Coal rock pre-failure precursor feature extraction method and system based on difference network

Abstract

This invention discloses a method and system for extracting precursor features of coal and rock failure under load based on differential networks, comprising: monitoring acoustic emission (AE) signals and electromagnetic radiation (EMR) signals during the coal and rock failure process, wherein AE and EMR sensors are arranged at different spatial positions on the sample; setting a sliding step size s and a time window length l, and processing the AE signal time series A t ={A i t‑l+1 ,...,A i t‑2 A i t‑1 A i t} and EMR signal time series E t ={E j t‑l+1 ,...,E j t‑2 E j t‑1 E j t} as sample data at time t; construct reference samples and perturbation samples, calculate the Pearson correlation coefficient between every two channels of each, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t Compare the differences in correlation networks at adjacent time points to construct difference network sequences (DNs) at different time points. t ; Calculate DN t The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of rapid increase in (t) is the critical point for coal and rock failure under load. This invention is of great significance for early warning and prevention of coal and rock dynamic disasters.

Description

Technical Field

[0001] This invention relates to the field of coal and rock load-induced damage monitoring and early warning technology, and in particular to a method and system for extracting precursor features of coal and rock load-induced damage based on differential networks. Background Technology

[0002] Rockbursts, coal and gas outbursts, and other coal and rock dynamic disasters seriously threaten the safe and efficient production of mines. Coal and rock failure under load exhibits obvious nonlinear rheological abrupt changes. During the dynamic development of the coal and rock system, there is a critical change phenomenon. When far from the abrupt change point, coal and rock failure is not obvious, but when the critical point is reached, the coal and rock suddenly become unstable and fail within a very short time, leading to coal and rock dynamic disasters. Exploring multi-dimensional monitoring data throughout the entire lifecycle of coal and rock failure under load and extracting precursor features applicable to the critical state of coal and rock failure is of great significance for early warning and prevention of coal and rock dynamic disasters. However, how to extract reasonable and effective precursor feature indicators for coal and rock failure under load remains an urgent problem to be solved. Summary of the Invention

[0003] This invention provides a method and system for extracting precursor features of coal and rock failure under load based on differential networks. The technical solution is as follows:

[0004] On the one hand, a method for extracting precursor features of coal and rock failure under load based on differential networks is provided, including:

[0005] S1. Acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock are monitored by sensors. The AE and EMR sensors are respectively arranged at different spatial positions of the sample.

[0006] S2. Set the sliding step size s and the time window length l. At time t, the AE signal time series A t ={A i t-l +1,...,A i t-2 A i t-1 A i t} and EMR signal time series E t ={E j t-l+1 ,...,E j t-2 E j t-1 E j t} As sample data at time t, i and j represent the number of channels of the AE and EMR sensors, respectively;

[0007] S3. Construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={N1,N2,...,N n};

[0008] S4. Compare the differences in correlation networks at adjacent time points and construct difference network sequences (DNs) at different time points. t ={DN2,DN3,...,DN n};

[0009] S5. Calculate the differential network sequence DN t ={DN2,DN3,...,DN n The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of sharp increase in (t) is the critical point for coal and rock failure under load.

[0010] Optionally, S3 specifically includes:

[0011] S31. Select the sample data corresponding to time point T = ts as the reference sample, and the sample data corresponding to time point T = t as the perturbation sample;

[0012] S32. Calculate the Pearson correlation coefficient between each pair of acoustic emission and electromagnetic radiation channels of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix.

[0013] S33. Construct a reference sample correlation network N with edge values ​​assigned using the Pearson correlation coefficient matrix. t-s =(G t-s E t-s W t-s ) and the correlation network N of perturbed samples t =(G t E t W t This forms a correlation network sequence N with edge assignments. t ={N1,N2,...,N n};

[0014] The correlation network is represented in the form N = (G, E, W), where G represents the set of nodes, E represents the set of edges, and W represents the set of weights of the edges connecting two nodes, representing the strength of the interaction between nodes. The correlation network uses AE and EMR sensor channels as nodes, and the Pearson correlation coefficient between each pair of channels represents the weight of the edge connecting the two nodes. Various elements are used to show the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels.

[0015] Optionally, S4 specifically includes:

[0016] Compare N t-s and N t The network structure, calculate N t-s and N t The difference in the Pearson correlation coefficients of corresponding sides of the two networks is taken as the difference network DN. t The weight of the edge, DN t =(G t E t W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges.

[0017] Among them, DN t The edges connecting two nodes in the network represent temporal difference correlations, representing the differential correlations between different channels at different spatial monitoring locations. The perturbation sample correlation network N... t Correlation network N with reference sample t-s The inconsistent edges reflect the difference between the coal and rock state at T=t and T=ts;

[0018] After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={DN2,DN3,...,DN n}, where the difference network DN n The weights of the edges are the correlations between adjacent time points in the network DN. n With DN n-1 The difference in edge weights.

[0019] Optionally, S5 specifically includes:

[0020] Calculate I(t) and its I according to the following formula. a (t):

[0021] I(t) = mean(W) t -W t-s )

[0022]

[0023] Optionally, during the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, the correlation network structure at adjacent time points is similar, and the difference network contains very few edges; when the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, the correlation network at adjacent time points has a large structural difference, and many difference edges appear in the difference network.

[0024] During the loading and failure process of different types of coal and rock, the coal and rock state is stable during the normal period, and the I(t) fluctuates little. Except for a few small fractures that cause slight fluctuations, the overall state is stable. Correspondingly, I a I(t) increases steadily over time; as the load increases, the coal and rock are in a critical warning period, during which I(t) increases significantly, reaching a maximum peak value. a (t) rises sharply, with the rate of change being the largest, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until complete failure.

[0025] On the other hand, a system for extracting precursor features of coal and rock failure under load based on differential networks is provided, the system comprising:

[0026] The monitoring module is used to monitor acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock through sensors. The AE and EMR sensors are arranged at different spatial positions of the sample.

[0027] The configuration module is used to set the sliding step size s, the time window length l, and at time t, the AE signal time series A t ={A i t-l+1 ,...,A i t-2 A i t-1 A i t} and EMR signal time series E t ={E j t-l+1 ,...,E j t-2 E j t-1 E j t} As sample data at time t, i and j represent the number of channels of the AE and EMR sensors, respectively;

[0028] The first construction module is used to construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={N1,N2,...,N n};

[0029] The second construction module is used to compare the differences in correlation networks at adjacent time points and construct the difference network sequence DN at different time points. t ={DN2,DN3,...,DN n};

[0030] The computation module is used to calculate the difference network sequence DN. t ={DN2,DN3,...,DN n The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of sharp increase in (t) is the critical point for coal and rock failure under load.

[0031] Optionally, the first construction module is specifically used for:

[0032] The sample data corresponding to time point T = ts is selected as the reference sample, and the sample data corresponding to time point T = t is selected as the perturbation sample;

[0033] Calculate the Pearson correlation coefficients between every two channels of acoustic emission and electromagnetic radiation of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix.

[0034] Using the Pearson correlation coefficient matrix, construct reference sample correlation networks N with edge assignments. t-s =(G t-s E t-s W t-s ) and the correlation network N of perturbed samples t =(G t E t W t This forms a correlation network sequence N with edge assignments. t ={N1,N2,...,N n};

[0035] The correlation network is represented in the form N = (G, E, W), where G represents the set of nodes, E represents the set of edges, and W represents the set of weights of the edges connecting two nodes, representing the strength of the interaction between nodes. The correlation network uses AE and EMR sensor channels as nodes, and the Pearson correlation coefficient between each pair of channels represents the weight of the edge connecting the two nodes. Various elements are used to show the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels.

[0036] Optionally, the second construction module is specifically used for:

[0037] Compare N t-s and N t The network structure, calculate N t-s and N t The difference in the Pearson correlation coefficients of corresponding sides of the two networks is taken as the difference network DN. t The weight of the edge, DN t =(G t E t W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges.

[0038] Among them, DN t The edges connecting two nodes in the network represent temporal difference correlations, representing the differential correlations between different channels at different spatial monitoring locations. The perturbation sample correlation network N... t Correlation network N with reference sample t-s The inconsistent edges reflect the difference between the coal and rock state at T=t and T=ts;

[0039] After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={DN2,DN3,...,DN n}, where the difference network DN n The weights of the edges are the correlations between adjacent time points in the network DN. n With DN n-1 The difference in edge weights.

[0040] Optionally, the computing module is specifically used for:

[0041] Calculate I(t) and its I according to the following formula. a (t):

[0042] I(t) = mean(W) t -W t-s )

[0043]

[0044] Optionally, during the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, the correlation network structure at adjacent time points is similar, and the difference network contains very few edges; when the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, the correlation network at adjacent time points has a large structural difference, and many difference edges appear in the difference network.

[0045] During the loading and failure process of different types of coal and rock, the coal and rock state is stable during the normal period, and the I(t) fluctuates little. Except for a few small fractures that cause slight fluctuations, the overall state is stable. Correspondingly, I a I(t) increases steadily over time; as the load increases, the coal and rock are in a critical warning period, during which I(t) increases significantly, reaching a maximum peak value. a (t) rises sharply, with the rate of change being the largest, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until complete failure.

[0046] On the other hand, an electronic device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, which is loaded and executed by the processor to implement the above-described method for extracting precursor features of coal and rock under load based on differential networks.

[0047] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-described method for extracting precursor features of coal and rock under load based on differential networks.

[0048] The beneficial effects of the technical solution provided by this invention include at least the following:

[0049] This invention first monitors acoustic emission and electromagnetic radiation signals during the loading and failure process of coal and rock. Second, it sets the sliding step size and time window length, using the time series of acoustic emission and electromagnetic radiation signals as sample data. Then, it constructs reference samples and perturbed samples, calculates the Pearson correlation coefficient between every two channels of the reference and perturbed samples, establishes the Pearson correlation coefficient matrix, and constructs a correlation network sequence with edge assignments. Next, it compares the differences in the correlation networks at adjacent time points, constructing a difference network sequence at different time points. Finally, it calculates the average value I(t) and cumulative value I of the edge weights of the difference network sequence. a (t) serves as a precursor characteristic, I(t) and I aThe period of sharp increase in I(t) is the critical point of coal and rock failure under load. This invention can extract the precursor features of the critical state of coal and rock failure, and creatively proposes precursor feature indices I(t) and I(t) based on differential networks. a (t), I(t) and I a (t) integrates the signal characteristics of different spatial monitoring points, which better reflects the spatiotemporal evolution process and damage state of coal and rock, and is of great significance for early warning and prevention of coal and rock dynamic disasters. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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 schematic diagram of the execution flow of a method for extracting precursor features of coal and rock under load based on differential networks, provided in an embodiment of the present invention.

[0052] Figure 2 This invention provides a synchronous acquisition system for coal and rock splitting load-induced destructive force, acoustic and electrical signals.

[0053] Figure 3 This is a schematic diagram of the Pearson correlation coefficient matrix provided in an embodiment of the present invention;

[0054] Figure 4 A dynamic change diagram of the correlation network provided in an embodiment of the present invention;

[0055] Figure 5 A dynamic change diagram of the differential network provided in an embodiment of the present invention;

[0056] Figure 6 This is a schematic diagram illustrating the precursory characteristics of different types of coal and rock splitting under load, provided in an embodiment of the present invention.

[0057] Figure 7 This is a block diagram of a coal and rock load-induced failure precursor feature extraction system based on differential networks, provided in an embodiment of the present invention.

[0058] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0059] Appendix Figure 2 Chinese Attachment Description:

[0060] 1-Pressure machine, 2-Splitting fixture, 3-Sample, 4-Electromagnetic radiation antenna, 5-Electromagnetic radiation amplifier, 6-Acoustic emission sensor, 7-Acoustic emission amplifier, 8-Electromagnetic shielding room, 9-High-speed data acquisition instrument, 10-Data storage and analysis computer. Detailed Implementation

[0061] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0062] This invention provides a method for extracting precursor features of coal and rock failure under load based on differential networks, including:

[0063] S1. Acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock are monitored by sensors. The AE and EMR sensors are respectively arranged at different spatial positions of the sample.

[0064] S2. Set the sliding step size s and the time window length l. At time t, the AE signal time series A t ={A i t-l +1,...,A i t-2 A i t-1 A i t} and EMR signal time series E t ={E j t-l+1 ,...,E j t-2 E j t-1 E j t} As sample data at time t, i and j represent the number of channels of the AE and EMR sensors, respectively;

[0065] S3. Construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={N1,N2,...,N n};

[0066] S4. Compare the differences in correlation networks at adjacent time points and construct difference network sequences (DNs) at different time points. t ={DN2,DN3,...,DN n};

[0067] S5. Calculate the differential network sequence DN. t={DN2,DN3,...,DN n The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of sharp increase in (t) is the critical point for coal and rock failure under load.

[0068] The process of coal and rock failure under load can be viewed as a systemic behavior. Coal and rock instability is not caused by individual coal and rock micro-elements, but by the correlation, regulation, and imbalance among large-scale coal and rock micro-elements. This invention views the occurrence and development of coal and rock failure under load as a dynamic process of spatiotemporal evolution of a complex network system, studying the precursor characteristics of coal and rock failure under load from a spatiotemporal perspective. Furthermore, geophysical methods such as acoustic emission and electromagnetic radiation exhibit advantages in real-time, non-destructive, and continuous dynamic monitoring, accurately perceiving the entire process of internal load-induced damage in coal and rock. These methods are widely used in laboratory experiments and early warning of coal and rock dynamic disasters in engineering sites. Based on this, this invention provides a method for extracting precursor characteristics of coal and rock failure under load based on differential networks, starting from acoustic emission and electromagnetic radiation monitoring data from laboratory coal and rock failure experiments. The following is a combination of... Figure 1-6 This invention provides a detailed description of a method for extracting precursor features of coal and rock failure under load based on differential networks, comprising:

[0069] S1. Acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock are monitored by sensors. The AE and EMR sensors are respectively arranged at different spatial positions of the sample.

[0070] The embodiments of the present invention are constructed as follows Figure 2 The coal and rock splitting failure force-acoustic-electric synchronous acquisition system shown was used to conduct splitting failure experiments on different types of coal and rock samples, monitoring the force-acoustic-electric characteristics throughout the entire splitting failure process. In the experiment, each sample was equipped with two acoustic emission sensors and three sets of pairwise orthogonal triaxial electromagnetic radiation loop antennas to collect electromagnetic radiation data in the X, Y, and Z directions. The entire experiment was conducted in a GP1-A type quick-assembly and disassembly shielded chamber to resist environmental electromagnetic interference. The experiment employed a displacement-controlled loading method with a loading rate of 5 μm / s. A preload of 0.5 kN was applied before the experiment began. The rock samples were prepared according to standard requirements, using standard Brazilian splitting test specimens (dimensions: Ф50 mm × 25 mm). Eight types of coal and rock samples were selected for the experiment, and their physical parameters are shown in the table below.

[0071]

[0072]

[0073] S2. Set the sliding step size s and the time window length l. At time t, the AE signal time series A t ={Ai t-l +1,...,A i t-2 A i t-1 A i t} and EMR signal time series E t ={E j t-l+1 ,...,E j t-2 E j t-1 E j t} As sample data at time t, i and j represent the number of channels of the AE and EMR sensors, respectively;

[0074] In this embodiment of the invention, the sliding step size s = 0.5s, the time window length l = 0.5s, the number of AE sensor channels i = 2, and the number of EMR sensor channels j = 9.

[0075] S3. Construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={N1,N2,...,N n};

[0076] Optionally, S3 specifically includes:

[0077] S31. Select the sample data corresponding to time point T = ts as the reference sample, and the sample data corresponding to time point T = t as the perturbation sample;

[0078] S32. Calculate the Pearson correlation coefficient between each pair of acoustic emission and electromagnetic radiation channels of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix.

[0079] Suppose two sequences x(x1, x2, ..., xn) of length n. n ), y(y1,y2,…,y n The formula for calculating the Pearson correlation coefficient between the two is:

[0080]

[0081] Taking coal sample C1 as an example, its Pearson correlation coefficient matrix at t = 0.5 s is as follows: Figure 3 As shown, the diagonal lines represent the correlation coefficient of 1 within the same channel. Figure 3 The depth of the shading for each data point represents the magnitude of the correlation coefficient; the larger the correlation coefficient, the deeper the shading.

[0082] S33. Construct a reference sample correlation network N with edge values ​​assigned using the Pearson correlation coefficient matrix. t-s =(G t-s E t-s W t-s ) and the correlation network N of perturbed samples t =(G t E t W t This forms a correlation network sequence N with edge assignments. t ={N1,N2,...,N n};

[0083] The correlation network is represented in the form N = (G, E, W), where G represents the set of nodes, E represents the set of edges, and W represents the set of weights of the edges connecting two nodes, representing the strength of the interaction between nodes. The correlation network uses AE and EMR sensor channels as nodes, and the Pearson correlation coefficient between each pair of channels represents the weight of the edge connecting the two nodes. Various elements (dot and line color, thickness, etc.) are used to display the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels.

[0084] The dynamic changes of the correlation network are shown in the figure below. Figure 4 As shown.

[0085] S4. Compare the differences in correlation networks at adjacent time points and construct difference network sequences (DNs) at different time points. t ={DN2,DN3,...,DN n};

[0086] Optionally, S4 specifically includes:

[0087] Compare N t-s and N t The network structure, calculate N t-s and N t The difference in the Pearson correlation coefficients of corresponding sides of the two networks is taken as the difference network DN. t The weight of the edge, DN t =(G t E t W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges.

[0088] Among them, DN t The edges connecting two nodes in the network represent temporal difference correlations, representing the differential correlations between different channels at different spatial monitoring locations. The perturbation sample correlation network N... tCorrelation network N with reference sample t-s The inconsistent edges reflect the difference between the coal and rock state at T=t and T=ts;

[0089] After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={DN2,DN3,...,DN n}, where the difference network DN n The weights of the edges are the correlations between adjacent time points in the network DN. n With DN n-1 The difference in edge weights.

[0090] The dynamic changes of the differential network are shown in the figure below. Figure 5 As shown.

[0091] During the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, and the correlation network structure at adjacent time points is similar, with very few edges in the difference network. When the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, and the correlation network at adjacent time points shows significant structural differences, with many difference edges appearing in the difference network.

[0092] S5. Calculate the differential network sequence DN. t ={DN2,DN3,...,DN n The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of sharp increase in (t) is the critical point for coal and rock failure under load.

[0093] Optionally, S5 specifically includes:

[0094] Calculate I(t) and its I according to the following formula. a (t):

[0095] I(t) = mean(W) t -W t-s )

[0096]

[0097] This invention proposes a method for extracting precursor features of coal and rock failure under load based on acoustic emission and electromagnetic radiation monitoring data from laboratory coal and rock loading failure experiments. The time-series monitoring data is converted into a difference network reflecting the spatiotemporal characteristics of coal and rock instability under load. The average value I(t) and cumulative value I of the edge weights of the difference network sequence are calculated. a (t) serves as a precursor characteristic, I(t) and I aThe period of sharp increase in (t) is the critical point for coal and rock failure under load. The precursory characteristics of splitting and load failure in different types of coal and rock are as follows: Figure 6 As shown.

[0098] Depend on Figure 6 It can be seen that during the splitting and loading failure process of different types of coal and rock, the coal and rock state is stable during the normal period, and I(t) fluctuates little. Except for a few small fractures that cause slight fluctuations, the overall state is stable. Correspondingly, I a I(t) increases steadily over time; as the load increases, the coal and rock are in a critical warning period, during which I(t) increases significantly, reaching a maximum peak value. a The rate of change of I(t) increases sharply, reaching its maximum, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until complete failure. Therefore, I(t) and I... a (t) can reflect the unstable state of coal and rock under load from a spatiotemporal perspective. It integrates the signal characteristics of different spatial monitoring points. Its sharp rise can be used as an early warning feature of coal and rock under load failure. It reflects the spatiotemporal evolution process and damage state of coal and rock during the process of coal and rock under load failure, and has a certain early warning effect on coal and rock under load instability.

[0099] like Figure 7 As shown, this embodiment of the invention also provides a system for extracting precursor features of coal and rock failure under load based on differential networks, the system comprising:

[0100] Monitoring module 710 is used to monitor acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock through sensors. The AE and EMR sensors are respectively arranged at different spatial positions of the sample.

[0101] The setting module 720 is used to set the sliding step size s, the time window length l, and at time t, the AE signal time series A t ={A i t-l+1 ,...,A i t-2 A i t-1 A i t} and EMR signal time series E t ={E j t-l+1 ,...,E j t-2 E j t-1 E j t} As sample data at time t, i and j represent the number of channels of the AE and EMR sensors, respectively;

[0102] The first construction module 730 is used to construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish a Pearson correlation coefficient matrix, and construct a correlation network sequence N with edge values ​​assigned. t ={N1,N2,...,N n};

[0103] The second construction module 740 is used to compare the differences in correlation networks at adjacent time points and construct difference network sequences DN at different time points. t ={DN2,DN3,...,DN n};

[0104] Module 750 is used to compute the differential network sequence DN. t ={DN2,DN3,...,DN n The average value I(t) and its cumulative value I of the edge weights a (t) serves as a precursor characteristic, I(t) and I a The period of rapid increase in (t) is the critical point for coal and rock failure under load.

[0105] Optionally, the first construction module is specifically used for:

[0106] The sample data corresponding to time point T = ts is selected as the reference sample, and the sample data corresponding to time point T = t is selected as the perturbation sample;

[0107] Calculate the Pearson correlation coefficients between every two channels of acoustic emission and electromagnetic radiation of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix.

[0108] Using the Pearson correlation coefficient matrix, construct the reference sample correlation network N with edge assignment. t-s =(G t-s E t-s W t-s ) and the correlation network N of perturbed samples t =(G t E t W t This forms a correlation network sequence N with edge assignments. t ={N1,N2,...,N n};

[0109] The correlation network is represented in the form N = (G, E, W), where G represents the set of nodes, E represents the set of edges, and W represents the set of weights of the edges connecting two nodes, representing the strength of the interaction between nodes. The correlation network uses AE and EMR sensor channels as nodes, and the Pearson correlation coefficient between each pair of channels represents the weight of the edge connecting the two nodes. Various elements are used to show the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels.

[0110] Optionally, the second construction module is specifically used for:

[0111] Compare N t-s and N t The network structure, calculate N t-s and N t The difference in the Pearson correlation coefficients of corresponding sides of the two networks is taken as the difference network DN. t The weight of the edge, DN t =(G t E t W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges.

[0112] Among them, DN t The edges connecting two nodes in the network represent temporal difference correlations, representing the differential correlations between different channels at different spatial monitoring locations. The perturbation sample correlation network N... t Correlation network N with reference sample t-s The inconsistent edges reflect the difference between the coal and rock state at T=t and T=ts;

[0113] After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={DN2,DN3,...,DN n}, where the difference network DN n The weights of the edges are the correlations between adjacent time points in the network DN. n With DN n-1 The difference in edge weights.

[0114] Optionally, the computing module is specifically used for:

[0115] Calculate I(t) and its I according to the following formula. a (t):

[0116] I(t) = mean(W) t -W t-s )

[0117]

[0118] Optionally, during the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, the correlation network structure at adjacent time points is similar, and the difference network contains very few edges; when the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, the correlation network at adjacent time points has a large structural difference, and many difference edges appear in the difference network.

[0119] During the loading and failure process of different types of coal and rock, the coal and rock state is stable during the normal period, with small fluctuations in I(t). Except for minor fluctuations when a few small fractures occur, the overall state is stable. Correspondingly, I a I(t) increases steadily over time; as the load increases, the coal and rock are in a critical warning period, during which I(t) increases significantly, reaching a maximum peak value. a (t) rises sharply, with the rate of change being the largest, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until complete failure.

[0120] The coal and rock load-induced failure precursor feature extraction system based on differential networks provided in this embodiment of the invention has a functional structure that corresponds to the coal and rock load-induced failure precursor feature extraction method based on differential networks provided in this embodiment of the invention, and will not be described again here.

[0121] Figure 8 This is a schematic diagram of the structure of an electronic device 800 provided in an embodiment of the present invention. The electronic device 800 may vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 801 and one or more memories 802. The memory 802 stores at least one instruction, which is loaded and executed by the processor 801 to implement the steps of the above-mentioned method for extracting precursor features of coal and rock loading failure based on differential networks.

[0122] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions that can be executed by a processor in a terminal to complete the aforementioned method for extracting precursor features of coal and rock under load based on differential networks. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0123] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for extracting precursor features of coal and rock failure under load based on differential networks, characterized in that, The method includes: S1. Acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock are monitored by sensors. The AE and EMR sensors are respectively arranged at different spatial positions of the sample. S2. Set the sliding step size s Time window length l At any moment t , the time series A of the AE signal t ={ A i t-l +1 ,..., A i t-2 , A i t-1 , A i t } and EMR signal time series E t ={ E j t-l+1 ,..., E j t-2 , E j t-1 , E j t } as a moment t Sample data, i , j These represent the number of channels for the AE and EMR sensors, respectively. S3. Construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={ N 1, N 2, ..., N n }; S3 specifically includes: S31, Select time point T= t - s The corresponding sample data is used as the reference sample, and the time point T = t The corresponding sample data is used as the perturbation sample; S32. Calculate the Pearson correlation coefficient between each pair of acoustic emission and electromagnetic radiation channels of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix. S33. Construct reference sample correlation networks with edge values ​​assigned using the Pearson correlation coefficient matrix. N t-s =( G t-s , E t-s , W t-s Correlation network of perturbed samples N t =( G t , E t , W t This forms a correlation network sequence N with edge assignments. t ={ N 1, N 2, ..., N n }; Among them, correlation networks are based on N =( G , E , W Represented in the form of ) G Represents a set of nodes. E Describe the set of edges. W The set of weights of the edges connecting two nodes represents the strength of the interaction between the nodes. The correlation network uses AE and EMR sensor channels as nodes. The Pearson correlation coefficient between each pair of channels represents the weight of the edges connecting the two nodes. Various elements are used to show the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels. S4. Compare the differences in correlation networks at adjacent time points and construct difference network sequences (DNs) at different time points. t ={ DN 2, DN 3,..., DN n }; S4 specifically includes: Compare N t-s and N t Network structure, computation N t-s and N t The difference in the Pearson correlation coefficients between the corresponding sides of the two networks is used as the difference network. DN t The weight of the edge. DN t =( G t , E t , W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges. in, DN t The edge connecting two nodes in the network represents the temporal difference correlation, indicating the difference correlation between different channels at different spatial monitoring locations, thus perturbing the sample correlation network. N t Correlation network with reference sample N t-s Inconsistent edges reflect T= t Coal and rock state and T= t - s Differences in coal and rock conditions over time; After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={ DN 2, DN 3, ..., DN n }, where the difference network DN n The weights of the edges are the correlation between adjacent time points in the network. DN n and DN n-1 The difference in edge weights; S5. Calculate the differential network sequence DN t ={ DN 2, DN 3, ..., DN n Average edge weights I ( t ) and its cumulative value I a ( t As a precursor characteristic, I ( t )and I a ( t The period of rapid increase in load is the critical point for coal and rock failure under load.

2. The method according to claim 1, characterized in that, S5 specifically includes: Calculate the following formula: I ( t )and I a ( t ): 。 3. The method according to claim 1, characterized in that, During the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, and the correlation network structure at adjacent time points is similar, with very few edges in the difference network. When the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, and the correlation network at adjacent time points shows significant structural differences, with many difference edges appearing in the difference network. During the loading and failure process of different types of coal and rock, the state of coal and rock is stable during the normal period. I ( t The fluctuations are relatively small, except for minor fluctuations when a few small fractures occur; overall, the situation remains stable. I a ( t It steadily increases over time; As the load increases, the coal and rock are in a critical warning period. I ( t It rose sharply, reaching a very high peak. I a ( t The rate of change rises sharply, reaching its maximum, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until they are completely destroyed.

4. A system for extracting precursor features of coal and rock failure under load based on differential networks, characterized in that, The system includes: The monitoring module is used to monitor acoustic emission (AE) and electromagnetic radiation (EMR) signals during the loading and failure process of coal and rock through sensors. The AE and EMR sensors are arranged at different spatial positions of the sample. The settings module is used to set the sliding step size. s Time window length l At any moment t , the time series A of the AE signal t ={ A i t-l +1 ,..., A i t-2 , A i t-1 , A i t } and EMR signal time series E t ={ E j t-l+1 ,..., E j t-2 , E j t-1 , E j t } as a moment t Sample data, i , j These represent the number of channels for the AE and EMR sensors, respectively. The first construction module is used to construct reference samples and perturbation samples based on the sample data, calculate the Pearson correlation coefficient between every two channels of the reference samples and perturbation samples respectively, establish the Pearson correlation coefficient matrix, and construct the correlation network sequence N with edge assignment. t ={ N 1, N 2, ..., N n }; The first construction module is specifically used for: Selected time point T= t - s The corresponding sample data is used as the reference sample, and the time point T = t The corresponding sample data is used as the perturbation sample; Calculate the Pearson correlation coefficients between every two channels of acoustic emission and electromagnetic radiation of the reference sample and the disturbed sample respectively, and establish the Pearson correlation coefficient matrix. Using the Pearson correlation coefficient matrix, construct reference sample correlation networks with edge assignments. N t-s =( G t-s , E t-s , W t-s Correlation network of perturbed samples N t =( G t , E t , W t This forms a correlation network sequence N with edge assignments. t ={ N 1, N 2,..., N n }; Among them, correlation networks are based on N =( G , E , W Represented in the form of ) G Represents a set of nodes. E Describe the set of edges. W The set of weights of the edges connecting two nodes represents the strength of the interaction between the nodes. The correlation network uses AE and EMR sensor channels as nodes. The Pearson correlation coefficient between each pair of channels represents the weight of the edges connecting the two nodes. Various elements are used to show the interaction between significantly correlated feature nodes in order to find the relationship between data from different channels. The second construction module is used to compare the differences in correlation networks at adjacent time points and construct the difference network sequence DN at different time points. t ={ DN 2, DN 3, ..., DN n }; The second construction module is specifically used for: Compare N t-s and N t Network structure, computation N t-s and N t The difference in the Pearson correlation coefficients between the corresponding sides of the two networks is used as the difference network. DN t The weight of the edge. DN t =( G t , E t , W t -W t-s The difference network is a network structure constructed by using the difference in the Pearson correlation coefficient of the correlation network at adjacent time points as the weight of the edges. in, DN t The edge connecting two nodes in the network represents the temporal difference correlation, indicating the difference correlation between different channels at different spatial monitoring locations, thus perturbing the sample correlation network. N t Correlation network with reference sample N t-s Inconsistent edges reflect T= t Coal and rock state and T= t - s Differences in coal and rock conditions over time; After the above steps, the time series monitoring data is converted into a differential network sequence (DN) reflecting the spatiotemporal characteristics of coal and rock instability under load. t ={ DN 2, DN 3, ..., DN n }, where the difference network DN n The weights of the edges are the correlation between adjacent time points in the network. DN n and DN n-1 The difference in edge weights; The calculation module is used to calculate the difference network sequence DN. t ={ DN 2, DN 3, ..., DN n Average edge weights I ( t ) and its cumulative value I a ( t As a precursor characteristic, I ( t )and I a ( t The period of rapid increase in load is the critical point for coal and rock failure under load.

5. The system according to claim 4, characterized in that, The computing module is specifically used for: Calculate the following formula: I ( t )and I a ( t ): 。 6. The system according to claim 4, characterized in that, During the stable phase of the coal-rock system, the correlation between data from different spatial monitoring locations remains relatively stable, and the correlation network structure at adjacent time points is similar, with very few edges in the difference network. When the coal-rock system is about to fracture, the correlation between data from different spatial monitoring locations changes drastically, and the correlation network at adjacent time points shows significant structural differences, with many difference edges appearing in the difference network. During the loading and failure process of different types of coal and rock, the state of coal and rock is stable during the normal period. I ( t The fluctuations are relatively small, except for minor fluctuations when a few small fractures occur; overall, the situation remains stable. I a ( t It steadily increases over time; As the load increases, the coal and rock are in a critical warning period. I ( t It rose sharply, reaching a very high peak. I a ( t The rate of change rises sharply, reaching its maximum, at which point the critical point for coal and rock failure under load is reached; subsequently, the coal and rock become unstable until they are completely destroyed.

Images