An AI-based Automatic Adjustment System and Method for Multi-Sensor Data

Abstract

The present invention discloses a multi-sensor data automatic adjustment system and method based on artificial intelligence, belonging to the technical field of data analysis. The present invention analyzes and classifies the environmental data required during work, constructs a sensor network combining multiple sensors; analyzes the environmental data in history, and finds the implicit relationships existing among all environmental data; extracts the characteristic data when the device fails, and calculates and judges the fault threshold of the device failure; collects the characteristic data of each sensor, and judges whether there is a fault in the sensors working in real time; when it is judged that the real-time sensor fails and cannot provide the corresponding environmental data, data speculation is carried out using the implicit relationships among the found environmental data to complete the missing environmental data; after supplementing the missing environmental data, the real-time environmental data is adjusted according to the environmental data measured by the sensor network.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and particularly to a multi-sensor data automatic adjustment system and method based on artificial intelligence. Background Art

[0002] With the continuous progress and development of sensor technology, various types of sensors have become more precise, sensitive, and reliable. Sensors can be used to sense various environmental parameters, such as temperature, humidity, pressure, light, etc., so as to provide real-time data feedback. The popularization and application of automatic control and information technology have made the acquisition, processing, and transmission of sensor data more efficient and convenient. Modern automatic control systems usually integrate sensors, controllers, and actuators, and are connected to other systems through networks or wireless communications to achieve remote monitoring and remote control. And the multi-sensor data automatic adjustment system is applied in various aspects, such as: fire protection, lighting, furniture, medical treatment, etc. The sensors used in different application fields are different. How to quickly and accurately construct a sensor network is crucial for the automatic adjustment system; and due to the development of sensor technology, the current automatic adjustment system has a relatively serious dependence on technology. When a sensor fails, the adjustment system will not be able to obtain real-time environmental data and will stop working. It can only work after the maintenance personnel repair the sensor normally; therefore, how to make the adjustment system continue to work when the sensor cannot measure environmental data can greatly improve the working efficiency of the system and reduce the technical dependence of the system. Summary of the Invention

[0003] The purpose of the present invention is to provide a multi-sensor data automatic adjustment system and method based on artificial intelligence to solve the problems raised in the above background art.

[0004] To solve the above technical problems, the present invention provides the following technical solutions:

[0005] A multi-sensor data automatic adjustment method based on artificial intelligence, the method comprising the following steps:

[0006] S100. Analyze and classify the environmental data required during work, and construct a sensor network combining multiple sensors;

[0007] Further, the specific steps for constructing a sensor network combining multiple sensors are as follows:

[0008] S101. Collect the environmental data required during work, extract the units of all environmental data, classify all environmental data according to the units of each environmental data, and select sensors according to the categories of environmental data, and obtain the types of sensors as {C 1 、C 2 、C 3...C n}, C 1 , C 2 , C 3 ...C n represent the selected 1st, 2nd, 3rd... nth sensors, where n is a positive integer;

[0009] S102. The area of the region to be monitored during the measurement work is m, and the measurement range of each sensor is {a 1 , a 2 , a 3 ... a n}; a 1 , a 2 , a 3 ... a n represent the measurement ranges of the 1st, 2nd, 3rd... nth sensors. Calculate the total number of sensors required to construct the sensor network and the placement distance of the sensors. The formula is:

[0010]

[0011] In the formula, S represents the number of sensors required to construct the sensor network, n represents the types of sensors, and the value range of i is from 1 to n; L i represents the relative distance for placing the ith sensor when constructing the sensor network, and π represents the pi. Place the sensors at the relative distances corresponding to each sensor within the working area according to the calculated placement distances of the n sensors to construct the sensor network.

[0012] By classifying the environmental data during work and selecting the types of sensors for measurement, and calculating the number of sensors and the placement distance based on the working area and the sensor measurement range, the measurement of the environmental data of the working environment can be completed with the least number of devices, and the constructed sensor network can comprehensively measure the environmental data;

[0013] S200. Collect the environmental data collected by each sensor in history, analyze the environmental data in history, and find the implicit relationships existing among all the environmental data;

[0014] Furthermore, the specific steps for finding the implicit relationships existing among all the environmental data are:

[0015] S201. Collect the environmental data measured by each sensor during work in history. Assume that there are k pieces of environmental data corresponding to each collected sensor. Extract two pieces of environmental data to draw a scatter plot, and draw scatter plots for all n pieces of environmental data;

[0016] S202. Calculate the regression curves of the two environmental data in each scatter plot. Let the equation of the regression curve be y = a×x + b, where x and y in the equation represent the two environmental data of the abscissa and ordinate in the scatter plot, and a and b represent the slope and intercept of the regression curve in the scatter plot respectively; calculate the values of a and b of the regression curve in each scatter plot using the formula:

[0017]

[0018] b = h_y p -a×h_x p

[0019] In the formula, h_y r represents the environmental data of the r-th ordinate in the scatter plot, h_x r represents the environmental data of the r-th abscissa in the scatter plot, k represents the number of points formed by the environmental data in the scatter plot, h_x p represents the average value of the abscissa environmental data in the scatter plot, h_y p represents the average value of the ordinate environmental data in the scatter plot, and the value of r ranges from 1 to k; r represents any one of the k points in the scatter plot, and calculate for the regression curves of the scatter plots drawn, and obtain the equations of the regression curves in each scatter plot;

[0020] S203. After calculating the equations of the regression curves in each scatter plot, extract the true values of the environmental data and the predicted values on the regression curves in each scatter plot. Let the true value of the point extracted from the scatter plot be (h_x_s r , h_y_s r ), and calculate the effectiveness of the regression curve equation in the scatter plot according to the extracted true values and predicted values. The formula is:

[0021]

[0022] In the formula, F represents the effectiveness of the regression curve equation in each scatter plot, h_x_s r represents the true value of the abscissa environmental data in the scatter plot, h_y_s r represents the true value of the ordinate environmental data in the scatter plot, k represents the number of environmental data points in the scatter plot; calculate the effectiveness of the regression curve equations in the scatter plots, and obtain effectiveness values of representing the effectiveness of the regression curves in the 1st, 2nd, 3rd... scatter plots, where

[0023] S204. Collect the environmental data with known relationships in history, calculate the error between the predicted value and the true value of the environmental data with known relationships as F_y, and use the error of the environmental data with known relationships in history as the threshold for judging the effectiveness of the regression curve. When F > F_y, it is judged that there is no implicit relationship between the two types of environmental data in the scatter plot. When F ≤ F_y, it is judged that there is an implicit relationship between the two types of environmental data in the scatter plot. Finally, the set of environmental data pairs with implicit relationships obtained through judgment is {(h_x, h_y) 1 , (h_x, h_y) 2 , (h_x, h_y) 3 ...(h_x, h_y) f}, (h_x, h_y) 1 , (h_x, h_y) 2 , (h_x, h_y) 3 ...(h_x, h_y) f represents the environmental data pairs with implicit relationships obtained through judgment, and f is a positive integer.

[0024] There may be an unclear relationship in the environmental data measured during work. Based on the value of one type of environmental data, the other type of environmental data can be speculated. Searching for the implicit relationship of environmental data can not only verify the environmental data measured by the sensor, but also use data speculation to supplement the missing environmental data when the sensor fails, ensuring the normal continuous operation of the system;

[0025] S300. Collect the device data when the sensor fails in history, extract the characteristic data when the device fails, and calculate the fault threshold for judging the device fault;

[0026] Further, the specific steps for calculating the characteristic threshold for judging the device fault are as follows:

[0027] S301. Collect the device data values before and after the failure of each type of sensor in history, calculate the difference in device data before and after the failure of each type of sensor, and the formula is:

[0028] E_c = |e - e'|

[0029] In the formula, E_c represents the difference in device data before and after the failure of the sensor in history, e represents the device data value after the failure of the sensor in history, and e' represents the device data value before the failure of the sensor in history; Select the device data with the largest difference as the characteristic data when the failure of each type of sensor occurs;

[0030] S302. Collect the characteristic data when each type of sensor fails in history, and calculate the fault threshold for each type of sensor according to the characteristic data. The formula is:

[0031]

[0032] In the formula, \(G_y\) represents the fault threshold of each sensor, \(B\) represents the number of characteristic data when each collected sensor fails, \(\epsilon\) represents any one of the \(B\) failures of each sensor, and \(T\) ε represents the characteristic data value when each sensor fails, and \(st\) represents the standard deviation of the \(B\) characteristic data when each collected sensor fails; after calculation, the fault thresholds of \(n\) sensors are \(\{G_y\) 1 、\(G_y\) 2 、\(G_y\) 3 ...\(G_y\) n \}, and \(G_y\) 1 、\(G_y\) 2 、\(G_y\) 3 ...\(G_y\) n represent the fault thresholds of the 1st, 2nd, 3rd... \(n\)th sensors in the sensor network.

[0033] S400. When using the constructed sensor network to collect environmental data in real time, measure the characteristic data of each sensor and determine whether there is a fault in the sensors working in real time;

[0034] Furthermore, the specific steps for determining whether there is a fault in the sensors working in real time are as follows:

[0035] S401. The characteristic data of each sensor collected in real time is

[0036]

[0037] which represents the real-time characteristic data of each sensor in the collected sensor network;

[0038] S402. Use the fault threshold of each sensor to judge the potential faults of the sensors working in real time. When , it is determined that there is a fault in the sensors working in real time and a maintenance warning is issued. When , it is determined that there is no fault in the sensors working in real time. Fault judgment is performed on all types of sensors in the sensor network, and the sensors with faults in the sensor network are obtained as \(\{D_g\) 1 、\(D_g\) 2 、\(D_g\) 3 ...\(D_g\) z \}, and \(D_g\) 1 、\(D_g\) 2 、\(D_g\) 3 ...\(D_g\) z represent the 1st, 2nd, 3rd... \(z\)th sensors with faults in real-time work, where \(z\) is a positive integer.

[0039] S500. When it is determined that the real-time sensor fails to provide the corresponding environmental data, use the implicit relationship between the retrieved environmental data to make data conjectures and complete the missing environmental data;

[0040] Further, the specific steps of using the implicit relationship between the retrieved environmental data to make data conjectures are as follows:

[0041] S501. After determining the sensor with a fault during real-time operation, extract the types of environmental data measured by the faulty sensor as the missing environmental data. Let the extracted missing environmental data be {H_z 1 、H_z 2 、H_z 3 ...H_z z}, where H_z 1 、H_z 2 、H_z 3 ...H_z z represent the missing environmental data measured by the 1st, 2nd, 3rd... zth faulty sensors; search for the types of missing environmental data H_z in the set of environmental data pairs with implicit relationships in S204, and extract the environmental data pairs in which the types of environmental data are the same as those of the missing environmental data H_z. The obtained environmental data pairs found for each H_z are

[0042] {(h_x, h_y) 1 、(h_x, h_y) 2 、(h_x, h_y) 3 ...(h_x, h_y) R},

[0043] (h_x, h_y) 1 、(h_x, h_y) 2 、(h_x, h_y) 3 ...(h_x, h_y) R represent the 1st, 2nd, 3rd... Rth environmental data pairs in which the H_z environmental data is found, and R is a positive integer;

[0044] S502. Make data conjectures about the missing environmental data H_z according to the regression curve equation of the extracted environmental data pairs. The formula is:

[0045]

[0046] In the formula, H_Z_S is the value of the missing environmental data obtained through data conjecture, R represents the number of retrieved environmental data pairs, h_x β represents the independent variable in the retrieved environmental function pair, and h_y βThe variable in the environmental data pair to be searched is represented by, β represents any one of the R environmental data pairs, a represents the slope of the regression curve of the environmental data pair to be searched, and b represents the intercept of the regression curve of the environmental data pair to be searched; or means that when the missing environmental data has the same type as the variable h_y in the searched environmental data pair, the formula a×h_x is used β +b; when the missing environmental data has the same type as the independent variable h_x in the searched environmental data pair, the formula

[0047] S503. Data speculation is performed on the missing environmental data measured by the z faulty sensors, and the data values obtained from the data speculation are used to supplement the missing environmental data caused by sensor failures in the sensor network, ultimately making the environmental data measured by the sensor network during real-time operation complete.

[0048] After determining the faulty sensors, the implicit relationship of the environmental data is used to speculate on the missing and unmeasurable environmental data, supplement the missing environmental data, obtain complete environmental data, ensure that the system can continue to work before and after maintenance by the staff, increase the working efficiency of the system, and reduce the technical dependence of the system;

[0049] S600. After supplementing the missing environmental data, the real-time environmental data is adjusted according to the environmental data measured by the sensor network.

[0050] Furthermore, using the complete environmental data measured by the sensor network, the environmental data is adjusted according to the standard value of each type of environmental data set in the work, and the value that each type of environmental data needs to be adjusted is calculated. The formula is: Ad = h_s - h_b, where Ad in the formula represents the value that each type of environmental data needs to be adjusted, h_s represents each type of environmental data in the complete environmental data provided by the sensor network, including the environmental data directly measured by the sensor and the speculated environmental data of the faulty sensor, and h_b represents the standard value of each type of environmental data set in the work; the environmental data is adjusted using the adjustment value Ad.

[0051] An automatic multi-sensor data adjustment system based on artificial intelligence. The automatic multi-sensor data adjustment system includes a data collection module, a sensor network construction module, an environmental data analysis module, a threshold calculation module, a fault detection module, a data speculation module, and a data adjustment module;

[0052] The data collection module is used to collect the environmental data measured by the sensors in history and its own device data;

[0053] The sensor network construction module is used to analyze the environmental data that needs to be measured during work, and calculate the types, quantities, and placement distances of the required sensors;

[0054] The environmental data analysis module is used to analyze the environmental data measured by sensors and determine whether there are implicit relationships between the environmental data;

[0055] The threshold calculation module is used to calculate a fault threshold for judging whether a sensor device has a fault according to the device data of the sensors that failed during operation in history;

[0056] The fault detection module is used to collect the characteristic data of each working sensor in real time and judge whether the sensor has a fault according to the fault threshold and the real-time characteristic data;

[0057] The data conjecture module is used to use the implicit relationship between environmental data to make a data conjecture and supplement the missing environmental data when it is judged that the measurement of environmental data is missing due to a sensor fault;

[0058] The data adjustment module is used to adjust the real-time environmental data by using the complete environmental data measured by the sensor network and the data conjecture.

[0059] The threshold calculation module includes a characteristic data search unit and a threshold calculation unit;

[0060] The characteristic data search unit is used to analyze the device data when the sensor fails in history and search for the characteristic data reflecting the sensor fault;

[0061] The threshold calculation unit is used to calculate a fault threshold for judging whether the sensor has a fault according to the characteristic data of each sensor found.

[0062] The data conjecture module is used to speculate on the missing environmental data according to the implicit relationship between environmental data due to sensor faults, obtain the conjectured environmental data, and supplement it, so that the sensor network delivers complete environmental data to the data adjustment module.

[0063] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0064] 1. By analyzing the working environment, calculating the types, quantities, and placement distances of sensors required, and constructing a sensor network, the present invention realizes the real-time monitoring of environmental data in the working environment, improves work efficiency, and reduces the cost required for measuring environmental data.

[0065] 2. The present invention calculates and analyzes the implicit relationship between environmental data, judges the faults of sensors in the sensor network, conjectures the missing environmental data before the environmental data is missing, supplements the environmental data missing due to sensor faults, ensures the normal operation of the system, improves work efficiency, and reduces the technical dependence on the data adjustment system. Description of the Drawings

[0066] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention.

[0067] In the accompanying drawings:

[0068] Figure 1 is a module distribution diagram of an artificial intelligence-based multi-sensor data automatic adjustment system of the present invention;

[0069] Figure 2 is a step schematic diagram of an artificial intelligence-based multi-sensor data automatic adjustment method of the present invention;

[0070] Figure 3 is a scatter plot of an artificial intelligence-based multi-sensor data automatic adjustment method of the present invention. Specific Embodiments

[0071] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0072] Please refer to Figures 1-3 , the present invention provides a technical solution:

[0073] An artificial intelligence-based multi-sensor data automatic adjustment method, the method includes the following steps:

[0074] S100. Analyze and classify the environmental data required during work, and construct a sensor network combining multiple sensors;

[0075] The specific steps for constructing a sensor network combining multiple sensors are as follows:

[0076] S101. Collect the environmental data required during work, extract the units of all environmental data, classify all environmental data according to the units of each environmental data, and select sensors according to the categories of environmental data, and obtain the types of sensors as {C 1 , C 2 , C 3 ... C n}, C 1 , C 2 , C 3 ... C n represents the 1st, 2nd, 3rd... nth types of sensors selected, and n is a positive integer;

[0077] S102. The area of the region to be monitored during the measurement work is m, and the measurement range of each type of sensor is {a 1 、a 2 、a 3 ...a n}, where a 1 、a 2 、a 3 ...a n represents the measurement ranges of the 1st, 2nd, 3rd... nth types of sensors. Calculate the total number of sensors required to build the sensor network and the sensor placement distance. The formula is:

[0078]

[0079] In the formula, S represents the number of sensors required to build the sensor network, n represents the types of sensors, and the value range of i is from 1 to n; L i represents the relative distance for placing sensors when building the sensor network, and π represents the pi. According to the calculated placement distances of the n types of sensors, place the sensors at the corresponding relative distances for each sensor within the working area range to build the sensor network.

[0080] By classifying the environmental data during work and selecting the types of sensors for measurement, and calculating the number of sensors and the placement distance based on the working area and the sensor measurement range, the measurement of the environmental data of the working environment can be completed with the least number of devices, and the built sensor network can comprehensively measure the environmental data;

[0081] S200. Collect the environmental data collected by each type of sensor in history, analyze the environmental data in history, and find the implicit relationships existing among all the environmental data;

[0082] Specific steps for finding the implicit relationships existing among all the environmental data:

[0083] S201. Collect the environmental data measured by each type of sensor during work in history. Assume that there are k environmental data corresponding to each type of sensor collected. Extract two pieces of environmental data to draw a scatter plot, and draw scatter plots for all n pieces of environmental data;

[0084] S202. Calculate the regression curve of the two pieces of environmental data in each scatter plot. Assume that the equation of the regression curve is y = a × x + b. In the equation, x and y represent the two pieces of environmental data of the abscissa and ordinate in the scatter plot, and a and b represent the slope and intercept of the regression curve in the scatter plot respectively. Use the formula to calculate the a and b values of the regression curve in each scatter plot. The formula is:

[0085]

[0086] b = h_y p -a×h_x p

[0087] In the formula, h_y r represents the environmental data of the r-th ordinate in the scatter plot, and h_x r represents the environmental data of the r-th abscissa in the scatter plot. k represents the number of points formed by the environmental data in the scatter plot, and h_x p represents the average value of the abscissa environmental data in the scatter plot, and h_y p represents the average value of the ordinate environmental data in the scatter plot. The value range of r is from 1 to k; r represents any one of the k points in the scatter plot, and the regression curves in the scatter plots are all calculated to obtain the equations of the regression curves in each scatter plot;

[0088] S203. After calculating the equations of the regression curves in each scatter plot, extract the true values of the environmental data and the predicted values on the regression curves in each scatter plot. Let the true value of the point extracted from the scatter plot be (h_x_s r , h_y_s r ). According to the extracted true values and predicted values, calculate the effectiveness of the regression curve equation in the scatter plot. The formula is:

[0089]

[0090] In the formula, F represents the effectiveness of the regression curve equation in each scatter plot, h_x_s r represents the true value of the abscissa environmental data in the scatter plot, h_y_s r represents the true value of the ordinate environmental data in the scatter plot, and k represents the number of environmental data points in the scatter plot; the effectiveness of the regression curve equations in the scatter plots are all calculated to obtain effectiveness values of representing the effectiveness of the regression curves in the 1st, 2nd, 3rd... scatter plots, where

[0091] is a positive integer; 1 S204. Collect the environmental data with known relationships in history, and calculate the error between the predicted value and the true value of the environmental data with known relationships in history as F_y. Use the error of the environmental data with known relationships in history as the threshold for judging the effectiveness of the regression curve. When F > F_y, it is judged that there is no implicit relationship between the two environmental data in the scatter plot. When F ≤ F_y, it is judged that there is an implicit relationship between the two environmental data in the scatter plot; finally, after judgment, the set of environmental data pairs with implicit relationships obtained is {(h_x, h_y) 2, (h_x, h_y) 3 ...(h_x, h_y) f}

[0092] (h_x, h_y) 1 , (h_x, h_y) 2 , (h_x, h_y) 3 ...(h_x, h_y) f represent environmental data pairs for which a hidden relationship is judged to exist, and f is a positive integer.

[0093] There may be an unobvious relationship in the environmental data measured during operation. Based on the value of one type of environmental data, another type of environmental data can be inferred. Searching for the hidden relationship of environmental data can not only verify the environmental data measured by the sensor, but also use data speculation to supplement missing environmental data when the sensor fails, ensuring the normal continued operation of the system;

[0094] S300. Collect the device data of the sensors that have failed in history, extract the characteristic data when the device fails, and calculate the fault threshold for judging the device failure;

[0095] The specific steps for calculating the characteristic threshold for judging the device failure are as follows:

[0096] S301. Collect the device data values of each type of sensor before and after failure in history, calculate the difference in device data before and after failure of each type of sensor, and the formula is:

[0097] E_c = |e - e'|

[0098] In the formula, E_c represents the difference in device data before and after failure of the sensor in history, e represents the device data value after the sensor fails in history, and e' represents the device data value before the sensor fails in history; select the device data with the largest difference as the characteristic data when each type of sensor fails;

[0099] S302. Collect the characteristic data when each type of sensor fails in history, and calculate the fault threshold for each type of sensor according to the characteristic data. The formula is:

[0100]

[0101] In the formula, G_y represents the fault threshold of each type of sensor, B represents the number of characteristic data collected when each type of sensor fails, ε represents any one of the B failures of each type of sensor, T ε represents the characteristic data value when each type of sensor fails, and st represents the standard deviation of the B characteristic data collected when each type of sensor fails; after calculation, the fault thresholds of n types of sensors are {G_y1 , \(G_y\) 2 , \(G_y\) 3 ... \(G_y\) n}, \(G_y\) 1 , \(G_y\) 2 , \(G_y\) 3 ... \(G_y\) n represent the fault thresholds of the 1st, 2nd, 3rd... nth sensors in the sensor network.

[0102] S400. When using the constructed sensor network to collect environmental data in real time, measure the characteristic data of each sensor and determine whether there is a fault in the sensors working in real time;

[0103] The specific steps for determining whether there is a fault in the sensors working in real time are as follows:

[0104] S401. The characteristic data of each sensor collected in real time is

[0105]

[0106] representing the real-time characteristic data of each sensor in the collected sensor network;

[0107] S402. Use the fault threshold of each sensor to judge the potential faults of the sensors working in real time. When , it is determined that there is a fault in the sensors working in real time and a maintenance warning is issued. When , it is determined that there is no fault in the sensors working in real time. Fault judgment is performed on all types of sensors in the sensor network, and the sensors with faults in the sensor network are obtained as \(\{D_g\) 1 , \(D_g\) 2 , \(D_g\) 3 ... \(D_g\) z \}, \(D_g\) 1 , \(D_g\) 2 , \(D_g\) 3 ... \(D_g\) z representing the 1st, 2nd, 3rd... zth sensors with faults in real-time work, where z is a positive integer.

[0108] S500. When it is determined that a real-time sensor fails and cannot provide the corresponding environmental data, use the implicit relationship between the found environmental data to make data conjectures and complete the missing environmental data;

[0109] The specific steps for making data conjectures using the implicit relationship between the found environmental data are as follows:

[0110] S501. When it is determined that there is a faulty sensor during real-time operation, extract the types of environmental data measured by the faulty sensor as missing environmental data. Let the extracted missing environmental data be {H_z 1 、H_z 2 、H_z 3 ...H_z z}, where H_z 1 、H_z 2 、H_z 3 ...H_z z represent the missing environmental data measured by the 1st, 2nd, 3rd... zth faulty sensors; search for the extracted types of missing environmental data H_z in the set of environmental data pairs with implicit relationships in S204, and extract the environmental data pairs in which the types of environmental data are the same as those of the missing environmental data H_z. The obtained environmental data pairs found for each H_z are

[0111] {(h_x, h_y) 1 、(h_x, h_y) 2 、(h_x, h_y) 3 ...(h_x, h_y) R},

[0112] (h_x, h_y) 1 、(h_x, h_y) 2 、(h_x, h_y) 3 ...(h_x, h_y) R represent the 1st, 2nd, 3rd... Rth environmental data pairs in which the H_z environmental data is found, and R is a positive integer;

[0113] S502. Make a data conjecture for the missing environmental data H_z according to the regression curve equation of the extracted environmental data pairs. The formula is:

[0114]

[0115] In the formula, H_Z_S is the value of the missing environmental data obtained through data conjecture, R represents the number of environmental data pairs found, h_x β represents the independent variable in the found environmental function pair, h_y β represents the variable in the found environmental data pair, β represents any one of the R environmental data pairs, a represents the slope of the regression curve of the found environmental data pair, and b represents the intercept of the regression curve of the found environmental data pair; or means that when the missing environmental data is the same as the variable h_y in the found environmental data pair after searching, use the formula a×h_x β +b; when the missing environmental data is the same as the independent variable h_x in the found environmental data pair after searching, use the formula

[0116] S503. For the missing environmental data measured by the z sensors with faults, data speculation is performed on all of them, and the missing environmental data in the sensor network caused by sensor faults is supplemented by using the data values obtained from the data speculation, so as to finally make the environmental data measured by the sensor network during real-time operation complete.

[0117] After determining the faulty sensors, the implicit relationship of the environmental data is used to speculate on the missing and unmeasurable environmental data, supplement the missing environmental data, obtain complete environmental data, ensure that the system can continue to work before and after the maintenance by the staff, improve the working efficiency of the system, and reduce the technical dependence of the system.

[0118] S600. After supplementing the missing environmental data, the real-time environmental data is adjusted according to the environmental data measured by the sensor network.

[0119] Using the complete environmental data measured by the sensor network, the environmental data is adjusted according to the standard value of each environmental data set in the work, and the value that each environmental data needs to be adjusted is calculated. The formula is: Ad = h_s - h_b, where Ad represents the value that each environmental data needs to be adjusted, h_s represents each environmental data in the complete environmental data provided by the sensor network, including the environmental data directly measured by the sensor and the speculated environmental data of the faulty sensor, and h_b represents the standard value of each environmental data set in the work; the environmental data is adjusted by using the adjustment value Ad.

[0120] An automatic multi-sensor data adjustment system based on artificial intelligence. The multi-sensor data automatic adjustment system includes a data collection module, a sensor network construction module, an environmental data analysis module, a threshold calculation module, a fault detection module, a data speculation module, and a data adjustment module.

[0121] The data collection module is used to collect the environmental data measured by the sensors in history and its own device data.

[0122] The sensor network construction module is used to analyze the environmental data that needs to be measured during work, and calculate the types, quantities, and placement distances of the required sensors.

[0123] The environmental data analysis module is used to analyze the environmental data measured by the sensors to determine whether there is an implicit relationship between the environmental data.

[0124] The threshold calculation module is used to calculate the fault threshold for judging whether there is a fault in the sensor device according to the device data of the sensors that had faults during operation in history.

[0125] The fault detection module is used to collect the characteristic data of each sensor in real time during operation, and determine whether there is a fault in the sensor according to the fault threshold and the real-time characteristic data;

[0126] The data conjecture module is used to utilize the implicit relationship between environmental data to make data conjectures and supplement the missing environmental data when it is determined that the sensor fails and causes the measurement of environmental data to be missing;

[0127] The data adjustment module is used to adjust the real-time environmental data by using the complete environmental data measured by the sensor network and the data conjectures.

[0128] The threshold calculation module includes a characteristic data search unit and a threshold calculation unit;

[0129] The characteristic data search unit is used to analyze the device data when the sensor fails in history and search for the characteristic data reflecting the sensor fault;

[0130] The threshold calculation unit is used to calculate the fault threshold for determining whether the sensor fails according to the characteristic data of each sensor found.

[0131] The data conjecture module is used to speculate on the missing environmental data due to sensor failure according to the implicit relationship between environmental data, obtain the conjectured environmental data, and supplement it so that the sensor network transmits the complete environmental data to the data adjustment module.

[0132] Embodiment 1

[0133] Perform calculations on the automatic intelligent air conditioning system in the house, extract the environmental data that the system needs to measure as temperature and humidity, and the sensors used are temperature sensors and humidity sensors; the working area of the system is 100, and the measurement ranges of the two sensors are 20 and 30 respectively; by calculation and rounding to the nearest integer, the numbers of the two sensors required are 5 and 3 respectively; the placement distances are 5 and 6 respectively; construct a sensor network according to the number of sensors and the placement distances;

[0134] Collect temperature and humidity environmental data, draw a scatter plot, and the temperature-humidity scatter plot is as Figure 3 shown, calculate the regression curve equation as h_y = 1.3786×h_x - 1.8929, and calculate the effectiveness of the temperature-humidity regression curve. The formula is:

[0135]

[0136] According to the formula, the effectiveness of the temperature-humidity regression curve is calculated to be approximately 0.6; the difference between the true value and the predicted value of the regression curve of the environmental data with an implicit relationship known in history is 1; it is determined that there is an implicit relationship between temperature and humidity, and the environmental data pair (temperature, humidity) is obtained;

[0137] Collect the device data when the temperature and humidity sensors in history malfunction. The characteristic data reflecting the malfunction are all the device measurement times. After calculation, the malfunction thresholds are 0.2 s and 0.5 s respectively; collect the real-time characteristic data as 1 s and 0.2 s respectively; determine that the temperature sensor has a malfunction.

[0138] Due to the malfunction of the temperature sensor, the sensor network only measures the humidity data as 25 and cannot measure the temperature data. Extract the environmental data corresponding to the temperature sensor as temperature, search for the temperature in the environmental data pair, and make a data conjecture according to the implicit relationship of the environmental data pair. Since the temperature is the independent variable of the environmental data pair, the formula is as follows:

[0139]

[0140] After calculation, the conjectured temperature data is 19.2. The air-conditioning system obtains the complete environmental data of temperature 19.2 and humidity 25, and adjusts the environment according to the set temperature of 26 and humidity of 20.

[0141] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0142] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for automatic adjustment of multi-sensor data based on artificial intelligence, characterized in that: The method comprises the following steps: S100, analyze and classify environmental data of different application scenarios at work, and build a sensor network combining multiple sensors; S200, collecting environmental data collected by each sensor in history, analyzing the environmental data in history, and finding implicit relationships between all environmental data; Specific steps to find the implicit relationship between all environmental data: S201, collect the environmental data measured by each sensor during operation in the history, assume that there are k environmental data corresponding to each sensor collected, extract two types of environmental data and draw a scatter plot, and draw all n types of environmental data to obtain A scatter plot; S202, calculate the regression curve of the two environmental data in each scatter plot, and assume that the equation of the regression curve is y=a×x+b, where x and y represent the two environmental data of the horizontal and vertical coordinates in the scatter plot, and a and b represent the slope and intercept of the regression curve in the scatter plot, respectively; calculate the a and b values ​​of the regression curve in each scatter plot using the formula, which is: b=h_y p -a×h_x p In the formula, h_y r Represents the environmental data of the rth ordinate in the scatter plot, h_x r represents the environmental data of the rth horizontal axis in the scatter plot, k represents the number of points that constitute the environmental data in the scatter plot, h_x p Represents the average value of the horizontal axis environmental data in the scatter plot, h_y p Represents the average value of the vertical coordinate environmental data in the scatter plot, and the value of r ranges from 1 to k; r represents any point among the k points in the scatter plot. The regression curves in each scatter plot are calculated to obtain the equation of the regression curve in each scatter plot; S203, after calculating the equation of the regression curve in each scatter plot, extract the true value of the environmental data in each scatter plot and the predicted value on the regression curve, and set the true value of the point in the extracted scatter plot to be (h_x_s r , h_y_s r ), the validity of the regression curve equation in the scatter plot is calculated based on the extracted true value and predicted value, and the formula is: In the formula, F represents the validity of the regression curve equation in each scatter plot, h_x_s r Indicates the true value of the horizontal coordinate environmental data in the scatter plot, h_y_s r represents the true value of the vertical coordinate environmental data in the scatter plot, and k represents the number of environmental data points in the scatter plot; The validity of the regression curve equations in each scatter plot is calculated, and we get The effectiveness is Indicates The validity of the regression curve in the scatter plot, is a positive integer; S204, collect environmental data with known relationships in history, calculate the error between the predicted value and the true value of the environmental data with known relationships as F_y, and use the error of the environmental data with known relationships in history as the threshold for judging the validity of the regression curve; when F>F_y, judge that there is no implicit relationship between the two environmental data in the scatter plot; when F≤F_y, judge that there is an implicit relationship between the two environmental data in the scatter plot; finally, the set of environmental data pairs with implicit relationships obtained through judgment is {(h_x, h_y)1, (h_x, h_y)2, (h_x, h_y)3...(h_x, h_y) f }, (h_x, h_y)1, (h_x, h_y)2, (h_x, h_y)3...(h_x, h_y) f It indicates that the environmental data pairs with implicit relationship are judged, and f is a positive integer; S300, collecting historical data of equipment where sensors have failed, extracting characteristic data when equipment fails, and calculating a fault threshold for determining equipment failure; S400, when collecting environmental data in real time using the constructed sensor network, measuring characteristic data of each sensor, and determining whether the sensor working in real time has a fault; S500, when it is determined that the real-time sensor fails and cannot provide corresponding environmental data, the implicit relationship between the searched environmental data is used to make data guesses to complete the missing environmental data; The specific steps of using the implicit relationship between the searched environmental data to make data guesses are: S501, when it is determined that a sensor with a fault in real-time operation is found, the type of environmental data measured by the sensor with the fault is extracted as missing environmental data, and the extracted missing environmental data is assumed to be {H_z1, H_z2, H_z3...H_z z }, H_z1, H_z2, H_z3...H_z z Indicates the missing environmental data measured by the 1st, 2nd, 3rd...zth faulty sensors; the extracted missing environmental data type H_z is searched in the set of environmental data pairs with implicit relationship in S204, and the environmental data pairs with the same type as the missing environmental data H_z are extracted, and the environmental data pairs found for each H_z are {(h_x, h_y)1, (h_x, h_y)2, (h_x, h_y)3...(h_x, h_y) R }, (h_x, h_y)1, (h_x, h_y)2, (h_x, h_y)3...(h_x, h_y) R It means that the 1st, 2nd, 3rd...Rth environmental data pairs with H_z environmental data are found, where R is a positive integer; S502, make a data guess for the missing environmental data H_z according to the regression curve equation of the extracted environmental data pair, the formula is: In the formula, H_Z_S is the missing environmental data value obtained through data conjecture, R represents the number of environmental data pairs to be found, and h_x β Represents the independent variable in the environment function pair to be searched, h_y β represents the variable in the searched environmental data pair, β represents any one of the R environmental data pairs, a represents the slope of the regression curve of the searched environmental data pair, and b represents the intercept of the regression curve of the searched environmental data pair; or represents that when the missing environmental data is the same as the variable h_y in the environmental data pair after the search, the formula a×h_x is used. β +b; when the missing environmental data is found to be the same type as the independent variable h_x in the environmental data pair, use the formula S503, performing data guessing on the missing environmental data measured by the z sensors with faults, and using the data values ​​of the data guessing to supplement the missing environmental data caused by the sensor faults in the sensor network, so as to finally complete the environmental data measured by the sensor network in real-time operation; S600: After supplementing the missing environmental data, adjust the real-time environmental data according to the environmental data measured by the sensor network.

2. The method for automatic adjustment of multi-sensor data based on artificial intelligence according to claim 1, characterized in that: The specific steps of constructing a sensor network combining multiple sensors in S100 are: S101, collect the environmental data needed for work, extract the units of all environmental data, classify all environmental data according to the units of each environmental data, select sensors according to the categories of environmental data, and obtain the types of sensors {C1, C2, C3...C n }, C1, C2, C3...C n Indicates the 1st, 2nd, 3rd...nth type of sensor selected, where n is a positive integer; S102, the area to be monitored during measurement is m, and the measurement range of each sensor is {a1, a2, a3...a n }, a1, a2, a3...a n Expressed as the measurement range of the 1st, 2nd, 3rd...nth sensors, the total number of sensors and the sensor placement distance required to build a sensor network are calculated using the formula: In the formula, S represents the number of sensors required to build a sensor network, n represents the type of sensor, and i ranges from 1 to n; L i It represents the relative distance of placing the i-th sensor when building a sensor network, and π represents the ratio of pi. According to the calculated placement distances of the n sensors, sensors are placed at the relative distance corresponding to each sensor within the working area to build a sensor network.

3. The method for automatic adjustment of multi-sensor data based on artificial intelligence according to claim 1, characterized in that: The specific steps of calculating the characteristic threshold for determining device failure in S300 are: S301, collecting the device data values ​​before and after each sensor fails in history, and calculating the difference between the device data before and after each sensor fails, the formula is: E_c=|e-e'| In the formula, E_c represents the difference between the device data before and after the sensor fails in history, e represents the device data value after the sensor fails in history, and e' represents the device data value before the sensor fails in history; the device data with the largest difference is selected as the characteristic data when the fault occurs for each sensor; S302, collecting characteristic data of each sensor when it fails in history, and calculating the fault threshold of each sensor according to the characteristic data, the formula is: In the formula, G_y represents the fault threshold of each sensor, B represents the number of characteristic data collected when each sensor fails, ε represents any one of the B failures of each sensor, and T ε represents the characteristic data value when each sensor fails, st represents the standard deviation of B characteristic data collected when each sensor fails; after calculation, the fault thresholds of n sensors are {G_y1, G_y2, G_y3...G_y n }, G_y1, G_y2, G_y3...G_y n Represents the fault threshold of the 1st, 2nd, 3rd, ...nth sensors in the sensor network.

4. The method for automatic adjustment of multi-sensor data based on artificial intelligence according to claim 1, characterized in that: The specific steps of determining whether the sensor working in real time has a fault in S400 are: S401, collect characteristic data of each sensor in real time: Represents the real-time characteristic data collected by each sensor in the sensor network; S402, using the fault threshold of each sensor to determine the potential faults of the sensors working in real time, When the sensor is faulty, it will issue a maintenance warning. When the real-time working sensor is judged to be free of faults, all sensors in the sensor network are judged to be faulty, and the faulty sensors in the sensor network are obtained as {D_g1, D_g2, D_g3...D_g z }, D_g1, D_g2, D_g3...D_g z It indicates the 1st, 2nd, 3rd, ... zth sensors that have faults in real-time operation, where z is a positive integer.

5. The method for automatic adjustment of multi-sensor data based on artificial intelligence according to claim 1, characterized in that: In S600, the complete environmental data measured by the sensor network is used to adjust the environmental data according to the standard value of each environmental data set in the work, and the value that needs to be adjusted for each environmental data is calculated. The formula is: Ad=h_s-h_b, where Ad represents the value that needs to be adjusted for each environmental data, h_s represents each environmental data in the complete environmental data provided by the sensor network, including environmental data directly measured by the sensor and guessed environmental data of the faulty sensor, and h_b represents the standard value of each environmental data set in the work; the environmental data is adjusted using the adjustment value Ad.

6. An artificial intelligence-based multi-sensor data automatic adjustment system using an artificial intelligence-based multi-sensor data automatic adjustment method according to any one of claims 1 to 5, characterized in that: The multi-sensor data automatic adjustment system includes a data collection module, a sensor network construction module, an environmental data analysis module, a threshold calculation module, a fault detection module, a data conjecture module and a data adjustment module; The data collection module is used to collect environmental data measured by sensors in history and its own equipment data; The sensor network building module is used to analyze the environmental data that needs to be measured during work and calculate the type, number and placement distance of the required sensors; The environmental data analysis module is used to analyze the environmental data measured by the sensor to determine whether there is an implicit relationship between the environmental data; The threshold calculation module is used to calculate the fault threshold for determining whether the sensor device has a fault based on the device data of the fault that occurred when the sensor was working in the history; The fault detection module is used to collect the characteristic data of each sensor in operation in real time, and determine whether the sensor has a fault according to the fault threshold and the real-time characteristic data; The data guessing module is used to make data guesses by using the implicit relationship between environmental data to supplement the missing environmental data when it is determined that the sensor fails and the measured environmental data is missing; The data adjustment module is used to adjust the real-time environmental data using the complete environmental data measured by the sensor network and data guessed.

7. The multi-sensor data automatic adjustment system based on artificial intelligence according to claim 6 is characterized in that: The threshold calculation module includes a feature data search unit and a threshold calculation unit; The characteristic data search unit is used to analyze the device data when the sensor fails in history and search for characteristic data reflecting the sensor failure; The threshold value calculation unit is used to calculate a fault threshold value for determining whether a sensor fails according to the characteristic data of each sensor found.

8. The multi-sensor data automatic adjustment system based on artificial intelligence according to claim 6 is characterized by: The data guessing module is used to guess the missing environmental data due to sensor failure based on the implicit relationship between environmental data, obtain the guessed environmental data, and supplement it so that the sensor network transmits complete environmental data to the data adjustment module.

Images