Intelligent control method based on Internet of Things

Abstract

The invention belongs to the field of intelligent control, and particularly relates to an intelligent control method based on the Internet of Things. Firstly, data collection is carried out, and different sensor signal data are fused. Then, a distributed MPC algorithm based on multi-agent cooperative reinforcement is adopted, a controlled object is modeled, a system is divided into corresponding agents of subsystems, a cost function containing state tracking errors and control input energy is set, the agents calculate and optimize based on the agents and interaction information, and rolling optimization is achieved. And finally, compressing signal data based on an improved discrete cosine transform technology. According to the method, the calculation burden is relieved, the control real-time performance and reliability are improved, the control delay is effectively reduced, and collaborative optimization of equipment can be better realized.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent control, and in particular relates to an intelligent control method based on the Internet of Things. Background Art

[0002] With the rapid development of Internet of Things technology, more and more devices and systems have achieved interconnection, which provides a broad application space for intelligent control. However, in practical applications, it faces many technical challenges. In the control of complex systems, traditional centralized control methods have certain limitations. For example, in large-scale industrial production processes or smart home systems, centralized control needs to process data from a large number of devices, which is too computationally burdensome and highly dependent on the overall performance of the system. Once a central node fails, the entire system may be paralyzed. At the same time, data transmission delay is also a problem that needs to be solved urgently. If the signal data cannot be effectively compressed during the transmission and processing of data collected by a large number of sensors, it will cause delayed execution of control instructions and affect the control effect. In addition, in the scenario of multi-device collaboration, the coupling relationship between the devices is complex, and traditional control methods are difficult to achieve efficient collaborative optimization and cannot give full play to the advantages of the Internet of Things system. Summary of the invention

[0003] In view of the technical problems existing in the above-mentioned background technology, the present invention proposes an intelligent control method based on the Internet of Things.

[0004] In order to achieve the above object, the technical solution adopted by the present invention comprises the following steps:

[0005] S1. First, collect data and fuse the signal data provided by different sensors;

[0006] S2, then use the distributed MPC algorithm based on multi-agent collaborative reinforcement to perform intelligent control. The specific implementation steps are:

[0007] S21. First, the controlled object is modeled and the state space model is used to describe its dynamic characteristics. The state equation is: k+1 =f(x k ,u k ), where x k is the state vector of the system at time k, u k is the control vector of the system at time k, and the f function represents the state transfer function of the system; the output equation is: k =g(x k ,u k ), where g is the output function;

[0008] S22, then divide the entire system into multiple N interrelated subsystems, each subsystem corresponds to an intelligent agent, and the subsystem state corresponding to intelligent agent i is represented by Indicates that the input express;

[0009] S23, set the cost function, including the state tracking error and control input energy, in the following form: in is the control input sequence of agent i starting from time k for the next H steps, is the corresponding state sequence, and the predicted time series is H, L i () is the stage cost function, F i () is the terminal cost function, which is used to evaluate the state of the endpoint of the prediction time domain;

[0010] S24. At the same time, the global cost function is the combination of all the local cost functions of the agents:

[0011] S25, then independently perform future state prediction and optimization calculation for each agent based on its own local model, current state information, and information obtained from interaction with adjacent agents;

[0012] S26. Finally, at each time k, each agent calculates the optimal control input sequence at the current time Then only the first control input of this sequence It is applied to the corresponding subsystem. As the system runs, at the next time k+1, the above distributed prediction, collaborative reinforcement and optimization calculation process is repeated again according to the new state information to perform rolling optimization.

[0013] S3. Finally, based on the improved discrete cosine transform technology, signal data is compressed to reduce the control delay in intelligent control.

[0014] Preferably, there is a coupling relationship between the agents in step S22, and the state evolution of agent i is affected by the states and inputs of other agents, which can be expressed as: in They represent the states and input sets of other agents except agent i respectively.

[0015] Preferably, the stage cost function L in step S23 is i ()for: Where Q i , R i are the state weight matrix and the control input weight matrix respectively, It is the reference state at time t, which is used to measure the control effect of each step.

[0016] Preferably, the terminal cost function F in step S23 i ()for: Where P i is the terminal weight matrix.

[0017] Preferably, in step S25, the agent i predicts the future state sequence at time k to satisfy: At the same time, under the condition of satisfying the system constraints, find the local cost function J i Minimum control input sequence It is expressed as:

[0018] Preferably, the system constraint is a state constraint Control Input Constraints

[0019] Preferably, the implementation of compressing the signal data based on the improved discrete cosine transform technology in step S3 is as follows:

[0020] S31. First, the signal of each agent Perform a discrete cosine transform to convert the signal from the time domain to the frequency domain:

[0021] S32. According to the characteristics of the signal, the low-frequency components are weighted, which is achieved through a weighting coefficient matrix: Where W is the weighting coefficient matrix;

[0022] S33, then retaining the low-frequency part after DCT transformation by setting a threshold, and quantizing the retained frequency domain coefficients;

[0023] S34. Finally, the quantized frequency domain coefficients are Huffman encoded to further compress the data and reduce the transmission load.

[0024] Compared with the prior art, the advantages and positive effects of the present invention are that, through the distributed MPC algorithm enhanced by multi-agent collaboration, the complex system is divided into multiple subsystems, each agent independently calculates and collaboratively optimizes, significantly reducing the computational burden and improving the real-time control. Compared with traditional centralized control, the risk of system paralysis due to central node failure is avoided, and the system reliability is enhanced. The signal data is compressed using the improved discrete cosine transform technology, effectively reducing the control delay. DETAILED DESCRIPTION

[0025] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0027] Embodiment, in the application scenario of smart home system, with the continuous development of Internet of Things technology, various devices in the home, such as lighting system, temperature control system, security system, etc., have been connected to the network and achieved interconnection. However, the traditional centralized control method exposes many problems when facing such a complex smart home system. These include the problem that centralized control needs to process data from a large number of devices, the calculation burden is too heavy, and the data transmission delay problem. In this regard, this application proposes an intelligent control method based on the Internet of Things.

[0028] First, in order to fully and accurately obtain the operating status and environmental information of various devices in the smart home system, multiple sensors are used for data collection and the data is fused. First, various sensors collect real-time data through wireless networks or wired interfaces. These sensors may come from different locations or work at different sampling frequencies, so unified data formats and time synchronization are required. Data fusion can be achieved through the following steps: Each sensor collects relevant signals in real time, such as temperature, humidity, pressure, etc., and transmits them to the central processing unit through the data acquisition module. The raw data is cleaned and standardized to eliminate the differences between the sensors. The Kalman filter fusion algorithm is used to integrate the data from multiple sensors to obtain a more accurate and comprehensive system state estimate. In this way, the fused data can be obtained to improve the accuracy and reliability of the system, thereby providing more accurate input for subsequent intelligent control and optimization.

[0029] Next, consider that traditional MPC needs to centrally process the data of all devices. As the number of devices increases, the computational complexity grows exponentially, resulting in excessive computational burden, prone to computational delays, and affecting the real-time performance of control. The present invention adopts a distributed MPC algorithm based on multi-agent collaborative reinforcement for intelligent control, divides the large system into multiple subsystems, and each subsystem corresponds to an agent for independent calculation, which greatly reduces the computational burden. Each agent operates relatively independently. When an agent or a local subsystem fails, it will not affect the basic operation of other agents and the entire system. In addition, by exchanging limited information between agents, the coupling relationship can be better handled, and collaborative reinforcement and optimization calculations can be performed to achieve the global optimal effect. First, the controlled object is modeled, and the state space model is used to describe its dynamic characteristics. The state equation is: x k+1 =f(x k ,u k ), where x kis the state vector of the system at time k, u k is the control vector of the system at time k, and the f function represents the state transfer function of the system; the output equation is: k =g(x k ,u k ), where g is the output function; then the entire system is divided into multiple N interrelated subsystems, each subsystem corresponds to an intelligent agent, and the subsystem state corresponding to intelligent agent i is expressed as Indicates that the input In addition, there is a coupling relationship between the agents, and the state evolution of agent i is affected by the states and inputs of other agents, which can be expressed as: in Represent the states and input sets of other agents except agent i. Set the cost function, including state tracking error and control input energy, in the following form: in is the control input sequence of agent i starting from time k for the next H steps, is the corresponding state sequence, and the predicted time series is H, L i () is the stage cost function: Where Q i , R i are the state weight matrix and the control input weight matrix respectively, is the reference state at time t, used to measure the control effect of each step, F i () is the terminal cost function, which is used to evaluate the state of the predicted time domain endpoint: Where P i is the terminal weight matrix. At the same time, the global cost function is the combination of all agent local cost functions: Then, each agent independently performs future state prediction and optimization calculation based on its own local model, current state information, and limited information obtained from interactions with neighboring agents. The future state sequence predicted by agent i at time k satisfies:

[0030] At the same time, under the condition of satisfying the system constraints, the system constraints are state constraints Control Input Constraints Find the local cost function J i Minimum control input sequence It is expressed as: Finally, at each time k, each agent calculates the optimal control input sequence at the current time Then only the first control input of this sequence It is applied to the corresponding subsystem. As the system runs, at the next moment k+1, the above distributed prediction, collaborative reinforcement and optimization calculation process is repeated again according to the new state information to perform rolling optimization.

[0031] Finally, in order to ensure effective data compression while reducing control delay, the present invention compresses signal data based on improved discrete cosine transform technology. First, the signal of each agent is Perform a discrete cosine transform to convert the signal from the time domain to the frequency domain: According to the characteristics of the signal, the low-frequency components are weighted, which is achieved through the weighting coefficient matrix: Where W is the weighted coefficient matrix; then by setting the threshold, the low-frequency part after DCT transformation is retained, and the retained frequency domain coefficients are quantized; finally, the quantized frequency domain coefficients are Huffman encoded to further compress the data and reduce the transmission load.

[0032] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent control method based on the Internet of Things, characterized in that: The following steps are involved: S1. First, collect data and fuse the signal data provided by different sensors; S2, then use the distributed MPC algorithm based on multi-agent collaborative reinforcement to perform intelligent control. The specific implementation steps are: S21. First, the controlled object is modeled and the state space model is used to describe its dynamic characteristics. The state equation is: k+1 =f(x k ,u k ), where x k is the state vector of the system at time k, u k is the control vector of the system at time k, and the f function represents the state transfer function of the system; the output equation is: k =g(x k ,u k ), where g is the output function; S22, then divide the entire system into multiple N interrelated subsystems, each subsystem corresponds to an intelligent agent, and the subsystem state corresponding to intelligent agent i is represented by Indicates that the input express; S23, set the cost function, including the state tracking error and control input energy, in the following form: in is the control input sequence of agent i starting from time k for the next H steps, is the corresponding state sequence, and the predicted time series is H, L i () is the stage cost function, F i () is the terminal cost function, which is used to evaluate the state of the endpoint of the prediction time domain; S24. At the same time, the global cost function is the combination of all the local cost functions of the agents: S25, then independently perform future state prediction and optimization calculation for each agent based on its own local model, current state information, and information obtained from interaction with adjacent agents; S26. Finally, at each time k, each agent calculates the optimal control input sequence at the current time Then only the first control input of this sequence It is applied to the corresponding subsystem. As the system runs, at the next time k+1, the above distributed prediction, collaborative reinforcement and optimization calculation process is repeated again according to the new state information to perform rolling optimization. S3. Finally, based on the improved discrete cosine transform technology, signal data is compressed to reduce the control delay in intelligent control.

2. According to the intelligent control method based on the Internet of Things according to claim 1, it is characterized in that: In step S22, there is a coupling relationship between the agents, and the state evolution of agent i is affected by the states and inputs of other agents, which can be expressed as: in They represent the states and input sets of other agents except agent i respectively.

3. The intelligent control method based on the Internet of Things according to claim 1 is characterized in that: In step S23, the stage cost function L i ()for: Where Q i , R i are the state weight matrix and the control input weight matrix respectively, It is the reference state at time t, which is used to measure the control effect of each step.

4. The intelligent control method based on the Internet of Things according to claim 1 is characterized in that: In step S23, the terminal cost function F i ()for: Where P i is the terminal weight matrix.

5. The intelligent control method based on the Internet of Things according to claim 1 is characterized in that: In step S25, agent i predicts the future state sequence at time k to satisfy: At the same time, under the condition of satisfying the system constraints, find the local cost function J i Minimum control input sequence It is expressed as:

6. The intelligent control method based on the Internet of Things according to claim 5, characterized in that: The system constraints are state constraints Control Input Constraints 7. The intelligent control method based on the Internet of Things according to claim 1, characterized in that: The implementation of compressing the signal data based on the improved discrete cosine transform technology in step S3 is as follows: S31. First, the signal of each agent Perform a discrete cosine transform to convert the signal from the time domain to the frequency domain: S32. According to the characteristics of the signal, the low-frequency components are weighted, which is achieved through a weighting coefficient matrix: Where W is the weighting coefficient matrix; S33, then retaining the low-frequency part after DCT transformation by setting a threshold, and quantizing the retained frequency domain coefficients; S34. Finally, the quantized frequency domain coefficients are Huffman encoded to further compress the data and reduce the transmission load.

Images