Fiber cloth cover plasterboard production whole process management and control method and system

Abstract

The invention relates to the technical field of production flow management and control, and particularly discloses a fiber cloth cover gypsum board production whole flow management and control method and system, and the method comprises the steps: obtaining a production flow chart, querying the participants of all flows in the production flow chart, and determining a monitor according to the participants; receiving production data acquired by the monitor in a preset time range, identifying the production data, and determining the abnormality of each participant; and counting the anomaly degree of each participant to obtain an abnormal array, and positioning an abnormal subject according to the abnormal array. The intelligent equipment is inserted into the production line based on the production flow chart, the data in the production process are obtained in real time according to the intelligent equipment, the data are analyzed, the state of each piece of production equipment is determined, then the abnormal main body is determined, workers only need to set some parameters in advance in the process, manual participation is not needed in actual use, and the working efficiency is improved. The automation degree is high, and the efficiency is high.

Description

Technical Field

[0001] The present invention relates to the technical field of production process control, and specifically to a full-process control method and system for the production of fiber cloth-faced gypsum boards. Background Art

[0002] At present, certain progress has been made in the production technology of fiber cloth-faced gypsum boards, but there are still some bottlenecks in the quality control of the entire production process. For example, the automation level of production equipment and the quality inspection standards need to be further improved. The existence of these problems not only affects the stability of product quality but also restricts the improvement of production efficiency. The key to these problems actually lies in the data collection and analysis process. Most of the existing data collection and analysis solutions are independent detection solutions for individual equipment, where staff members individually control each production equipment. This method is feasible and convenient for defining responsibilities, but the efficiency is not high and requires a large number of personnel to participate. Excessive personnel participation actually also affects the automation level of the entire production line. Therefore, how to provide a quality control solution for the entire production process based on intelligent devices is the technical problem that the present invention aims to solve. Summary of the Invention

[0003] The purpose of the present invention is to provide a full-process control method and system for the production of fiber cloth-faced gypsum boards to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides the following technical solutions:

[0005] A full-process control method for the production of fiber cloth-faced gypsum boards, the method comprising:

[0006] Obtain a production flow chart, query the participating entities in each process of the production flow chart, and determine a monitor according to the participating entities; the monitor includes at least a sensor and a camera;

[0007] Receive production data obtained by the monitor within a preset time range, identify the production data, and determine the abnormality degree of each participating entity;

[0008] Statistically analyze the abnormality degree of each participating entity to obtain an abnormality array, and locate the abnormal entity according to the abnormality array;

[0009] Among them, the abnormality degree of the participating entity is jointly determined by a sensing abnormality degree and a visual abnormality degree. The sensing abnormality degree is calculated from the data obtained by the sensor, and the visual abnormality degree is calculated from the data obtained by the camera.

[0010] As a further solution of the present invention: the step of obtaining a production flow chart, querying the participating entities in each process of the production flow chart, and determining a monitor according to the participating entities includes:

[0011] Obtain the production flow chart and query the participating entities of each process in the production flow chart;

[0012] Query the sensors of each participating entity, regularly obtain the monitoring data of each sensor within a preset test period, and determine the information volume of the monitor according to the monitoring data;

[0013] Select sensors according to the information volume, and determine the total information volume of each participating entity according to the selection process;

[0014] Insert cameras between adjacent participating entities according to the total information volume, and establish a corresponding relationship between the cameras and the participating entities; wherein, the corresponding relationship is: the camera corresponds to the previous participating entity among the adjacent participating entities;

[0015] The calculation process of the information volume is: I = ασ; where I is the information volume, α is a preset coefficient, and σ is the standard deviation of the monitoring data within the test period;

[0016] The content of selecting sensors according to the information volume and determining the total information volume of each participating entity according to the selection process includes:

[0017] Determine the selection probability of each sensor according to the information volume, randomly select sensors according to the selection probability until at least one sensor in each participating entity is selected;

[0018] Count the selected sensors of each participating entity, sum up the information volume, and obtain the total information volume;

[0019] Among them, the calculation process of the selection probability is:

[0020] where P i is the selection probability of the i-th sensor, T i is the eigenvalue of the i-th sensor, β is a preset correction coefficient, I i is the information volume of the i-th sensor, N is the total number of unselected sensors, M is the total number of selected sensors, I j is the information volume of the j-th selected sensor, and Δ represents the difference in the entity numbers of the j-th selected sensor and the i-th sensor. When two sensors belong to the same participating entity, the difference in entity numbers is zero; the numbers of the sensors and the entities are all known data and are determined by the production flow chart in the preprocessing stage.

[0021] As a further solution of the present invention: the steps of receiving the production data within a preset time range obtained by the monitor, identifying the production data, and determining the abnormality degree of each participating entity include:

[0022] Receive the sensing data containing the first identity tag uploaded by all sensors, and count the sensing data according to the identity tag to obtain the data sequence corresponding to each sensor; the first identity tag is used to characterize which sensor uploads the sensing data;

[0023] Read the data sequences of each sensor of the same participating entity, perform Fourier transform on each data sequence to obtain a frequency domain graph;

[0024] Determine the sensing abnormality degree of the participating entity according to the frequency domain graph;

[0025] Receive the images containing the second identity tag uploaded by all cameras, query the participating entity corresponding to the camera according to the position tag, read the sensing abnormality degree of the participating entity, and determine the recognition accuracy of the image; the recognition accuracy is used to adjust the amount of recognition resources; the second identity tag is used to characterize which camera uploads the image;

[0026] Based on the recognition accuracy, recognize the image to determine the visual abnormality degree of the participating entity;

[0027] Determine the abnormality degree of the participating entity according to the sensing abnormality degree and the visual abnormality degree.

[0028] As a further solution of the present invention: the step of determining the sensing abnormality degree of the participating entity according to the frequency domain graph includes:

[0029] Compare the frequency domain graphs of different sensors of the same participating entity, extract the characteristic frequency domain graph, and synchronously determine the fidelity rate of the characteristic frequency domain graph;

[0030] Input the characteristic frequency domain graph into the trained frequency domain recognition model, output the stability, and correct the stability according to the fidelity rate;

[0031] Calculate the sensing abnormality degree according to the corrected stability, and the sensing abnormality degree is inversely proportional to the corrected stability;

[0032] The step of recognizing the image based on the recognition accuracy to determine the visual abnormality degree of the participating entity includes:

[0033] Sequentially select convolution kernels in the preset convolution kernel table according to the recognition accuracy, and perform matching recognition on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and outlier items; each data item in the convolution kernel table is arranged in descending order based on the outlier value;

[0034] Query the outlier value corresponding to the convolution kernel with successful matching recognition, and accumulate all the outlier values obtained by matching;

[0035] Determine the visual abnormality degree of the participating entity according to the accumulated outlier values;

[0036] The steps of comparing the frequency-domain diagrams of different sensors of the same participating entity, extracting the characteristic frequency-domain diagram, and synchronously determining the fidelity rate of the characteristic frequency-domain diagram include:

[0037] Query all sensors of the same participating entity. For any sensor, compare its frequency-domain diagram with the frequency-domain diagrams of other sensors of the same participating entity to obtain the similarity. Calculate the average value of the similarities to obtain the average similarity. Select the frequency-domain diagram of the sensor with the maximum average similarity as the characteristic frequency-domain diagram, and use the average similarity as the fidelity rate.

[0038] As a further solution of the present invention: The steps of receiving the production data within a preset time range obtained by the monitor, identifying the production data, and determining the abnormality degree of each participating entity further include:

[0039] Adjust the image acquisition frequency of the corresponding camera according to the abnormality degree of the participating entity.

[0040] As a further solution of the present invention: The steps of statistically calculating the abnormality degree of each participating entity to obtain an abnormality array and positioning the abnormal entity according to the abnormality array include:

[0041] Query the serial number of the participating entity in the production process;

[0042] Read the abnormality degree of the participating entity. Taking the serial number as the independent variable and the abnormality degree as the dependent variable, determine the abnormality curve and curve function of the production model;

[0043] Calculate the derivative function of the curve function, intercept the derivative function according to a preset threshold line, and locate the abnormal entity according to the interception result.

[0044] The technical solution of the present invention also provides a full-process control system for the production of fiber cloth-faced gypsum boards. The system includes:

[0045] A monitor determination module, configured to obtain a production flow chart, query the participating entities in each process of the production flow chart, and determine a monitor according to the participating entities; the monitor includes at least a sensor and a camera;

[0046] An abnormality degree calculation module, configured to receive the production data within a preset time range obtained by the monitor, identify the production data, and determine the abnormality degree of each participating entity;

[0047] An abnormal entity positioning module, configured to statistically calculate the abnormality degree of each participating entity to obtain an abnormality array, and locate the abnormal entity according to the abnormality array;

[0048] Among them, the abnormality degree of the participating entity is jointly determined by the sensing abnormality degree and the visual abnormality degree. The sensing abnormality degree is calculated from the data obtained by the sensor, and the visual abnormality degree is calculated from the data obtained by the camera.

[0049] As a further solution of the present invention: the monitor determination module includes:

[0050] A participating entity query unit, configured to obtain a production flow chart and query the participating entities of each process in the production flow chart;

[0051] An information quantity calculation unit, configured to query the sensors of each participating entity, regularly obtain the monitoring data of each sensor within a preset test period, and determine the information quantity of the monitor according to the monitoring data;

[0052] A sensor selection unit, configured to select sensors according to the information quantity and determine the total information quantity of each participating entity according to the selection process;

[0053] A camera selection unit, configured to insert cameras between adjacent participating entities according to the total information quantity and establish a corresponding relationship between the cameras and the participating entities; wherein, the corresponding relationship is: the camera is corresponding to the previous participating entity among the adjacent participating entities;

[0054] The calculation process of the information quantity is: I = ασ; where I is the information quantity, α is a preset coefficient, and σ is the standard deviation of the monitoring data within the test period;

[0055] The content of selecting sensors according to the information quantity and determining the total information quantity of each participating entity according to the selection process includes:

[0056] Determine the selection probability of each sensor according to the information quantity, randomly select sensors according to the selection probability until at least one sensor in each participating entity is selected;

[0057] Count the selected sensors of each participating entity, sum up the information quantity to obtain the total information quantity;

[0058] Wherein, the calculation process of the selection probability is:

[0059] In the formula, P i is the selection probability of the i-th sensor, T i is the eigenvalue of the i-th sensor, β is a preset correction coefficient, I i is the information quantity of the i-th sensor, N is the total number of unselected sensors, M is the total number of selected sensors, I j is the information quantity of the j-th selected sensor, and Δ represents the difference in the entity numbers of the j-th selected sensor and the i-th sensor. When two sensors belong to the same participating entity, the difference in entity numbers is zero; the numbers of the sensors and the entities are both known data and are determined by the production flow chart in the preprocessing stage.

[0060] As a further solution of the present invention: The abnormality calculation module includes:

[0061] A data sequence query unit, configured to receive the sensing data containing the first identity tag uploaded by all sensors, count the sensing data according to the identity tag, and obtain the data sequence corresponding to each sensor; the first identity tag is used to represent which sensor uploads the sensing data;

[0062] A frequency domain graph generation unit, configured to read the data sequences of each sensor of the same participating entity, perform Fourier transform on each data sequence, and obtain a frequency domain graph;

[0063] A first calculation unit, configured to determine the sensing abnormality degree of the participating entity according to the frequency domain graph;

[0064] An identification accuracy determination unit, configured to receive the images containing the second identity tag uploaded by all cameras, query the participating entity corresponding to the camera according to the position tag, read the sensing abnormality degree of the participating entity, and determine the identification accuracy of the image; the identification accuracy is used to adjust the amount of identification resources; the second identity tag is used to represent which camera uploads the image;

[0065] A second calculation unit, configured to identify the image based on the identification accuracy and determine the visual abnormality degree of the participating entity;

[0066] A comprehensive calculation unit, configured to determine the abnormality degree of the participating entity according to the sensing abnormality degree and the visual abnormality degree.

[0067] As a further solution of the present invention: The content of determining the sensing abnormality degree of the participating entity according to the frequency domain graph includes:

[0068] Compare the frequency domain graphs of different sensors of the same participating entity, extract the characteristic frequency domain graph, and synchronously determine the fidelity rate of the characteristic frequency domain graph;

[0069] Input the characteristic frequency domain graph into the trained frequency domain recognition model, output the stability, and correct the stability according to the fidelity rate;

[0070] Calculate the sensing abnormality degree according to the corrected stability, and the sensing abnormality degree is inversely proportional to the corrected stability;

[0071] The content of identifying the image based on the identification accuracy and determining the visual abnormality degree of the participating entity includes:

[0072] Sequentially select convolution kernels in a preset convolution kernel table according to the identification accuracy, and perform matching identification on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and outlier items; each data item in the convolution kernel table is arranged in descending order based on the outlier value;

[0073] Query the outliers corresponding to the convolution kernels with successful matching recognition, and accumulate all the outliers obtained by matching.

[0074] Determine the visual abnormality degree of the participating entity according to the accumulated outliers.

[0075] The steps of comparing the frequency domain diagrams of different sensors of the same participating entity, extracting the characteristic frequency domain diagram, and synchronously determining the fidelity rate of the characteristic frequency domain diagram include:

[0076] Query all sensors of the same participating entity. For any sensor, compare its frequency domain diagram with the frequency domain diagrams of other sensors of the same participating entity to obtain the similarity; calculate the average value of the similarities to obtain the average similarity, and select the frequency domain diagram of the sensor with the largest average similarity as the characteristic frequency domain diagram, and use the average similarity as the fidelity rate.

[0077] Compared with the prior art, the beneficial effects of the present invention are:

[0078] The present invention inserts intelligent devices into the production line based on the production flow chart, obtains data in the production process in real time according to the intelligent devices, analyzes the data to determine the states of various production devices, and then determines the abnormal entities. This process only requires staff to preset some parameters in advance. During actual use, no manual participation is required, and the degree of automation is high and the efficiency is relatively high.

[0079] The present invention establishes a real-time monitoring and data analysis system, monitors and collects data in real time for each link in the production process. Through data analysis, abnormal situations in the production process are discovered in time, and then uploaded to the master control terminal, and the master control terminal takes corresponding corrective measures to ensure the stability of the production process and the consistency of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0081] Figure 1 It is a flow block diagram of the full-process control method for the production of fiber cloth-faced gypsum board.

[0082] Figure 2 It is the first sub-flow block diagram of the full-process control method for the production of fiber cloth-faced gypsum board.

[0083] Figure 3 It is the second sub-flow block diagram of the full-process control method for the production of fiber cloth-faced gypsum board.

[0084] Figure 4 It is the third sub-flow block diagram of the full-process control method for the production of fiber cloth-faced gypsum board.

[0085] Figure 5 It is a block diagram of the composition structure of the full-process control system for the production of fiber cloth surface gypsum board. Specific implementation manners

[0086] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0087] Figure 1 It is a flowchart of the full-process control method for the production of fiber cloth surface gypsum board. In an embodiment of the present invention, a full-process control method for the production of fiber cloth surface gypsum board, the method includes:

[0088] Step S100: Obtain a production flow chart, query the participating entities of each process in the production flow chart, and determine monitors according to the participating entities; the monitors at least include sensors and cameras;

[0089] Regardless of the product produced, the producer will pre-create a production flow chart, which contains each process. Query the participating entities corresponding to each process. The participating entities corresponding to each process may not be unique, and the participating entities are mostly production equipment; select or install some monitors according to the queried participating entities to obtain data during the production process. The monitors include sensors and cameras; specifically, the sensors are generally sensors in the production equipment, and the cameras are generally cameras installed at the outlet of the production equipment to obtain the working results of the production equipment. The working quality of the production equipment can be indirectly judged by the sensors, and the working quality of the production equipment can be directly judged by the working results. That is, the working condition of the production equipment is determined by the working results obtained by the equipment.

[0090] Step S200: Receive the production data within a preset time range obtained by the monitors, identify the production data, and determine the abnormality degree of each participating entity;

[0091] Receive the data uploaded by the monitor, which is called production data. By identifying the production data, the abnormality levels of each participating entity can be determined. When identifying the production data, it includes identifying the data obtained by the sensors to get the sensing abnormality level, and also includes identifying the images obtained by the cameras to get the visual abnormality level. By combining the sensing abnormality level and the visual abnormality level, the abnormality level of the participating entity is comprehensively obtained. In short, the abnormality level of the participating entity is jointly determined by the sensing abnormality level and the visual abnormality level. The sensing abnormality level is calculated from the data obtained by the sensors, and the visual abnormality level is calculated from the data obtained by the cameras. The process of analyzing the data obtained by the sensors belongs to a numerical analysis process, which is one-dimensional, and the process of analyzing the data obtained by the cameras belongs to an image analysis process, which is two-dimensional.

[0092] Step S300: Statistically analyze the abnormality levels of each participating entity to obtain an abnormality array, and locate the abnormal entity according to the abnormality array.

[0093] After the abnormality level of each participating entity is calculated, an abnormality array is obtained. The element number of the abnormality array corresponds to the participating entity, and the element value corresponds to the abnormality level of the corresponding participating entity. By analyzing the abnormality array, the participating entity with problems can be determined, which is called the abnormal entity. After obtaining the abnormal entity, a warning message is reported to achieve the purpose of actual full-process control.

[0094] It should be noted that the execution process of this method is a timed execution process. Every preset period, such as five minutes, the abnormal entity is determined. Specifically, every preset period, a acquisition instruction is generated to acquire the production data once. The way to acquire the production data is: based on the generation time of the acquisition instruction, acquire the data within a preset period of time before, and this time is also preset, such as one hour.

[0095] Figure 2 This is the first sub-process block diagram of the full-process control method for fiber cloth-faced gypsum board production. The steps of obtaining the production flow chart, querying the participating entities in each process of the production flow chart, and determining the monitor according to the participating entity include:

[0096] Step S101: Obtain the production flow chart and query the participating entities in each process of the production flow chart.

[0097] Step S102: Query the sensors of each participating entity, regularly obtain the monitoring data of each sensor within the preset test period, and determine the information volume of the monitor according to the monitoring data.

[0098] Step S103: Select sensors according to the information volume, and determine the total information volume of each participating entity according to the selection process.

[0099] Step S104: Insert cameras between adjacent participating entities according to the total amount of information, and establish a corresponding relationship between the cameras and the participating entities; wherein, the corresponding relationship is that the camera corresponds to the previous participating entity among the adjacent participating entities.

[0100] In an example of the technical solution of the present invention, the determination process of the monitor is described. Obtain the production flow chart, query the participating entities in each process of the production flow chart. In a fully automated production workshop, the participating entities are production equipment. Query the sensors of each participating entity, first conduct tests, determine a test time span, regularly obtain the monitoring data of the sensors within the test time span, calculate the amount of information according to the obtained monitoring data, select sensors according to the amount of information. After selecting the sensors, query the amount of information of the selected sensors in each participating entity, sum up the amount of information to obtain the total amount of information, insert cameras between adjacent participating entities according to the total amount of information, and at the same time establish a corresponding relationship between the cameras and the participating entities. The camera corresponds to the previous participating entity among the adjacent participating entities.

[0101] Further, some simple explanations are needed for the above content. Under normal conditions, there are multiple production equipment in the intelligent fiber cloth-faced gypsum board production workshop. These production equipment are sorted according to the production process. After the raw materials enter the production equipment, they are processed and come out of the production equipment, and then enter the next production equipment. There is a transportation process between production equipment, and the distance of the transportation process may be far or near. Install cameras during the transportation process to obtain the processing situation of the product. Analyzing the processing situation of the product, what is obtained is the operation situation of the previous production equipment. Based on this, the camera actually corresponds to the semi-finished product, and it corresponds to the source (production equipment) of the semi-finished product.

[0102] The calculation process of the amount of information is: I = ασ; where I is the amount of information, α is a preset coefficient, and σ is the standard deviation of the monitoring data during the test period;

[0103] The content of selecting sensors according to the amount of information and determining the total amount of information of each participating entity includes:

[0104] Determine the selection probability of each sensor according to the amount of information, randomly select sensors according to the selection probability until at least one sensor is selected in each participating entity;

[0105] Count the selected sensors of each participating entity, sum up the amount of information to obtain the total amount of information;

[0106] Among them, the calculation process of the selection probability is:

[0107] where P iis the selection probability of the i-th sensor, T i is the characteristic value of the i-th sensor, β is the preset correction coefficient, I i is the information amount of the i-th sensor, N is the total number of sensors that have not been selected, M is the total number of sensors that have been selected, and I j is the information amount of the j-th selected sensor, Δ represents the difference in subject serial numbers between the j-th selected sensor and the i-th sensor. When the two sensors belong to the same participating subject, the subject serial number difference is zero. The serial numbers of the sensors and the subjects are both known data and are determined by the production flow chart in the preprocessing stage.

[0108] The above content explains the sensor selection process. First, during the testing phase, the sensor's information content is determined. The greater the fluctuation in the sensor data, the greater its information content is considered to be. Therefore, it is directly proportional to the standard deviation. Then, for each sensor, the eigenvalue is calculated. The eigenvalue is directly proportional to its own information content and inversely proportional to the information content of other sensors. The closer the other sensors are to the current sensor, the smaller their influence. Finally, the calculated eigenvalues are combined to calculate the selection probability of each sensor.

[0109] It should be noted that both α and β are positive values to ensure the positive and negative relationship between the parameters. In addition, each time a sensor is selected, the N value needs to be updated and the probability of sensor selection needs to be recalculated; the loop is executed until at least one sensor is selected from each participating entity.

[0110] Figure 3 This is a flowchart of the second sub-process of the method for controlling the entire production process of fiber-covered gypsum board. The steps of receiving production data within a preset time range acquired by a monitor, identifying the production data, and determining the abnormality of each participating entity include:

[0111] Step S201: receiving sensor data uploaded by all sensors containing a first identity tag, and counting the sensor data according to the identity tag to obtain a data sequence corresponding to each sensor; the first identity tag is used to identify which sensor uploaded the sensor data;

[0112] Step S202: reading the data sequences of each sensor of the same participant, performing Fourier transform on each data sequence, and obtaining a frequency domain graph;

[0113] Step S203: determining the sensor abnormality of the participant according to the frequency domain graph;

[0114] Step S204: Receive the images containing the second identity tag uploaded by all cameras, query the participating entity corresponding to the camera according to the location tag, read the sensing abnormality degree of the participating entity, and determine the recognition accuracy of the image; the recognition accuracy is used to adjust the amount of recognition resources; the second identity tag is used to indicate which camera uploads the image;

[0115] Step S205: Recognize the image based on the recognition accuracy, and determine the visual abnormality degree of the participating entity;

[0116] Step S206: Determine the abnormality degree of the participating entity according to the sensing abnormality degree and the visual abnormality degree.

[0117] In an example of the technical solution of the present invention, for each sensor, it is necessary to obtain sensing data in real time according to a preset frequency. There is no order in the uploading process of the sensors. The execution entity of this method may receive the uploading data of different sensors within the same second. In addition, due to network fluctuation problems, the sensing data received by the execution entity of this method is very messy. Therefore, when the sensor uploads sensing data, it is necessary to insert a first identity tag to indicate which sensor uploads the sensing data. In addition, the sensing data also needs to have a time tag. After receiving it, the execution entity of this method arranges the sensing data according to the time tag to obtain a data sequence; thus, each sensor corresponds to a data sequence.

[0118] Similar to the above content, the images uploaded by the cameras also need to indicate which camera uploads the images. Therefore, when the cameras upload images, a second identity tag is inserted. Similarly, the images themselves also contain time information. The execution entity of this method sorts the images of each camera to obtain an image sequence.

[0119] Furthermore, for the data sequence of the sensor, it is essentially a discrete array in the time domain. Perform a frequency domain transformation on it to obtain frequency domain features. According to the frequency domain features, the abnormality degree of the sensing data can be determined, and then the sensing abnormality degree of the participating entity can be calculated; on this basis, the present application determines the recognition accuracy of the image according to the sensing abnormality degree. The greater the sensing abnormality degree, the higher the recognition accuracy required. Recognize the image based on the recognition accuracy to determine the visual abnormality degree of the participating entity, and comprehensively consider the sensing abnormality degree and the visual abnormality degree to obtain the abnormality degree of the participating entity.

[0120] Specifically, the sensing anomaly degree is a feature of the participating entity obtained from the data sequences of multiple sensors, which affects the image recognition process of the corresponding camera. That is, the process of determining the recognition accuracy of the image based on the sensing anomaly degree is to determine the recognition accuracy of the camera corresponding to a certain participating entity according to the sensing anomaly degree of that entity. After the image is recognized based on the recognition accuracy, the visual anomaly degree is obtained. The visual anomaly degree is the abnormal condition of the product coming out of the corresponding production equipment, which also reflects the abnormal condition of the production equipment. Therefore, both the sensing anomaly degree and the visual anomaly degree are parameters reflecting the abnormal conditions of the same participating entity.

[0121] In an example of the technical solution of the present invention, the step of determining the sensing anomaly degree of the participating entity according to the frequency domain diagram includes:

[0122] Compare the frequency domain diagrams of different sensors of the same participating entity, extract the characteristic frequency domain diagram, and synchronously determine the fidelity rate of the characteristic frequency domain diagram;

[0123] Input the characteristic frequency domain diagram into the trained frequency domain recognition model, output the stability degree, and correct the stability degree according to the fidelity rate;

[0124] Calculate the sensing anomaly degree according to the corrected stability degree, and the sensing anomaly degree is inversely proportional to the corrected stability degree.

[0125] The sensor is an internal component of the participating entity. For the same participating entity, there may be multiple sensors. After the data sequence of each sensor undergoes frequency domain conversion, a frequency domain diagram is obtained. By comparing the frequency domain diagrams of different sensors of the same participating entity, one frequency domain diagram is extracted as the standard, which is called the characteristic frequency domain diagram, and the fidelity rate of the characteristic frequency domain diagram is synchronously determined; input the characteristic frequency domain diagram into the trained frequency domain recognition model, and output the stability degree. The recognition process of the frequency domain recognition model is more refined. One of the models is the comparison model. First, some standard frequency domain diagrams are pre - statistically obtained, and the characteristic frequency domain diagram is compared and analyzed with the standard frequency domain diagrams. The more similar the comparison result is, the more standard its working process is, and the higher the stability degree is; combined with the determined fidelity rate, the stability degree is adjusted. Generally, direct multiplication is sufficient; finally, the sensing anomaly degree is calculated according to the corrected stability degree, and this process can be achieved with the help of a subtraction function.

[0126] In an example of the technical solution of the present invention, the step of recognizing the image based on the recognition accuracy and determining the visual anomaly degree of the participating entity includes:

[0127] Sequentially select convolution kernels in the preset convolution kernel table according to the recognition accuracy, and perform matching recognition on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and anomaly value items; each data item in the convolution kernel table is arranged in descending order based on the anomaly value;

[0128] Query the outliers corresponding to the convolution kernels with successful matching recognition, and accumulate all the outliers obtained by matching.

[0129] Determine the visual abnormality degree of the participating entity according to the accumulated outliers.

[0130] The above content describes the visual recognition process. A convolution kernel table is preset and created by the staff. The convolution kernel table includes convolution kernel items and outlier items. The various data items in the convolution kernel table are arranged in descending order based on the outliers. The number of data items in the convolution kernel table is very large, and it is necessary for the staff to count historical data, extract and count the convolution kernels. This occurs in the preprocessing stage and belongs to known data for this application.

[0131] Furthermore, select convolution kernels sequentially in the preset convolution kernel table according to the recognition accuracy. The higher the recognition accuracy, the more convolution kernels are selected. Match and recognize the image according to the selected convolution kernels. The match and recognition process belongs to the convolution recognition process, which will not be elaborated in this application; then, query the outliers corresponding to the convolution kernels with successful matching recognition, accumulate all the outliers obtained by matching, and determine the visual abnormality degree of the participating entity according to the accumulated outliers. The greater the accumulated outliers, the higher the visual abnormality degree.

[0132] The steps of comparing the frequency domain graphs of different sensors of the same participating entity, extracting the characteristic frequency domain graph, and synchronously determining the fidelity rate of the characteristic frequency domain graph include:

[0133] Query all the sensors of the same participating entity. For any sensor, compare its frequency domain graph with the frequency domain graphs of other sensors of the same participating entity to obtain the similarity; calculate the average value of the similarities to obtain the average similarity, and select the frequency domain graph of the sensor with the largest average similarity as the characteristic frequency domain graph, and use the average similarity as the fidelity rate.

[0134] It is worth mentioning that since the sensing data is discrete data, the obtained frequency domain graphs are also discrete data. The comparison process between frequency domain graphs is actually the comparison process of arrays, and existing array similarity calculation schemes can be used.

[0135] As a preferred embodiment of the technical solution of the present invention, the steps of receiving the production data within a preset time range obtained by the monitor, identifying the production data, and determining the abnormality degree of each participating entity further include:

[0136] Adjust the image acquisition frequency of the corresponding camera according to the abnormality degree of the participating entity.

[0137] In an example of the technical solution of the present invention, after calculating the abnormality degree of the participating entity, adjust the image acquisition frequency of the corresponding camera according to the abnormality degree of the participating entity. The higher the abnormality degree, the higher the image acquisition frequency, and the longer the image sequence obtained in the above content.

[0138] It should be noted that in this application, the participating entities and the cameras are not necessarily in one-to-one correspondence. One camera definitely corresponds to one participating entity, but one participating entity may correspond to multiple cameras. For example, when two cameras are installed between adjacent participating entities, both of these two cameras correspond to the same participating entity.

[0139] Figure 4 It is the third sub-process block diagram of the full-process control method for the production of fiber cloth-faced gypsum board. The steps of statistically analyzing the abnormality degrees of each participating entity to obtain an abnormality array and locating the abnormal entity according to the abnormality array include:

[0140] Step S301: Query the serial number of the participating entity in the production process;

[0141] Step S302: Read the abnormality degree of the participating entity. Taking the serial number as the independent variable and the abnormality degree as the dependent variable, determine the abnormality curve and the curve function of the production model;

[0142] Step S303: Calculate the derivative function of the curve function, intercept the derivative function according to a preset threshold line, and locate the abnormal entity according to the interception result.

[0143] In an example of the technical solution of the present invention, the process of locating the abnormal entity is described. Query the serial number of the participating entity in the production process. The serial number can be the order of each production equipment passed by the product. Read the abnormality degree of the participating entity. Taking the serial number as the independent variable and the abnormality degree as the dependent variable, determine the abnormality curve and the curve function of the production model. Take the derivative of the curve function to determine whether there is a sudden change in the abnormality degree. If the sudden change reaches the preset threshold, the participating entity corresponding to the serial number at the mutation can be used as the abnormal entity. Taking the threshold as a constant function relative to the serial number to obtain the threshold line, intercept the derivative function, and the serial number corresponding to the function line above the derivative function is the serial number of the abnormal entity.

[0144] It should be noted that the underlying meaning of the abnormality evaluation process based on the change situation is that changes with small amplitudes and gradual changes can be ignored. In this application, neither of these two will be recognized as an abnormal entity. In actual situations, it is also the case. If the abnormality degrees of each device are rising slowly, it is very likely that it is a misjudgment caused by the environment or working hours. For example, as the working hours increase, the device temperature gradually rises.

[0145] Figure 5 It is the block diagram of the composition structure of the full-process control system for the production of fiber cloth-faced gypsum board. In an embodiment of the present invention, a full-process control system for the production of fiber cloth-faced gypsum board, the system 10 includes:

[0146] A monitor determination module 11, configured to obtain a production flow chart, query the participating entities of each process in the production flow chart, and determine a monitor according to the participating entities; the monitor includes at least a sensor and a camera;

[0147] An abnormality degree calculation module 12, configured to receive production data within a preset time range obtained by the monitor, identify the production data, and determine the abnormality degree of each participating entity;

[0148] An abnormal entity location module 13, configured to count the abnormality degrees of each participating entity to obtain an abnormal array, and locate the abnormal entity according to the abnormal array;

[0149] Wherein, the abnormality degree of a participating entity is jointly determined by a sensing abnormality degree and a visual abnormality degree, the sensing abnormality degree is calculated from the data obtained by the sensor, and the visual abnormality degree is calculated from the data obtained by the camera.

[0150] Further, the monitor determination module 11 includes:

[0151] A participating entity query unit, configured to obtain a production flow chart and query the participating entities of each process in the production flow chart;

[0152] An information quantity calculation unit, configured to query the sensors of each participating entity, regularly obtain the monitoring data of each sensor within a preset test period, and determine the information quantity of the monitor according to the monitoring data;

[0153] A sensor selection unit, configured to select sensors according to the information quantity and determine the total information quantity of each participating entity according to the selection process;

[0154] A camera selection unit, configured to insert a camera between adjacent participating entities according to the total information quantity and establish a corresponding relationship between the camera and the participating entity; wherein, the corresponding relationship is: the camera is corresponding to the previous participating entity among the adjacent participating entities;

[0155] The calculation process of the information quantity is: I = ασ; where I is the information quantity, α is a preset coefficient, and σ is the standard deviation of the monitoring data within the test period;

[0156] The content of selecting sensors according to the information quantity and determining the total information quantity of each participating entity according to the selection process includes:

[0157] Determine the selection probability of each sensor according to the information quantity, randomly select sensors according to the selection probability until at least one sensor is selected in each participating entity;

[0158] Count the selected sensors of each participating entity, sum up the information quantity, and obtain the total information quantity;

[0159] Among them, the calculation process of the selection probability is as follows:

[0160] In the formula, P i is the selection probability of the i-th sensor, T i is the eigenvalue of the i-th sensor, β is a preset correction coefficient, I i is the information amount of the i-th sensor, N is the total number of sensors not selected, M is the total number of sensors already selected, I j is the information amount of the j-th selected sensor, Δ represents the difference in the main body serial numbers between the j-th selected sensor and the i-th sensor. When two sensors belong to the same participating main body, the difference in the main body serial numbers is zero; the serial numbers of the sensors and the serial numbers of the main bodies are all known data, which are determined by the production flow chart in the preprocessing stage.

[0161] Specifically, the anomaly degree calculation module 12 includes:

[0162] A data sequence query unit, which is used to receive the sensing data containing the first identity tag uploaded by all sensors, count the sensing data according to the identity tag, and obtain the data sequence corresponding to each sensor; the first identity tag is used to represent which sensor uploads the sensing data;

[0163] A frequency domain graph generation unit, which is used to read the data sequences of each sensor of the same participating main body, perform Fourier transform on each data sequence, and obtain a frequency domain graph;

[0164] A first calculation unit, which is used to determine the sensing anomaly degree of the participating main body according to the frequency domain graph;

[0165] An identification accuracy determination unit, which is used to receive the images containing the second identity tag uploaded by all cameras, query the participating main body corresponding to the camera according to the position tag, read the sensing anomaly degree of the participating main body, and determine the identification accuracy of the image; the identification accuracy is used to adjust the amount of identification resources; the second identity tag is used to represent which camera uploads the image;

[0166] A second calculation unit, which is used to identify the image based on the identification accuracy and determine the visual anomaly degree of the participating main body;

[0167] A comprehensive calculation unit, which is used to determine the anomaly degree of the participating main body according to the sensing anomaly degree and the visual anomaly degree.

[0168] Furthermore, the content of determining the sensing anomaly degree of the participating main body according to the frequency domain graph includes:

[0169] Compare the frequency domain graphs of different sensors of the same participating main body, extract the characteristic frequency domain graph, and synchronously determine the fidelity rate of the characteristic frequency domain graph;

[0170] Input the characteristic frequency domain graph into the trained frequency domain recognition model to output the stability, and correct the stability according to the fidelity rate;

[0171] Calculate the sensing abnormality degree according to the corrected stability, and the sensing abnormality degree is inversely proportional to the corrected stability;

[0172] The content of identifying the visual abnormality degree of the participating subject based on the recognition accuracy includes:

[0173] Sequentially select convolution kernels in the preset convolution kernel table according to the recognition accuracy, and perform matching recognition on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and outlier items; each data item in the convolution kernel table is arranged in descending order based on the outlier;

[0174] Query the outlier corresponding to the convolution kernel with successful matching recognition, and accumulate all the matched outliers;

[0175] Determine the visual abnormality degree of the participating subject according to the accumulated outliers;

[0176] The steps of comparing the frequency domain graphs of different sensors of the same participating subject, extracting the characteristic frequency domain graph, and synchronously determining the fidelity rate of the characteristic frequency domain graph include:

[0177] Query all sensors of the same participating subject. For any sensor, compare its frequency domain graph with the frequency domain graphs of other sensors of the same participating subject to obtain the similarity; calculate the mean value of the similarities to obtain the average similarity, and select the frequency domain graph of the sensor with the largest average similarity as the characteristic frequency domain graph, and use the average similarity as the fidelity rate.

[0178] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A full-process control method for the production of fiber cloth-faced gypsum boards, characterized in that, The method includes: Obtain a production flow chart, query the participating entities of each process in the production flow chart, and determine monitors according to the participating entities; the monitors at least include sensors and cameras; Receive production data within a preset time range obtained by the monitors, identify the production data, and determine the abnormality degree of each participating entity; Statistically analyze the abnormality degree of each participating entity to obtain an abnormality array, and locate the abnormal entity according to the abnormality array; Among them, the abnormality degree of the participating entity is jointly determined by the sensing abnormality degree and the visual abnormality degree. The sensing abnormality degree is calculated from the data obtained by the sensor, and the visual abnormality degree is calculated from the data obtained by the camera.

2. The full-process control method for the production of fiber cloth-faced gypsum boards according to claim 1, wherein, The step of obtaining a production flow chart, querying the participating entities of each process in the production flow chart, and determining monitors according to the participating entities includes: Obtain a production flow chart and query the participating entities of each process in the production flow chart; Query the sensors of each participating entity, regularly obtain the monitoring data of each sensor within a preset test period, and determine the information volume of the monitor according to the monitoring data; Select sensors according to the information volume, and determine the total information volume of each participating entity according to the selection process; Insert cameras between adjacent participating entities according to the total information volume, and establish a corresponding relationship between the cameras and the participating entities; among them, the corresponding relationship is: the camera is corresponding to the previous participating entity among the adjacent participating entities; The calculation process of the information volume is: I = ασ; where I is the information volume, α is a preset coefficient, and σ is the standard deviation of the monitoring data within the test period; The content of selecting sensors according to the information volume and determining the total information volume of each participating entity according to the selection process includes: Determine the selection probability of each sensor according to the information volume, randomly select sensors according to the selection probability until at least one sensor in each participating entity is selected; Statistically analyze the selected sensors of each participating entity, sum up the information volume, and obtain the total information volume; Among them, the calculation process of the selection probability is: Wherein, P i is the selection probability of the i-th sensor, T i is the eigenvalue of the i-th sensor, β is a preset correction coefficient, I i is the information amount of the i-th sensor, N is the total number of sensors not selected, M is the total number of sensors already selected, I j is the information amount of the j-th selected sensor, Δ represents the difference in the main body serial numbers between the j-th selected sensor and the i-th sensor. When the two sensors belong to the same participating entity, the difference in the main body serial numbers is zero; the serial numbers of the sensors and the serial numbers of the entities are all known data, which are determined by the production flow chart in the preprocessing stage.

3. The full-process control method for the production of fiber cloth-faced gypsum boards according to claim 1, characterized in that, The step of receiving production data within a preset time range obtained by the monitors, identifying the production data, and determining the abnormality degree of each participating entity includes: Receive the sensing data containing the first identity tag uploaded by all sensors, statistically analyze the sensing data according to the identity tag, and obtain the data sequence corresponding to each sensor; the first identity tag is used to represent which sensor uploads the sensing data; Read the data sequences of the sensors of the same participating entity, perform Fourier transform on each data sequence, and obtain a frequency domain diagram; Determine the sensing abnormality degree of the participating entity according to the frequency domain diagram; Receive the images containing the second identity tag uploaded by all cameras, query the participating entity corresponding to the camera according to the position tag, read the sensing abnormality degree of the participating entity, and determine the recognition accuracy of the image; the recognition accuracy is used to adjust the recognition resource amount; the second identity tag is used to represent which camera uploads the image; Based on the recognition accuracy, identify the image and determine the visual abnormality degree of the participating entity; Determine the abnormality degree of the participating entity according to the sensing abnormality degree and the visual abnormality degree.

4. The full-process control method for the production of fiber cloth-faced gypsum boards according to claim 3, wherein, The steps of determining the sensing abnormality degree of the participating entity according to the frequency domain diagram include: Comparing the frequency domain diagrams of different sensors of the same participating entity, extracting the characteristic frequency domain diagram, and synchronously determining the fidelity rate of the characteristic frequency domain diagram; Inputting the characteristic frequency domain diagram into the trained frequency domain recognition model to output the stability degree, and correcting the stability degree according to the fidelity rate; Calculating the sensing abnormality degree according to the corrected stability degree, and the sensing abnormality degree is inversely proportional to the corrected stability degree; The steps of recognizing the image based on the recognition accuracy and determining the visual abnormality degree of the participating entity include: Sequentially selecting convolution kernels in a preset convolution kernel table according to the recognition accuracy, and performing matching recognition on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and outlier items; the data items in the convolution kernel table are arranged in descending order based on the outlier values; Querying the outlier values corresponding to the convolution kernels with successful matching recognition, and accumulating all the outlier values obtained by matching; Determining the visual abnormality degree of the participating entity according to the accumulated outlier values; The steps of comparing the frequency domain diagrams of different sensors of the same participating entity, extracting the characteristic frequency domain diagram, and synchronously determining the fidelity rate of the characteristic frequency domain diagram include: Querying all the sensors of the same participating entity. For any sensor, comparing its frequency domain diagram with the frequency domain diagrams of other sensors of the same participating entity to obtain the similarity; calculating the mean value of the similarities to obtain the average similarity, and selecting the frequency domain diagram of the sensor with the largest average similarity as the characteristic frequency domain diagram, and taking the average similarity as the fidelity rate.

5. The full-process control method for the production of fiber cloth-faced gypsum boards according to claim 3, characterized in that, The steps of receiving the production data within a preset time range obtained by the monitor, recognizing the production data, and determining the abnormality degree of each participating entity further include: Adjusting the image acquisition frequency of the corresponding camera according to the abnormality degree of the participating entity.

6. The full-process control method for the production of fiber cloth-faced gypsum boards according to claim 1, wherein, The steps of counting the abnormality degree of each participating entity to obtain an abnormality array and positioning the abnormal entity according to the abnormality array include: Querying the serial number of the participating entity in the production process; Reading the abnormality degree of the participating entity, using the serial number as the independent variable and the abnormality degree as the dependent variable to determine the abnormality curve and the curve function of the production model; Calculating the derivative function of the curve function, intercepting the derivative function according to a preset threshold line, and positioning the abnormal entity according to the interception result.

7. A full-process control system for the production of fiber cloth-faced gypsum boards, characterized in that, The system includes: A monitor determination module for obtaining the production flow chart, querying the participating entities in each process of the production flow chart, and determining the monitor according to the participating entities; the monitor includes at least sensors and cameras; An abnormality degree calculation module for receiving the production data within a preset time range obtained by the monitor, recognizing the production data, and determining the abnormality degree of each participating entity; An abnormal entity positioning module for counting the abnormality degree of each participating entity to obtain an abnormality array and positioning the abnormal entity according to the abnormality array; Among them, the abnormality degree of the participating entity is jointly determined by the sensing abnormality degree and the visual abnormality degree. The sensing abnormality degree is calculated from the data obtained by the sensor, and the visual abnormality degree is calculated from the data obtained by the camera.

8. The full-process control system for the production of fiber cloth-faced gypsum boards according to claim 7, characterized in that, The monitor determination module includes: A participating entity query unit for obtaining the production flow chart and querying the participating entities in each process of the production flow chart; An information quantity calculation unit, configured to query sensors of each participating entity, regularly obtain monitoring data of each sensor within a preset test period, and determine the information quantity of the monitor according to the monitoring data; A sensor selection unit, configured to select sensors according to the information quantity, and determine the total information quantity of each participating entity according to the selection process; A camera selection unit, configured to insert cameras between adjacent participating entities according to the total information quantity, and establish a corresponding relationship between the cameras and the participating entities; wherein, the corresponding relationship is: the camera corresponds to the previous participating entity among the adjacent participating entities; The calculation process of the information quantity is: I = ασ; where I is the information quantity, α is a preset coefficient, and σ is the standard deviation of the monitoring data within the test period; The content of selecting sensors according to the information quantity and determining the total information quantity of each participating entity according to the selection process includes: Determine the selection probability of each sensor according to the information quantity, randomly select sensors according to the selection probability until at least one sensor in each participating entity is selected; Count the selected sensors of each participating entity, sum up the information quantities, and obtain the total information quantity; Among them, the calculation process of the selection probability is: Wherein, P i is the selection probability of the i-th sensor, T i is the eigenvalue of the i-th sensor, β is a preset correction coefficient, I i is the information quantity of the i-th sensor, N is the total number of sensors not selected, M is the total number of sensors already selected, I j is the information quantity of the j-th selected sensor, Δ represents the difference in the main body serial numbers between the j-th selected sensor and the i-th sensor. When the two sensors belong to the same participating entity, the difference in the main body serial numbers is zero; the serial numbers of the sensors and the serial numbers of the entities are all known data, which are determined by the production flow chart in the preprocessing stage.

9. The full-process control system for the production of fiber cloth-faced gypsum boards according to claim 7, wherein The anomaly calculation module includes: A data sequence query unit, configured to receive sensing data containing a first identity tag uploaded by all sensors, count the sensing data according to the identity tag, and obtain a data sequence corresponding to each sensor; the first identity tag is used to indicate which sensor uploads the sensing data; A frequency domain graph generation unit, configured to read the data sequences of each sensor of the same participating entity, perform Fourier transform on each data sequence, and obtain a frequency domain graph; A first calculation unit, configured to determine the sensing anomaly of the participating entity according to the frequency domain graph; An identification accuracy determination unit, configured to receive images containing a second identity tag uploaded by all cameras, query the participating entity corresponding to the camera according to the position tag, read the sensing anomaly of the participating entity, and determine the identification accuracy of the image; the identification accuracy is used to adjust the amount of identification resources; the second identity tag is used to indicate which camera uploads the image; A second calculation unit, configured to identify the image based on the identification accuracy and determine the visual anomaly of the participating entity; A comprehensive calculation unit, configured to determine the anomaly of the participating entity according to the sensing anomaly and the visual anomaly.

10. The full-process control system for the production of fiber cloth-faced gypsum boards according to claim 9, characterized in that, The content of determining the sensing anomaly of the participating entity according to the frequency domain graph includes: Compare the frequency domain graphs of different sensors of the same participating entity, extract the characteristic frequency domain graph, and synchronously determine the fidelity of the characteristic frequency domain graph; Input the characteristic frequency domain graph into a trained frequency domain recognition model, output the stability, and correct the stability according to the fidelity; Calculate the sensing anomaly according to the corrected stability, and the sensing anomaly is inversely proportional to the corrected stability; The content of identifying the image based on the identification accuracy and determining the visual anomaly of the participating entity includes: Select convolution kernels sequentially in a preset convolution kernel table according to the recognition accuracy, and perform matching recognition on the image according to the selected convolution kernels; the convolution kernel table includes convolution kernel items and outlier items; each data item in the convolution kernel table is arranged in descending order based on the outlier. Query the outlier corresponding to the convolution kernel with successful matching recognition, and accumulate all the matched outliers. Determine the visual abnormality degree of the participating subject according to the accumulated outliers. The steps of comparing the frequency domain graphs of different sensors of the same participating subject, extracting the characteristic frequency domain graph, and synchronously determining the fidelity rate of the characteristic frequency domain graph include: Query all sensors of the same participating subject. For any sensor, compare its frequency domain graph with the frequency domain graphs of other sensors of the same participating subject to obtain the similarity; calculate the mean of the similarities to obtain the average similarity, and select the frequency domain graph of the sensor with the largest average similarity as the characteristic frequency domain graph, and use the average similarity as the fidelity rate.

Images