Designing an improved geoacoustic event location algorithm in the "Prognoz-ADS" system

. The article shows complex algorithm for lo cating geoacoustic events. Location algorithms problems are considered and methods for their solution are proposed. The requirement list is suggested for the developed algorithm, and the complex algorithm key stages are formulated.


Introduction
Obtaining reliable acoustic emission events coordinates is one of the main tasks of seismicacoustic m onitoring i n a rock m ass [1,2]. At prese nt, a number of algorit hms and calculation m ethods for determining t he l ocation h ave been de veloped by t he M ining Institute of th e Far Eastern Branch of th e Russian Acad emy o f Scien ces. Man y of t hem have been introduced into t he "Prognoz-ADS" software package and are successfully used to assess the geomechanical state of mine fields [3][4][5]. The developed methods are based on the analysis of the signal registration tim e by each of the receive rs. The re are a lso approaches based on other principles [6]. Each of the methods used has its own advantages and di sadvantages. This i mposes re strictions o n t heir use. It i s proposed t o develop approaches for creating a universal location algorithm in the presented work, in order to use it for m ost receiving antennas configurations. Proven analytical methods will be applied to solve t his prob lem. Also , so me sig nificant im provements to i ncrease of t he locat ion accuracy and obtain additional information about the recorded signal will be used to rea ch this target [7].
At present, location algorithms based on the coordinate descent and brute force methods are used in most cases to determine the seismic-acoustic events coordinates [8].
The e vents places and em pirical discrepa ncy value are t he calculation results by each algorithm. The di screpancy allows evaluating accuracy of determining the location. There is a way to combine several algorithms in Geoacoustic-ADS program. This method consists in stage d s olving the determining c oordinate's tas k using each algorithm and c hoosing a solution with the smallest discrepancy value.
The choice of the methods used and t he setting of their parameters exist, including, for example, setting the starting point for the coordinate descent method and the grid step for the brute force method.
The c onstant value o f t he c alculated s peed of s ound propagation i n t he rock m ass, regardless of t he signal recei ver, is a disadvantage of all the algor ithms use d [9]. Some modifications of t he presented m ethods al low varying t his val ue, but o nly f or al l se nsors together. This approach leads to th e sound speed averaging in the rock mass. This distorts the information about the registered signal sources.
At th e sam e t ime, th e relativ e co mplexity o f setting parameters b efore p erforming calculations i s an other drawback of e xisting m ethods. T his i mposes a number o f requirements to the monitoring system operator qualification.
There a re problems of wrong location determination i n th e case of a flat receiving antenna and the lack of an a ssessment of s ignal receiver's effective ness in eac h s pecific case in Geoacoustic-ADS program [10].
The authors propose to formulate and develop a number of mathematical and s oftware tools to so lve these problems. Th ese tools will lay th e fou ndation for creatin g a un iversal complex hi gh-precision an d hi gh-performance al gorithm for determining t he sei smicacoustic ev ents lo cation. Th e d eveloped app roach will pro vide more accu rate information about signals propagation. This will serv e as an additional tool for studying the geological anisotropy of rock mass.

Complex algorithm conception
The propagation speed of acoustic waves will be different not only for each receiver, but also for one receiver along the entire signal path due to geol ogical anisotropy and t he presence of empty spaces in the researched zones [11]. Therefore, it is proposed to vary the calculated signal propagation speed for the elements of the observing network in relation to each case of si gnal registration. This will make it possi ble to obtain a velocity map and to determine the eve nts c oordinates with a higher accurac y [12]. After time, data on the velocity d istribution will allo w no t only to in crease th e p erformance of h igh-precision location determination, but also to assess the rock massif state [13].
Time of si gnal regi stration varying i s proposed as an other al ternative ap proach t o improving the quality of determining the events location [14]. This approach will not allow getting close to obtaining real data on the longitudinal waves propagation, but can be used to assess th e co ntribution of th e observational n etwork elem ents fo r th e calcu lation accuracy. This will allow more efficient t uning of t he i nitial data for performing further complex algorithm iterations.
Obtaining in accurate data on th e lo cation of th e sign al sou rce is po ssible wh en using one-component sensors. Therefore, it is proposed to perform calculation with exclusion of such sensors from the receivi ng antenna one-by-one, if the requirem ent for their m inimum number i s c onfirmed. T his a pproach a nd i nformation on t he ge ometric o rientation of t he receivers will allow y ou t o get a re present a bout se nsors efficiency a nd m ore accura tely calculate the observation network sensitivity in the controlled zones [15].
The final list of proposed solutions includes:  varying the sound speed in the rock mass for each receiver;  calculation of the si gnal prop agation m ap by recalculating t he s peed values a fter registering a new event;  varying of the signal registration tim e to estim ate the weight of eac h receiving antenna element a by the contribution to the total value of the residual;  changing the quantitative composition of the recei ving antenna to assess the quality indicators of elements.

Consideration of the designed algorithm characteristics
The developed algorithm should the requirements of reliability, validity and efficiency of calculations and complies with the aspects presented in Figure 1. Accuracy provides for obtaining a so lution with a minimum d iscrepancy value characterizing a seismic-acoustic event clos e to the act ual lo cation. Ac hievement of the specified accuracy will b e carried out using the effective selection of variable parameters, reconfiguring the receiving antenna and the order of the mathematical methods involved in the calculation.
The universality of the algorithm means the possibility of using it fo r the configuration of a flat antenna. It is necessary to provide for obtaining s everal solutions with t he lowest discrepancy values to solve the flat antenna problem at the initial stage, and in the future, to obtain the most reliable position of the signal source, taking into account the orientation of the receivers in space and the calculated sensitivity zone.
Automation will reduce the work of the monitoring system operator on setting the initial parameters and on analyzing the resulting solution. It is planned to increase the degree of automation by using pilot industrial tests. Th e choice of a set of mathematical tools for the operation of the complex algorithm, the adjustment of the initial parameters will be carried out according to the test results. In a ddition, a unique methodology should be cre ated for processing the signals recorded by the monitoring system for each controlled object [16].
Performance is cha racterized by the location speed decision and is associated with the qualitative assignment of the initial data and preliminary information about the location of the acoustic s ignal s ource a s ge ometric c oordinates of the p ossible l ocation zo ne. In addition, it is planned to implement an adaptive step of the varied parameters depending on the results obtained at each iteration to increase the efficiency of calculations.

Logical structure of the complex algorithm
Let is consider the main stages of the proposed algorithm.
Calculation of location using the brute force method and obtaining a set of coordinates of the seismic-acoustic signal source probable location.
Initialization of th e cu rrent or creation of a b ase m ap of th e velocity d istribution an d recalculation of the empirical discrepancy using available data and the signal trajectory (1).
• ∑ , (2) where -d istance between the well with the installed i-th sensor and the seismoacoustic event, m; -dista nce from a seis moacoustic event t o the nearest sensor that registered a si gnal, m; -signal propagation speed, m/sec. The distortion coefficient is determined for each signal receiver by varying the tim e of signal arri val and estim ating th e ch ange in th e residual value. T his c oefficient ca n t ake values from 0 to 1, and the total value will be 1. The receivers stack is formed according to the results of this stage, wher e the sensor with the hi ghest distortion coefficient placed on the list peak.
Then the trace from the recei ver with the maxim um absolute signal di stortion to t he current location of the source is constructed and the geometric domain list is formed along the si gnal p ath. Sign al propagation velocities for these domains vary and the em pirical discrepancy is recalculated taking into account the velocity map. The selection of velocities continues until the discrepancy reaches a threshold critical value.
Further, the distortion coe fficients are reca lculated for t he receive rs re maining in t he stack. As a result, a new linked list is formed and the selection of velocities continues along the trajectory of the next element of the receiving antenna.
The m ost accurate c oordinates of the si gnal s ource a nd the parameters of this calculation are saved during for all iterations of the complex algorithm.
The solutions are derived for the case of a flat antenna to be processed by the operator. If the sensitivity zone of the controlled area is known, the solution is selected automatically.
The velocity distrib ution m ap is rebuilt as a resu lt of analysis o f obtained so lutions, specifying information on the propagation of sound waves in the rock mass.
Simplified block diagram of the developed algorithm is shown in Figure 2. The m ain co mponents of t he algo rithm wit h a simplified estimate o f th eir co mplexity are presented in Table 1. The number of receivers in comparison with the number of calculated spatial domains is much less. Therefo re, regardless of th e so rting al gorithm used, t he greatest load on t he computing power will be the number of calculated domain and them brute force operation to determine event coordinates with complexity O (n).

Conclusions
The approaches discussed in the article will solve a number of location methods problems used in th e "Progn oz-ADS" system cu rrently. A co mplex lo cation al gorithm will allo w solving the problem of a flat receiving a ntenna a nd obtaining information about the propagation o f s ound waves i n a r ock m ass from a sei smic-acoustic s ignal source. The proposed ap proaches will be ach ieved by m eeting th e requ irements fo r precision, universality, au tomation and p erformance. The in itial co ncept of a co mplex lo cation algorithm was form ulated in th e presen ted work an d th e au thors will carry out its implementation and further development.
The reported study was funded by RFBR a nd NSFC according to the research project No. 20-55-53028.