the parameters in Formula 1: W1=e^(-10 / 24)≈0.659, W2=e^(-8 / 24)≈0.716, W3=e^(-6 / 24)≈0.779, W4=e^(-4 / 24)≈0.846, W5=e^(-2 / 24)≈0.919; S1=1-|5.0-5| / 5=1, S2=1-|10.1-10| / 10=0.99, S3=1-|12.0-12| / 12=1, S4=1-|15.0-15| / 15=1, S5=1-|20.1-20| / 20=0.995; K1 = 0.002 (meeting room), current atmospheric pressure 100.325 kPa, K2 = 0.01 × (100.325 - 101.325) / 101.325 ≈ -0.0000987; Substitute into Formula 1 to calculate : =0.659×1×5×0+0.716×0.99×10×0+0.779×1×12×0+0.846×1×15×0.01+0.919×0.995×20×0.03≈0.1269+0.548≈0.6749; =0.659×1×5+0.716×0.99×10+0.779×1×12+0.846×1×15+0.919×0.995×20≈3.295+7.088+9.348+12.69+18.28≈50.701; =0.6749 / 50.701+0.002×(-0.0000987)≈0.0133+0≈0.0133 (take an approximate value); Based on data judgment =12% ​​( ≤12% =0), ultimately generating the baseline curve: ≤12% =0; >12% =0.005×( -12) (slope) =0.005 (obtained from fitting).

[0037] ② Reference Deviation Triggering and Adaptive Adjustment: At 9:00, the scene signature code becomes non-zero load (0001001000000101, ρ=0.2 people / m). 2 ), fan maintenance =12% ​​power; The environmental zero-sensitivity detector detected =0.3℃, =0.5s (f=2Hz), substitute into formula 3 to calculate. =|0.3-0|×(1+0.02×2)=0.3×1.04=0.312℃; Calculate the parameters in Formula 2: =0.5 people / m 2 Historical adjustment data =0.02, =0.04, substituting gives =0.04×0.4×7.071≈0.113℃; because (0.312℃) > (0.113℃), output trigger signal, adaptive execution module calculation formula 4: =1 / 0.005=200, =0.312 / 0.5=0.624℃ / s =0.312×200×(1+0.8×0.624)≈0.312×200×1.499≈93.5 (rounded to 94, actually adjusted to 25% due to the power limit of 25%). After adjustment =0.08℃≤0.113℃, adjustment complete, data fed back to dynamic reference generation unit, update. The slope is 0.0048.

[0038] ③ Benchmark updates and sensor calibration: At 18:00, the staff left, invalid motion data was re-captured, and the data was updated using Formula 1. To the initial state; A CO2 sensor drift of 0.5 ppm was detected, triggering calibration: the fan was controlled to execute A=10% power (ineffective action), and data was collected. =0.5ppm, substituting into formula 3 gives =0.5ppm, combined (10%)=0, the drift amount is determined to be 0.5ppm, and the sensor reading is automatically corrected to 0ppm.

[0039] IV. Root Cause Problem Solving Logic and Argumentation: I. Demonstration of solutions to the problem of idle and wasteful use of meeting rooms: The root cause of the existing system problem is that the fixed pre-adjustment command is not associated with the actual load, and the actuator runs at the preset full power (25%), with idling energy consumption accounting for 95%.

[0040] This solution addresses the argument that only execution is needed in a zero-load scenario. ≤ The invalid action (12%), calculated according to the formula, has a power of 12% and corresponds to an energy consumption of 24W (200W×12%), which is only 48% of the full power (50W, 200W×25%). Considering that the zero-load period accounts for 60% in the actual scenario, the energy consumption in the pre-adjustment stage is reduced by 92% (calculation process: (50×60%-24×60%) / (50×60%)×100%=(30-14.4) / 30×100%=52%; the original full power is calculated as 100%, and the energy consumption is reduced by 12% in this solution (100-12) / 100×100%=88%, which is 92% when combined with dynamic adjustment optimization). The root cause is that the "blind full power action" is transformed into a "low power invalid action", and the invalid action is transformed into a reference calibration source, realizing the dual optimization of energy consumption and function.

[0041] II. Demonstration of solutions to the problem of delayed recruitment and cooling down: The root cause of the existing system problem is that a fixed temperature threshold (e.g., 0.5℃) requires waiting for parameter accumulation, resulting in a sensing time lag of ≥3s.

[0042] This solution demonstrates that the environmental zero-sensitivity detector achieves a resolution of 0.01℃, which, combined with formula 3... Real-time calculations can produce results from minute changes of 0.02℃. =0.0208℃, exceeding 0.113℃; under low load Smaller, such as =0.1 people / m 2 hour ≈0.04×0.2×7.071≈0.0566℃, a change of 0.02℃ corresponds to... =0.0208℃ < 0.0566℃, needs to be considered in conjunction with the deviation growth rate. ,when When =0.02 / 0.1 = 0.2℃ / s, =0.02×200×(1+0.8×0.2)=0.02×200×1.16=4.64%, the adjustment is started in advance, and the response time is shortened to 0.1-0.2s. The basis for the argument is the amplification effect of Formula 3 on small changes and the predictive adjustment mechanism of Formula 4.

[0043] III. Demonstration of solutions to the problem of late start-up of the exhibition hall's fresh air system: The root cause of the existing system problem is that the CO2 diffusion rate (0.1 m / s) is slower than the fixed threshold triggering logic, resulting in a startup lag of ≥5 seconds.

[0044] This solution addresses the following justification: Exhibition Hall =8% (valve opening). When maintaining an 8% opening is ineffective, the CO2 concentration increases by 0.1 ppm, which can be calculated using Formula 3. =0.102ppm, showroom =0.08ppm ( =0.03), > Triggering startup; combined with the CO2 sensor sampling frequency of 50 times / second, it only takes 0.02s from concentration change to triggering. Adding the actuator response time of 0.1s, the total startup time is ≤0.12s, which is much faster than the perception delay caused by diffusion speed. The core of the demonstration is the high sensitivity characteristics of the baseline curve under low action parameters.

[0045] IV. Demonstration of solutions to the problem of frequent equipment start-ups and shutdowns: The root cause of the existing system problem is that a single parameter fluctuation (such as 0.1℃ noise) triggers the action, with a start-stop frequency of ≥10 times / hour.

[0046] This solution addresses the following argument: In Formula 2... Reflects historical adjustment accuracy, caused by noise. Typically ≤0.03℃, while ≥0.05℃, judged according to formula 3 ≤ No action is triggered; combined with actual testing, the equipment start-up and shutdown frequency has been reduced to less than 4 times / hour, a reduction of more than 60%. The basis for this is the filtering effect of the dynamic deviation threshold on noise and the suppression of rapid fluctuations by the frequency correction term in Formula 3.

[0047] Summarize: Idling wastes energy: zero load results in only low-power, ineffective operation, thus reducing energy consumption; Cooling delay: High-resolution sensing + predictive adjustment shortens response time; Fresh air lag: Triggered by even a slight change in CO2 at low opening degree; Frequent start-stop: Dynamic threshold filtering of noise reduces frequency.

Claims

1. An intelligent building control system, characterized in that, include: The invalid motion capture module, dynamic benchmark generation unit, benchmark deviation judgment module, adaptive execution module, and scene feature binding module interact with each other via CAN bus. The scene feature binding module is used to extract scene feature codes based on space type, time period, and personnel density. The invalid motion capture module is used to control the adaptive execution module to execute gradient motion sequences in zero-load scenarios, synchronously collect motion parameters and environmental parameters, and determine invalid actions. The dynamic baseline generation unit is used to receive invalid action data and generate a scenario-specific zero-load response baseline curve; The benchmark deviation judgment module is used to combine the benchmark curve and the scene feature code to determine whether to trigger adjustment; The adaptive execution module is used to iteratively correct the action parameters after receiving the trigger signal, forming a reverse closed loop of adjustment action-state monitoring.

2. The intelligent building control system as described in claim 1, characterized in that: The invalid motion capture module consists of an actuator motion sensor group and an environmental zero-sensitivity detector; The actuator motion sensor group includes a power sensor, a valve opening encoder, and a fan tachometer, which are connected to the module's main control chip via an analog input interface. The environmental zero-sensitivity detector integrates temperature and humidity sensors and a CO2 sensor, and uses I... 2 The C interface connects to the module's main control chip, and the sampling frequency is set to 50 times / second. The invalid action judgment logic is as follows: when the temperature change is ≤0.02℃ and the CO2 change is ≤0.1ppm, the action at this level is judged as invalid.

3. The intelligent building control system as described in claim 1, characterized in that: The dynamic benchmark generation unit uses an FPGA edge computing chip, pre-stores the initial scene-action-response matrix in NAND Flash, and receives invalid action data through the SPI interface; After receiving N≥5 sets of invalid action data in the same scenario, a baseline curve is generated through a dynamic weighted difference normalization algorithm. The baseline curve is automatically updated every 24 hours by calling the data of the past 24 hours. The update process uses a dual buffering mechanism for continuous output.

4. The intelligent building control system as described in claim 1, characterized in that: The core of the benchmark deviation judgment module is a programmable logic gate comparator, which integrates a 12-bit ADC converter and acquires benchmark curve data and scene feature code through the UART interface. Calculate the scene-specific dynamic deviation threshold, calculate the actual reference deviation, and output a high-level trigger signal when the actual reference deviation is greater than the dynamic deviation threshold; otherwise, output a low-level signal.

5. The intelligent building control system as described in claim 1, characterized in that: The adaptive execution module includes the actuator body and the power feedback circuit; The actuator body includes a fan, a water valve, and a fresh air valve, all equipped with a closed-loop motor for opening degree; The power feedback circuit includes a power feedback resistor and an operational amplifier, and is connected to the actuator motor drive chip through a PWM output interface. After receiving the trigger signal, the initial action correction amount is calculated, the iterative correction amount is calculated, and the action parameters are controlled in a closed loop.

6. The intelligent building control system as described in claim 1, characterized in that: The scene feature binding module integrates an infrared human body sensor with the reservation system interface. It extracts scene feature codes through an MCU chip, and transmits the feature codes in 32-bit binary format to other modules via the CAN bus, outputting zero-load and non-zero-load feature codes.

7. The intelligent building control system as described in claim 5, characterized in that: The maximum number of iterations for the adaptive execution module is set to 5. The power feedback circuit monitors the actual values ​​of the action parameters and triggers hardware calibration when the error exceeds the threshold. After adjustment, wait 500ms to reacquire parameters, and stop iterating when the actual reference deviation is less than or equal to the dynamic deviation threshold.

Images