Fire-fighting platform digital twin construction method under artificial intelligence assistance

Abstract

The present application relates to the technical field of data processing, in particular to a fire-fighting platform digital twin construction method assisted by artificial intelligence, comprising: obtaining each item of source data of a collection source point at each time; screening each item of source data to obtain a differentiated source point and a fire-fighting event vector thereof at each time; obtaining an independent test event vector through deviation of the fire-fighting event vector, and obtaining fire-fighting event independence of the differentiated source point by using differences between the fire-fighting event vector and the independent test event vector and time distribution closeness of the independent test event vector; analyzing distribution of the fire-fighting event independence of all differentiated source points after the collection source point is converted into the differentiated source point to obtain a contribution weight; predicting a modeling support weight as a data adoption proportion of each collection source point to construct a fire-fighting platform digital twin. The present application aims to solve the problem that different sources of data have different influence weights on the model when constructing the fire-fighting platform digital twin, resulting in inharmonious data fusion.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a fire platform digital twin construction method assisted by artificial intelligence. BACKGROUND

[0002] The fire platform digital twin is a digital building structure model constructed by comprehensively collecting data models such as BIM, GIS, LOT, etc. Real-time sensor data is obtained through the fire platform digital twin to plan the response strategy for the fire event, thereby dynamically simulating and planning the fire event response strategy for the entity building.

[0003] When using real-time sensor data input into the fire platform digital twin for fire event response strategy planning, different sensor data is integrated and matched by a weighted information transmission method. The weight of different sensor data depends on the prior decision result. When different structure buildings are digitally modeled, the prior decision result is difficult to fit into a new building, resulting in abnormal calibration of sensor data in the new building, and causing the judgment accuracy and real-time performance of the fire event response strategy planning to decline. SUMMARY

[0004] The present application provides a fire platform digital twin construction method assisted by artificial intelligence to solve the existing problems.

[0005] The fire platform digital twin construction method assisted by artificial intelligence of the present application adopts the following technical scheme:

[0006] An embodiment of the present application provides a fire platform digital twin construction method assisted by artificial intelligence, which comprises the following steps:

[0007] A plurality of sensors arranged at each collection source point in the building model are used to obtain each item of source data of each collection source point at each time; wherein the last collection time is recorded as the current time;

[0008] The prior abnormal threshold of each item of source data is used to filter each item of source data to obtain a plurality of differentiated source points at the current time, and a fire event vector of each differentiated source point at each time is constructed. The deviation of the fire event vector of the same differentiated source point at the current time from other times is filtered to obtain an independent test event vector of each differentiated source point at the current time. The time distribution of the independent test event vector in all times, and the difference between the fire event vector and the independent test event vector of each differentiated source point at the current time are used to obtain the fire event independence of each differentiated source point at the current time.

[0009] analyze the distribution of the independence of the fire events of all the differentiated source points at each time, obtain the relative independence of the fire events of each differentiated source point at each time, analyze the change of the relative independence of the fire events after the collection source point is changed into the differentiated source point, and obtain the contribution weight of each differentiated source point at each time;

[0010] obtain the modeling support weight of each collection source point at the next time by using the contribution weight of each differentiated source point at all times; the modeling support weight is used as the data adoption proportion of each collection source point, and a digital twin of the fire platform is constructed.

[0011] Preferably, the prior abnormal threshold of each item of source data is used to filter each item of source data, obtain a plurality of differentiated source points at the current time, and construct the fire event vector of each differentiated source point at each time, which includes:

[0012] If there is any one source data of all items of source data of a collection source point greater than or equal to the prior abnormal threshold of the source data at the current time, the collection source point is recorded as a differentiated source point;

[0013] the difference between each item of source data of each differentiated source point at each time and the prior abnormal threshold thereof is recorded as the abnormal performance degree of each item of source data of each differentiated source point at each time;

[0014] the difference between each item of source data of each differentiated source point at each time and the prior abnormal threshold thereof is recorded as the abnormal performance degree of each item of source data of each differentiated source point at each time;

[0015] obtain the abnormal urgency degree of each item of source data of each differentiated source point at each time, and the abnormal urgency degree is in a positive proportional relationship with the abnormal performance degree and the urgency performance degree;

[0016] integrate the abnormal urgency degrees of all items of source data of each differentiated source point at each time to construct the fire event vector of each differentiated source point at each time.

[0017] Preferably, the specific obtaining step of the independent test event vector includes:

[0018] use the fire event vector of each differentiated source point at all times to construct a multi-source data space of each differentiated source point at the current time;

[0019] In the multi-source data space of each differentiated source point, a plurality of independent test event vectors of each differentiated source point at the current time are screened according to the difference between the fire event vector at the current time and the fire event vector at other times.

[0020] Preferably, the plurality of source data spaces of each differentiated source point are filtered according to the differences between the fire event vector at the current time and the fire event vectors at other times to obtain a plurality of independent inspection event vectors of each differentiated source point at the current time, which includes:

[0021] In the plurality of source data spaces of each differentiated source point, the spatial distance between the fire event vector at the current time and the fire event vectors at each time except the current time is obtained;

[0022] After the spatial distance is bipolarized and divided, a minimum difference time set of each differentiated source point at the current time is obtained;

[0023] The fire event vector corresponding to each time in the minimum difference time set is taken as a plurality of independent inspection event vectors of each differentiated source point at the current time.

[0024] Preferably, the specific obtaining step of the minimum difference time set includes:

[0025] The spatial distances between each differentiated source point at the current time and the fire event vectors at each time except the current time are arranged in descending order to obtain a spatial distance ascending sequence of each differentiated source point at the current time;

[0026] The difference between each sequence value and the next sequence value in the spatial distance ascending sequence is calculated, and is recorded as the first difference value of each sequence value in the spatial distance ascending sequence of each differentiated source point at the current time;

[0027] The sequence value with the largest first difference value is recorded as a secondary segmentation sequence value;

[0028] A set composed of the time corresponding to all sequence values in the closed interval composed of the first sequence value and the secondary segmentation sequence value is recorded as the minimum difference time set of each differentiated source point at the current time.

[0029] Preferably, the specific obtaining step of the fire event independence includes:

[0030] The time sequence distribution closeness of each differentiated source point at the current time is obtained according to the number and distribution time range of the independent inspection event vectors at all times;

[0031] The source data dispersion degree of each differentiated source point at the current time is obtained according to the extreme value of the source data in the fire event vector of each differentiated source point at the current time and the data average dispersion of each item of source data in the independent inspection event vector;

[0032] obtaining the fire event independence of each differentiated source point at the current time, which is in direct proportion to the time sequence distribution tightness and the source data dispersion degree.

[0033] Preferably, the specific acquisition step of the time sequence distribution tightness comprises:

[0034] obtaining the time sequence distribution tightness of each differentiated source point at the current time, which is in direct proportion to the number of all independent test event vectors of each differentiated source point at the current time, and is in inverse proportion to the range of the time of the independent test event vector.

[0035] Preferably, the specific acquisition step of the source data dispersion degree comprises:

[0036] taking the maximum value of all item source data in the fire event vector of each differentiated source point at the current time as the extreme value of each differentiated source point at the current time;

[0037] taking the average value of the standard deviation of each item source data of each independent test event vector of each differentiated source point at the current time as the data average dispersion degree of each item source data in the independent test event vector;

[0038] obtaining the source data dispersion degree of each differentiated source point at the current time, which is in direct proportion to the extreme value and is in inverse proportion to the data average dispersion degree.

[0039] Preferably, the specific acquisition step of the fire event relative independence comprises:

[0040] obtaining the maximum value and the minimum value of the fire event independence of all differentiated source points at the same time;

[0041] taking the difference between the fire event independence of each differentiated source point and the minimum value of the fire event independence, and the ratio of the difference between the maximum value and the minimum value of the fire event independence, as the fire event relative independence of each differentiated source point at each time.

[0042] Preferably, the specific acquisition step of the contribution weight comprises:

[0043] taking the length of the differentiated time course of the cth differentiated source point as ;

[0044] taking the fire event relative independence of the cth differentiated source point at the tth time as ; ;

[0045] taking the number of differentiated source points at the tth time as ; ;

[0046] The contribution weight of the c-th differentiated source point at the current moment is calculated as follows:

[0047]

[0048] in, For the c-th differential source point in the th... The relative independence of fire incidents at any given moment; For the first The number of differential source points at any given time.

[0049] The beneficial effects of the technical solution of this invention are as follows: This invention acquires various source data at each time point by arranging several sensors at each acquisition source point in the building model; it filters the source data using prior anomaly thresholds to obtain several differentiated source points at the current time, thereby finding the acquisition source points that exhibit differences at the current time and using them to subsequently determine the degree of inconsistency of the source points in the data fusion process; it constructs fire event vectors for each differentiated source point at each time point; the fire event vectors can be used to reflect the anomalies of various source data at each time point, thereby showing the differences in the source data at each time point; by filtering the deviation of the fire event vectors of the same differentiated source point at the current time relative to other times, it obtains the independent test event vectors for each differentiated source point at the current time; and it utilizes the close temporal distribution of the independent test event vectors across all times, as well as the difference between the fire event vectors and the independent test event vectors of each differentiated source point at the current time. The process involves obtaining the fire event independence of each differentiated source point at the current moment; finding an independent test event vector approximating the current moment and analyzing its temporal distribution to determine whether the current fire event vector is an occasional anomaly or due to incompatible data fusion; analyzing the distribution of fire event independence across all differentiated source points at each moment to obtain the relative independence of fire events for each differentiated source point; analyzing the change in relative independence of fire events after a data source point is transformed into a differentiated source point to obtain the contribution weight of each differentiated source point at each moment; using the contribution weight to represent the incompatible data fusion of each data source point, which is then used to adjust the data adoption ratio of the data source point in constructing the fire platform digital twin; and using the contribution weight of each differentiated source point at all moments to predict the modeling support weight of each data source point in the next moment. This modeling support weight serves as the data adoption ratio of each data source point to construct the fire platform digital twin. This approach addresses the problem of incompatible data fusion caused by different weights of data from different sources affecting the model when constructing the fire platform digital twin. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the steps of a method for constructing a digital twin of a fire-fighting platform with the assistance of artificial intelligence, as described in this invention. Detailed Implementation

[0052] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for constructing a digital twin of a fire-fighting platform with artificial intelligence assistance proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0053] Unless otherwise defined, 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 invention pertains.

[0054] The following description, in conjunction with the accompanying drawings, details the specific scheme of the method for constructing a digital twin of a fire-fighting platform with artificial intelligence assistance provided by the present invention.

[0055] Please see Figure 1 The diagram illustrates a flowchart of a method for constructing a digital twin of a fire-fighting platform with artificial intelligence assistance, according to an embodiment of the present invention. The method includes the following steps:

[0056] Step S001: Using several sensors arranged at each acquisition source point in the building model, acquire various source data of each acquisition source point at each time.

[0057] It should be noted that the prior decision structure described above is not applicable to modeling entirely new building structures. Therefore, this embodiment analyzes the differentiated performance of data collected from each data acquisition point during the modeling process and applies different modeling support weights to different data acquisition points. This results in a higher proportion of data adoption from data acquisition points with different characteristics when modeling the digital twin of the fire protection platform, thus eliminating the problem of the inapplicability of prior decision results.

[0058] Therefore, the first step is to use several sensors placed at each data acquisition point in the building model to acquire various source data at each time point.

[0059] Specifically, a plurality of collection source points are selected in the entity building, and a plurality of sensors are installed at each collection source point. The deployed sensors are connected to the control computer, and the raw data collected by each sensor at each time point is standardized so that the data distribution range of each sensor at each collection source point is within The standardized data is used as each item of source data of each collection source point at each time point.

[0060] It should be noted that the sensors deployed at different collection source points are connected to the control computer through wireless ap, wired connection or other means, so that the data collected by each sensor at each time point is transmitted to the control computer for recording. In order to align the data return frequency of different sensors and realize simultaneous collection, the control computer transmits time pulses to each sensor through the built-in hardware clock. The pulse frequency selected in this embodiment is 10 Hz. When each sensor receives the pulse, it collects data once and returns it to the control computer.

[0061] Further, for different sensors, the prior abnormal threshold of each item of source data that can be collected by each sensor is obtained by checking the abnormal threshold of different sensors in the control computer.

[0062] As an example, for a temperature sensor, the prior abnormal threshold recorded by the control computer is 280 degrees Celsius. When the temperature exceeds the prior temperature threshold, it indicates that the data collected by the temperature sensor is abnormal.

[0063] Step S002, using the prior abnormal threshold of each item of source data to screen each item of source data, obtaining a plurality of differentiated source points at the current time, and constructing a fire event vector of each differentiated source point at each time; by the deviation of the fire event vector of the same differentiated source point at the current time from other time, the independent inspection event vector of each differentiated source point at the current time is screened, and the time distribution of the independent inspection event vector in all time is used. The tightness and the difference between the fire event vector and the independent inspection event vector of each differentiated source point at the current time are used to obtain the fire event independence of each differentiated source point at the current time.

[0064] Based on the above description, the collection source points that show abnormality are fused in the process of building the building model, which belongs to the inharmonious data fusion. Therefore, the data adoption weight of the collection source points that show abnormality is adjusted in this embodiment.

[0065] Specifically, the time of the last collection is recorded as the current time. If there is any one source data in all items of source data of a collection source point that is greater than or equal to the prior abnormal threshold of the source data at the current time, the collection source point is recorded as a differentiated source point.

[0066] It should be noted that when the sensor obtains an abnormal value, it indicates that the sensor has an abnormality at the position corresponding to the differentiated source point. If the abnormal data is used for data fusion to construct the digital twin of the fire platform, the sensitivity of the sensor alarm at this position is higher when a fire event occurs or is triggered by mistake. At the same time, the greater the source data amplitude of the same sensor at the adjacent time, the more obvious the change in the detection value of the source data, and the stronger the performance of the data fusion. Therefore, by analyzing the abnormal performance of the source data obtained by the sensor, the abnormal urgency of each source data in each differentiated source point at the current time can be obtained.

[0067] Further, since one differentiated source point corresponds to the same collection position in the building structure, the abnormal urgency performance of all sensors at this position can reflect the rationality of the data collected by the collection source point. Therefore, by integrating the abnormal urgency of all source data of the same differentiated source point at the same time, the overall inharmonious performance of the differentiated source point can be easily fed back.

[0068] Preferably, the specific steps for constructing the fire event vector of each differentiated source point at each time are as follows:

[0069] Analyze the difference between each source data of each differentiated source point at each time and the prior abnormal threshold, and the difference between the source data at each time and the source data at the previous time, to obtain the abnormal urgency of each source data of each differentiated source point at each time;

[0070] Integrate the abnormal urgency of all source data of each differentiated source point at each time to construct the fire event vector of each differentiated source point at each time.

[0071] Specifically, the specific way to analyze the difference between each source data of each differentiated source point at each time and the prior abnormal threshold, and the difference between the source data at each time and the source data at the previous time, to obtain the abnormal urgency of each source data of each differentiated source point at each time, is as follows:

[0072] The difference between each source data of each differentiated source point at each time and the prior abnormal threshold is denoted as the abnormal performance degree of each source data of each differentiated source point at each time;

[0073] The difference between each source data of each differentiated source point at each time and the source data at the previous time is denoted as the urgency performance degree of each source data of each differentiated source point at each time;

[0074] Obtain the abnormal urgency of each source data of each differentiated source point at each time, which is in a positive proportional relationship with the abnormal performance degree and the urgency performance degree.

[0075] As an example, the ratio of each item of source data of each differentiated source point at each time to its prior abnormal threshold is recorded as the abnormal performance degree of each item of source data of each differentiated source point at each time; the difference between each item of source data of each differentiated source point at each time and its previous time is recorded as the urgency performance degree of each item of source data of each differentiated source point at each time.

[0076] The product of the abnormal performance degree and the urgency performance degree is recorded as the abnormal urgency degree of each item of source data of each differentiated source point at each time.

[0077] Further, the specific way of constructing the fire event vector of each differentiated source point at each time by integrating the abnormal urgency degrees of all items of source data of each differentiated source point at each time is as follows:

[0078] The fire event vector of each differentiated source point at each time is a 1*n dimensional vector, where n represents the number of items of source data in each differentiated source point; each element in the fire event vector is the abnormal urgency degree of each item of source data of each differentiated source point at each time.

[0079] As an example, the current time is recorded as the t-th time, any one differentiated source point is recorded as the c-th differentiated source point, and the abnormal urgency degree of the x-th item of source data of the c-th differentiated source point at the t-th time is recorded as ; then the expression of the fire event vector of the c-th differentiated source point at the t-th time is , where is the vector transposition symbol.

[0080] It should be noted that the evolution of the fire event over time is the current situation represented by the source data collected by different sensors after the occurrence of the fire event. Different fire event states are formed by changing the values of different items of source data, that is, the values of the fire event vector at different times are different. The difference between the fire event states is used to evaluate the event development speed, so as to design the development process of the fire event in the current building, so that the detection of the fire event by the digital twin fire platform of the current building is more accurate.

[0081] Therefore, the fire event vectors of the same differentiated source point at all times can be integrated, and then the deviation of the fire event vector at the current time from other times is analyzed, the performance difference of the fire event at the current time from the fire event at other times in history is analyzed, and the independence of the fire event of each differentiated source point at the current time is obtained to reflect the fire event state of each differentiated source point at the current time. If the deviation degree of the fire event state at the current time is greater, it indicates that the data collected by the differentiated source point is less compatible in the fusion process of all data, and the weight of the differentiated source point in the fusion is adjusted, so that the digital twin of the fire platform is more inclined to the data of the differentiated source point, that is, the data of the differentiated source point is adopted in a higher proportion when the digital twin of the fire platform is modeled and iterated.

[0082] Preferably, the specific steps of screening the independent test event vector of each differentiated source point at the current time through the deviation of the fire event vector of the same differentiated source point at the current time from other times are as follows:

[0083] A multi-source data space of each differentiated source point at the current time is constructed by using the fire event vector of each differentiated source point at all times.

[0084] In the multi-source data space of each differentiated source point, a number of independent test event vectors of each differentiated source point at the current time are screened according to the difference between the fire event vectors at the current time and other times.

[0085] Specifically, the multi-source data space of each differentiated source point at the current time is constructed by using the fire event vector of each differentiated source point at all times in the following specific manner:

[0086] The abnormal urgency of each source data in the fire event vector of each differentiated source point is taken as a dimension, and the increasing direction of the abnormal urgency is taken as the positive direction of each dimension to obtain a multi-dimensional space of each differentiated source point.

[0087] The fire event vector of each differentiated source point at all times is projected into the multi-dimensional space to obtain a multi-source data space of each differentiated source point at the current time.

[0088] It should be noted that after the multi-source data space of each differentiated source point at the current time is constructed, the state description of the fire event reflected by each differentiated source point at different times is quantized into a vector in the space. A number of independent test event vectors of each differentiated source point at the current time can be screened through the difference between the fire event vectors at the current time and other times in the multi-source data space. The independent test event vector represents a fire event similar to the fire event vector at the current time, so that the independence of the fire event at the current time is verified by using the independent test event.

[0089] Further, in the multi-source data space of each differentiated source point, according to the difference between the fire event vector of the current time and the fire event vector of other times, the specific manner of screening a plurality of independent test event vectors of each differentiated source point at the current time is as follows:

[0090] In the multi-source data space of each differentiated source point, according to the difference between the fire event vector of the current time and the fire event vector of each time except the current time, the spatial distance between the fire event vector of each differentiated source point at the current time and the fire event vector of each time except the current time is obtained.

[0091] After the spatial distance is bipolarized and divided, the maximum difference time set and the minimum difference time set of each differentiated source point at the current time are obtained.

[0092] The fire event vector corresponding to each time in the minimum difference time set is taken as a plurality of independent test event vectors of each differentiated source point at the current time.

[0093] It should be noted that the embodiment reflects the difference between the fire event vector of the current time and the fire event vector of each time except the current time by using cosine similarity as the spatial distance between the fire event vector of the current time and the fire event vector of each time except the current time. Other embodiments can calculate the spatial distance in other ways, which are not limited in the embodiment. The cosine similarity calculation vector difference is a known technology, and will not be described in detail in the embodiment.

[0094] It should be noted that since the data of the same differentiated source point at all times reflects different fire event states, and the fire event vectors of similar fire events are similar, when judging the independence of the fire event state of the current time, the fire event vector corresponding to the time similar to the fire event vector of the current time is selected as the independent test event vector. Based on the above, the similar fire events are similar to the fire event of the current time, and therefore the difference between their fire event vectors is much larger than that of non-similar fire events. Therefore, the embodiment is bipolarized and divided.

[0095] Specifically, after the spatial distance is bipolarized and divided, the specific manner of obtaining the maximum difference time set and the minimum difference time set of each differentiated source point at the current time is as follows:

[0096] The spatial distances of each differentiated source point at the current time and the fire event vector of each time except the current time are arranged in descending order to obtain a spatial distance ascending sequence of each differentiated source point at the current time.

[0097] The difference between each sequence value and the next sequence value in the spatial distance ascending sequence is calculated, and is recorded as the first difference value of each sequence value in the spatial distance ascending sequence of each differentiated source point at the current time.

[0098] The sequence value with the largest first difference value is recorded as a secondary segmentation sequence value;

[0099] A set of time points corresponding to all sequence values in the closed interval formed by the first sequence value and the secondary segmentation sequence value in the spatial distance ascending sequence is recorded as a minimum difference time point set of each difference source point at the current time point; a set of time points corresponding to all sequence values in the left-open right-closed interval formed by the secondary segmentation sequence value and the last sequence value in the spatial distance ascending sequence is recorded as a maximum difference time point set of each difference source point at the current time point;

[0100] It should be noted that if the independent inspection event vector of each difference source point at the current time point is continuously distributed and gathered in a short time, and is unevenly distributed at all time points, it indicates that the anomaly reflected by the difference source point at the current time point is new, rather than due to the random scattered abnormal data collected by the difference source point.

[0101] Preferably, the specific steps of obtaining the fire event independence of each difference source point at the current time point by using the time distribution of the independent inspection event vector at all time points and the difference between the fire event vector and the independent inspection event vector of each difference source point at the current time point are as follows:

[0102] According to the number and distribution time range of the independent inspection event vector at all time points, the time sequence distribution tightness of each difference source point at the current time point is obtained.

[0103] According to the extreme value of the source data in the fire event vector of each difference source point at the current time point and the data average dispersion of each item of source data in the independent inspection event vector, the source data dispersion degree of each difference source point at the current time point is obtained.

[0104] The fire event independence of each difference source point at the current time point is obtained by combining the time sequence distribution tightness and the source data dispersion degree.

[0105] Specifically, the specific way of obtaining the time sequence distribution tightness of each difference source point at the current time point according to the number and distribution time range of the independent inspection event vector at all time points is as follows:

[0106] The time sequence distribution tightness of each difference source point at the current time point is obtained, and the time sequence distribution tightness is in direct proportion to the number of all independent inspection event vectors of each difference source point at the current time point, and is in inverse proportion to the range of time points of the independent inspection event vector.

[0107] As an example, a ratio of a number of all independent inspection event vectors of each differentiated source point at a current time to a range of time of the independent inspection event vectors is recorded as a time sequence distribution closeness of each differentiated source point at the current time.

[0108] It is required to be explained that the range of time of the independent inspection event vectors represents a difference between a maximum collection time and a minimum collection time in all independent inspection event vectors of each differentiated source point at the current time; the difference is used to represent a distribution time range of the independent inspection event vectors, and a smaller value indicates that the distribution is more uneven at all collection times, and more reflects that the fire event at the approximate current time is more independent; at the same time, the more independent inspection events appear in the smaller distribution time, the more important the fire event at the current time, and the greater the adjustment demand of the data adoption proportion of the differentiated source point.

[0109] Further, according to an extreme value of source data in the fire event vector of each differentiated source point at the current time and a data average dispersion of each item of source data in the independent inspection event vector, a specific manner of obtaining a source data dispersion degree of each differentiated source point at the current time is as follows:

[0110] The maximum value of all items of source data in the fire event vector of each differentiated source point at the current time is recorded as an extreme value of each differentiated source point at the current time.

[0111] The average value of the standard deviation of each item of source data of each independent inspection event vector of each differentiated source point at the current time is recorded as a data average dispersion of each item of source data in the independent inspection event vector.

[0112] The source data dispersion degree of each differentiated source point at the current time is obtained, and the source data dispersion degree is in a positive proportional relationship with the extreme value and in an inverse proportional relationship with the data average dispersion.

[0113] As an example, a ratio of the extreme value to the data average dispersion is recorded as the source data dispersion degree of each differentiated source point at the current time.

[0114] It is required to be explained that the extreme value represents an abnormal performance of the fire event vector at the current time, and a larger value can better represent the abnormality, and the data average dispersion reflects the distribution of each item of source data in the independent inspection event vector; if the extreme value at the current time is larger, the distribution of each item of source data is smaller, which can better indicate that the fire event state at the current time is more different from the fire event state corresponding to the approximate independent inspection event vector, and better reflects the independence of the fire event state at the current time.

[0115] Specifically, the fire event independence of each differentiated source point at the current moment is obtained in combination with the time sequence distribution tightness and the source data dispersion degree, wherein the fire event independence is in direct proportion to both the time sequence distribution tightness and the source data dispersion degree.

[0116] As an example, the product of the time sequence distribution tightness and the source data dispersion degree is recorded as the fire event independence of each differentiated source point at the current moment.

[0117] Step S003: The distribution of the fire event independence of all differentiated source points at each moment is analyzed to obtain the relative fire event independence of each differentiated source point at each moment, and the change of the relative fire event independence after the collection source point is converted into the differentiated source point is analyzed to obtain the contribution weight of each differentiated source point at each moment.

[0118] It should be noted that after the fire event occurs, it is not limited to a single differentiated source point. Since the data of the differentiated source point will be gradually adopted in the modeling iteration, the digital twin of the fire platform deviates from the normal range of other data, so that other collection source points are also converted into differentiated source points, that is, the anomaly of a single differentiated source point gradually spreads. Therefore, the relative fire event independence is obtained by analyzing the fire event independence of all differentiated source points at each moment, and then the change of the relative fire event independence after each collection source point is converted into a differentiated source point is analyzed to reflect the influence on other collection source points, and the contribution weight of each differentiated source point at each moment is obtained.

[0119] Preferably, the specific steps of analyzing the distribution of the fire event independence of all differentiated source points at each moment to obtain the relative fire event independence of each differentiated source point at each moment, and analyzing the change of the relative fire event independence after the collection source point is converted into the differentiated source point to obtain the contribution weight of each differentiated source point at each moment are as follows:

[0120] At the same moment, the relative fire event independence of each differentiated source point at each moment is obtained according to the distribution of the fire event independence of each differentiated source point in the fire event independence of all differentiated source points;

[0121] The time interval between the moment when each differentiated source point is converted into a differentiated source point for the first time and the current moment is recorded as the differentiated time course of each differentiated source point.

[0122] The change trend of the relative fire event independence of each differentiated source point in the differentiated time course and the appearance and disappearance of other differentiated source points are analyzed to obtain the contribution weight of each differentiated source point at each moment.

[0123] Specifically, at the same time, based on the distribution of the fire event independence of each differentiated source point among the fire event independence of all differentiated source points, the specific method for obtaining the relative independence of fire events at each time point is as follows:

[0124] At the same time, obtain the maximum and minimum values ​​of the independence of fire events at all differentiated source points;

[0125] The ratio of the difference between the fire event independence of each differentiated source point and the minimum value of the fire event independence to the difference between the maximum value and the minimum value of the fire event independence is denoted as the relative fire event independence of each differentiated source point at each time.

[0126] Furthermore, the analysis of the changing trends in the relative independence of fire events and the visibility of other differentiated sources during the differentiated time history of each source is as follows: The specific method for obtaining the contribution weight of each differentiated source at each moment is as follows:

[0127] Based on the increase in the relative independence of fire events at each moment compared to the previous moment, and the change in the number of differentiated sources at each moment compared to the previous moment, the contribution weight of each differentiated source at each moment is obtained.

[0128] As an example, let the length of the differential time history of the c-th differential source point be denoted as... ;

[0129] The c-th source of differentiation is in the... The relative independence of fire incidents at any given moment is denoted as ;

[0130] The first The number of differentiated source points at time t is denoted as ;

[0131] The contribution weight of the c-th differentiated source point at the current moment is calculated as follows:

[0132]

[0133] in, For the c-th differential source point in the th... The relative independence of fire incidents at any given moment; For the first The number of differential source points at any given time.

[0134] It should be noted that, The value indicates the first time compared to the first The greater the value of the increase in the relative independence of the fire event of the cth differentiated source point at the moment, the greater the data differentiation performance of the cth differentiated source point gradually increases over time, indicating that the data collected by the cth differentiated source point is less compatible when fused. The greater the number of collection source points that change into differentiated source points over time, the greater the influence of the cth differentiated source point on the digital twin of the fire platform, causing the data of other collection source points to no longer match the original digital twin of the fire platform, resulting in being judged as abnormal by the digital twin of the fire platform.

[0135] It should be further explained that, as the iteration of the collection data on the digital twin of the fire platform tends to be more inclined to the data collected by the cth differentiated source point, The value can be negative, but at the same time, as the influence of the digital twin of the fire platform gradually decreases, the number of differentiated source points will decrease , reflecting the influence of the cth differentiated source point on the digital twin of the fire platform, so when they are all negative, the contribution weight will still increase, so the greater the value of the contribution weight.

[0136] Step S004, using the contribution weight of each differentiated source point at all moments to predict the modeling support weight of each collection source point at the next moment; the modeling support weight is used as the data adoption ratio of each collection source point to construct the digital twin of the fire platform.

[0137] Based on the above steps, the contribution weight of each differentiated source point at each moment is obtained, and the contribution weight of each differentiated source point at all moments is input into the unscented Kalman filter to obtain the modeling support weight of each differentiated source point at the next moment;

[0138] It should be noted that if the collection source point is not a differentiated source point, the modeling support weight of each collection source point at the next moment is obtained by predicting the weight according to the prior decision; wherein the unscented Kalman filter described in the present embodiment is a known technology, and the present embodiment will not be described again.

[0139] Further, the modeling support weight of each differentiated source point at the next moment and the modeling support weight of each collection source point at the next moment are used as the data adoption ratio of each source data in each collection source point collected at the next moment when constructing the digital twin of the fire platform.

[0140] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. An artificial intelligence assisted fire platform digital twin construction method, characterized in that, The method comprises the following steps: Obtaining each item of source data of each acquisition source point at each time point by using a plurality of sensors arranged at each acquisition source point in the building model; wherein the time point of the last acquisition is recorded as the current time point; Screening each item of source data by using the prior abnormal threshold of each item of source data, obtaining a plurality of differentiated source points at the current time point, and constructing a fire event vector of each differentiated source point at each time point; screening an independent inspection event vector of each differentiated source point at the current time point by the deviation of the fire event vector of the same differentiated source point at the current time point from other time points, and obtaining the fire event independence of each differentiated source point at the current time point by using the time distribution closeness of the independent inspection event vector in all time points and the difference between the fire event vector and the independent inspection event vector of each differentiated source point at the current time point; Analyzing the distribution of the fire event independence of all differentiated source points at each time point, obtaining the relative independence of the fire event of each differentiated source point at each time point, analyzing the change of the relative independence of the fire event after the acquisition source point is changed into the differentiated source point, and obtaining the contribution weight of each differentiated source point at each time point; Obtaining the modeling support weight of each acquisition source point at the next time point by using the contribution weight of each differentiated source point at all time points; the modeling support weight is used as the data adoption proportion of each acquisition source point to construct the digital twin of the fire platform; The screening of each item of source data by using the prior abnormal threshold of each item of source data, the obtaining of a plurality of differentiated source points at the current time point, and the construction of the fire event vector of each differentiated source point at each time point comprise: If there is any one source data of all items of source data of a certain acquisition source point greater than or equal to the prior abnormal threshold of the source data at the current time point, the acquisition source point is recorded as a differentiated source point; The difference between each item of source data of each differentiated source point at each time point and the prior abnormal threshold thereof is recorded as the abnormal performance degree of each item of source data of each differentiated source point at each time point; The difference between each item of source data of each differentiated source point at each time point and the prior abnormal threshold thereof is recorded as the abnormal performance degree of each item of source data of each differentiated source point at each time point; The abnormal urgency degree of each item of source data of each differentiated source point at each time point is obtained, and the abnormal urgency degree is in a positive proportional relationship with the abnormal performance degree and the urgency performance degree; The abnormal urgency degree of each item of source data of each differentiated source point at each time point is obtained, and the abnormal urgency degree is in a positive proportional relationship with the abnormal performance degree and the urgency performance degree; 2. The method as claimed in claim 1, wherein the method is based on artificial intelligence assistance for a fire platform digital twin construction. The abnormal urgency degree of each item of source data of each differentiated source point at each time point is obtained, and the abnormal urgency degree is in a positive proportional relationship with the abnormal performance degree and the urgency performance degree. The specific obtaining steps of the independent inspection event vector comprise: Constructing a multi-source data space of each differentiated source point at the current time point by using the fire event vector of each differentiated source point at all time points; 3. The method of claim 2, wherein the method is performed by the system of claim 1. In the multi-source data space of each differentiated source point, a plurality of independent inspection event vectors of each differentiated source point at the current time point are screened according to the difference between the fire event vector at the current time point and the fire event vector at other time points. In the multi-source data space of each differentiated source point, a plurality of independent inspection event vectors of each differentiated source point at the current time point are screened according to the difference between the fire event vector at the current time point and the fire event vector at other time points. In the multi-source data space of each differentiated source point, according to the difference between the current time and each time except the current time, the spatial distance of each differentiated source point at the current time from the fire event vector of each time except the current time is obtained; After bipolar division of the spatial distance, a minimum difference time set of each differentiated source point at the current time is obtained; The fire event vector corresponding to each time in the minimum difference time set is taken as a number of independent inspection event vectors of each differentiated source point at the current time.

4. The method of claim 3, wherein the method is characterized by: The specific acquisition steps of the minimum difference time set include: The spatial distance of each differentiated source point at the current time from the fire event vector of each time except the current time is arranged in descending order to obtain a spatial distance ascending sequence of each differentiated source point at the current time; The difference between each sequence value and the next sequence value in the spatial distance ascending sequence is calculated, and the first difference value of each sequence value in the spatial distance ascending sequence of each differentiated source point at the current time is recorded; The sequence value with the maximum first difference value is recorded as a secondary segmentation sequence value; The set of time points corresponding to all sequence values in the closed interval formed by the first sequence value and the secondary segmentation sequence value in the spatial distance ascending sequence is recorded as the minimum difference time set of each differentiated source point at the current time.

5. The method of claim 1, wherein the method is a method of constructing a digital twin of a fire-fighting platform with the aid of artificial intelligence. The specific acquisition steps of the fire event independence include: According to the number and distribution time range of the independent inspection event vectors in all time points, the time sequence distribution tightness of each differentiated source point at the current time is obtained; According to the extreme value of the source data in the fire event vector of each differentiated source point at the current time and the data average dispersion of each item of source data in the independent inspection event vector, the source data dispersion degree of each differentiated source point at the current time is obtained; The fire event independence of each differentiated source point at the current time is obtained, and the fire event independence is in direct proportion to the time sequence distribution tightness and the source data dispersion degree.

6. The method of claim 5, wherein the method is performed by the system of claim 1. The specific acquisition steps of the time sequence distribution tightness include: The time sequence distribution tightness of each differentiated source point at the current time is obtained, and the time sequence distribution tightness is in direct proportion to the number of all independent inspection event vectors of each differentiated source point at the current time and is in inverse proportion to the range of time points of the independent inspection event vectors.

7. The method of claim 5, wherein the method further comprises: The specific acquisition steps of the source data dispersion degree include: The maximum value of all items of source data in the fire event vector of each differentiated source point at the current time is recorded as the extreme value of each differentiated source point at the current time; The average of the standard deviations of each item of source data in each independent inspection event vector of each differentiated source point at the current time is recorded as the data average dispersion of each item of source data in the independent inspection event vector; The source data dispersion degree of each differentiated source point at the current time is obtained, and the source data dispersion degree is in direct proportion to the extreme value and is in inverse proportion to the data average dispersion.

8. The method of claim 1, wherein the method is a method of constructing a digital twin of a fire-fighting platform with the aid of artificial intelligence. The specific acquisition steps of the relative independence of the fire event include: At the same time, the maximum value and the minimum value of the fire event independence of all differentiated source points are obtained; The difference between the fire event independence of each differentiated source point and the minimum value of the fire event independence is recorded as the fire event relative independence of each differentiated source point at each time, and the ratio of the difference between the maximum value and the minimum value of the fire event independence is recorded as the fire event relative independence of each differentiated source point at each time.

9. The method of claim 1, wherein the method is a method of constructing a digital twin of a fire-fighting platform with the aid of artificial intelligence. The specific obtaining step of the contribution weight comprises: Let the length of the differentiated time history of the cth differentiated source point be denoted as ; The cth differentiated source point is relative independence of the fire event at the moment is recorded as ; The first The number of differentiated source points at the moment is denoted as ; The calculation mode of the contribution weight of the cth differentiated source point at the current time is: wherein, is the fire event relative independence of the cth differentiated source point at the time instant; is the number of differentiated source points at the time instant.

Images