A method and system for calibrating a sensor array with a mobile robot

Abstract

The present application relates to the technical field of sensor calibration, and particularly relates to a sensor array calibration method and system matched with a mobile robot, which comprises the following steps: setting up a calibration environment, arranging a mobile robot carrying a standard gas sensor in the calibration environment to collect standard concentration data; training the calibration concentration data and the standard concentration data in a convolutional neural network model by using a terminal server to obtain calibration models of a plurality of calibration gas sensors and calibrate the plurality of calibration gas sensors; the terminal server generates calibration accuracy of each calibration gas sensor according to a gas distribution cloud map and calibrated concentration data, and determines calibration effect according to the calibration accuracy to perform a plurality of correction measures; the present application quickly completes the calibration of the sensor array on the basis of keeping detection of the fixed calibration gas sensor by introducing the mobile robot, thereby effectively improving the calibration efficiency while ensuring the calibration accuracy.

Description

Technical Field

[0001] This invention relates to the field of sensor calibration technology, and in particular to a sensor array calibration method and system for use with mobile robots. Background Technology

[0002] In the context of the Internet of Everything, intelligence has become the main theme of the times. Intelligent applications are closely related to sensors. For the calibration methods of traditional indoor air environment detection or harmful gas monitoring sensors, the existing common method is to build a pressure chamber and place the gas sensors inside for calibration one by one.

[0003] Chinese Patent Publication No. CN112683836B discloses a calibration method and system for a carbon dioxide sensor based on a backpropagation (BP) neural network. The method includes: acquiring a first acquisition voltage of each initial gas sensor; dividing all the first acquisition voltages of the initial gas sensors into a training set, a validation set, and a test set, and sequentially training, validating, and testing the BP neural network to be tested to obtain candidate BP neural network models for the initial gas sensors; acquiring a second acquisition voltage of the gas sensor to be calibrated; inputting the second acquisition voltage and the corresponding ambient temperature into the candidate BP neural network model of each initial gas sensor to obtain an output value; calculating the difference between the output value and the corresponding carbon dioxide concentration; and calculating the sum of the absolute values ​​of all differences; and using the candidate BP neural network model corresponding to the smallest sum of absolute values ​​as the calibration BP neural network model for the gas sensor to be calibrated.

[0004] However, the above methods cannot effectively guarantee calibration efficiency while ensuring calibration accuracy. Summary of the Invention

[0005] Therefore, this invention provides a sensor array calibration method and system for use with mobile robots. This overcomes the problem in existing technologies that cannot effectively guarantee calibration efficiency while ensuring calibration accuracy.

[0006] To achieve the above objectives, in one aspect, the present invention provides a sensor array calibration method for use with a mobile robot, comprising:

[0007] Step S1: Set up a calibration environment and arrange several calibration gas sensors in a preset array in the calibration environment to collect calibration concentration data;

[0008] Step S2: Deploy a mobile robot equipped with a standard gas sensor in the calibration environment and operate it with several operating strategies to collect standard concentration data;

[0009] Step S3: Use the terminal server to train the calibration concentration data and the standard concentration data in a convolutional neural network model to obtain the calibration model of the plurality of calibration gas sensors and calibrate the plurality of calibration gas sensors.

[0010] Step S4: Collect calibrated concentration data using several calibrated gas sensors, and obtain a gas distribution cloud map in the calibration environment using a CFD simulation program in the terminal server. The terminal server generates the calibration accuracy of each calibrated gas sensor based on the gas distribution cloud map and the calibrated concentration data, and determines the calibration effect based on the calibration accuracy to implement several correction measures.

[0011] The temperature and gas concentration of the calibration environment can be adjusted, and the pressure, temperature and humidity of the calibration environment remain constant during a single acquisition. The preset array ensures that the acquisition of adjacent calibration gas sensors with the same distance in both the longitudinal and transverse directions does not interfere with each other. The operation strategy includes the operation path, operation speed and operation posture.

[0012] Furthermore, in step S2, the terminal server determines the initial operating strategy of the mobile robot based on the density of the preset array;

[0013] The initial operating strategy satisfies the following conditions: the mobile robot's operating posture is the sensor convergence posture, the operating speed is the initial speed, the operating path is the shortest path that can pass through the locations of all calibrated gas sensors, and the mobile robot stays within a preset range of each calibrated gas sensor location for a preset duration to collect the standard concentration data.

[0014] The sensor convergence posture satisfies the condition that the standard gas sensor is mounted on the surface of the mobile robot, and the initial velocity is negatively correlated with the density of the preset array and positively correlated with the volume of the calibration environment.

[0015] Furthermore, in step S2, when the mobile robot moves to the preset range of any of the calibrated gas sensors, the mobile robot obtains the gas flow velocity at its location through its onboard flow velocity sensor and determines the gas disturbance characteristics within the preset range.

[0016] If no gas disturbance characteristics are observed within the preset range, the mobile robot collects the standard concentration data for the preset duration.

[0017] The gas disturbance characteristic satisfies the condition that there is an airflow with a flow velocity greater than a preset velocity within the preset range.

[0018] Furthermore, in step S2, when gas disturbance characteristics exist within the preset range, the mobile robot executes a data acquisition stabilization strategy.

[0019] If the gas disturbance characteristics exist within the preset range after the preset time, the terminal server adjusts the mobile robot's operating posture to a sensor protrusion posture and collects the standard concentration data for the preset time.

[0020] The data acquisition stabilization strategy involves the mobile robot stopping operation and waiting for the gas disturbance characteristics to disappear. The sensor protrusion posture involves the standard gas sensor detaching from the surface of the mobile robot and protruding in the direction of the calibrated gas sensor. Furthermore, when the sensor is in the protrusion state, the mobile robot reduces its operating speed according to the protrusion amount of the standard gas sensor.

[0021] Furthermore, in step S4, the terminal server determines the accuracy level of each calibration gas sensor based on the calibration accuracy of the plurality of calibration gas sensors.

[0022] The terminal server controls the mobile robot at a low accuracy level to correct the operation strategy and execute steps S2 and S3.

[0023] The accuracy levels include low accuracy, calibration fault, and high accuracy. The low accuracy level is satisfied that there are calibration gas sensors with calibration accuracy less than the accuracy threshold and the number of calibration gas sensors with calibration accuracy less than the accuracy threshold is less than a preset number. The corrected operation strategy is satisfied that the operating speed is the initial speed and the operating path is the shortest path that can pass through the locations of all calibration gas sensors with accuracy less than the accuracy threshold.

[0024] Further, in step S4, the terminal server enables the pre-robot to run and execute steps S2 to S4 under the initial running strategy at the calibrated fault level, and generates a comparison dataset to determine the cause of the calibrated fault.

[0025] The calibration fault level is satisfied that there are calibration gas sensors with calibration accuracy less than the accuracy threshold and the number of calibration gas sensors with calibration accuracy less than the accuracy threshold is greater than or equal to a preset number. The comparison dataset includes the accuracy of each calibration gas sensor after performing steps S2 to S4. The calibration fault causes include mobile robot faults and preset array deployment errors.

[0026] Furthermore, in step S4, when the terminal server determines that the cause of the fault is a mobile robot fault, it uses the calibration result of the pre-robot as the accurate calibration value.

[0027] The cause of the malfunction is that when the mobile robot malfunctions, the accuracy of the comparison dataset is at either the low accuracy level or the high accuracy level.

[0028] The high accuracy level is achieved when the number of calibrated gas sensors with a calibration accuracy less than the accuracy threshold is 0.

[0029] Furthermore, in step S4, when the terminal server identifies the cause of the fault as an incorrect deployment of the preset array, it adjusts the preset array and the initial operating strategy based on the comparison dataset.

[0030] The cause of the malfunction is that when the mobile robot malfunctions, the accuracy of the comparison dataset is at either the low accuracy level or the high accuracy level.

[0031] On the other hand, the present invention provides a sensor array calibration system for use with a mobile robot, comprising:

[0032] The space module is used to provide a calibration environment for several calibration gas sensors;

[0033] A mobile acquisition module, which is connected to the space module, includes several mobile acquisition units for mobile acquisition of standard concentration data of the several calibration gas sensors;

[0034] The central control module is connected to the space module and the mobile acquisition module respectively, and is used to train a convolutional neural network model based on the calibration concentration data of the plurality of calibration gas sensors and the standard concentration data to obtain the calibration model of the plurality of calibration gas sensors and to calibrate the plurality of calibration gas sensors.

[0035] An encapsulated integration module is connected to the space module, the mobile acquisition module and the central control module respectively, in order to integrate several calibrated gas sensors, the mobile acquisition module and the central control module into a gas concentration monitoring platform for gas concentration monitoring in several application scenarios.

[0036] The calibration environment is characterized by constant pressure, temperature, and humidity, and the temperature and gas concentration are adjustable.

[0037] Furthermore, the mobile acquisition unit is equipped with a standard gas sensor for acquiring the standard concentration data and a flow rate sensor for acquiring the gas flow rate;

[0038] The standard gas sensor has a higher acquisition accuracy than the calibration gas sensor, and the mobile acquisition unit is equipped with several mounting postures for the standard gas sensor.

[0039] The aforementioned mounting postures include a sensor convergence posture in which the standard gas sensor is mounted on the surface of the mobile robot, and a sensor extension posture in which the standard gas sensor detaches from the surface of the mobile robot and extends in the direction of the calibration gas sensor.

[0040] Compared with the prior art, the beneficial effect of the present invention is that the mobile robot equipped with a standard gas sensor overcomes the disadvantages of traditional sensor calibration methods. While maintaining the detection of the fixed calibration gas sensor, it can quickly complete the calibration of the sensor array, thereby effectively improving the calibration efficiency.

[0041] Furthermore, the terminal server of this invention determines the initial operating strategy of the mobile robot based on the density of the preset array and judges the gas disturbance characteristics. Considering that excessive speed during the movement of the mobile robot will affect the detection of gas concentration, and that the higher the density, the more times the robot stops, maintaining a relatively low speed can reduce the impact on the gas while ensuring smooth operation. When there are airflow disturbance characteristics, it indicates that the gas distribution is uneven and will affect the calibration. Executing a stable acquisition strategy can effectively improve the accuracy of data. If the waiting time is too long, it will prolong the calibration cycle and affect the calibration efficiency. By adjusting the running posture, the consistency of the data collected by the standard gas sensor and the calibration gas sensor on the mobile robot can be improved without effectively avoiding the impact on the gas concentration detection, thereby further improving the calibration efficiency.

[0042] Furthermore, the terminal server of this invention trains the calibration concentration data and standard concentration data in a convolutional neural network model to obtain the calibration model of the plurality of calibration gas sensors and calibrates the plurality of calibration gas sensors, thereby further improving the calibration efficiency.

[0043] Furthermore, this invention utilizes a CFD simulation program within a terminal server to obtain a gas distribution cloud map within the calibration environment. The terminal server generates the calibration accuracy of each calibrated gas sensor based on the gas distribution cloud map and the calibrated concentration data, and determines the calibration effect based on the calibration accuracy to execute several correction measures. This effectively ensures the accuracy of the data. Moreover, in the event of a calibration failure, the cause of the calibration failure can be determined through a comparison dataset collected by a pre-built robot, enabling timely correction of the calibration failure and further improving calibration efficiency.

[0044] Furthermore, the temperature and gas concentration of the calibration environment built by this invention can be adjusted. By calibrating the calibration gas sensor under different temperatures and gas concentrations, the versatility of the gas concentration monitoring platform after the packaged integrated module is effectively improved.

[0045] Furthermore, this invention reduces manual operation and eliminates the need to retrieve sensors for calibration in a pressure chamber. The online calibration method shortens the calibration cycle and can quickly address calibration issues such as sensor data drift caused by changes in the external environment, thus ensuring the continuous acquisition of accurate real-time data. Literature indicates that even sensors from the same manufacturer and of the same model can exhibit significant differences in their test results. Therefore, using fixed correction models and coefficients for sensors from the same batch is very crude. This invention effectively addresses the need for repetitive calibration of multiple sensors, significantly shortening the calibration cycle, improving calibration efficiency, and enhancing the ease of use of low-cost multi-sensor systems.

[0046] Furthermore, after being packaged and integrated, this invention can be used in various human factors engineering projects such as air conditioning regulation in large shopping malls and independent temperature and humidity control in factories, effectively reducing the high cost of hardware facilities. Attached Figure Description

[0047] Figure 1 This is a flowchart of the sensor array calibration method for mobile robots according to the present invention;

[0048] Figure 2 This is a preset array arrangement diagram according to an embodiment of the present invention;

[0049] Figure 3 This is a structural block diagram of the sensor array calibration system for mobile robots according to the present invention;

[0050] Figure 4 This is a schematic diagram of data regression prediction based on a BP neural network according to an embodiment of the present invention;

[0051] Figure 5 This is a comparison chart of the test set residual results of an embodiment of the present invention;

[0052] Figure 6 This is a comparison chart of the training set residual results in an embodiment of the present invention;

[0053] Figure 7 This is a block diagram of the BP neural network according to an embodiment of the present invention. Detailed Implementation

[0054] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0055] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0056] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0057] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0058] Please see Figure 1 As shown, it is a flowchart of the sensor array calibration method for mobile robots according to the present invention, including:

[0059] Step S1: Set up a calibration environment and arrange several calibration gas sensors in a preset array in the calibration environment to collect calibration concentration data;

[0060] Step S2: Deploy the mobile robot equipped with a standard gas sensor in the calibration environment and operate it with several operating strategies to collect standard concentration data.

[0061] Step S3: Use the terminal server to train the calibration concentration data and standard concentration data in the convolutional neural network model to obtain the calibration model of several calibration gas sensors and calibrate the several calibration gas sensors.

[0062] Step S4: Collect calibrated concentration data using several calibrated gas sensors, and obtain a gas distribution cloud map in the calibration environment using the CFD simulation program in the terminal server. The terminal server generates the calibration accuracy of each calibrated gas sensor based on the gas distribution cloud map and the calibrated concentration data, and determines the calibration effect based on the calibration accuracy to implement several correction measures.

[0063] The calibration environment's temperature and gas concentration are adjustable, and the calibration environment's pressure, temperature, and humidity remain constant during a single acquisition. The preset array ensures that adjacent calibration gas sensors with the same longitudinal and lateral distances do not interfere with each other's acquisitions. The operating strategy includes the operating path, operating speed, and operating posture.

[0064] It is understandable that the purpose of the preset array deployment method in this embodiment is to simplify the operation path of the mobile robot while avoiding data acquisition interference, thereby reducing the device's processing time. The mobile robot equipped with high-precision sensors overcomes the disadvantages of traditional sensor calibration methods, rapidly completing the calibration of the sensor array while maintaining detection by a fixed calibration gas sensor, thus effectively improving calibration efficiency.

[0065] Specifically, in step S2, the terminal server determines the initial operating strategy of the mobile robot based on the density of the preset array;

[0066] The initial operating strategy satisfies the following conditions: the mobile robot's operating posture is the sensor convergence posture, the operating speed is the initial speed, the operating path is the shortest path that can pass through the locations of all calibrated gas sensors, and the mobile robot stays within a preset range of each calibrated gas sensor location for a preset time to collect standard concentration data.

[0067] The sensor convergence posture meets the standard. The gas sensor is mounted on the surface of the mobile robot. The initial velocity is negatively correlated with the density of the preset array and positively correlated with the volume of the calibration environment.

[0068] Optional, please refer to Figure 2 As shown, it is a preset array arrangement diagram of an embodiment of the present invention. For a number of calibration gas sensors distributed in this array, the preset duration is 10 minutes.

[0069] Specifically, in step S2, when the mobile robot runs to the preset range of any calibrated gas sensor, the mobile robot obtains the gas flow rate at the location of the mobile robot through its onboard flow rate sensor and determines the gas disturbance characteristics within the preset range. When there are no gas disturbance characteristics within the preset range, it starts collecting standard concentration data for a preset duration.

[0070] The gas disturbance characteristics satisfy the condition that there is an airflow with a flow velocity greater than the preset velocity within a preset range.

[0071] Understandably, the preset duration is related to the acquisition capability of the standard gas sensor and the computing power of the terminal server, the preset range is the acquisition range within which the calibrated gas sensor can guarantee its acquisition accuracy, and the preset speed is the minimum speed at which the airflow velocity affects the acquisition.

[0072] Optional, please refer to Figure 2 As shown, it is a preset array arrangement diagram of an embodiment of the present invention. For a plurality of calibration gas sensors distributed in this array, the preset duration is 10 minutes, the preset range is a fan-shaped range with a radius of 0.5m centered on the calibration gas sensor in its maximum acquisition direction, and the preset speed is 0.5m / s.

[0073] Specifically, in step S2, when the mobile robot has gas disturbance characteristics within a preset range, it executes a data acquisition stabilization strategy. If gas disturbance characteristics still exist within a preset range after a preset time, the terminal server adjusts the mobile robot's running posture to a sensor protrusion posture and collects standard concentration data for a preset time.

[0074] The data acquisition stabilization strategy involves the mobile robot stopping and waiting for gas disturbance characteristics to disappear. The sensor protrusion posture involves the standard gas sensor detaching from the mobile robot's surface and protruding towards the calibration gas sensor. While the sensor is protruding, the mobile robot's speed is reduced according to the protrusion amount of the standard gas sensor. The terminal server determines the initial operating strategy of the mobile robot based on the preset array density and judges gas disturbance characteristics. Considering that excessively high speed during mobile robot movement can affect gas concentration detection, and that higher density leads to more robot stops, maintaining a relatively low speed minimizes the impact on the gas while ensuring smooth operation. The presence of airflow disturbance characteristics indicates uneven gas distribution and can affect calibration. Implementing the data acquisition stabilization strategy effectively improves data accuracy. Excessive waiting time prolongs the calibration cycle and affects calibration efficiency. Adjusting the operating posture improves the consistency of data collected by the standard and calibration gas sensors on the mobile robot while effectively avoiding interference with gas concentration detection, further improving calibration efficiency.

[0075] Specifically, in step S4, the terminal server determines the accuracy level of several calibration gas sensors based on the calibration accuracy of several calibration gas sensors.

[0076] The terminal server controls the mobile robot at a low accuracy level to correct the operating strategy and execute steps S2 and S3.

[0077] Accuracy levels include low accuracy, calibration failure, and high accuracy. Low accuracy is defined as the presence of calibration gas sensors with calibration accuracy less than the accuracy threshold and the number of such sensors being less than a preset number. The corrected operating strategy is defined as the operating speed being the initial speed and the operating path being the shortest path that passes through the locations of all calibration gas sensors with accuracy below the accuracy threshold.

[0078] Optionally, the calibration accuracy of each calibration gas sensor is calculated using the following steps:

[0079] Step S001: Calculate the calibration error. Calibration error = |Standard data - Test data|;

[0080] Step S002, calculate the accuracy: Accuracy (%) = (1 - calibration error) x 100%.

[0081] Optional, for Figure 2 The array shown is configured with an accuracy threshold of 95% according to production standards, and a preset quantity of 3.

[0082] Specifically, in step S4, the terminal server enables the pre-robot to run with the initial operation strategy under the calibrated fault level and executes steps S2 to S4, and generates a comparison dataset to determine the cause of the calibrated fault.

[0083] The calibration fault level is defined as having a calibration gas sensor with a calibration accuracy less than the accuracy threshold and the number of calibration gas sensors with a calibration accuracy less than the accuracy threshold is greater than or equal to a preset number. The comparison dataset includes the accuracy of each calibration gas sensor after executing steps S2 to S4. The calibration fault causes include mobile robot faults and preset array layout errors.

[0084] Specifically, in step S4, when the terminal server determines that the cause of the fault is a mobile robot fault, it uses the calibration result of the pre-robot as the accurate calibration value.

[0085] The cause of the malfunction is that the accuracy of the comparison dataset is either low or high when the mobile robot malfunctions.

[0086] The number of calibration gas sensors whose calibration accuracy is less than the accuracy threshold is 0, meeting the high accuracy level requirement.

[0087] Specifically, in step S4, when the terminal server determines that the cause of the fault is an incorrect deployment of the preset array, it adjusts the preset array and the initial operation strategy based on the comparison dataset.

[0088] The cause of the malfunction is that the accuracy of the comparison dataset is either low or high when the mobile robot malfunctions.

[0089] The CFD simulation program in the terminal server is used to obtain the gas distribution cloud map in the calibration environment. The terminal server generates the calibration accuracy of each calibration gas sensor based on the gas distribution cloud map and the concentration data after calibration, and judges the calibration effect based on the calibration accuracy to implement several correction measures. This effectively ensures the accuracy of the data. In addition, when calibration failure occurs, the cause of the calibration failure can be determined by the comparison dataset collected by the robot in advance, and the calibration failure can be corrected in time, thereby further improving the calibration efficiency.

[0090] Please see Figure 3 As shown, it is a structural block diagram of a sensor array calibration system for a mobile robot provided by the present invention, including: a space module, which is used to provide a calibration environment for several calibration gas sensors;

[0091] The mobile acquisition module, which is connected to the space module, includes several mobile acquisition units for mobile acquisition of standard concentration data of several calibration gas sensors;

[0092] The central control module is connected to the space module and the mobile acquisition module respectively. It is used to train the convolutional neural network model based on the calibration concentration data and standard concentration data of several calibration gas sensors to obtain the calibration model of several calibration gas sensors and to calibrate several calibration gas sensors.

[0093] The packaged integration module is connected to the space module, the mobile acquisition module and the central control module respectively, in order to integrate several calibrated gas sensors, the mobile acquisition module and the central control module into a gas concentration monitoring platform for gas concentration monitoring in several application scenarios.

[0094] The temperature and gas concentration of the calibration environment can be adjusted. By calibrating the calibration gas sensor under different temperatures and gas concentrations, the versatility of the gas concentration monitoring platform after the packaged integrated module is effectively improved.

[0095] The calibration environment is characterized by constant pressure, temperature, and humidity, and the temperature and gas concentration can be adjusted.

[0096] Specifically, the mobile acquisition unit is equipped with a standard gas sensor for acquiring standard concentration data and a flow rate sensor for acquiring gas flow rate.

[0097] Among them, the acquisition accuracy of the standard gas sensor is greater than that of the calibrated gas sensor, and the mobile acquisition unit is equipped with several mounting postures of the standard gas sensor.

[0098] Several mounting postures include a sensor convergence posture in which a standard gas sensor is mounted on the surface of a mobile robot, and a sensor protrusion posture in which the standard gas sensor detaches from the surface of the mobile robot and protrudes in the direction of the calibrated gas sensor.

[0099] Example 1: This example uses a CO2 gas concentration sensor. Step S1 includes:

[0100] Step S11: Local area network setup and preset array configuration;

[0101] An ESP8266 NodeMCU development board is used as the server to receive data in the terminal server. The calibration environment is a closed experimental chamber with dimensions of 5.92m × 6.00m × 2.80m.

[0102] The ground of the enclosed experimental chamber was divided into a uniform grid, and nine DNI R sensor nodes were arranged, named node1, node2, node3...node9 respectively. The height of the sensors was 0.7m, which is equivalent to the height of an adult's mouth and nose when sitting at an office. Each sensor was also equipped with an ESP8266NodeMCU development board as a client to request data. The sensors and the development board communicated via serial port. A 5V lithium battery was used to power the development board and the sensors, thus completing the local area setup.

[0103] Step S12: Establish experimental condition groups;

[0104] The sealed experimental chamber was maintained at standard pressure and standard humidity (40%-60%), and five different experimental chamber temperatures were set: 22℃, 24℃, 26℃, 28℃, and 30℃; four different carbon dioxide gas source release concentrations were set: 0L / min, 0.25L / min, 0.6L / min, and 1.0L / min; clean and dry N2 was added at the same time, and different experimental conditions were combined for testing.

[0105] In step S13, the host computer in the terminal server synchronously receives real-time data from each node to complete the setup of the calibration environment.

[0106] Step S2 includes:

[0107] Step S21: The standard gas sensor is calibrated in a constant temperature, constant pressure and constant humidity chamber.

[0108] Step S22: Set the ambient chamber temperature to 20℃ and the gas source concentration to 0L / min. After the fixed sensor array is powered on and the flow field stabilizes, the mobile robot equipped with a standard gas sensor enters the network and collects data for 5 minutes at each node (node1, node2, node3...node9) according to the pre-planned route of the SLAM automatic navigation algorithm.

[0109] During the 5-minute detection period, the data collected includes: ① real-time temperature T per second; ② analog voltage value V received from a fixed sensor once per second; ③ concentration value C received from a standard sensor once per second. Simultaneously, timestamps are set to ensure that the real-time temperature of the node, the data from the moving standard gas sensor, and the corresponding fixed node sensor remain synchronized every second.

[0110] Step S23: Set the ambient chamber temperature (20℃) to remain constant, increase the indoor gas release concentration by 0.25L / min, and follow step S22 to detect at each node 1, node 2... node 9 for 5 minutes and record the data.

[0111] Step S24: Set the ambient chamber temperature (20℃) to remain constant, increase the indoor gas release concentration by 0.6L / min, and follow step S22 to detect at each node (node1, node2...node9) for 5 minutes and record the data.

[0112] Step S25: Set the ambient chamber temperature (20℃) to remain constant, increase the indoor gas release concentration by 1.0L / min, and follow step S22 to detect at each node 1, node 2... node 9 for 5 minutes and record the data.

[0113] Step S26: In the same manner as above, when the ambient temperature of the chamber is kept constant at 22°C, under four different indoor air source release conditions of 0L / min, 0.25L / min, 0.6L / min, and 1.0L / min, the data of node1, node2...node9 are detected and recorded in sequence.

[0114] Step S27: In the same manner as above, at the set ambient chamber temperatures of 24℃, 26℃, 28℃, and 30℃, release four sets of gas sources at 0L / min, 0.25L / min, 0.6L / min, and 1.0L / min respectively, and complete the four sets of tests in the order of the nodes.

[0115] Thus far, the data acquisition work has been completed under 5 different temperature conditions (22℃, 24℃, 26℃, 28℃, 30℃) and corresponding indoor carbon dioxide concentrations, collecting the voltage values ​​V of 9 fixed sensors, the concentration value C of standard sensors, and the real-time temperature T of different nodes.

[0116] Understandably, the reason for collecting real-time temperatures at different nodes every second is that, after setting the indoor temperature, various factors such as the indoor space layout cause the temperatures at different nodes to be inconsistent, even vastly different. To ensure accurate model training later, real-time temperatures at different nodes are collected every second.

[0117] Understandably, setting different indoor concentration ranges is to simulate indoor carbon dioxide concentration levels when there are different numbers of people indoors. This is to prevent the indoor carbon dioxide concentration from becoming too high or too low when there are too many people indoors or when there is no one indoors, which could cause the fixed sensor to drift even after calibration.

[0118] Step S31, please refer to Figure 4The diagram illustrates the data regression prediction based on a BP neural network according to an embodiment of the present invention. In this step, regression training is performed on the data of each node to obtain nine different fixed sensor calibration models. The following will take the calibration of the Node1 fixed sensor as an example to describe the model training process in detail. Step 1: Through the above data acquisition process, the simulated voltage value V of the Node1 fixed sensor, the real-time temperature value T of the node, and the standard sensor concentration value C are obtained under different indoor carbon dioxide concentrations within the temperature fluctuation range of 22-30℃. Step 2: The real-time temperature value T of the node and the fixed sensor node voltage value V are used as inputs, and the standard sensor concentration value C is used as the output; 80% of the dataset is used as the training set and 20% as the test set. After data normalization, the data is substituted into the BP neural network to train the calibration model. The hyperparameters are set as follows: 1000 iterations, error threshold of 1e-6, and learning rate of 0.01. Step 3: The regression comparison chart of the true value and the predicted value and the RMSE value are obtained. Then the weight parameters are adjusted and recalculated until the RMSE reaches the accuracy requirement. At this point, the trained model is the calibration model for the Node1 fixed sensor, completing the calibration of the Node1 fixed sensor.

[0119] Please see Figure 5 and Figure 6 As shown, Figure 5 This is a comparison chart of the test set residual results of an embodiment of the present invention. Figure 6 The image shows a comparison of the training set residuals in this embodiment of the invention. The RMSE of the test set is 0.24908, and the RMSE of the training set is 0.14447. The small RMSE indicates that the trained calibration model performs well.

[0120] Step S32, please refer to Figure 7 The diagram shown is a structural block diagram of the BP neural network according to an embodiment of the present invention. In this step, the sample data, i.e., the calibrated concentration data, is input into the BP neural network for processing. The BP neural network consists of an input layer, several hidden layers, and an output layer. Each layer has several nodes, and nodes in different layers are connected by weights. The BP neural network includes two parts: forward propagation and backward propagation. During forward propagation, the input data is processed layer by layer. When the output layer does not obtain the expected value, backward propagation returns the error signal through the original neuron pathway, modifies the weights of each layer, reduces the error, and then returns to forward propagation. This process is repeated iteratively until the expected value, i.e., the calibrated gas concentration data, is output.

[0121] The terminal server trains the calibration concentration data and standard concentration data in a convolutional neural network model to obtain calibration models for several calibration gas sensors and calibrates these sensors, further improving calibration efficiency.

[0122] Step S4: Collect calibrated concentration data using several calibrated gas sensors, and obtain a gas distribution cloud map in the calibration environment using the CFD simulation program in the terminal server. The terminal server generates the calibration accuracy of each calibrated gas sensor based on the gas distribution cloud map and the calibrated concentration data, and determines the calibration effect based on the calibration accuracy to implement several correction measures.

[0123] Optionally, step S5 is added to this embodiment, which integrates several calibrated gas sensors, a mobile acquisition module, and a central control module into a gas concentration monitoring platform to monitor gas concentration in several application scenarios. By inputting the grid model of the space to be measured, the sensor array model, and the calibration route of the mobile robot into the gas concentration testing platform, it is possible to effectively detect whether the arrangement of the sensor array and the selection of the calibration route of the vehicle-mounted sensor are reasonable under different conditions, thereby achieving the optimal combination. This reduces manual operation and eliminates the need to retrieve the sensors for calibration in a pressure chamber. The online calibration method shortens the calibration cycle and can quickly respond to a series of calibration problems such as sensor data drift caused by changes in the external environment, thereby continuously obtaining accurate real-time data. Literature shows that even for sensors from the same manufacturer and of the same model, there are considerable differences in the data results they detect. Therefore, using a fixed correction model and coefficients for sensors produced in the same batch is very coarse. This invention precisely compensates for the repetitive calibration work of multiple sensors, significantly shortens the calibration cycle, improves calibration efficiency, and enhances the ease of use of low-cost multi-sensor systems. Furthermore, once packaged and integrated, it can be used in various human factors engineering projects such as air conditioning regulation in large shopping malls and independent temperature and humidity control in factories, effectively reducing the high cost of hardware facilities.

[0124] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0125] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention; various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calibrating a sensor array for use with a mobile robot, characterized in that, include: Step S1: Set up a calibration environment and arrange several calibration gas sensors in a preset array in the calibration environment to collect calibration concentration data; Step S2 involves deploying a mobile robot equipped with a standard gas sensor in the calibration environment and operating it using several strategies to collect standard concentration data, wherein... When the mobile robot moves into the preset range of any of the calibrated gas sensors, the mobile robot obtains the gas flow velocity at its location through its onboard flow velocity sensor and determines the gas disturbance characteristics within the preset range. If no gas disturbance features are observed within the preset range, the mobile robot collects the standard concentration data for a preset duration; the gas disturbance features are defined as the presence of an airflow with a flow velocity greater than a preset velocity within the preset range. When the mobile robot exhibits gas disturbance characteristics within a preset range, it executes a data acquisition stabilization strategy. If gas disturbance characteristics still exist within the preset range after a preset time, the terminal server adjusts the mobile robot's operating posture to a sensor protrusion posture and collects standard concentration data for a preset time. Step S3: Use the terminal server to train the calibration concentration data and the standard concentration data in a convolutional neural network model to obtain the calibration model of the plurality of calibration gas sensors and calibrate the plurality of calibration gas sensors. Step S4: Collect calibrated concentration data using several calibrated gas sensors, and obtain a gas distribution cloud map in the calibration environment using a CFD simulation program in the terminal server. The terminal server generates the calibration accuracy of each calibrated gas sensor based on the gas distribution cloud map and the calibrated concentration data, and determines the calibration effect based on the calibration accuracy to implement several correction measures. The temperature and gas concentration of the calibration environment can be adjusted, and the pressure, temperature and humidity of the calibration environment remain constant during a single acquisition. The preset array ensures that the acquisition of adjacent calibration gas sensors with the same distance in both the longitudinal and transverse directions does not interfere with each other. The operation strategy includes the operation path, operation speed and operation posture.

2. The sensor array calibration method for a mobile robot according to claim 1, characterized in that, In step S2, the terminal server determines the initial operating strategy of the mobile robot based on the density of the preset array; The initial operating strategy satisfies the following conditions: the mobile robot's operating posture is the sensor convergence posture, the operating speed is the initial speed, the operating path is the shortest path that can pass through the locations of all calibrated gas sensors, and the mobile robot stays within a preset range of each calibrated gas sensor location for a preset duration to collect the standard concentration data. The sensor convergence posture satisfies the condition that the standard gas sensor is mounted on the surface of the mobile robot, and the initial velocity is negatively correlated with the density of the preset array and positively correlated with the volume of the calibration environment.

3. The sensor array calibration method for a mobile robot according to claim 2, characterized in that, In step S2, The data acquisition stabilization strategy involves the mobile robot stopping operation and waiting for the gas disturbance characteristics to disappear. The sensor protrusion posture involves the standard gas sensor detaching from the surface of the mobile robot and protruding in the direction of the calibrated gas sensor. Furthermore, when the sensor is in the protrusion state, the mobile robot reduces its operating speed according to the protrusion amount of the standard gas sensor.

4. The sensor array calibration method for a mobile robot according to claim 3, characterized in that, In step S4, the terminal server determines the accuracy level of each calibration gas sensor based on the calibration accuracy of the plurality of calibration gas sensors. The terminal server controls the mobile robot at a low accuracy level to correct the operation strategy and execute steps S2 and S3. The accuracy levels include low accuracy, calibration fault, and high accuracy. The low accuracy level is satisfied that there are calibration gas sensors with calibration accuracy less than the accuracy threshold and the number of calibration gas sensors with calibration accuracy less than the accuracy threshold is less than a preset number. The corrected operation strategy is satisfied that the operating speed is the initial speed and the operating path is the shortest path that can pass through the locations of all calibration gas sensors with accuracy less than the accuracy threshold.

5. The sensor array calibration method for a mobile robot according to claim 4, characterized in that, In step S4, the terminal server enables the pre-operation robot to run with the initial operation strategy under the calibrated fault level and executes steps S2 to S4, and generates a comparison dataset to determine the cause of the calibrated fault. The calibration fault level is satisfied that there are calibration gas sensors with calibration accuracy less than the accuracy threshold and the number of calibration gas sensors with calibration accuracy less than the accuracy threshold is greater than or equal to a preset number. The comparison dataset includes the accuracy of each calibration gas sensor after performing steps S2 to S4. The calibration fault causes include mobile robot faults and preset array deployment errors.

6. The sensor array calibration method for a mobile robot according to claim 5, characterized in that, In step S4, when the terminal server determines that the cause of the fault is a mobile robot fault, it uses the calibration result of the pre-robot as the accurate calibration value. The cause of the malfunction is that when the mobile robot malfunctions, the accuracy of the comparison dataset is at either the low accuracy level or the high accuracy level. The high accuracy level is achieved when the number of calibrated gas sensors with a calibration accuracy less than the accuracy threshold is 0.

7. The sensor array calibration method for a mobile robot according to claim 6, characterized in that, In step S4, when the terminal server determines that the cause of the fault is an incorrect deployment of the preset array, it adjusts the preset array and the initial operation strategy based on the comparison dataset. The cause of the malfunction is that when the mobile robot malfunctions, the accuracy of the comparison dataset is at either the low accuracy level or the high accuracy level.

8. A sensor array calibration system for a mobile robot equipped with the method described in any one of claims 1-7, characterized in that, include: The space module is used to provide a calibration environment for several calibration gas sensors; A mobile acquisition module, which is connected to the space module, includes several mobile acquisition units for mobile acquisition of standard concentration data of the several calibration gas sensors; The central control module is connected to the space module and the mobile acquisition module respectively, and is used to train a convolutional neural network model based on the calibration concentration data of the plurality of calibration gas sensors and the standard concentration data to obtain the calibration model of the plurality of calibration gas sensors and to calibrate the plurality of calibration gas sensors. An encapsulated integration module is connected to the space module, the mobile acquisition module and the central control module respectively, in order to integrate several calibrated gas sensors, the mobile acquisition module and the central control module into a gas concentration monitoring platform for gas concentration monitoring in several application scenarios. The calibration environment is characterized by constant pressure, temperature, and humidity, and the temperature and gas concentration are adjustable.

9. The sensor array calibration system for a mobile robot according to claim 8, characterized in that, The mobile acquisition unit is equipped with a standard gas sensor for acquiring the standard concentration data and a flow rate sensor for acquiring the gas flow rate. The standard gas sensor has a higher acquisition accuracy than the calibration gas sensor, and the mobile acquisition unit is equipped with several mounting postures for the standard gas sensor. The aforementioned mounting postures include a sensor convergence posture in which the standard gas sensor is mounted on the surface of the mobile robot, and a sensor protrusion posture in which the standard gas sensor detaches from the surface of the mobile robot and protrudes in the direction of the calibration gas sensor.

Images