A chemical safety inspection system and method based on multi-data fusion

Abstract

The present application relates to the technical field of chemical safety inspection, and particularly relates to a chemical safety inspection system and method based on multi-data fusion, which is applied to a safety inspection robot and comprises the following steps: constructing a mobile platform digital model and a mechanical arm digital model; constructing a corresponding perception device digital model; establishing a communication connection among the safety inspection robot, the perception device and a central server; integrating a perception device data monitoring module one and a state monitoring module one in the mobile platform digital model, and integrating a perception device data monitoring module two and a state monitoring module two in the mechanical arm digital model; monitoring the perception device data through the perception device data monitoring module one and the perception device data monitoring module two; monitoring the mobile platform running state data and the mechanical arm running state data through the state monitoring module one and the state monitoring module two; monitoring the chemical equipment data, the mobile platform running state data and the mechanical arm running state data through the monitoring model, and sending an alarm signal when an abnormality is detected.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of chemical safety inspection, and particularly relates to a chemical safety inspection system and method based on multi-data fusion. BACKGROUND

[0002] The chemical production environment has high requirements for safety monitoring due to its complexity and potential danger. The traditional inspection method relies on manual inspection, which is not only inefficient but also has safety risks. Therefore, many chemical companies use chemical safety inspection robots to replace manual inspection. The chemical safety inspection robot inspects the inspection area according to the preset inspection route, monitors the environmental data and equipment operation data around the chemical equipment, and transmits the data to the monitoring center for processing. However, the traditional inspection system cannot predict the failure of the inspection robot itself, which causes the data transmitted back to be unable to guarantee that it is obtained under the condition that the inspection robot is working normally, affecting the accuracy of monitoring.

[0003] Therefore, we propose a chemical safety inspection system and method based on multi-data fusion. SUMMARY

[0004] The present application is based on multi-data fusion technology, which can monitor the running state of the inspection robot in real time, and at the same time considers the running data of the inspection equipment (safety inspection robot) itself during the monitoring process, optimizes the prediction model, and improves the accuracy of monitoring.

[0005] The technical scheme of the present application is as follows: a chemical safety inspection method based on multi-data fusion, the method comprising:

[0006] The safety inspection robot comprises a mobile platform, a mechanical arm and a plurality of sensing devices, characterized in that the method comprises:

[0007] Obtain the physical data and historical running data of the mobile platform and the mechanical arm, and construct the mobile platform digital model and the mechanical arm digital model respectively;

[0008] Obtain the physical data and historical running data of the plurality of sensing devices, and construct the corresponding sensing device digital model;

[0009] Establish a communication connection between the safety inspection robot, the sensing device and the center server;

[0010] Integrate the sensing device data monitoring module one and the state monitoring module one in the mobile platform digital model, and integrate the sensing device data monitoring module two and the state monitoring module two in the mechanical arm digital model;

[0011] The perception device data monitoring module one and the perception device data monitoring module two respectively acquire the perception device data from the center server, and output the perception device data after preprocessing;

[0012] The state monitoring module one and the state monitoring module two acquire the mobile platform running state data and the mechanical arm running state data from the center server, and output the mobile platform running state data and the mechanical arm running state data after preprocessing;

[0013] The monitoring model monitors the perception device data, the mobile platform running state data and the mechanical arm running state data, and outputs an abnormal alarm signal and regulates the mechanical arm running state.

[0014] Preferably, the perception device comprises an environment perception module and a running state perception module, the environment perception module comprises a sound collection device, a gas concentration sensor, an infrared camera and a high-definition camera, and the running state perception module comprises a distance sensor, a speed sensor and angle encoders one and two installed on the outer side of the mobile platform.

[0015] The angle encoders one and two are respectively used to acquire the horizontal rotation angle data and the pitch angle data of the mechanical arm, and the perception device digital model comprises a sound collection device digital model, a gas concentration sensor digital model, an infrared camera digital model, a high-definition camera digital model, a distance sensor digital model, a speed sensor digital model, an angle encoder one digital model and an angle encoder two digital model.

[0016] The perception device data monitoring module one and the perception device data monitoring module two respectively acquire the perception device data from the center server, comprising the following steps:

[0017] The perception device data monitoring module one and the perception device data monitoring module two are connected with the center server;

[0018] The center server acquires the monitoring data from the environment perception module according to a preset acquisition frequency, adds a collection time stamp one, and constitutes a monitoring data set;

[0019] The perception device data monitoring module one extracts the sound signal and the gas concentration data from the monitoring data set;

[0020] The perception device data monitoring module two extracts the infrared image data and the high-definition image data from the monitoring data set.

[0021] Preferably, the perception device data monitoring module one and the perception device data monitoring module two respectively acquire the perception device data from the center server, comprising the following steps:

[0022] The obtained gas concentration data is de-duplicated, denoised and outlier-removed to form a gas concentration dataset;

[0023] The sound signal is converted to the frequency domain through Fourier transform, and the frequency mean and standard deviation frequency values are extracted to form a voiceprint dataset;

[0024] A plurality of display tags one is established, and the plurality of display tags one is associated with the perception device digital model;

[0025] The infrared image data and the high-definition image data are denoised to form an infrared image dataset and a high-definition image dataset;

[0026] The voiceprint dataset, the gas concentration dataset, the infrared image dataset and the high-definition image dataset are respectively associated with the corresponding display tags one;

[0027] The display tags one enable the perception device digital model to display the corresponding perception device data information.

[0028] Preferably, the mobile platform running state data and the mechanical arm running state data are obtained from the center server by the state monitoring module one and the state monitoring module two, including the following steps:

[0029] The state monitoring module one and the state monitoring module two are connected with the center server;

[0030] The center server obtains the running data from the running state perception module according to a preset acquisition frequency, and adds a second acquisition time stamp to form a state dataset;

[0031] The distance data and the mobile platform movement data are extracted from the state dataset by the state monitoring module one;

[0032] The horizontal rotation angle data of the mechanical arm and the pitch angle data of the mechanical arm are extracted from the state dataset by the state monitoring module two;

[0033] The obtained mobile platform running state data and the mechanical arm running state data are preprocessed and output, including the following steps:

[0034] The obtained distance data, mobile platform movement data, horizontal rotation angle data of the mechanical arm and pitch angle data of the mechanical arm are preprocessed respectively to form a distance dataset, a mobile speed dataset, a horizontal angle dataset and a pitch angle dataset;

[0035] A plurality of display tags two and display tags three are established, the plurality of display tags two are associated with the mobile platform digital model, and the plurality of display tags three are respectively associated with the mechanical arm digital model;

[0036] associate the distance dataset and the moving speed dataset with the corresponding display label two respectively;

[0037] associate the horizontal angle dataset and the pitch angle dataset with the corresponding display label three respectively;

[0038] display the moving speed and the distance data information of the distance to the obstacle of the mobile platform digital model through the display label two;

[0039] display the horizontal angle and the pitch angle data information of the mechanical arm digital model through the display label three.

[0040] Preferably, the monitoring the perception device data, the mobile platform running state data and the mechanical arm running state data through the monitoring model comprises the following steps:

[0041] constructing the monitoring model ; wherein, , respectively represent intercept one, intercept two, intercept three and intercept four, respectively represent error term one, error term two, error term three and error term four; are all multi-dimensional vectors, and respectively represent regression coefficient vector one, regression coefficient vector two, regression coefficient vector three and regression coefficient vector four;

[0042] represents the device failure probability prediction value one, represents the device failure probability prediction value two; represents the mobile platform failure probability prediction value, represents the mechanical arm failure probability prediction value, , , and represent the input vector;

[0043] constructing the sound feature vector set, the gas concentration feature vector set, the mobile platform speed feature vector set and the mechanical arm motion state feature vector set;

[0044] after normalizing the data in the voiceprint key feature vector set, input the data to ; ; respectively represent sound coefficient one and sound coefficient two;

[0045] after normalizing the data in the gas concentration feature vector set, input the data to ; ; respectively represent gas concentration coefficient one and gas concentration coefficient two;

[0046] after normalizing the data in the mobile platform speed feature vector set, input the data to respectively are the first and second speed coefficients;

[0047] The data in the mechanical arm motion state feature vector set is normalized and input to respectively are the first and second angle coefficients;

[0048] Set the device safety probability threshold and the inspection robot safety probability threshold

[0049] If , it is judged that the device has a safety failure, a safety alarm signal is output, and the infrared images and high-definition images 5 seconds before and after the timestamp are extracted from the infrared image data set and the high-definition image data set; and the display color of the sound collection device digital model and the gas concentration sensor digital model is changed;

[0050] If , it is judged that the inspection robot has a failure, a maintenance alarm signal is sent, and the display color of the speed sensor digital model, the first angle encoder digital model, and the second angle encoder digital model is changed; wherein respectively are the first, second, third, and fourth weight values.

[0051] Preferably, the construction of the sound feature vector set, the gas concentration feature vector set, the mobile platform speed feature vector set, and the mechanical arm motion state feature vector set includes the following steps:

[0052] Set the sliding window one;

[0053] Select multiple frequency means and standard deviations from the voiceprint data set through the sliding window one, normalize to form a voiceprint feature vector, and obtain multiple voiceprint feature vectors by moving the sliding window one multiple times, and the multiple voiceprint feature vectors form the sound feature vector set;

[0054] Select multiple gas concentration values from the gas concentration data set through the sliding window one, extract the average gas concentration value and the highest gas concentration value, normalize to form a gas concentration feature vector; obtain multiple gas concentration key feature vectors by moving the sliding window one multiple times, and the multiple gas concentration key feature vectors form the gas concentration feature vector set;

[0055] Set the sliding window two;

[0056] ​​​​​The highest speed and average speed are extracted from the moving speed dataset through sliding window two, and after normalization, they form the moving platform speed feature vector. Multiple moving sliding window one is used to obtain multiple moving platform speed feature vectors, and multiple moving platform speed feature vectors form the moving platform speed feature vector set.

[0057] Multiple horizontal angles are selected from the horizontal angle dataset using sliding window two, and the average rate of change of the horizontal angles is calculated. Multiple pitch angle data are obtained from the pitch angle dataset using sliding window two, and the average rate of change of the pitch angles is calculated. The average rate of change of the horizontal angles and the average rate of change of the pitch angles are normalized to form the motion state vector of the robotic arm.

[0058] Multiple robotic arm motion state vectors are obtained by sliding windows multiple times, and these multiple robotic arm motion state vectors constitute a set of robotic arm motion state feature vectors.

[0059] Preferably, the calculation yields the first The average gas concentration value within each sliding window and highest gas concentration value Multiple moves of the sliding window constitute a set of gas concentration feature vectors:

[0060] ;

[0061] Calculate to obtain the first The average of the frequency mean within a sliding window and the average of the frequency values ​​of the standard deviation Multiple moves of the sliding window constitute a set of sound feature vectors:

[0062] ;

[0063] Calculate to obtain the first The highest speed within the second sliding window and average speed Multiple moves of the sliding window constitute a set of velocity feature vectors for the mobile platform:

[0064] ;

[0065] Calculate to obtain the first Average rate of change of horizontal angle within sliding window two:

[0066] and the average rate of change of pitch angle Multiple moves of the sliding window constitute a set of velocity feature vectors for the mobile platform:

[0067] ;in, the number of times of moving of the first sliding window and the second sliding window;

[0068] , wherein the first horizontal angle value, the first horizontal angle value, the first horizontal angle value, the first horizontal angle value, .

[0069] Preferably, the historical safety alarm signal output data and the historical maintenance alarm signal output data are obtained from the database, the number of times of issuing the historical maintenance alarm signal within 2 minutes before and after the historical safety alarm signal output is counted, and the confidence degree is calculated , which is used for reflecting the probability of causing the device safety alarm due to the failure of the inspection robot, so as to optimize the monitoring model, and the optimized .

[0070] The application further provides a chemical safety inspection system based on multi-data fusion.

[0071] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the chemical safety inspection method based on multi-data fusion.

[0072] The application has the following beneficial effects:

[0073] 1、The application obtains physical data and historical operation data of a mobile platform, a mechanical arm and a plurality of sensing devices, constructs corresponding digital models, and provides support for subsequent monitoring and management, integrates a sensing device data monitoring module and a state monitoring module in the mobile platform digital model and the mechanical arm digital model respectively, improves the maintainability and expandability of the system, and enables each module to work independently and not interfere with each other. The monitoring model is used for monitoring chemical equipment environment data and the state (heat or vibration) of the sensing inspection area, and mobile platform operation state data and mechanical arm operation state data in real time, and outputs an abnormal alarm signal when an abnormality is detected, which helps to discover potential safety hazards or fault problems in the inspection process. According to the output result of the monitoring model, the operation state of the mechanical arm can also be monitored to determine whether the inspection robot is faulty, discover the fault of the inspection robot in time, and avoid affecting the normal inspection function. The action of the mechanical arm can also be remotely adjusted according to actual needs, unnecessary collision, damage or inspection action out of place, so that the camera and the sensor cannot effectively obtain environment and equipment data, so as to improve the inspection efficiency.

[0074] 2、The present application realizes real-time data acquisition of the mobile platform digital model and the mechanical arm digital model by establishing data connection between the perception device data monitoring module one and the perception device data monitoring module two and the center server, ensures timely update of the data, enables the digital model to quickly reflect the environmental change and the equipment state of the inspection area, realizes intuitive display of the data by establishing a plurality of display tags one and associating the tags with the perception device digital model, enables the user to intuitively understand the running state and the monitoring data of each equipment, enhances user experience, and facilitates quick positioning of problems and decision-making.

[0075] 3、In the present application, the data input to the monitoring model includes voiceprint feature data, gas concentration feature data, mobile platform speed feature data and mechanical arm motion state feature data, the data of a plurality of sensors are fused, the association of the voiceprint and the gas concentration feature and the chemical equipment fault is established, the association of the mobile platform speed and the mechanical arm rotation angle change and the mechanical arm state is established, and the association of the inspection robot fault and the equipment safety alarm is established through statistical analysis of historical data, so that the monitoring model is optimized, the optimized monitoring model is obtained, and the accuracy of the early warning of the monitoring model is improved. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 The flowchart of the chemical safety inspection method based on multi-data fusion. DETAILED DESCRIPTION

[0077] The following description is provided to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only examples, and other obvious modifications can be conceived by those skilled in the art. The basic principles of the present application defined in the following description can be applied to other embodiments, modifications, improvements, equivalents and other technical solutions without departing from the spirit and scope of the present application.

[0078] It can be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of one element can be one, and in another embodiment, the number of the element can be multiple, and the term "one" cannot be understood as a limitation on the number.

[0079] Nowadays, chemical safety inspection robots are mostly used instead of manual inspection in chemical plants, the robots perform inspection in corresponding areas of the chemical plant according to a preset inspection route, and the present application is also based on the existing chemical safety inspection robots.

[0080] REFERENCE Figure 1The technical scheme provided by the present application is: a chemical safety inspection method based on multi-data fusion, applied to a safety inspection robot, the safety inspection robot comprising a mobile platform, a mechanical arm and a plurality of sensing devices, the sensing devices comprising an environment sensing module and a running state sensing module, the environment sensing module comprising a sound collecting device, a gas concentration sensor, an infrared camera and a high-definition camera; the running state sensing module comprising a distance sensor, a speed sensor and angle encoders one and two installed on the outer side of the mobile platform;

[0081] The angle encoders one and two are respectively used to acquire horizontal rotation angle data and pitch angle data of the mechanical arm.

[0082] The sensing device digital models comprise a sound collecting device digital model, a gas concentration sensor digital model, an infrared camera digital model, a high-definition camera digital model, a distance sensor digital model, a speed sensor digital model, an angle encoder one digital model and an angle encoder two digital model, all the digital models can be combined with the mobile platform digital model and the mechanical arm digital model according to actual positions on the safety inspection robot, or can be independently displayed without being combined with the mobile platform digital model and the mechanical arm digital model.

[0083] The method comprises the following steps:

[0084] Step one, acquiring physical data and historical running data of the mobile platform and the mechanical arm, and constructing a mobile platform digital model and a mechanical arm digital model respectively;

[0085] Step two, acquiring physical data and historical running data of the plurality of sensing devices, and constructing corresponding sensing device digital models;

[0086] Step three, establishing a communication connection of the safety inspection robot, the sensing devices and a center server;

[0087] Step four, integrating a sensing device data monitoring module one and a state monitoring module one in the mobile platform digital model, and integrating a sensing device data monitoring module two and a state monitoring module two in the mechanical arm digital model;

[0088] Step five, acquiring sensing device data from the center server by the sensing device data monitoring module one and the sensing device data monitoring module two respectively, and outputting the acquired sensing device data after preprocessing, specifically comprising the following steps:

[0089] Establishing a data connection of the sensing device data monitoring module one and the sensing device data monitoring module two and the center server;

[0090] The central server obtains monitoring data from the environment perception module according to a preset acquisition frequency, and adds an acquisition timestamp I to form a monitoring data set;

[0091] The sound signal and gas concentration data are extracted from the monitoring data set by the perception device data monitoring module I;

[0092] The infrared image data and high-definition image data are extracted from the monitoring data set by the perception device data monitoring module II;

[0093] The obtained gas concentration data is processed by removing duplicates, noise reduction, and removing outliers to form a gas concentration data set;

[0094] The sound signal is converted to the frequency domain through Fourier transform, and the frequency mean and standard deviation frequency values are extracted to form a voiceprint data set;

[0095] A plurality of display labels I are established, and the plurality of display labels I are associated with the perception device digital model;

[0096] The infrared image data and high-definition image data are processed by removing noise to form an infrared image data set and a high-definition image data set;

[0097] The voiceprint data set, the gas concentration data set, the infrared image data set, and the high-definition image data set are respectively associated with the corresponding display labels I;

[0098] The perception device digital model can display the corresponding perception device data information through the display labels I.

[0099] In this embodiment, the association of the display labels I with the perception device can be realized through python programming. Specifically, a dictionary named datasets is first created, including different types of perception device data. Then, a dictionary named display_labels is created, mapping each data set type to a corresponding display label I. A function named display_data is then defined, which accepts a display label I as a parameter and retrieves the corresponding data from the datasets dictionary based on the label and prints it out. Finally, when the user selects different display labels I, the display_data function is called to display the corresponding data.

[0100] Step six, obtain the mobile platform running state data and the mechanical arm running state data from the central server through the state monitoring module I and the state monitoring module II, and output the mobile platform running state data and the mechanical arm running state data after preprocessing, specifically including the following steps:

[0101] Establish data connection between the state monitoring module I, the state monitoring module II, and the central server;

[0102] The central server obtains the running data from the running state perception module according to a preset acquisition frequency, and adds an acquisition time stamp II to form a state data set. The distance data and the mobile platform movement data are extracted from the state data set by the state monitoring module I; the horizontal rotation angle data of the mechanical arm and the pitch angle data of the mechanical arm are extracted from the state data set by the state monitoring module II.

[0103] The obtained distance data, mobile platform movement data, horizontal rotation angle data of the mechanical arm and pitch angle data of the mechanical arm are preprocessed respectively to form a distance data set, a mobile speed data set, a horizontal angle data set and a pitch angle data set. A plurality of display labels II and a plurality of display labels III are established, the plurality of display labels II are associated with the mobile platform digital model, and the plurality of display labels III are respectively associated with the mechanical arm digital model. The distance data set and the mobile speed data set are respectively associated with the corresponding display labels II. The horizontal angle data set and the pitch angle data set are respectively associated with the corresponding display labels III. The mobile platform digital model can display the mobile speed and the distance data information of the distance to the obstacle through the display labels II. The mechanical arm digital model can display the horizontal angle and the pitch angle data information through the display labels III.

[0104] Step seven, monitoring the perception device data, the mobile platform running state data and the mechanical arm running state data through the monitoring model, outputting an abnormal alarm signal and regulating the mechanical arm running state. The monitoring model in the embodiment is constructed according to a regression algorithm, and specifically includes the following steps:

[0105] including the following steps:

[0106] Constructing a monitoring model ; wherein, , respectively represent intercept I, intercept II, intercept III and intercept IV, respectively represent error term I, error term II, error term III and error term IV; are all multi-dimensional vectors, respectively representing regression coefficient vector I, regression coefficient vector II, regression coefficient vector III and regression coefficient vector IV;

[0107] It should be noted that the specific training process of the monitoring model in the embodiment can refer to the training process of the existing regression algorithm model, and the training process will not be described here. The following model prediction failure process all uses the trained monitoring model.

[0108] represents a device failure probability prediction value I, represents a device failure probability prediction value II; represents a mobile platform failure probability prediction value, represents a mechanical arm failure probability prediction value, 、 、 and represents an input vector;

[0109] The sound feature vector set, the gas concentration feature vector set, the mobile platform speed feature vector set and the mechanical arm motion state feature vector set are constructed, specifically:

[0110] The method comprises the following steps:

[0111] A sliding window one is set, a plurality of frequency mean values and standard deviations are selected from the voiceprint data set through the sliding window one, and after normalization, a voiceprint feature vector is formed. A plurality of voiceprint feature vectors are obtained by moving the sliding window one multiple times, and the plurality of voiceprint feature vectors form a sound feature vector set. A plurality of gas concentration values are selected from the gas concentration data set through the sliding window one, and the average gas concentration value and the highest gas concentration value are extracted. After normalization, a gas concentration feature vector is formed. A plurality of gas concentration key feature vectors are obtained by moving the sliding window one multiple times, and the plurality of gas concentration key feature vectors form a gas concentration feature vector set.

[0112] A sliding window two is set, the highest speed and the average speed are extracted from the mobile speed data set through the sliding window two, and after normalization, a mobile platform speed feature vector is formed. A plurality of mobile platform speed feature vectors are obtained by moving the sliding window one multiple times, and the plurality of mobile platform speed feature vectors form a mobile platform speed feature vector set. A plurality of horizontal angles are selected from the horizontal angle data set through the sliding window two, and the horizontal angle average change rate is calculated. A plurality of pitch angle data are obtained from the pitch angle data set through the sliding window two, and the pitch angle average change rate is calculated. The horizontal angle average change rate and the pitch angle average change rate are normalized to form a mechanical arm motion state vector. A plurality of mechanical arm motion state vectors are obtained by moving the sliding window two multiple times, and the plurality of mechanical arm motion state vectors form a mechanical arm motion state feature vector set.

[0113] In some preferred embodiments, the average value of the gas concentration values in the first sliding window one is calculated and the highest gas concentration value , and the sliding window one is moved multiple times to form a gas concentration feature vector set:

[0114] ;

[0115] The average value of the frequency mean values in the first sliding window one is calculated and the average value of the standard deviation frequency values , and the sliding window one is moved multiple times to form a sound feature vector set:

[0116] ​​ ;

[0117] Calculate to obtain the first The highest speed within the second sliding window and average speed Multiple moves of the sliding window constitute a set of velocity feature vectors for the mobile platform:

[0118] ;

[0119] Calculate to obtain the first Average rate of change of horizontal angle within sliding window two:

[0120] and the average rate of change of pitch angle Multiple moves of the sliding window constitute a set of velocity feature vectors for the mobile platform:

[0121] ;in, This indicates the number of times the sliding window 1 and the sliding window 2 have moved;

[0122] , ,in Indicates the first A horizontal angle value. Indicates the first One pitch angle value, .

[0123] After normalizing the data in the voiceprint key feature vector set, it is input into... ; ; These are sound coefficient one and sound coefficient two, respectively.

[0124] After normalizing the data in the gas concentration feature vector set, it is input into... ; ; These are gas concentration coefficient one and gas concentration coefficient two, respectively.

[0125] After normalizing the data in the mobile platform speed feature vector set, it is input into... ; ; These are speed coefficient one and speed coefficient two, respectively.

[0126] After normalizing the data in the feature vector set of the robotic arm's motion state, it is input into... ; ; These are angle coefficient one and angle coefficient two, respectively.

[0127] Set device safety probability threshold and the safety probability threshold of the inspection robot ;

[0128] If , it is judged that the device has a safety failure, a safety alarm signal is output, and the infrared image and the high-definition image of the timestamp 5 seconds before and after are extracted from the infrared image data set and the high-definition image data set; and the display color of the sound collection device digital model and the gas concentration sensor digital model is changed; Specifically, the display color of the sound collection device digital model and the gas concentration sensor digital model can be changed by modifying the properties of the related models. For example, the color property of the model can be set to red or other warning color, which is used to remind the staff that the corresponding sensor data is abnormal.

[0129] If , it is judged that the inspection robot has a failure, a maintenance alarm signal is sent, and the display color of the speed sensor digital model, the angle encoder one digital model and the angle encoder two digital model is changed; wherein, weight value one, weight value two, weight value three and weight value four respectively.

[0130] The historical safety alarm signal output data and the historical maintenance alarm signal sending data are obtained from the database, the number of times of sending the historical maintenance alarm signal within 2 minutes before and after the historical safety alarm signal is counted, and the confidence degree is calculated, which is used to reflect the probability of device safety alarm (false alarm) caused by the failure of the inspection robot, for example, the collected data is inaccurate due to the incomplete rotation angle of the mechanical arm, or the collected data is inaccurate due to the inappropriate distance between the mobile platform and the chemical equipment in the inspection area. The monitoring model is optimized, the accuracy of monitoring is improved, and the optimized , the rest of the expressions in the monitoring model remain unchanged.

[0131] The application also provides a chemical safety inspection system based on multi-data fusion, which is used to execute the chemical safety inspection method based on multi-data fusion.

[0132] The application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the chemical safety inspection method based on multi-data fusion.

[0133] The processes described above with reference to the flowcharts can be implemented as computer software programs in accordance with embodiments of the present disclosure. Embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising program code for performing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication section, and / or installed from a detachable medium. When the computer program is executed by a central processing unit (CPU), the above-described functions defined in the methods of the present application are performed. It should be noted that the computer readable medium of the present application can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, but not limited to, be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device. In the present application, the computer readable signal medium can include a data signal carried in a baseband or as part of a carrier wave, in which the computer readable program code is carried. Such a propagated data signal can take many forms, including but not limited to, an electromagnetic signal, an optical signal or any suitable combination of the above. The computer readable signal medium can also be any computer readable medium that can send, propagate or transfer a program for use by or in connection with an instruction execution system, apparatus or device. The program code contained on the computer readable medium can be transmitted by any suitable medium, including but not limited to, wireless, wire, optical cable, RF or the like, or any suitable combination of the above.

[0134] The computer program product of the present application can be a computer program product comprising a computer-readable medium bearing computer program code embodied therein for use with a computer. The computer program code can be code defining and / or implementing the present application. The computer program code can be written in any suitable computer readable programming language. The computer program code can be stored in a computer- readable storage medium, such as, but not limited to, any type of disk including an optical disk, a CD-ROM, a CD-R, a CD-RW, a DVD, a flash memory, a ROM, a RAM, a magnetic disk or hard drive, or any other suitable type of medium including a medium that holds the software for a particular or specialized computing purpose, or any suitable combination of media. The computer program product can be a computer program product distributed to end users, whether as a stand-alone program, as part of a physical system, or as a software download. The computer program product can be distributed on a physical medium, such as, but not limited to, a floppy disk, a CD-ROM, a CD-R, a CD-RW, a DVD, a flash memory, a ROM, a RAM, a magnetic disk or hard drive, or any other suitable type of medium, or any suitable combination of media. The computer program product can be distributed from a program distribution center, either as a tangible medium or via electronic delivery, such as from a Web site via the Internet, or from one computer to another via electronic transfer, such as by e-mail. The computer program product can be distributed in an encrypted manner, such as via encryption or via password protection.

[0135] Those skilled in the art will understand that the application described above and illustrated in the accompanying drawings is presented by way of example only and is not limiting as to the present application. The intent is to cover all modifications and alternatives of the present application falling within the scope of the application.

Claims

1. A chemical safety inspection method based on multi-data fusion, applied to a safety inspection robot, the safety inspection robot comprising a mobile platform, a mechanical arm and a plurality of sensing devices, characterized in that, The method comprises: acquiring physical data and historical operation data of the mobile platform and the mechanical arm, and constructing a digital model of the mobile platform and a digital model of the mechanical arm respectively; acquiring physical data and historical operation data of a plurality of sensing devices, and constructing corresponding digital models of the sensing devices; establishing a communication connection between the safety inspection robot, the sensing devices and the central server; integrating a sensing device data monitoring module one and a state monitoring module one in the digital model of the mobile platform, and integrating a sensing device data monitoring module two and a state monitoring module two in the digital model of the mechanical arm; acquiring sensing device data from the central server through the sensing device data monitoring module one and the sensing device data monitoring module two respectively, and outputting the acquired sensing device data after preprocessing; acquiring mobile platform operation state data and mechanical arm operation state data from the central server through the state monitoring module one and the state monitoring module two respectively, and outputting the acquired mobile platform operation state data and mechanical arm operation state data after preprocessing; The monitoring model is used for monitoring the perception device data, the mobile platform running state data and the mechanical arm running state data, outputting an abnormal alarm signal, and regulating the mechanical arm running state, comprising the following steps ; wherein , respectively represent intercept one, intercept two, intercept three and intercept four, respectively represent error term one, error term two, error term three and error term four; are all multi-dimensional vectors, and respectively represent regression coefficient vector one, regression coefficient vector two, regression coefficient vector three and regression coefficient vector four; represents a device failure probability prediction value one, represents a device failure probability prediction value two; represents a mobile platform failure probability prediction value, represents a robotic arm failure probability prediction value, , , and represent input vectors, respectively; constructing a sound feature vector set, a gas concentration feature vector set, a mobile platform speed feature vector set and a mechanical arm motion state feature vector set; The data in the voiceprint key feature vector set is normalized and input to ; ; respectively are sound coefficient one and sound coefficient two; The data in the gas concentration feature vector set is normalized and input to ; ; respectively are the gas concentration coefficient one and the gas concentration coefficient two; The data in the mobile platform speed feature vector set is normalized and input to ; ; respectively as the first speed coefficient and the second speed coefficient The data in the mechanical arm motion state feature vector set is normalized and input to ; ; respectively, angle coefficient one and angle coefficient two Setting a device security probability threshold and a patrol robot security probability threshold ; If , it is judged that the device has a safety failure, a safety alarm signal is output, the infrared image and the high-definition image of the timestamp 5 seconds before and after are extracted from the infrared image dataset and the high-definition image dataset; and the display color of the sound collection device digital model and the gas concentration sensor digital model is changed; If , it is judged that the inspection robot has a failure, a maintenance alarm signal is sent, and the display color of the speed sensor digital model, the angle encoder one digital model and the angle encoder two digital model is changed; wherein, are weight value one, weight value two, weight value three and weight value four, respectively.

2. The chemical safety inspection method based on multi-data fusion according to claim 1, characterized in that, the sensing device comprises an environment sensing module and an operation state sensing module, the environment sensing module comprises a sound collection device, a gas concentration sensor, an infrared camera and a high-definition camera; the operation state sensing module comprises a distance sensor, a speed sensor and angle encoders one and two installed on the outer side of the mobile platform; the angle encoders one and two are respectively used for acquiring horizontal rotation angle data and pitch angle data of the mechanical arm; the digital model of the sensing device comprises a digital model of the sound collection device, a digital model of the gas concentration sensor, a digital model of the infrared camera, a digital model of the high-definition camera, a digital model of the distance sensor, a digital model of the speed sensor, a digital model of the angle encoder one and a digital model of the angle encoder two; the acquiring of the sensing device data from the central server through the sensing device data monitoring module one and the sensing device data monitoring module two respectively comprises the following steps: establishing a data connection between the sensing device data monitoring module one, the sensing device data monitoring module two and the central server; the central server acquires monitoring data from the environment sensing module according to a preset acquisition frequency, adds a first acquisition time stamp, and constitutes a monitoring data set; the sensing device data monitoring module one extracts sound signals and gas concentration data from the monitoring data set; the sensing device data monitoring module two extracts infrared image data and high-definition image data from the monitoring data set.

3. The chemical safety inspection method based on multi-data fusion according to claim 2, characterized in that, the preprocessing and outputting of the acquired sensing device data comprises the following steps: performing de-duplication, noise reduction and abnormal value removal processing on the acquired gas concentration data, and constituting a gas concentration data set; performing Fourier transform on the sound signals to convert them to the frequency domain, extracting frequency mean and standard deviation frequency values to constitute a voiceprint data set; establishing a plurality of display tags one, and associating the plurality of display tags one with the digital model of the sensing device; performing noise removal processing on the infrared image data and the high-definition image data, and constituting an infrared image data set and a high-definition image data set; The voiceprint dataset, the gas concentration dataset, the infrared image dataset and the high-definition image dataset are respectively associated with corresponding display labels one; The display label one enables the perception device digital model to display corresponding perception device data information.

4. The chemical safety inspection method based on multi-data fusion according to claim 3, characterized in that, The mobile platform running state data and the mechanical arm running state data are obtained from the center server by the state monitoring module one and the state monitoring module two, and the following steps are included: The state monitoring module one, the state monitoring module two and the center server are connected; The center server obtains the running data from the running state perception module according to a preset acquisition frequency, adds a second acquisition time stamp, and constitutes a state dataset; The distance data and the mobile platform movement data are extracted from the state dataset by the state monitoring module one; The horizontal rotation angle data and the pitch angle data of the mechanical arm are extracted from the state dataset by the state monitoring module two; The obtained distance data, mobile platform movement data, horizontal rotation angle data and pitch angle data of the mechanical arm are respectively preprocessed to form a distance dataset, a mobile speed dataset, a horizontal angle dataset and a pitch angle dataset; A plurality of display labels two and display labels three are established, the plurality of display labels two are associated with the mobile platform digital model, and the plurality of display labels three are respectively associated with the mechanical arm digital model; The distance dataset and the mobile speed dataset are respectively associated with corresponding display labels two; The horizontal angle dataset and the pitch angle dataset are respectively associated with corresponding display labels three; The display label two enables the mobile platform digital model to display the mobile speed and the distance data information of the distance to the obstacle; The display label three enables the mechanical arm digital model to display the horizontal angle and the pitch angle data information. The sound feature vector set, the gas concentration feature vector set, the mobile platform speed feature vector set and the mechanical arm motion state feature vector set are constructed, and the following steps are included:

5. The chemical safety inspection method based on multi-data fusion according to claim 4, characterized in that, A sliding window one is set; A plurality of frequency means and standard deviations are selected from the voiceprint dataset by the sliding window one, and a voiceprint feature vector is constituted after normalization; a plurality of voiceprint feature vectors are obtained by moving the sliding window one multiple times, and the plurality of voiceprint feature vectors constitute a sound feature vector set; A plurality of gas concentration values are selected from the gas concentration dataset by the sliding window one, and an average gas concentration value and a highest gas concentration value are extracted; a gas concentration feature vector is constituted after normalization; a plurality of gas concentration key feature vectors are obtained by moving the sliding window one multiple times, and the plurality of gas concentration key feature vectors constitute a gas concentration feature vector set; A sliding window two is set; The highest speed and the average speed are extracted from the mobile speed dataset by the sliding window two, and a mobile platform speed feature vector is constituted after normalization; a plurality of mobile platform speed feature vectors are obtained by moving the sliding window one multiple times, and the plurality of mobile platform speed feature vectors constitute a mobile platform speed feature vector set; ​ The horizontal angle average change rate is calculated by selecting multiple horizontal angles from the horizontal angle data set through the sliding window two; the pitch angle average change rate is calculated by obtaining multiple pitch angle data from the pitch angle data set through the sliding window two; and the horizontal angle average change rate and the pitch angle average change rate are normalized to form a mechanical arm motion state vector; Multiple mechanical arm motion state vectors are obtained through multiple sliding windows two, and the multiple mechanical arm motion state vectors form a mechanical arm motion state feature vector set.

6. The chemical safety inspection method based on multi-data fusion according to claim 5, characterized in that, The average value of the gas concentration values in the first sliding window is calculated and the highest gas concentration value in the first sliding window The first sliding window is moved a plurality of times to form a gas concentration feature vector set ； The average of the mean values of the frequencies within the first sliding window is calculated and the average of the standard deviation frequency values The set of sound feature vectors is formed by multiple moving sliding windows ； The maximum speed and average speed in each sliding window are calculated The moving platform speed feature vector set is constituted by moving the sliding window multiple times​​ ； The horizontal angle average change rate in the first sliding window is calculated as follows: The horizontal angle average change rate in the second sliding window is calculated as follows: and pitch angle average change rate , multiple moving sliding window one constitutes a moving platform speed feature vector set: ; wherein, represents the number of times of movement of the sliding window one and the sliding window two; , wherein denotes the th horizontal angle value, denotes the th elevation angle value, .

7. The chemical safety inspection method based on multi-data fusion according to claim 6, characterized in that, Obtain historical safety alarm signal output data and historical maintenance alarm signal output data from the database, count the number of times of historical maintenance alarm signal output within 2 minutes before and after the historical safety alarm signal output, and calculate the confidence degree , for reflecting the probability of causing the safety alarm of the equipment due to the failure of the inspection robot, so as to optimize the monitoring model, and the optimized , wherein, is a correction coefficient, .

8. A chemical safety inspection system based on multi-data fusion, characterized in that, The system is used to execute the chemical safety inspection method based on multi-data fusion in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the chemical safety inspection method based on multi-data fusion in any one of claims 1-7.

Images