An automatic control method for keyboard matching multiple devices

Abstract

The present invention relates to the field of computer technology, and in particular to an automatic control method for matching multiple devices with a keyboard, which comprises the following steps: the keyboard receives identification numbers, device data information, sensor data and user-selected modes of multiple devices to form input information; the keyboard uses the input information and the user-selected mode to define a corresponding reward function and train a reinforcement learning algorithm based on the received input information, thereby obtaining weight parameters trained for the corresponding mode and saving the model; the keyboard calls the trained model of the corresponding mode according to the user's selected mode based on the received input information, and inputs the input information into the reinforcement learning algorithm; the reinforcement learning algorithm outputs multiple control instruction information of the corresponding mode based on the user's selected mode and corresponding sensor data, device and other information. The present invention can automatically control networked devices, save manpower and material resources, and improve work efficiency.

Description

Technical Field

[0001] The invention relates to the field of computer technology, and in particular to an automatic control method for matching a keyboard with multiple devices. Background Art

[0002] With the development of science and technology, the number of networked devices continues to increase, and further, management and control will become more complex and changeable. Since human energy is limited, facing so many networked devices, a set of automated and intelligent control methods is urgently needed. However, the control methods currently used are relatively traditional, which are only some threshold adjustments, and artificial intelligence related methods have not been further used for adjustment and control. Artificial intelligence methods can obtain control models of corresponding modes through training. In view of this, it is necessary to provide an automatic control method for keyboard matching multiple devices to solve the above problems, thereby saving labor costs and time costs for users and improving work efficiency. Summary of the invention

[0003] In view of the problems of the prior art, the present invention provides an automatic control method for matching a keyboard with multiple devices. The method has an ingenious design and allows users to switch between different modes according to their needs. The present invention defines a reward function for a corresponding mode based on the received input information and can train a model of the required mode, thereby enabling automatic control of networked devices, saving manpower and material resources and improving work efficiency.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] The present invention provides an automatic control method for keyboard matching multiple devices, comprising the following steps:

[0006] S1, the keyboard receives identification numbers of multiple devices, device data information, sensor data and a mode selected by the user to form input information;

[0007] S2. The keyboard defines the corresponding reward function based on the received input information and the mode selected by the user and performs training of the reinforcement learning algorithm, thereby obtaining the weight parameters trained for the corresponding mode and saving the model;

[0008] S3, the keyboard calls the trained model of the corresponding mode according to the user's selected mode based on the received input information, and inputs the input information into the reinforcement learning algorithm;

[0009] S4. The reinforcement learning algorithm outputs multiple control instruction information of the corresponding mode according to the user's selection mode and corresponding sensor data, equipment and other information.

[0010] Among them, the modes selected by the user include safety mode, energy-saving mode, quiet mode, entertainment mode and air purification mode.

[0011] Wherein, a mode switching knob is provided on the keyboard, and the user can switch between different modes through the mode switching knob.

[0012] Among them, the reinforcement learning algorithm defines the reward function of the corresponding pattern according to the received input information, so as to train the model of the required pattern. In deep reinforcement learning, reinforcement learning is composed of a deep neural network, which maps multiple inputs to multiple outputs.

[0013] The sensor data includes temperature value, humidity value, PM2.5 value, light value and sound volume value.

[0014] The device data information includes the current time and the number of devices currently turned on.

[0015] The multiple control command information includes a switch command of the air conditioner. K1 With temperature control instructions Curtain motor rotation command O R , Audio volume adjustment command O Y , Air purifier switch command O K2 Air Purifier Adjustment Instructions

[0016] The reinforcement learning algorithm adopts the DQN algorithm, through the switch I of the air conditioner K1 , air conditioning temperature adjustment I T , temperature sensor value I v1 , air conditioning power I q1 , Audio switch Volume I Y , sound sensor value I v2 , audio power Air Purifier Switch I K3 , air purifier purification intensity I A , PM2.5 gas detection sensor value I v3 , air purifier power RGB Light Adjustment I rgb , RGB light power Smoke Sensor I smoke ,The input information of the algorithm model is:

[0017] And through the switch of the air conditioner O K1 , air conditioning temperature adjustment O T , Audio switch Volume Y , air purifier switch O K3 , air purifier purification intensity O A , RGB light value Orgb , extinguishing gas and dry powder output value O powder , exhaust fan speed adjustment O fan , the output information of the algorithm model is Update the neural network parameters to p = p + a (OO y ), where O y is the actual output value of the device corresponding to O, and α is the learning rate, both of which are set to 0.1.

[0018] Beneficial effects of the present invention:

[0019] The present invention is cleverly designed, and users can switch between different modes according to their needs. The present invention defines a reward function for a corresponding mode based on received input information, and can train a model for a desired mode, thereby being able to automatically control networked devices, saving manpower and material resources, and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The present invention is a flowchart of an automatic control method for matching a keyboard with multiple devices. DETAILED DESCRIPTION

[0021] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with the embodiments and drawings, and the contents mentioned in the embodiments are not intended to limit the present invention. The present invention is described in detail below in conjunction with the drawings.

[0022] An automatic control method for keyboard matching multiple devices comprises the following steps:

[0023] S1, the keyboard receives identification numbers of multiple devices, device data information, sensor data and a mode selected by the user to form input information;

[0024] S2. The keyboard defines the corresponding reward function based on the received input information and the mode selected by the user and performs training of the reinforcement learning algorithm, thereby obtaining the weight parameters trained for the corresponding mode and saving the model;

[0025] S3, the keyboard calls the trained model of the corresponding mode according to the user's selected mode based on the received input information, and inputs the input information into the reinforcement learning algorithm;

[0026] S4. The reinforcement learning algorithm outputs multiple control instruction information of the corresponding mode according to the user's selection mode and corresponding sensor data, equipment and other information.

[0027] Specifically, the present invention is cleverly designed, and users can switch between different modes according to their needs. The present invention defines a reward function for a corresponding mode based on the received input information, and can train a model for the desired mode, thereby being able to automatically control networked devices, save manpower and material resources, and improve work efficiency.

[0028] Among them, the modes selected by the user include safety mode, energy-saving mode, quiet mode, entertainment mode and air purification mode.

[0029] Wherein, a mode switching knob is provided on the keyboard, and the user can switch between different modes through the mode switching knob.

[0030] Among them, the reinforcement learning algorithm defines the reward function of the corresponding pattern according to the received input information, so as to train the model of the required pattern. In deep reinforcement learning, reinforcement learning is composed of a deep neural network, which maps multiple inputs to multiple outputs.

[0031] The sensor data includes temperature value, humidity value, PM2.5 value, light value and sound volume value.

[0032] The device data information includes the current time and the number of devices currently turned on.

[0033] The multiple control command information includes a switch command of the air conditioner. K1 With temperature control instructions Curtain motor rotation command O R , Audio volume adjustment command O Y , Air purifier switch command O K2 Air Purifier Adjustment Instructions

[0034] The reinforcement learning algorithm adopts the DQN algorithm, through the switch I of the air conditioner K1 , air conditioning temperature adjustment I T , temperature sensor value I v1 , air conditioning power I q1 , Audio switch Volume I Y , sound sensor value I v2 , audio power Air Purifier Switch I K3 , air purifier purification intensity I A , PM2.5 gas detection sensor value I v3 , air purifier power RGB Light Adjustment I rgb , RGB light power Smoke Sensor I smoke ,The input information of the algorithm model is:

[0035] And through the switch of the air conditioner O K1 , air conditioning temperature adjustment O T , Audio switch Volume Y , air purifier switch O K3 , air purifier purification intensity O A , RGB light value O rgb , extinguishing gas and dry powder output value O powder , exhaust fan speed adjustment O fan , the output information of the algorithm model is Update the neural network parameters to p = p + a (OO y ), where O y is the actual output value of the device corresponding to O, and α is the learning rate, both of which are set to 0.1.

[0036] In the safety mode combined with the energy-saving mode, the input information of the algorithm model is Define the reward function to minimize the amount of power:

[0037] Among them, W smoke Set to 10, because safety issues require greater sensitivity to the value of smoke and gas. The goal is to make the reward value r as large as possible. When the reward value r is stable, the model parameters converge and the corresponding instructions are output. That is, shut down unnecessary equipment and reduce the power consumption of corresponding equipment as much as possible, put the corresponding equipment into energy-saving mode, and turn on the fan to deliver dry powder to the target area when the smoke value reaches the concentration determined by the model;

[0038] In the air purification mode combined with the safety mode and the energy-saving mode, define the reward function of minimizing the power consumption and minimizing the PM2.5 value:

[0039] Among them, W smoke Set to 10, W v3 Set to 5, because this mode is more sensitive to the PM2.5 gas value;

[0040] In the combination of quiet mode, safety mode and energy-saving mode, define the reward function of minimizing the power consumption and the sound value as small as possible:

[0041] Among them, W smoke Set to 10, W v2 Set to 5, because this mode is more sensitive to the value of the sound;

[0042] In the entertainment mode combined with the safety mode and the energy-saving mode, define the reward function of minimizing the power consumption and the sound value as small as possible:

[0043] Among them, W smoke Set to 10, W rgb Set to 5, because this mode is more sensitive to the value of the RGB light.

[0044] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention is disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes by using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent changes and modifications made to the above embodiments according to the technology of the present invention are within the scope of the technical solution of the present invention without departing from the content of the technical solution of the present invention.

Claims

1. An automatic control method for keyboard matching multiple devices, It is characterized in that The following steps are involved: S1, the keyboard receives identification numbers of multiple devices, device data information, sensor data and a mode selected by the user to form input information; S2. The keyboard defines the corresponding reward function based on the received input information and the mode selected by the user and performs training of the reinforcement learning algorithm, thereby obtaining the weight parameters trained for the corresponding mode and saving the model; S3, the keyboard calls the trained model of the corresponding mode according to the user's selected mode based on the received input information, and inputs the input information into the reinforcement learning algorithm; S4, the reinforcement learning algorithm outputs the corresponding mode of multi-control instruction information according to the user's selection mode and the corresponding sensor data and device information; The multiple control command information includes the switch command of the air conditioner With temperature control instructions , Curtain motor rotation command , Audio volume adjustment instructions , Air purifier switch instructions Air Purifier Adjustment Instructions ; The reinforcement learning algorithm adopts the DQN algorithm, through the switch of the air conditioner , air conditioning temperature adjustment , temperature sensor value , air conditioning power , Audio switch , volume level , sound sensor value , audio power , air purifier switch , air purifier purification intensity , PM2.5 gas detection sensor value , air purifier power , RGB lighting adjustment , RGB light power , smoke sensor ,The input information of the algorithm model is: , and through the switch of the air conditioner , air conditioning temperature adjustment , Audio switch , volume level , air purifier switch , air purifier purification intensity , RGB light value , extinguishing gas and dry powder output values , exhaust fan speed adjustment , the output information of the algorithm model is , update the neural network parameters to ,in, For the corresponding The actual output value of the device in is the learning rate, all set to 0.1; The user-selected mode includes a safety mode, an energy-saving mode, a quiet mode, an entertainment mode, and an air purification mode; The keyboard is provided with a mode switching knob, and the user can switch between different modes by using the mode switching knob; The reinforcement learning algorithm defines the reward function of the corresponding mode according to the received input information, and can train the model of the required mode. In deep reinforcement learning, reinforcement learning is composed of a deep neural network that maps multiple inputs to multiple outputs; In the safety mode combined with the energy-saving mode, the input information of the algorithm model is , define the reward function of power: , in, Set to 10, let the reward value The bigger the better, when the reward value After stabilization, the model parameters converge and the corresponding instructions are output , let the corresponding equipment enter the energy-saving mode, and when the smoke value reaches the concentration determined by the model, turn on the fan to deliver the dry powder to the target area; In the air purification mode combined with the safety mode and the energy saving mode, the reward function for power and PM2.5 values ​​is defined as follows: ,in, Set to 10, Set to 5; In the quiet mode combined with the safe mode and the energy-saving mode, define the reward function for the power and sound values: ,in, Set to 10, Set to 5; In the entertainment mode combined with the safety mode and the energy-saving mode, define the reward function for the power and sound values: ,in, Set to 10, Set to 5.

2. According to claim 1, a keyboard matching multiple devices automatic control method, Features: The sensor data includes temperature value, humidity value, PM2.5 value, light value and sound volume value.

3. According to claim 1, a keyboard matching multiple devices automatic control method, Features: The device data information includes the current time and the number of devices currently turned on.

Images