A parameter abnormality monitoring method, system and medium based on horizontal continuous casting process

Abstract

The present application relates to the technical field of horizontal continuous casting, and specifically to a parameter abnormality monitoring method, system and medium based on the horizontal continuous casting process, which specifically includes: installing multiple temperature sensors on each device of the horizontal continuous casting process production line, obtaining the spatial coordinates of each temperature sensor and the temperature data collected at each moment; obtaining each temperature mutation point according to the difference between the temperature data at adjacent moments; constructing the adaptive K value of each temperature mutation point according to the difference between the temperature mutation point of the same temperature sensor and the temperature mutation point in the adjacent time period, and the difference between the spatial coordinates of the temperature sensors corresponding to the temperature mutation point at the same moment; obtaining the LOF value of each temperature mutation point based on the adaptive K value combined with the LOF algorithm, and performing temperature abnormality detection based on the LOF value. The problem of errors in abnormal temperature data identification caused by the use of a fixed K value by the traditional LOF algorithm is avoided, and the accuracy of abnormal temperature data identification is improved.

Description

Technical Field

[0001] The present application relates to the technical field of horizontal continuous casting, and in particular to a method, system and medium for monitoring parameter anomalies based on a horizontal continuous casting process. Background Art

[0002] Horizontal continuous casting technology plays a key role in the metal forming process with its high efficiency and continuity. It is particularly suitable for manufacturing long castings with uniform cross-sections, such as steel billets and iron pipes. Austempered ductile iron, as an advanced engineering material, has achieved microstructure optimization and significantly improved material properties through precise heat treatment and alloying. However, the horizontal continuous casting process is extremely sensitive to temperature control, and any abnormality may affect product quality. Therefore, strict monitoring of key parameters such as temperature during the continuous casting process is crucial to ensuring product quality, which helps to promptly detect and correct deviations in production.

[0003] The LOF algorithm (Local Outlier Factor) is an anomaly detection algorithm based on the local density of data, which is used to identify abnormal points in a data set. In the production process of horizontal continuous casting ductile iron, the LOF algorithm can be effectively applied to parameter anomaly monitoring. The K value in the LOF algorithm is a key parameter, which defines the number of nearest neighbor data points considered when calculating the local density of each data point. The traditional LOF algorithm uses a fixed K value and cannot be adjusted according to changes in the data, which may cause errors in the identification of abnormal data. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a parameter abnormality monitoring method, system and medium based on the horizontal continuous casting process. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for monitoring abnormal parameters based on a horizontal continuous casting process, the method comprising the following steps:

[0006] Install multiple temperature sensors on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each time;

[0007] For the temperature data of any temperature sensor, each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments;

[0008] The temperature mutation density of each temperature mutation point is constructed according to the difference between each temperature mutation point of the temperature sensor and the temperature mutation points in the adjacent time period; the temperature space density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to each temperature mutation point at the same time; the adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature space density;

[0009] The LOF value of each temperature mutation point is obtained based on the adaptive K value combined with the LOF algorithm, and temperature anomaly detection is performed based on the LOF value.

[0010] In one embodiment, the process of obtaining each temperature mutation point is as follows:

[0011] Taking any moment as the last moment of the neighborhood window at any moment, constructing the neighborhood windows at each moment;

[0012] Construct the cumulative temperature difference of the temperature at each moment based on the difference between the temperature at each moment and the temperature at all moments in the corresponding neighborhood window;

[0013] The average value of all the accumulated temperature differences of any temperature sensor is recorded as a first average value; and the temperature at which the accumulated temperature difference is greater than the first average value is recorded as a temperature mutation point.

[0014] In one embodiment, the expression of the cumulative temperature difference is:

[0015] , where represents the cumulative temperature difference at the ath moment, M represents the number of elements in the neighborhood window at each moment, represents the temperature value at the ath moment, Represents the temperature value at the i-th moment in the neighborhood window at the a-th moment.

[0016] In one embodiment, the process of obtaining the temperature mutation density is as follows:

[0017] For each temperature sensor, the sequence composed of all elements in the neighborhood window at the time when each temperature mutation point of the temperature sensor is located is recorded as the neighborhood temperature sequence;

[0018] The temperature mutation density of the kth temperature mutation point of the temperature sensor is recorded as , The calculation expression is:

[0019] , where N represents the number of temperature mutation points in the preset time period before the kth temperature mutation point of the temperature sensor, represents the time interval between the kth temperature mutation point of the temperature sensor and the jth temperature mutation point in the corresponding preset time period, Indicates the similarity between the neighborhood temperature sequence of the kth temperature mutation point of the temperature sensor and the neighborhood temperature sequence of the jth temperature mutation point in the corresponding preset time period, is the normalization function.

[0020] In one embodiment, the process of obtaining the temperature spatial density is:

[0021] The coordinates of the temperature sensors at each temperature mutation point are used as the coordinates of each temperature mutation point;

[0022] The set of temperature mutation points of all temperature sensors at each acquisition moment is taken as the mutation point set at each acquisition moment;

[0023] The inverse of the distance between any temperature mutation point and the coordinates of each temperature mutation point in the corresponding mutation point set is calculated and recorded as the first inverse; and the fusion value of all the first inverses of any temperature mutation point is used as the temperature space density of any temperature mutation point.

[0024] In one embodiment, the calculation expression of the adaptive K value of each temperature mutation point is:

[0025] , where represents the adaptive K value of the qth temperature mutation point, Indicates rounding. represents the temperature distribution density of the qth temperature mutation point, represents the temperature space density of the qth temperature mutation point, are preset parameters.

[0026] In one embodiment, the LOF value of each temperature mutation point is obtained based on the adaptive K value combined with the LOF algorithm, specifically:

[0027] All temperature data collected by all sensors are used as the input of the LOF algorithm, and the adaptive K value of each temperature mutation point is used as the K value of each temperature mutation point in the LOF algorithm, and the output is the LOF value of each temperature mutation point.

[0028] In one embodiment, the temperature anomaly detection based on the LOF value is specifically:

[0029] If the LOF value of each temperature mutation point is greater than the preset abnormal threshold, then each temperature mutation point is an abnormal temperature fluctuation point; otherwise, each temperature mutation point is a normal temperature fluctuation point.

[0030] In a second aspect, the embodiment of the present application further provides a parameter abnormality monitoring system based on a horizontal continuous casting process, comprising:

[0031] Data acquisition module: multiple temperature sensors are installed on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each time;

[0032] Data processing module: for the temperature data of any temperature sensor, each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments;

[0033] The temperature mutation density of each temperature mutation point is constructed according to the difference between each temperature mutation point of the temperature sensor and the temperature mutation points in the adjacent time period; the temperature space density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to each temperature mutation point at the same time; the adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature space density;

[0034] Anomaly detection module: Based on the adaptive K value combined with the LOF algorithm, the LOF value of each temperature mutation point is obtained, and temperature anomaly detection is performed based on the LOF value.

[0035] In the third aspect, an embodiment of the present application also provides a parameter abnormality monitoring medium based on a horizontal continuous casting process, comprising: a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of the method described in the first aspect above are implemented.

[0036] The embodiments of the present application have at least the following beneficial effects:

[0037] The present application obtains the temperature mutation point through temperature fluctuation, and obtains the adaptive K value of the temperature mutation point by considering the distribution density of the temperature mutation point in time and space, so that the LOF algorithm can be adaptively adjusted according to the data density, avoiding the problem that the traditional LOF algorithm uses a fixed K value to cause errors in the identification of abnormal temperature data;

[0038] A plurality of temperature sensors are installed on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each moment; each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments; the temperature mutation density of each temperature mutation point is constructed according to the difference between the temperature mutation point of the same temperature sensor and the temperature mutation point in the adjacent time period; the temperature spatial density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to the temperature mutation points at the same moment; an adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature spatial density; the LOF value of each temperature mutation point is obtained based on the adaptive K value combined with the LOF algorithm, and temperature anomaly detection is performed based on the LOF value, thereby improving the accuracy of abnormal temperature data recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of a method for monitoring abnormal parameters of a horizontal continuous casting process provided by an embodiment of the present application;

[0041] Figure 2 Schematic diagram of the process of obtaining the adaptive K value of each temperature mutation point;

[0042] Figure 3 It is a structural schematic diagram of a parameter abnormality monitoring system based on horizontal continuous casting process. DETAILED DESCRIPTION

[0043] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, describes in detail a parameter abnormality monitoring method, system and medium based on a horizontal continuous casting process proposed in the present application, its specific implementation method, structure, features and effects. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0045] The following is a detailed description of a specific scheme of a parameter abnormality monitoring method, system and medium based on a horizontal continuous casting process provided by the present application in conjunction with the accompanying drawings.

[0046] See also Figure 1 , which shows a flow chart of a method for monitoring abnormal parameters based on a horizontal continuous casting process provided by an embodiment of the present application, the method comprising the following steps:

[0047] Step S1, installing various temperature sensors on the horizontal continuous casting production line, obtaining the spatial coordinates of each temperature sensor and the temperature data collected at each moment, and preprocessing the data.

[0048] The equipment on the horizontal continuous casting process production line includes a material holding barrel, an intermediate tank, a quenching furnace, a crystallizer, a secondary nozzle, a embryo pulling device, a synchronous cutter and a cooling bed, and Q thermistor temperature sensors are installed on each device. The sensor is installed on the vertical center axis of the equipment, and the distance between each sensor on the same device is the same. With the height as the z-axis, the direction perpendicular to the horizontal continuous casting process production line as the y-axis direction, the direction parallel to the horizontal continuous casting process production line as the x-axis direction, and the starting point of the horizontal continuous casting process production line as the origin, a spatial rectangular coordinate system is constructed to obtain the spatial position coordinates of each sensor. The temperature data of each device in the production process is collected in real time. This application adopts the maximum and minimum value normalization method to standardize all the acquired data. The maximum and minimum value normalization method is a well-known technology, and the specific process will not be repeated.

[0049] Preferably, in one embodiment of the present application, the value of Q is set to 5, and the sampling frequency of each sensor is set to 10 Hz. In other embodiments of the present application, the value of Q and the sampling frequency of the sensor can be set by the implementer according to actual conditions.

[0050] Step S2, obtaining each temperature mutation point according to the difference between the temperature data at adjacent moments.

[0051] In the horizontal continuous casting process, temperature control is a key process parameter, which directly affects the microstructure, physical properties and quality of the final product. During the continuous casting process, the temperature of the molten metal determines its fluidity and ability to fill the mold. During the continuous casting process, a small change in temperature may lead to potential quality problems. For example, overheating may cause cracks in the casting, while too low a temperature may cause an uneven internal structure of the casting.

[0052] Temperature monitoring is very important in the horizontal continuous casting process of isothermal quenching ductile iron. The temperature is maintained by the heating device during metal smelting or quenching. The power fluctuation of the heating device may cause changes in the heating efficiency, resulting in temperature fluctuations. In addition, the error of the temperature sensor itself may also cause errors in the collected temperature. In order to avoid the temperature fluctuations caused by these influencing factors affecting the detection of real temperature anomaly data, the K value of the LOF algorithm can be adjusted to make the selection of the nearest neighbor point more reasonable.

[0053] The key parameter K value of the LOF algorithm is used to calculate the number of nearest neighbor data points considered. When the K value is small, the algorithm focuses more on the very close distance range around the data point. A larger K value means that the algorithm considers a wider range of nearest neighbor points when evaluating the local density of data points.

[0054] The LOF algorithm mainly relies on the local density of the data to detect abnormal points. Therefore, the K value can be calculated based on the local density of the temperature data during the horizontal continuous casting process. However, the data collected by the sensor is temperature data, so it is necessary to first find the temperature mutation point and determine the K value through the density of the temperature mutation point, so as to use the LOF algorithm to detect temperature abnormal points from the temperature mutation point.

[0055] Temperature mutation points usually refer to significant jumps or drops in continuously monitored temperature data. Such mutations may indicate abnormal conditions in the horizontal continuous casting process, but temperature fluctuations may also occur under normal circumstances. In order to identify normal temperature fluctuations in the horizontal continuous casting process and fluctuations caused by equipment failures, it is necessary to first find the temperature mutation points.

[0056] (1) Take each acquisition moment as the last moment of its neighborhood window and construct a 1*M neighborhood window of each acquisition moment. For any acquisition moment, the 1st to M-1st elements in its neighborhood window are the M-1st acquisition moments before any acquisition moment.

[0057] Preferably, in one embodiment of the present application, the value of M can be set to 10. In other embodiments of the present application, the value of M can be set by the implementer according to the actual situation, and the present application does not make any specific limitation.

[0058] When performing the following calculations, if the number of collection moments in the neighborhood window of each collection moment is less than M, the following calculations will not be performed for that collection moment.

[0059] (2) For each temperature sensor, the cumulative temperature difference of the temperature sensor at each time is constructed based on the difference between the temperature of the temperature sensor at each acquisition time and the temperature of each acquisition time in the corresponding neighborhood window. The expression is:

[0060] , where represents the cumulative temperature difference of the temperature sensor at the ath moment, M represents the number of elements in the neighborhood window at each moment, represents the temperature value of the temperature sensor at the ath moment, Represents the temperature value at the i-th moment in the neighborhood window of the temperature sensor at the a-th moment.

[0061] The greater the cumulative temperature difference of the temperature value, the greater the cumulative degree of temperature mutation of the temperature sensor at that moment, and the greater the possibility that it is a temperature mutation point, otherwise the possibility that the current processing point becomes a temperature mutation point is smaller.

[0062] (3) For any temperature sensor, based on the above steps, the cumulative temperature difference of the temperature at each collection moment except the 1st to M-1th collection moments is obtained; the mean of all the cumulative temperature differences of any temperature sensor is calculated and recorded as the first mean; among the temperatures of all moments of any temperature sensor, the temperature whose cumulative temperature difference is greater than the first mean is recorded as the temperature mutation point.

[0063] Step S3, constructing an adaptive K value for each temperature mutation point according to the difference between each temperature mutation point and the temperature mutation points in the adjacent time period, and the difference between the spatial coordinates of the temperature sensors corresponding to the temperature mutation points at the same moment.

[0064] (1) The temperature mutation density of each temperature mutation point is constructed based on the difference between each temperature mutation point and the temperature mutation points in the adjacent time period:

[0065] The temperature mutation point represents a relatively large temperature change at a certain moment, but a single temperature mutation point may be a normal temperature fluctuation in the horizontal continuous casting process. In order to exclude these normal temperature fluctuations, for each temperature sensor, the temperature mutation density of each temperature mutation point of the temperature sensor is calculated based on the number of temperature mutation points in the preset time period and the time interval between temperature mutation points:

[0066] The sequence composed of all elements in the neighborhood window at the time when each temperature mutation point of the temperature sensor is located is recorded as the neighborhood temperature sequence;

[0067] The expression of the temperature mutation density of each temperature mutation point of the temperature sensor is:

[0068] , where represents the temperature mutation density of the kth temperature mutation point of the temperature sensor, N represents the number of temperature mutation points in the n-minute time period before the kth temperature mutation point of the temperature sensor. It should be noted that the value of n is set to 5 in the embodiment of the present application. In other embodiments of the present application, the implementer can set the value of n according to the actual production situation. It represents the time interval between the kth temperature mutation point of the temperature sensor and the jth temperature mutation point in the corresponding n-minute time period. Indicates the similarity between the neighborhood temperature sequence of the kth temperature mutation point of the temperature sensor and the neighborhood temperature sequence of the jth temperature mutation point in the corresponding n-minute time period, is the normalization function.

[0069] It should be noted that, for the calculation of the similarity between the neighborhood temperature sequences, the present application uses the Pearson correlation coefficient between the neighborhood temperature sequences as the similarity between the neighborhood temperature sequences. In other embodiments of the present application, the implementer may also use other similarity algorithms to obtain the similarity between the neighborhood temperature sequences.

[0070] The temperature mutation density takes into account the time interval between each temperature mutation point and other temperature mutation points. The smaller the time interval, the greater the data density near the temperature mutation point. At the same time, the similarity of the temperature data sequences in the neighborhood window where the two temperature mutation points are located is taken into account. The greater their similarity, the closer the causes of the temperature mutation at these two time points are, and it is difficult to distinguish them when performing LOF detection. More neighboring points should be considered, so the value of K should be increased, and the temperature mutation density will be greater at this time.

[0071] The greater the temperature mutation density, the more temperature mutation points there are near the current temperature mutation point, and the smaller the similarity between the mutation points. At this time, a smaller K value should be selected to make the LOF algorithm focus on a smaller data range and detect abnormal temperature data points with smaller differences.

[0072] (2) The temperature spatial density of each temperature mutation point is constructed based on the difference between the spatial coordinates of the temperature sensors corresponding to the temperature mutation points at the same time:

[0073] The temperature mutation density calculated based on the above only takes into account the temporal distribution density of the temperature mutation points collected by each sensor. The K value calculated based on this can only detect abnormal points for different sensors separately, and does not take into account the relationship between the data collected by each sensor during the horizontal continuous casting process. Therefore, this application introduces the spatial density distribution of the temperature mutation points collected by each sensor.

[0074] In the horizontal continuous casting production line, since the continuous casting equipment is installed on a horizontal line and the continuous casting process is also carried out in the horizontal direction, when installing the sensors, their y-axis coordinates are very small and can be ignored. The sensor coordinates are collected as x-axis and z-axis coordinates.

[0075] The coordinates of the sensor where each temperature mutation point is located are used as the coordinates of the temperature mutation point;

[0076] The set of temperature mutation points of all temperature sensors at each acquisition moment is taken as the mutation point set at the acquisition moment;

[0077] Calculate the reciprocal of the distance between any temperature mutation point and the coordinates of each temperature mutation point in the corresponding mutation point set, and record it as the first reciprocal; and use the fusion value of all the first reciprocals of any temperature mutation point as the temperature space density of any temperature mutation point:

[0078] The inverse of the distance between any temperature mutation point and the coordinates of each temperature mutation point in the corresponding mutation point set is calculated and recorded as the first inverse; and the fusion value of all the first inverses of any temperature mutation point is used as the temperature space density of any temperature mutation point.

[0079] It should be noted that the fusion described in this application is to combine multiple variables, which can be specifically addition, multiplication, etc. The specific fusion method can be determined according to actual conditions during the application process, and this application does not impose any special restrictions.

[0080] Preferably, in one embodiment of the present application, the calculation formula for the temperature space density of any temperature mutation point may be:

[0081] , where represents the temperature space density of the qth temperature mutation point, represents the number of temperature mutation points in the mutation point set where the qth temperature mutation point is located. , Respectively represent the x-coordinate and z-coordinate of the qth temperature mutation point, , They represent the x-coordinate and z-coordinate of the m-th temperature mutation point in the set of mutation points where the q-th temperature mutation point is located. is the normalization function.

[0082] In other embodiments of the present application, the temperature spatial density of any temperature mutation point may be a normalized value of the product of all the first inverses of any temperature mutation point.

[0083] The temperature space density takes into account the spatial distribution of temperature mutation points. The larger its value is, the more temperature mutation points there are near the temperature mutation point in space, and the more likely it is that the device where the temperature sensor that collects the current temperature mutation point is located has an abnormality. At this time, the smaller the K value should be, the more abnormal temperature data will be detected. Otherwise, a larger K value should be selected.

[0084] (3) Construct the adaptive K value of each temperature mutation point based on the temperature mutation density and temperature space density:

[0085] Based on the above analysis, the K value can be calculated from the temperature distribution density and the temperature space density. One of the calculation methods is:

[0086]

[0087] In the formula, represents the adaptive K value of the qth temperature mutation point, Indicates rounding. represents the temperature distribution density of the qth temperature mutation point, represents the temperature space density of the qth temperature mutation point, is a preset parameter, the purpose of which is to adjust the value of K to an appropriate range. Preferably, in one embodiment of the present application The value of is set to 10. In other embodiments of the present application, The implementer can set the value according to the actual situation.

[0088] The value of K is related to the temperature distribution density and temperature spatial density of the temperature mutation point. The greater the time density and spatial density of the temperature mutation point, the higher the possibility of temperature anomaly, and the smaller the value of K should be. At this time, the algorithm is more sensitive to small density changes and can detect more subtle anomalies. On the contrary, a larger K value should be used to avoid false detection.

[0089] In particular, when the adaptive K value of the temperature mutation point is less than 3, the final adaptive K value of the temperature mutation point is set to 3.

[0090] Step S4, obtaining the LOF value of each temperature mutation point based on the adaptive K value combined with the LOF algorithm, and performing temperature anomaly detection based on the LOF value.

[0091] All temperature data collected by all sensors are used as the input of the LOF algorithm, wherein the adaptive K value of each temperature mutation point is used as the K value of each temperature mutation point in the LOF algorithm, and the output is the LOF value of each temperature mutation point.

[0092] An abnormal threshold is set. Preferably, in one embodiment of the present application, the abnormal threshold is set to 1. In other embodiments of the present application, the implementer can set the value of the abnormal threshold according to the actual situation. When the LOF value of each temperature mutation point is greater than the abnormal threshold, the temperature mutation point is a data point when the temperature fluctuation is abnormal. The time and coordinates corresponding to the temperature mutation point are obtained, and the time point and coordinate information of the abnormal temperature are fed back to the technician to provide data reference for the technician. If the LOF value of each temperature mutation point is less than or equal to the abnormal threshold, the temperature mutation point is a normal temperature fluctuation data point.

[0093] The schematic diagram of the process of obtaining the adaptive K value of each temperature mutation point is as follows: Figure 2 shown.

[0094] See also Figure 3 , Figure 3 Schematic diagram of the structure of a parameter abnormality monitoring system based on a horizontal continuous casting process provided in an embodiment of the present application. In this embodiment, each unit included in the terminal is used to execute each step in an embodiment corresponding to a parameter abnormality monitoring method based on a horizontal continuous casting process. Figure 3 , the parameter abnormality monitoring system includes:

[0095] Data acquisition module: multiple temperature sensors are installed on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each time;

[0096] Data processing module: for the temperature data of any temperature sensor, each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments;

[0097] The temperature mutation density of each temperature mutation point is constructed according to the difference between each temperature mutation point of the temperature sensor and the temperature mutation points in the adjacent time period; the temperature space density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to each temperature mutation point at the same time; the adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature space density;

[0098] Anomaly detection module: Based on the adaptive K value combined with the LOF algorithm, the LOF value of each temperature mutation point is obtained, and temperature anomaly detection is performed based on the LOF value.

[0099] Based on the same inventive concept as the above method, an embodiment of the present application also provides a parameter abnormality monitoring medium based on a horizontal continuous casting process, including: a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned parameter abnormality monitoring methods based on a horizontal continuous casting process are implemented.

[0100] In summary, the embodiment of the present application provides a parameter anomaly monitoring method based on a horizontal continuous casting process, wherein a temperature mutation point is obtained through temperature fluctuation, and an adaptive K value of the temperature mutation point is obtained by considering the distribution density of the temperature mutation point in time and space, so that the LOF algorithm can be adaptively adjusted according to the data density, thereby avoiding the problem that the traditional LOF algorithm uses a fixed K value to cause errors in the identification of abnormal temperature data;

[0101] A plurality of temperature sensors are installed on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each moment; each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments; the temperature mutation density of each temperature mutation point is constructed according to the difference between the temperature mutation point of the same temperature sensor and the temperature mutation point in the adjacent time period; the temperature spatial density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to the temperature mutation points at the same moment; an adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature spatial density; the LOF value of each temperature mutation point is obtained based on the adaptive K value combined with the LOF algorithm, and temperature anomaly detection is performed based on the LOF value, thereby improving the accuracy of abnormal temperature data recognition.

[0102] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of the present application are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] The various embodiments in the present application are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0104] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A parameter abnormality monitoring method based on horizontal continuous casting process, characterized in that: The method comprises the following steps: Install multiple temperature sensors on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each time; For the temperature data of any temperature sensor, each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments; The temperature mutation density of each temperature mutation point is constructed according to the difference between each temperature mutation point of the temperature sensor and the temperature mutation points in the adjacent time period; the temperature space density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to each temperature mutation point at the same time; the adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature space density; Based on the adaptive K value combined with the LOF algorithm, the LOF value of each temperature mutation point is obtained, and temperature anomaly detection is performed based on the LOF value; The calculation expression of the adaptive K value of each temperature mutation point is: , where represents the adaptive K value of the qth temperature mutation point, Indicates rounding. represents the temperature distribution density of the qth temperature mutation point, represents the temperature space density of the qth temperature mutation point, is the preset parameter; The process of obtaining each temperature mutation point is as follows: Taking any moment as the last moment of the neighborhood window at any moment, constructing the neighborhood windows at each moment; Construct the cumulative temperature difference of the temperature at each moment based on the difference between the temperature at each moment and the temperature at all moments in the corresponding neighborhood window; Recording the average of all the accumulated temperature differences of any temperature sensor as a first average value; recording the temperature at which the accumulated temperature difference is greater than the first average value as a temperature mutation point; The process of obtaining the temperature mutation density is as follows: For each temperature sensor, the sequence composed of all elements in the neighborhood window at the time when each temperature mutation point of the temperature sensor is located is recorded as the neighborhood temperature sequence; The temperature mutation density of the kth temperature mutation point of the temperature sensor is recorded as , The calculation expression is: , where N represents the number of temperature mutation points in the preset time period before the kth temperature mutation point of the temperature sensor, represents the time interval between the kth temperature mutation point of the temperature sensor and the jth temperature mutation point in the corresponding preset time period, Indicates the similarity between the neighborhood temperature sequence of the kth temperature mutation point of the temperature sensor and the neighborhood temperature sequence of the jth temperature mutation point in the corresponding preset time period, is the normalization function; The process of obtaining the temperature space density is as follows: The coordinates of the temperature sensors at each temperature mutation point are used as the coordinates of each temperature mutation point; The set of temperature mutation points of all temperature sensors at each acquisition moment is taken as the mutation point set at each acquisition moment; The inverse of the distance between any temperature mutation point and the coordinates of each temperature mutation point in the corresponding mutation point set is calculated and recorded as the first inverse; and the fusion value of all the first inverses of any temperature mutation point is used as the temperature space density of any temperature mutation point.

2. A parameter abnormality monitoring method based on a horizontal continuous casting process according to claim 1, characterized in that: The expression of the cumulative temperature difference is: , where represents the cumulative temperature difference at the ath moment, M represents the number of elements in the neighborhood window at each moment, represents the temperature value at the ath moment, Represents the temperature value at the i-th moment in the neighborhood window at the a-th moment.

3. A parameter abnormality monitoring method based on a horizontal continuous casting process as claimed in claim 1, characterized in that: The LOF value of each temperature mutation point is obtained based on the adaptive K value combined with the LOF algorithm, specifically: All temperature data collected by all sensors are used as the input of the LOF algorithm, and the adaptive K value of each temperature mutation point is used as the K value of each temperature mutation point in the LOF algorithm, and the output is the LOF value of each temperature mutation point.

4. A parameter abnormality monitoring method based on a horizontal continuous casting process as claimed in claim 1, characterized in that: The temperature anomaly detection based on the LOF value is specifically as follows: If the LOF value of each temperature mutation point is greater than the preset abnormal threshold, then each temperature mutation point is an abnormal temperature fluctuation point; otherwise, each temperature mutation point is a normal temperature fluctuation point.

5. A parameter abnormality monitoring system based on a horizontal continuous casting process, implementing the method as claimed in claim 1, characterized in that: The system comprises: Data acquisition module: multiple temperature sensors are installed on each device of the horizontal continuous casting process production line to obtain the spatial coordinates of each temperature sensor and the temperature data collected at each time; Data processing module: for the temperature data of any temperature sensor, each temperature mutation point is obtained according to the difference between the temperature data at adjacent moments; The temperature mutation density of each temperature mutation point is constructed according to the difference between each temperature mutation point of the temperature sensor and the temperature mutation points in the adjacent time period; the temperature space density of each temperature mutation point is constructed according to the difference between the spatial coordinates of the temperature sensors corresponding to each temperature mutation point at the same time; the adaptive K value of each temperature mutation point is constructed based on the temperature mutation density and the temperature space density; Anomaly detection module: Based on the adaptive K value combined with the LOF algorithm, the LOF value of each temperature mutation point is obtained, and temperature anomaly detection is performed based on the LOF value.

6. A parameter abnormality monitoring medium based on a horizontal continuous casting process, comprising: A memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the methods of claims 1 to 4 when executing the computer program.

Images