An edge-computing-based building control method, system and storage medium

Abstract

The application discloses a kind of based on edge computing's building control method, system and storage medium, the control method includes: system initialization step, edge computing node loads configuration and strategy, establishes equipment topological relationship;Data acquisition and pre-processing step, sensor data are collected and are filtered, denoised and standardized processing;Edge intelligent decision-making step, control logic is executed based on preset rule and real-time data, generates and issues control instruction;Equipment cooperative control step, according to linkage strategy trigger relevant equipment cooperative action;State reporting and synchronization step, control result is reported to cloud and is synchronized with other edge nodes.The application deeply fuses edge computing technology and building equipment distributed autonomous control, response delay is reduced to millisecond level from traditional several seconds, significantly improve the reliability and user experience of intelligent building.

Description

Technical Field

[0001] This invention belongs to the field of smart building and Internet of Things technology, and in particular relates to a building control method, system and storage medium based on edge computing. Background Technology

[0002] With the continuous advancement of smart city construction, smart buildings have become an important part of modern urban development. However, the applicant has discovered that the current traditional building equipment control system mainly adopts a centralized architecture, which has the following technical defects: (1) Response delay problem: All control commands need to be uploaded to the cloud server for processing before being sent down for execution. For scenarios that require real-time response (such as fire alarm, security linkage, etc.), there is a significant time delay, which makes it difficult to meet the millisecond-level response requirements. (2) Strong network dependence: Traditional control systems are highly dependent on network connections. Once the network is interrupted or fluctuates, the entire control system will be paralyzed, which will seriously affect the normal operation of the building. (3) Low data processing efficiency: All massive sensor data is uploaded to the cloud for centralized processing, which consumes huge bandwidth and puts enormous computing pressure on the cloud server, resulting in low overall system efficiency. (4) Lack of local decision-making capability: Existing systems generally lack local intelligent decision-making capability, and cannot complete complex logical judgments and autonomous control at the edge, relying entirely on cloud instructions; (5) Difficulty in equipment linkage: It is difficult to achieve efficient linkage between equipment from different manufacturers and of different types. Each system is independent of the others, forming information silos; Meanwhile, some current so-called "edge gateway" products only implement simple protocol conversion and data collection functions, and are far from reaching the level of "autonomous control" intelligence. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a building control method, system and storage medium based on edge computing, which deeply integrates edge computing technology with distributed autonomous control of building equipment, reducing response latency from the traditional several seconds to the millisecond level, and significantly improving the reliability of smart buildings and user experience.

[0004] The objective of this invention is achieved through the following technical solution: A building control method based on edge computing includes: System initialization steps: loading configurations and policies for edge computing nodes, and establishing device topology relationships; The data acquisition and preprocessing steps involve acquiring sensor data and performing filtering, noise reduction, and standardization. Edge intelligent decision-making steps execute control logic based on preset rules and real-time data, generating and issuing control commands; The equipment collaborative control steps involve triggering collaborative actions of relevant equipment based on the linkage strategy. The status reporting and synchronization steps report the control results to the cloud and synchronize them with other edge nodes.

[0005] In one implementation, the preset rules in the edge intelligent decision-making step include edge node scheduling rules set in the local rule engine, and the scheduling weight calculation formula for the edge node scheduling is as follows: Wi=α Ci+β Mi+γ Li+δ Di; Where Wi is the scheduling weight value of the i-th edge node, α, β, γ and δ are weight coefficients, Ci is the CPU utilization rate, Mi is the memory utilization rate, Li is the communication link quality score, and Di is the physical distance factor.

[0006] In one implementation, in the device collaborative control step, the linkage strategy includes a device linkage priority strategy, and the priority calculation formula for the device linkage priority strategy is: Pi=αi Ui+βi Si+γi Ti; Where Ui represents the urgency level, Si represents the safety relevance level, and Ti represents the time decay factor.

[0007] In one implementation, the preset rules in the edge intelligent decision-making step also include a network degradation decision-making algorithm, a load prediction algorithm based on LSTM neural network, and a fault early warning algorithm based on multi-parameter fusion.

[0008] In one implementation, the network degradation decision algorithm formula is as follows: Nscore=α Rlatency+β Bbandwidth+γ Jjitter+δ Ppacketloss; Where Nscore is the network health score, Rlatency is the latency score, Bbandwidth is the bandwidth utilization score, Jjitte is the jitter score, Ppacketloss is the packet loss rate score, and α, β, γ and δ are network characteristic coefficients.

[0009] In one implementation, the load forecasting algorithm based on LSTM neural networks is formulated as follows: =f(Qt,Qt 1,...,Qt n,Tout,Tin,I,N); in, The historical load sequence is given, where Tout is the outdoor temperature, Tin is the indoor set temperature, I is the light intensity, and N is the population density. Load prediction is performed using an LSTM neural network based on f.

[0010] In one implementation, the formula for the fault early warning algorithm based on multi-parameter fusion is: Hi=w1 Tscore+w2 Vscore+w3 Escore+w4 Cscore; Where Hi is the health index of the i-th device, Tscore is the temperature health score, Vscore is the vibration health score, Escore is the electrical parameter health score, Cscore is the communication quality health score, and w1, w2, w3 and w4 are coefficients.

[0011] The present invention also provides a building control system based on edge computing, which applies the above-described building control method based on edge computing and includes: The edge computing layer includes an edge autonomous control engine deployed within each edge computing node inside the building; The equipment access layer is used for the access and management of building equipment; The sensing and acquisition layer, which is a multi-sensor network deployed in various areas of the building, is responsible for collecting environmental data and equipment status information in real time. The device access layer is connected to the edge computing layer and the perception and acquisition layer respectively. The edge autonomous control engine includes an event input layer and a decision layer connected to the event input layer. The decision layer is connected to an output execution layer. The decision layer has a built-in local rule engine and a local learning module for the control logic to be executed on the edge side. In one implementation, the multiple edge computing nodes are linked together through a communication protocol hierarchy; The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a data processing unit, implements the steps of the above-described edge computing-based building control method.

[0012] The beneficial effects of this invention are as follows: By deeply integrating edge computing technology with distributed autonomous control of building equipment, three core innovations are proposed: Edge Autonomous Control Engine (EACE), distributed collaborative control, and offline autonomous operation mechanism. The system can maintain the basic operational capabilities of the building when the network is interrupted or the cloud fails, and the response latency is reduced from the traditional several seconds to the millisecond level, which significantly improves the reliability of smart buildings and user experience. Attached Figure Description

[0013] The invention will now be described in more detail with reference to embodiments and the accompanying drawings. Figure 1 A schematic diagram of the system architecture of the present invention is shown; Figure 2 A schematic diagram of the autonomous control engine module of the present invention is shown; Figure 3 The distributed collaborative control flowchart of the present invention is shown; Figure 4 This diagram illustrates the switching between the three operating modes of the present invention. Figure 5 A schematic diagram of an embodiment of HVAC edge control is shown; Figure 6 The timing diagram of the fire protection and security edge linkage is displayed; In the accompanying drawings, the same parts use the same reference numerals. The drawings are not to scale. Detailed Implementation

[0014] The invention will now be further described with reference to the accompanying drawings.

[0015] This invention provides a building control method based on edge computing, comprising: System initialization steps: loading configurations and policies for edge computing nodes, and establishing device topology relationships; The data acquisition and preprocessing steps involve acquiring sensor data and performing filtering, noise reduction, and standardization. Edge intelligent decision-making steps execute control logic based on preset rules and real-time data, generating and issuing control commands; The equipment collaborative control steps involve triggering collaborative actions of relevant equipment based on the linkage strategy. The status reporting and synchronization steps report the control results to the cloud and synchronize them with other edge nodes; It should be noted that the local rule engine has a rich set of control policy templates, which can support the execution of complex logical judgments and condition triggers on the edge side without relying on cloud commands. Based on the event stream processing architecture, it can achieve millisecond-level event response and command issuance. The response chain of a traditional centralized architecture is as follows: The link steps are as follows: User operation (click "Turn on air conditioner") - Internet transmission - Cloud server receives request - Property management logic processing - Cloud server - Device receives - Device status reporting; The process is explained as follows: User actions (HTTP requests); Internet transmission (with latency, typically 50-200ms, higher between cities); The cloud server accepts the request (routing and forwarding, 10-30ms). Property logic processing (authentication + permission verification + instruction construction, 50-150ms); Cloud server (forwarding by IoT platform, 30-100ms); The device receives the data (execution time 50-200ms). Device status reporting (returning to the cloud via the original path, then feeding back to the APP, and refreshing the display); The traditional centralized architecture takes 5-15 seconds in total (unstable and greatly affected by the network). It is not a single slow link, but the delay of 6-8 links is superimposed, and each link has uncertainty (network jitter, platform rate limiting, high concurrency queuing). In this invention, the link steps are as follows: Mobile APP user operation (click "turn on air conditioner") - LAN transmission - edge gateway directly parses the command - device end receives - device status local receipt; The process is explained as follows: User operations (edge ​​gateway, WiFi / wired); Local area network transmission (latency <5ms); The edge gateway directly parses the commands (and sends them to the device); The device receives (executes an action, 50-150ms, physical execution time); Local device status feedback (pushed from the edge gateway to the APP for display); It should be noted that, compared to traditional centralized architecture: Network hop count: Traditional centralized architecture (2-3 hops for public network round trip), this invention (direct LAN connection, 0 hops), saving 90%; For business logic processing, the traditional centralized architecture (cloud authentication + routing + forwarding) is replaced by this invention (edge ​​local determination, no authentication required), saving 90%; Ttotal=Tcollect+Tprocess+Tdecision+Texecute; Where: Ttotal is the total end-to-end response time, Tcollect is the sensor data acquisition time, Tprocess is the data preprocessing time (including filtering and noise reduction), Tdecision is the edge decision time (rule matching + execution), and Texecute is the instruction execution time (sent to the device). Through data preprocessing optimization and decision algorithm optimization, the end-to-end response time is controlled within 200 milliseconds, that is, the total time of this invention is <200ms, and the response latency is reduced from the traditional several seconds to the millisecond level, which significantly improves the reliability of smart buildings and user experience. Specifically, in the edge intelligent decision-making step, the preset rules include the edge node scheduling rules set in the local rule engine, and the calculation formula for the scheduling weight of the edge node is as follows: Wi=α Ci+β Mi+γ Li+δ Di; Where Wi is the scheduling weight value of the i-th edge node, α, β, γ and δ are weight coefficients, Ci is CPU utilization, Mi is memory utilization, Li is communication link quality score and Di is physical distance factor; It should be noted that, as Figure 3 As shown, an edge node intelligent scheduling algorithm is used for cross-node task scheduling. When a node is overloaded, some tasks are automatically transferred to adjacent nodes. The device status and alarm information between nodes are synchronized every 100ms to ensure global consistency. When the main node fails, the backup node takes over the control task within 50ms to achieve seamless switching. Furthermore, in the equipment collaborative control step, the linkage strategy includes an equipment linkage priority strategy, and the priority calculation formula for the equipment linkage priority strategy is as follows: Pi=αi Ui+βi Si+γi Ti; Where Ui represents the urgency level, Si represents the safety relevance level, and Ti represents the time decay factor; It should be noted that prioritizing data improves the user experience. Furthermore, in the edge intelligent decision-making process, the pre-defined rules also include a network degradation decision-making algorithm, a load prediction algorithm based on LSTM neural networks, and a fault early warning algorithm based on multi-parameter fusion. Specifically, the formula for the network degradation decision algorithm is: Nscore=α Rlatency+β Bbandwidth+γ Jitter + δ P packet loss; Where, N score is the network health score, R latency is the latency score, B bandwidth is the bandwidth utilization score, J jitte is the jitter score, P packet loss is the packet loss rate score, and α, β, γ, and δ are network characteristic coefficients; It should be noted that, as Figure 4 shown, that is, an offline autonomous operation mechanism is adopted, and the cloud-edge collaboration mode > the edge autonomous mode > the device local protection mode, which is automatically switched according to the network status; Among them: when the latency < 50ms, the packet loss rate < 1%, and the network health score N score > 80, the cloud-edge collaboration mode is executed, and the control entity is the cloud + edge collaborative decision-making; 50ms < latency < 200ms, 1% < packet loss rate < 3%, that is, when 50 < N score < 80, the edge autonomous mode is executed, and the control entity is the edge-side independent decision-making; The latency > 200ms, the packet loss rate > 5%, that is, when N score < 50, the device protection mode is executed, and the control entity is the device's own protection logic; Furthermore, the load prediction algorithm formula based on the LSTM neural network is: = f(Qt, Qt 1,..., Qt n, Tout, Tin, I, N); Where, is the historical load sequence, Tout is the outdoor temperature, Tin is the indoor set temperature, I is the light intensity, N is the personnel density, and the LSTM neural network load prediction is performed through f; The fault warning algorithm formula based on multi-parameter fusion is: Hi = w1 T score + w2 V score + w3 E score + w4 C score; Where, Hi is the health index of the i-th device, T score is the temperature health score, V score is the vibration health score, E score is the electrical parameter health score, C score is the communication quality health score, and w1, w2, w3, and w4 are coefficients; It should be noted that the LSTM neural network load prediction is performed through f to achieve energy consumption optimization, and the health assessment based on multi-parameter fusion is used to achieve early warning of device faults; The present invention also provides a building control system based on edge computing, such as Figure 1 and Figure 2 As shown, it includes: The edge computing layer includes an edge autonomous control engine deployed within each edge computing node inside the building; The equipment access layer is used for the access and management of building equipment; The sensing and acquisition layer, which is a multi-sensor network deployed in various areas of the building, is responsible for collecting environmental data and equipment status information in real time. The device access layer is connected to the edge computing layer and the perception and acquisition layer respectively. The edge autonomous control engine includes an event input layer and a decision layer connected to the event input layer. The decision layer is connected to the output execution layer. The decision layer has a built-in local rule engine and a local learning module for the control logic to be executed on the edge side. Specifically, it also includes cloud servers, bidirectional data synchronization between the edge computing layer and cloud servers, and hierarchical linkage between multiple edge computing nodes through communication protocols; It should be noted that the Edge Autonomous Control Engine (EACE) is deployed on edge computing nodes. It has a rich set of pre-built control strategy templates, which can support the execution of complex logical judgments and condition triggers on the edge side without relying on cloud commands. Based on the event stream processing architecture, it can achieve millisecond-level event response and command issuance, maintain the device state machine model, support the switching of multiple operating states of the device and anomaly protection, and achieve predictive control and parameter optimization based on historical data through local machine learning algorithms. Meanwhile, multiple edge nodes are deployed in a distributed manner. Each edge node achieves state synchronization, load balancing, fault tolerance, and hierarchical linkage through a lightweight communication protocol. That is, it synchronizes device status, control policies, and alarm information in real time, dynamically allocates control tasks according to the load of each node, and automatically transfers control tasks to other healthy nodes when a node fails. It supports cross-regional and cross-system device linkage control. In one embodiment, such as Figure 5 As shown, taking edge control of HVAC as an example: Edge computing gateways are deployed in the weak current shafts of each floor, with built-in EACE control engines. They are connected to fan coil units, fresh air units, air conditioning boxes and other equipment through protocols such as BACnet, Modbus, and KNX. Temperature and humidity sensors, CO2 sensors and human presence sensors are deployed in each air conditioning zone. The built-in control strategies include: (1) automatically adjusting the air supply volume according to the indoor personnel density; (2) optimizing the operating parameters based on the outdoor temperature and indoor load; (3) realizing independent zone control to meet the comfort needs of different areas; and (4) automatically entering the energy-saving mode when no one is around at night. All control logic is executed at the edge, reducing response time from the traditional 5-10 seconds to less than 200 milliseconds; By employing a load forecasting algorithm and optimized control based on LSTM neural networks, operating parameters are adjusted in advance based on historical data and meteorological information, achieving energy savings of 15%-25%. In one embodiment, edge linkage of a barrier gate access control system is taken as an example: The barrier gate, access control, and elevator control systems are all connected to the same edge node, and scene linkage includes: (1) When the owner swipes his card to open the unit door, the elevator is automatically called to the corresponding floor. (2) After the vehicle is recognized, the barrier is automatically raised and the garage lights are turned on. (3) After the visitor makes a successful reservation, the temporary access permission is automatically issued to the gate and access control. The device linkage priority algorithm is used to sort the priorities to ensure that fire linkage (priority 100) > security linkage (priority 80) > daily access (priority 20). The edge node monitors the status of each device in real time. In case of abnormality, an alarm is immediately triggered and the relevant device is locked. When the network is interrupted, the edge node can still perform basic personnel and vehicle access control. The network degradation intelligent decision algorithm is used to switch modes. In one embodiment, such as Figure 6 The diagram shows the timing sequence of edge linkage between fire protection and security systems, enabling millisecond-level edge linkage response between the fire protection and security systems. After a smoke alarm signal is triggered, the edge node completes the following actions within 50 milliseconds: (1) Turn on the emergency evacuation instructions; (2) Turn off normal lighting and turn on emergency lighting; (3) Turn on the smoke control and exhaust system; (4) Notify relevant personnel to evacuate; (5) Lock the elevators and gates in the relevant areas; After an intrusion alarm is triggered, the edge node immediately: (1) Activate the on-site audible and visual alarm; (2) Real-time recording is performed in conjunction with the video system; (3) Push the image to the security center; (4) Lock down the relevant entrances and exits; The entire linkage process does not rely on the cloud, and is unaffected even if the cloud platform fails. It uses a health assessment algorithm based on multi-parameter fusion to assess the health of key equipment such as fire control panels and alarm detectors, with a fault early warning accuracy rate of over 90%.

[0016] In the description of this invention, it should be understood that the terms "upper", "lower", "bottom", "top", "front", "rear", "inner", "outer", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0017] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other embodiments.

Claims

1. A building control method based on edge computing, characterized in that, include: System initialization steps: loading configurations and policies for edge computing nodes, and establishing device topology relationships; The data acquisition and preprocessing steps involve acquiring sensor data and performing filtering, noise reduction, and standardization. Edge intelligent decision-making steps execute control logic based on preset rules and real-time data, generating and issuing control commands; The equipment collaborative control steps involve triggering collaborative actions of relevant equipment based on the linkage strategy. The status reporting and synchronization steps report the control results to the cloud and synchronize them with other edge nodes.

2. The building control method based on edge computing according to claim 1, characterized in that, In the edge intelligent decision-making process, the preset rules include edge node scheduling rules set in the local rule engine. The calculation formula for the scheduling weight of the edge node scheduling is as follows: Wi=a Ci+β Mi+g Li+d Di; Where Wi is the scheduling weight value of the i-th edge node, α, β, γ and δ are weight coefficients, Ci is the CPU utilization rate, Mi is the memory utilization rate, Li is the communication link quality score, and Di is the physical distance factor.

3. The building control method based on edge computing according to claim 1, characterized in that, In the equipment collaborative control step, the linkage strategy includes an equipment linkage priority strategy, and the priority calculation formula for the equipment linkage priority strategy is as follows: Pi=αi Ui+βi Si+γi Ti; Where Ui represents the urgency level, Si represents the safety relevance level, and Ti represents the time decay factor.

4. The building control method based on edge computing according to claim 1, characterized in that, In the edge intelligent decision-making process, the pre-defined rules also include a network degradation decision-making algorithm, a load prediction algorithm based on LSTM neural network, and a fault early warning algorithm based on multi-parameter fusion.

5. The building control method based on edge computing according to claim 4, characterized in that, The formula for the network degradation decision algorithm is: Nscore=α Rlatency+β Bbandwidth+γ Jjitter+δ Ppacketloss; Where Nscore is the network health score, Rlatency is the latency score, Bbandwidth is the bandwidth utilization score, Jjitte is the jitter score, Ppacketloss is the packet loss rate score, and α, β, γ and δ are network characteristic coefficients.

6. The building control method based on edge computing according to claim 4, characterized in that, The formula for the load forecasting algorithm based on LSTM neural network is: =f(Qt,Qt 1,...,Qt n,Tout,Tin,I,N); in, The historical load sequence is given, where Tout is the outdoor temperature, Tin is the indoor set temperature, I is the light intensity, and N is the population density. Load prediction is performed using an LSTM neural network based on f.

7. The building control method based on edge computing according to claim 4, characterized in that, The formula for the fault early warning algorithm based on multi-parameter fusion is: Hi=w1 Tscore+w2 Vscore+w3 Escore+w4 Cscore; Where Hi is the health index of the i-th device, Tscore is the temperature health score, Vscore is the vibration health score, Escore is the electrical parameter health score, Cscore is the communication quality health score, and w1, w2, w3 and w4 are coefficients.

8. A building control system based on edge computing, employing the building control method according to any one of claims 1 to 7, characterized in that, include: The edge computing layer includes an edge autonomous control engine deployed within each edge computing node inside the building; The equipment access layer is used for the access and management of building equipment; The sensing and acquisition layer, which is a multi-sensor network deployed in various areas of the building, is responsible for collecting environmental data and equipment status information in real time. The device access layer is connected to the edge computing layer and the perception and acquisition layer respectively. The edge autonomous control engine includes an event input layer and a decision layer connected to the event input layer. The decision layer is connected to an output execution layer. The decision layer has a built-in local rule engine and a local learning module for the control logic to be executed on the edge side.

9. A building control system based on edge computing according to claim 8, characterized in that, Multiple edge computing nodes are linked together through a communication protocol layer.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the data processing unit, it implements the steps of the edge computing-based building control method according to any one of claims 1 to 7.

Images