CN115081156B - Self-perception, self-decision and self-execution intelligent ventilation control platform and control method for mine - Google Patents

Abstract

The invention belongs to the technical field of intelligent ventilation control of mines, and particularly relates to an intelligent ventilation control platform and a control method for a mine, which realize intelligent control of automatic sensing acquisition of ventilation parameters, automatic analysis and decision of ventilation states and automatic feedback execution of ventilation facilities.

Description

Mine intelligent ventilation control platform and control method with self-perception, self-decision and self-execution

technical field

The invention belongs to the technical field of mine intelligent ventilation management and control, and specifically relates to a self-sensing, self-decision-making, and self-executing mine intelligent ventilation management and control platform and control method, which realizes automatic sensing and collection of ventilation parameters, automatic analysis and decision-making of ventilation status, and automatic ventilation facilities. Intelligent management and control of feedback execution.

Background technique

The intelligentization of the energy industry is already an inevitable trend in the current energy development. Coal is the main energy source in our country, and its intelligent development has been put on the agenda. However, compared with other industries, most coal mining enterprises in my country are still in the state of manual and semi-manual mining, and the level of intelligence is not high. From the current cases of coal mine safety accidents in my country, it is found that most of the major accidents are generally related to the ventilation system, but the current research on mine ventilation is still in the traditional stage, and it is impossible to realize real-time, reliable and stable on-demand adjustment, and it is also impossible to achieve The catastrophe emergency intelligent control of the ventilation system during the catastrophe period is not to mention intelligent.

From the perspective of the development trend of coal intelligence, ventilation intelligence is an important content of coal intelligence. The core technology of ventilation intelligence is to realize the unmanned autonomy of mine intelligent ventilation processes such as accurate data perception, intelligent decision-making of schemes, and remote adjustment and control. implement. Among them, pattern recognition and intelligent decision-making algorithms are the core of realizing the above technical content. Mine ventilation pattern recognition is an important algorithm for ventilation intelligence, and this part is less studied. In 2014, Professor Lu Xinming proposed a mine ventilation system state identification method and a multi-state automatic identification method. In this method, the roadway is graded, and the change of roadway wind resistance is identified through the wind speed, temperature and humidity sensors at key locations. This method is mainly aimed at the identification of wind resistance changes in the mine ventilation system, and cannot identify other types of ventilation state changes. In 2018, Liang Qichao and others proposed a multi-level pattern recognition method for mine ventilation. Multi-level pattern recognition adopts hierarchical indicators such as disaster resistance ability, wind flow stability, fan stability, fan operation efficiency, mine ventilation cost per ton, ventilation engineering cost, etc., and then realizes pattern recognition through cluster analysis, weighted distance, fuzzy algorithm, etc. judge. This method cannot effectively use the sensor data of the intelligent mine, and can only be applied to the optimization of manual adjustment schemes, and cannot identify various early warnings, faults, and catastrophes of the mine. Mine ventilation intelligent decision-making algorithm is one of the core algorithms of mine intelligent ventilation. The main methods of traditional mine ventilation scheme decision-making are divided into three categories: ① methods based on graph theory topological relationship construction, such as loop method, path method, etc.; ② methods based on nonlinear mathematical programming, such as Lagrange multiplier method etc.; ③ The solution method based on evolutionary computation. Several methods have their own advantages and disadvantages. The adjustment process of the graph theory topological relationship solution method relies on the participation of technicians, and the actual adjustment effect cannot reach the ideal value due to the change of the total wind resistance of the mine after a variable resistance, and it needs to be continuously adjusted until it meets the requirements. . The solution method of mathematical programming is slow, and there is also the problem of local optimum. In recent years, with the development of evolutionary computing, intelligent algorithms such as genetic algorithm (GA), simulated annealing algorithm (SA), and particle swarm optimization (PSO) have also been applied to mine ventilation, but the decision-making speed and accuracy for large-scale networks are low. It is difficult to apply in real time in actual field. The lack of existing technology has led to the failure of the key algorithm nodes of mine intelligent ventilation, and the inability to realize self-perception, self-decision-making, and self-execution.

Contents of the invention

In order to solve the problem that there is no intelligent ventilation management and control platform in the field of mine intelligent ventilation that can automatically sense and collect parameters, analyze decision-making, and remote feedback adjustment, the present invention provides a mine intelligent ventilation management and control platform that can sense, make, and execute itself, as well as its management and control. method.

The present invention adopts following technical scheme to realize:

A self-perception, self-decision and self-execution mine intelligent ventilation management and control platform, including:

The precise sensing module is used to collect various ventilation parameters of the ventilation pipe network in real time, and clean the collected parameters to form basic data, which is uploaded to the mine ventilation brain system;

The mine ventilation brain system is used to analyze the basic data provided by the precise perception module in real time, identify the operation mode of the ventilation system and generate a decision-making plan, and transmit the decision-making plan to the feedback adjustment module,

The mine ventilation brain system includes a ventilation pattern recognition module and a decision-making module,

The feedback adjustment module is used to receive a decision-making scheme, control the ventilation facilities and adjust the ventilation system.

Further, the precise perception module includes:

The ventilation parameter monitoring module is used to monitor the wind speed, wind pressure, gas type and gas temperature and humidity in the ventilation pipe network;

The ventilation power monitoring module is used to monitor the operating parameters and control parameters of ventilation facilities, including operating air volume, operating negative pressure, operating power, motor power, revolutions, and start-stop information;

The ventilation structure monitoring module is used to monitor the switch status, opening area, air leakage and pressure difference of the air door;

The dust monitoring module is used to monitor the dust concentration of the operating site and the return air tunnel.

Further, the ventilation parameter monitoring module includes a wind speed sensor, a wind pressure sensor, a gas sensor and a gas temperature and humidity sensor installed in the mine ventilation pipe network; the ventilation power monitoring module includes a wind speed sensor, a wind pressure sensor installed on the fan attachment Sensors, motor parameter sensors, speed sensors and start-stop sensors.

Further, the mine ventilation brain system also includes:

The dynamic visualization module of the 3D stereogram of the ventilation system is used to realize the front-end display of the true 3D ventilation system, realize the refined modeling and rendering of the 3D graphic elements and the dynamic real-time display of the ventilation sensor data,

The network calculation simulation module is used to realize the simulation calculation of the wind direction and air volume of the whole mine roadway and the operating condition of the fan according to the roadway wind resistance or sensor data, and provide another basic data for the ventilation mode recognition.

Further, the ventilation pattern recognition module is composed of a multi-channel parallel fuzzy detector, and the multi-channel parallel fuzzy detector includes a ventilation power detector, a ventilation resistance detector, a ventilation structure detector, an air volume supply-demand ratio detector, Roadway deformation detectors, fire detectors, dust detectors and gas detectors.

Further, the decision-making module includes a scheme optimization decision-making module and an emergency plan decision-making module,

The scheme optimization decision-making module is used for the regulation and control when the ventilation is abnormal with the adjustment of the air volume as the solution. By initializing the objective function and calculating the ventilation system adjustment scheme through the backbone particle swarm algorithm of parallel computing, the ventilation function of the ventilation system is realized. Find the optimal solution between power consumption and required air volume;

The emergency plan decision-making module is used for regulating the air volume as a solution for ventilation anomalies or disasters that are difficult to solve.

Further, the mathematical model (1) of the ventilation pattern recognition module is expressed as follows:

（1）

(1)

In the formula, χ _det is the ventilation mode output by the detector det , including normal, early warning, catastrophe and failure modes; ∨ is the maximum membership degree mode; σ _det ^{[ mod ]} is the membership degree function of the detector det mode mod ; X is the input feature, different detectors read different sensor data as input values.

Further, the scheme optimization decision-making module includes the following calculation process,

为方案优化决策的目标函数的表达式，由加权的经济项和安全项组成，通过求解min

得到最优方案；ω为通风功耗项的权重，0<ω<1；ζ为目标函数的经济项表达式；ψ为目标函数的安全项表达式；k为通风机的数量；l为用风地点数量；q _af 为通风机a的风量，m³/s；

为通风机a的风压特性函数；

为通风机a的功率特性函数；ε为一个避免分母为0的极小值；q _bs 为用风地点b的供风量，m³/s；q _bl 为用风地点b的风量下限，m³/s；q _bu 为用风地点b的风量上限，m³/s；q _br 为用风地点b的需风量，m³/s。In the formula,

The expression of the objective function for the optimal decision-making of the scheme, which is composed of weighted economic items and security items, by solving min

The optimal solution is obtained; ω is the weight of the ventilation power consumption item, 0< ω <1; ζ is the expression of the economic item of the objective function; ψ is the expression of the security item of the objective function; k is the number of fans ; The number of wind locations; q _af is the air volume of fan a , m ³ /s;

is the wind pressure characteristic function of fan a ;

is the power characteristic function of the fan a ; ε is a minimum value that avoids the denominator being 0; q _bs is the air supply volume of the wind-using site b , m ³ /s; q _bl is the lower limit of the air volume of the wind-using site b , m ³ /s; q _bu is the upper limit of wind volume at wind-using site b , m ³ /s; q _br is the required air volume at wind-using site b , m ³ /s.

，所有粒子通过不断向最优位置聚拢实现寻优，巷道数目设为m，待调风门数目设为s，粒子数目设为z，Constantly update the position of particle i and calculate the objective function of the corresponding position

, all particles are optimized by continuously gathering to the optimal position, the number of lanes is set to m , the number of dampers to be adjusted is set to s , the number of particles is set to z ,

The position of particle i is composed of the equivalent wind resistance of s dampers to be adjusted:

In the formula, R _i ( t ) is the position of particle i when it is updated for the tth time; r ₁ , r ₂ ,…, r _s are the equivalent wind resistances of all dampers to be adjusted,

The t +1th update equation of a single particle is as follows:

In the formula, R _i ( t + 1) is the sampling position of particle i at the t + 1th update; N is a normal distribution; μ _i ( t ) is the mean value of the normal distribution after the tth update; σ _i ( t ) is the variance of the normal distribution after the tth update;

Among them, the mean and variance of the normal distribution are obtained by the following formula:

In the formula, R _i ^* ( t ) is the optimal position of particle i at the t -th update; R _g ^* ( t ) is the optimal position of the population at the t -th update,

Use the ventilation network calculation algorithm to solve the air volume of all roadways of particle i at the t +1th update:

为通风网络解算算法；In the formula, Q _i ( t+1 ) is the air volume of all tunnels at the t + 1st calculation of particle i , composed of q ₁ , q ₂ , … , q _m , m ³ /s;

Solving algorithms for ventilation networks;

Calculate the global optimal position of the population and the optimal position of particle i after the t +1th update:

R _g ^* ( t + 1) is the global optimal position of the population after the t + 1th update, R _i ^* ( t + 1) is the optimal position of particle i after the t + 1th update, R is the particle position,

达到一定精度后，停止计算，将此时的种群最优位置对应的等效风阻R _g ^*转换为风门的开度大小，形成优化方案。When the calculation reaches the maximum number of iterations max t or the value of the objective function

After reaching a certain accuracy, the calculation is stopped, and the equivalent wind resistance R _g ^* corresponding to the optimal position of the population at this time is converted into the opening of the damper to form an optimization scheme.

Further, using a parallel computing architecture to parallelize the calculation process of the solution optimization decision, the parallel computing architecture includes a particle layer, a computing layer and a sharing layer,

及最优位置

，i=1, 2, …, z；The particle layer is composed of different processes. Each process contains several particles as the basic unit. There is no direct communication between the particle group processes. This layer stores the current position of all particles after the t -th update

and optimal location

, i =1, 2, ..., z ;

和全局最优值

，从粒子层中读取各粒子的最优位置

和粒子最优值

；(2)粒子位置与目标函数值计算，使用粒子的最优位置

和全局最优位置

计算t+1时刻粒子的位置

，进而计算该位置的目标函数值

；(3) 最优位置与最优值更新，若目标函数值

小于全局最优值

，则将第t+1次全局最优位置

设置为

，全局最优值设置为

；若目标函数值

小于粒子最优值

，则将第t+1次粒子最优位置

设置为

，粒子最优值

设置为

；The calculation layer connects the particle layer and the sharing layer, collects information from the sharing layer and the particle layer respectively, realizes data exchange and high-speed parallel computing between the two layers, and feeds back better calculation results to the particle layer and the sharing layer; The specific process of t + 1 update is: (1) Information collection, each process in the calculation layer collects the global optimal position from the shared layer

and the global optimum

, read the optimal position of each particle from the particle layer

and particle optimum

; (2) Calculation of particle position and objective function value, using the optimal position of the particle

and the global optimal position

Calculate the position of the particle at time t +1

, and then calculate the objective function value of the position

; (3) Update the optimal position and optimal value, if the objective function value

less than the global optimum

, then the t +1th global optimal position

Set as

, the global optimum is set to

; if the objective function value

less than particle optimal value

, then the optimal particle position of the t +1th time

Set as

, the optimal value of the particle

Set as

;

The shared layer is responsible for saving the global optimal position and global optimal value. It is a shared memory between multiple processes, and a process lock ensures the security of inter-process communication. Only one process can operate the shared memory at a time.

Furthermore, the parallel computing architecture is deployed on the cloud server to achieve high-speed decision-making calculations based on server load balancing.

Further, the mine intelligent management and control platform also includes:

The knowledge database module is used to store the structured and unstructured data of the ventilation system, receive and store the real-time data collected by the accurate perception module in real time, and the decision-making plan data of the decision-making module.

Further, the mine intelligent management and control platform also includes a ventilation pipe network model, and the ventilation pipe network model is a solid model built according to the scale of the real mine, and is used for intelligent ventilation teaching and scientific research.

A self-perception, self-decision and self-execution mine intelligent ventilation control method is completed based on a self-perception, self-decision, and self-execution mine intelligent ventilation control platform, including the following steps:

S1. The precise sensing module collects various ventilation parameters of the ventilation pipe network in real time, cleans the collected parameters, and uploads them to the mine ventilation brain system;

S2. The mine ventilation brain system analyzes and identifies the ventilation mode in real time, generates a decision-making plan, and transmits the decision-making plan to the feedback adjustment module;

S3. The feedback adjustment module receives the decision-making scheme, and controls the ventilation facilities to adjust the ventilation system.

Further, the ventilation mode recognition in the step S2 obtains the ventilation modes output by each detector through the mathematical model (1), including normal, early warning, failure, catastrophe and so on.

Further, when the operation mode of the ventilation system in the step S2 is identified as an abnormal ventilation that can be solved by adjusting the air volume, a decision-making scheme for optimization is generated; the operation mode of the ventilation system is identified as it is difficult to adjust the air volume as a solution When solving ventilation abnormalities or disasters, generate decision-making plans for emergency plans.

The mine ventilation brain system described in the present invention integrates intelligent ventilation algorithms and visualization software. The precise sensing module collects ventilation system parameters in real time and uploads them to the mine ventilation brain system for decision analysis. The feedback adjustment module executes the adjustment instructions of the mine ventilation brain system, realizing real-time Basic requirements for intelligent ventilation data perception, analysis and decision-making, remote execution, and knowledge storage. Obtain real-time parameters through the precise perception module, analyze and identify the status of the ventilation system through the mine ventilation brain system, give a decision-making plan, and simulate various mine ventilation methods, working face ventilation methods, mine normal ventilation and whole mine, local by adjusting the switch state of the damper Ventilation processes such as reverse wind can restore disaster relief decisions in the process of faults and disasters.

The invention can realize the perception and decision-making process of mine intelligent ventilation. The platform has the advantages of being practical, wide in use, strong applicability, and strong expandability, and provides ideas and products for the research and application of mine intelligent ventilation.

The following is a detailed description of the change of the ventilation mode of the intelligent ventilation management and control platform and the realization of local anti-wind through the accompanying drawings and related implementation plans.

Description of drawings

Figure 1 is a logical diagram of the various parts of the intelligent ventilation management and control platform.

Figure 2 is a flow chart of ventilation pattern recognition.

Figure 3 is a diagram of the parallel computing architecture for ventilation intelligent decision-making.

Figure 4 is a flow chart of ventilation scheme optimization decision-making.

Figure 5 shows the logic of perception, decision-making, and execution scenarios of the intelligent ventilation management and control platform.

Figure 6 shows the central side-by-side ventilation mode and its main ventilation routes of the intelligent ventilation control platform.

Figure 7 shows the diagonal ventilation mode of the two wings of the intelligent ventilation control platform and its main ventilation routes.

Figure 8 shows the ventilation route before the local backwind of the working face.

Figure 9 shows the ventilation route after local backwind on the working face.

In the picture:

1—wind tunnel, ventilator and accessories; 2—remote control damper;

3—closeable air shaft (open); 4—closeable air shaft (closed).

Detailed ways

The present invention will be clearly and completely described below in conjunction with the drawings and embodiments. Apparently, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

A self-sensing, self-decision-making, and self-executing mine intelligent ventilation management and control platform, including a mine ventilation brain system, a precise perception module, a feedback adjustment module, and a knowledge database.

The mine ventilation brain system is the core of the entire ventilation control platform, providing intelligent ventilation algorithms and visualization software support for the platform. The precise perception module collects various ventilation parameters of the ventilation pipe network in real time, cleans the collected parameters, and uploads them to the mine ventilation brain system and knowledge database. The mine ventilation brain system analyzes and identifies the operation mode of the ventilation system in real time and generates a decision-making plan, and transmits the analyzed mode and plan to the feedback adjustment module. The feedback adjustment module controls the ventilation facilities or fans to adjust the ventilation system, and the knowledge database stores the ventilation system. Structured and unstructured data, real-time reception and storage of sensor multi-source heterogeneous real-time data collected by the precise perception module, and decision-making plan data of the decision-making control module.

The mine ventilation brain system integrates functions such as dynamic visualization of three-dimensional stereograms of the ventilation system, network calculation simulation, ventilation pattern recognition, scheme optimization decision-making, and emergency plan decision-making. Among them, ventilation mode recognition and scheme optimization decision-making are the core functions of the system.

1. Ventilation pattern recognition

Ventilation pattern recognition is an algorithm that detects the operation pattern of the ventilation system in real time based on sensor data.

Figure 2 shows the flow chart of ventilation pattern recognition. Pattern recognition consists of multi-channel parallel fuzzy detectors, including ventilation power detectors, ventilation resistance detectors, ventilation structure detectors, air volume supply-demand ratio detectors, roadway deformation detectors, fire detectors, dust damage detectors, and gas detectors , each detector is composed of different nonlinear membership functions, which are used to detect different modes such as normal, early warning, fault, catastrophe, etc. The mathematical model is expressed as follows:

In the formula, χ _det is the ventilation mode output by the detector det , including normal, warning, fault and catastrophe modes corresponding to each detector; ∨ is the maximum membership mode; σ _det ^{[ mod ]} is the det mode mod of the detector Membership function; X is the input feature, and different detectors read different sensor data as input values.

The ventilation mode output by the detector det includes χ _{ventilation power} = "normal", "warning", "catastrophe" or "fault", χ _{ventilation resistance} = "normal", "warning", "catastrophe" or "fault", χ _{ventilation Structure} = "normal", "warning", "catastrophe" or "fault" , χ _{air volume supply and demand ratio} = "normal", "warning", "catastrophe" or "fault" , χ _{roadway deformation} = "normal", "warning" , "catastrophe" or "fault" , χ _fire = "normal", "early warning", "catastrophe" or "fault", χ _{dust damage} = _" normal", "early warning", "catastrophe" or "fault", χgas ="Normal", "Warning", "Catastrophe" or "Fault".

After the ventilation pattern recognition is completed, the knowledge database will use rule retrieval, matching or reasoning to select the method of scheme optimization decision-making or emergency plan decision-making. The scheme optimization decision is used to control the ventilation abnormal mode that can take the adjustment of the air volume as the solution, including the abnormality of the ventilation power detector χ _{ventilation power} = "fault" and the abnormality of the ventilation supply-demand ratio detector χ _{air volume supply-demand ratio} = "fault", And _other ventilation abnormalities χgas =“fault” _, χdust =“fault” that take the size of the air volume as a solution. Example 3 is the abnormality of the ventilation power detector and the abnormality of the ventilation supply-demand ratio detector. The emergency plan decision-making is the expert experience entered in the knowledge base, and it is the ventilation control for the abnormal ventilation or catastrophe mode that is difficult to solve by the scheme optimization decision-making algorithm, including χ _{ventilation resistance} = "fault", χ _{ventilation structure} = "fault", χ _{roadway Deformation} = "fault", χ _fire = "fault", χ _{ventilation power} = "catastrophe", χ _{ventilation resistance} = "catastrophe", χ _{ventilation structure} = "catastrophe", χ _{air volume supply and demand ratio} = "catastrophe", χ _{roadway deformation} = " catastrophe ", x _fire =" catastrophe ", x _gas =" catastrophe ", x _dust =" catastrophe "; Embodiment 1 is that ventilation resistance detector is abnormal, and embodiment 2 is that fire detector is abnormal.

The rest of the normal and warning will not trigger actions, but the warning will display the warning information on the 3D interface.

, Program optimization decision

With the advancement and succession of mining at the working face, it is difficult to ensure that the original ventilation scheme is still applicable to the changed ventilation system. It is difficult to take into account the power consumption of ventilation and the air volume required by the location of the ventilation system when formulating mine ventilation system schemes. Unreasonable mine ventilation schemes will cause huge waste of resources and even bring safety hazards. The plan optimization decision is to find the optimal solution between the ventilation power consumption and the required air volume of the ventilation system when adjusting the air volume daily, so that the system is always in the optimal state. When the ventilation power detector detects the abnormal power consumption of the main ventilator, or the ventilation supply-demand ratio detector detects the abnormal air supply volume of the wind location, the scheme optimization decision algorithm will be called to optimize the ventilation scheme. The content of the algorithm is as follows:

Construct an objective function based on economy (ventilation power consumption) and safety (air volume requirement):

为方案优化决策的目标函数的表达式，由加权的经济项和安全项组成，通过求解min

得到最优方案；ω为通风功耗项的权重，0<ω<1；ζ为目标函数的经济项表达式；ψ为目标函数的安全项表达式；k为通风机的数量；l为用风地点数量；q _af 为通风机a的风量，m³/s；

为通风机a的风压特性函数；

为通风机a的功率特性函数；ε为一个避免分母为0的极小值；q _bs 为用风地点b的供风量，m³/s；q _bl 为用风地点b的风量下限，m³/s；q _bu 为用风地点b的风量上限，m³/s；q _br 为用风地点b的需风量，m³/s。In the formula,

The expression of the objective function for the optimal decision-making of the scheme, which is composed of weighted economic items and security items, by solving min

The optimal solution is obtained; ω is the weight of the ventilation power consumption item, 0< ω <1; ζ is the expression of the economic item of the objective function; ψ is the expression of the security item of the objective function; k is the number of fans ; The number of wind locations; q _af is the air volume of fan a , m ³ /s;

is the wind pressure characteristic function of fan a ;

is the power characteristic function of the fan a ; ε is a minimum value that avoids the denominator being 0; q _bs is the air supply volume of the wind-using site b , m ³ /s; q _bl is the lower limit of the air volume of the wind-using site b , m ³ /s; q _bu is the upper limit of wind volume at wind-using site b , m ³ /s; q _br is the required air volume at wind-using site b , m ³ /s.

，所有粒子通过不断向最优位置聚拢实现寻优。巷道数目设为m，待调风门数目设为s，粒子数目设为z。The optimal decision-making scheme is obtained by using the improved backbone particle swarm optimization algorithm (BBPSO) to solve. The algorithm constantly updates the position of particle i and calculates the objective function of the corresponding position

, all particles achieve optimization by continuously gathering to the optimal position. The number of roadways is set to m , the number of dampers to be adjusted is set to s , and the number of particles is set to z .

The position of particle i is composed of the equivalent wind resistance of s dampers to be adjusted:

In the formula, R _i ( t ) is the position of particle i when it is updated for the tth time; r ₁ , r ₂ ,…, rs _are the equivalent wind resistances of all the dampers to be adjusted.

The t +1th update equation of a single particle is as follows:

In the formula, R _i ( t + 1) is the sampling position of particle i at the t + 1th update; N is a normal distribution; μ _i ( t ) is the mean value of the normal distribution after the tth update; σ _i ( t ) is the variance of the normal distribution after the tth update;

Among them, the mean and variance of the normal distribution are obtained by the following formula:

In the formula, R _i ^* ( t ) is the optimal position of particle i at the tth update; R _g ^* ( t ) is the optimal position of the population at the tth update.

Use the ventilation network calculation algorithm to solve the air volume of all roadways of particle i at the t +1th update:

为通风网络解算算法；In the formula, Q _i ( t+1 ) is the air volume of all tunnels at the t + 1st calculation of particle i , composed of q ₁ , q ₂ , … , q _m , m ³ /s;

Solving algorithms for ventilation networks;

Calculate the global optimal position of the population and the optimal position of particle i after the t +1th update:

In the formula: R _g ^* ( t + 1) is the global optimal position of the population after the t + 1th update, R _i ^* ( t + 1) is the optimal position of particle i after the t + 1th update, and R is particle position,

达到一定精度后，停止计算。将此时的种群最优位置对应的等效风阻R _g ^*转换为风门的开度大小，形成优化方案。When the calculation reaches the maximum number of iterations max t or the value of the objective function

When a certain accuracy is reached, the calculation is stopped. The equivalent wind resistance R _g ^* corresponding to the optimal position of the population at this time is converted into the opening degree of the damper to form an optimization scheme.

Using the process pool and shared memory technology, the parallel computing architecture of the backbone particle swarm optimization decision-making algorithm is designed to parallelize the above-mentioned calculation process, greatly improve the solution speed, and realize the real-time calculation of ventilation decision-making. As shown in Figure 3, the architecture includes a particle layer, a computing layer and a sharing layer.

及最优位置

，i=1, 2, …, z；The particle layer is composed of different processes. Each process contains several particles as the basic unit. There is no direct communication between the particle group processes. This layer stores the current position of all particles after the t -th update

and optimal location

, i =1, 2, ..., z ;

和全局最优值

，从粒子层中读取各粒子的最优位置

和粒子最优值

。(2)粒子位置与目标函数值计算。使用粒子的最优位置

和全局最优位置

计算t+1时刻粒子的位置

，进而计算该位置的目标函数值

。(3) 最优位置与最优值更新。若目标函数值

小于全局最优值

，则将第t+1次全局最优位置

设置为

，全局最优值设置为

；若目标函数值

小于粒子最优值

，则将第t+1次粒子最优位置

设置为

，粒子最优值

设置为

。The calculation layer connects the particle layer and the sharing layer, collects information from the sharing layer and the particle layer respectively, realizes data exchange and high-speed parallel computing between the two layers, and feeds back better calculation results to the particle layer and the sharing layer; The specific process of t +1 update is: (1) Information collection. Each process in the calculation layer collects the global optimal position from the shared layer

and the global optimum

, read the optimal position of each particle from the particle layer

and particle optimum

. (2) Calculation of particle position and objective function value. Optimal Positions for Using Particles

and the global optimal position

Calculate the position of the particle at time t +1

, and then calculate the objective function value of the position

. (3) Update the optimal position and optimal value. If the objective function value

less than the global optimum

, then the t +1th global optimal position

Set as

, the global optimum is set to

; if the objective function value

less than particle optimal value

, then the optimal particle position of the t +1th time

Set as

, the optimal value of the particle

Set as

.

The shared layer is responsible for saving the global optimal position and global optimal value. It is a shared memory between multiple processes, and a process lock ensures the security of inter-process communication. Only one process can operate the shared memory at a time.

Figure 4 is a flow chart of ventilation scheme optimization decision-making. The data collected by the sensor is recognized by the ventilation mode, and it is found that the air volume of the wind-using site is insufficient or the power consumption of the ventilation is unreasonable. Start the plan optimization decision calculation, initialize the objective function, calculate the ventilation system adjustment plan through the backbone particle swarm algorithm of parallel computing, and distribute instructions to the remote control equipment.

The feedback adjustment module takes the PLC control system as the core, and consists of a host computer, a controller, a sensor, etc., and is used to remotely execute the commands of the mine ventilation brain system, and also supports manual control through the control box, including frequency conversion main ventilators, Frequency conversion local ventilator, remote control damper, sealable air intake shaft. The mine ventilation brain system analyzes and accurately perceives the ventilation parameters collected by the cleaning module. The mine ventilation brain system analyzes and decides to give a ventilation system adjustment plan, and distributes instructions to the feedback adjustment module. The system can realize real-time adjustment and on-demand distribution of the air volume of the ventilation system by changing the working condition of the fan, the opening and closing of the air door, and the opening and closing state of the air shaft.

The ventilation parameter monitoring module is installed in the mine ventilation pipe network. The ventilation pipe network is the carrier of ventilation air flow and includes various types of roadways, such as mining working faces, transportation alleys, and inlet and return air roadways. The ventilation pipe network is equipped with hardware equipment such as precise sensing module, feedback adjustment module, disaster simulation module, industrial ring network and transmission substation. The disaster simulation module can control mine dust damage, mine heat damage, gas explosion, coal and gas outburst, mine fire and other processes at specific locations in the pipeline network. The industrial ring network and transmission substation are used for equipment communication and provide reliable channels for data transmission. When the mine intelligent ventilation management and control platform described in the present invention is used for intelligent ventilation teaching and scientific research, it also includes a ventilation pipe network model, which is a solid model built according to the scale of the real mine, generally 1:10.

Through the mine ventilation brain system, the ventilator and damper are remotely controlled, and different combination changes of fan blade angle and damper switch mode can be realized, which can realize a variety of mine ventilation methods, working face ventilation methods, and mine reverse wind simulation.

1. Various mine ventilation modes can be realized by changing the state of the corresponding damper: ① Diagonal type ② Regional type ③ Central side-by-side type ④ Mixed type.

2. By changing the state of the corresponding damper, it can simulate a variety of working face ventilation methods, such as: U-type, Z-type, Y-type.

3. By changing the angle of the fan blades and the state of the corresponding air door, the full mine reverse wind and partial reverse wind can be realized.

The above instructions depend on the designed damper arrangement scheme of the ventilation pipe network.

The knowledge database accesses the real-time sensor data and knowledge data generated by various mine information systems in a unified format, and provides them to various application systems through a unified data access interface in the form of a service to complete real-time and historical knowledge data. Quick exchange.

Example 1: Change of ventilation mode - central side-by-side type → two-wing diagonal type

The technical route of Example 1 is: accurate data sensing and collection → ventilation pattern recognition → abnormal ventilation resistance detector → knowledge database matching → emergency plan decision-making

Generally, newly-built mines adopt a central parallel ventilation method, and both the inlet and outlet air shafts are located in the industrial square. However, with the gradual mining, the working face is getting farther and farther away from the industrial square, resulting in longer ventilation routes and increased ventilation resistance. At this time, it is necessary to change the ventilation method to two-wing diagonal type to reduce ventilation resistance. The intelligent ventilation control platform can simulate this scenario.

As shown in Figure 6, the ventilation mode of the intelligent ventilation management and control platform is central parallel. The central side-by-side ventilation method is a mine ventilation method in which both the inlet and outlet air shafts are located in the center of the system. Two air intake shafts are opened, and one air intake shaft is closed. Through a specific damper control scheme, two air intake shafts in the center of the system are formed, and the right-wing fan is connected to the return air shaft in the center of the system. Central parallel ventilation mode. According to the main ventilation route in the figure, it can be seen that the ventilation route is longer and the resistance is greater.

Step 1: The precise sensing module obtains the time-series data of wind speed and wind pressure through the wind speed and wind pressure sensors deployed on the main ventilators and main ventilation routes, and performs data cleaning on the wind speed and wind pressure sensor data. The cleaning process mainly includes data outlier detection and filtering smoothing.

Step 2: Upload the cleaned data to the mine ventilation brain system, and call the ventilation pattern recognition algorithm to identify the status of the ventilation system. When the monitoring values of the wind speed and wind pressure sensors are input into the ventilation resistance detector, the following formula can be used to judge whether the resistance of the entire ventilation pipe network meets the reasonable resistance distribution range:

In the formula, χ1 is the ventilation resistance detector; V is the monitoring value of the wind speed sensor, m/ _s ; P is the monitoring value of the wind pressure sensor, Pa.

When χ ₁ = "warning" or "fault", it means that the ventilation resistance is about to or has exceeded the reasonable resistance range.

Step 3: Input the result of state recognition into the knowledge database, and trigger the emergency plan event in which the ventilation mode of the whole mine is changed from the central parallel type to the two-wing diagonal type through retrieval, matching or reasoning rules.

Step 4: Distribute the action commands to the relevant ventilation facilities in the ventilation pipe network through the feedback adjustment module. After the execution is completed, the execution results and the executed ventilation system data are fed back to the mine ventilation brain system, and stored in the knowledge database.

As shown in Figure 7, the ventilation mode of the intelligent ventilation control platform is a two-wing diagonal style. Two air intake shafts are opened, and one air intake shaft is closed. Through a specific air door control scheme (the air door on the ventilation route in the figure is opened), the air intake shaft is located in the center of the system, and the return air shaft is located at the opposite corners of the left and right wings. type ventilation. According to the main ventilation route in the figure, it can be seen that the ventilation route is greatly shortened and the resistance is reduced.

Example 2: Partial anti-wind at the working face during the fire period

The technical route of Embodiment 2 is: Accurate data sensing and collection→ventilation pattern recognition→fire detector abnormality→knowledge database matching→emergency plan decision-making

Local backwind usually occurs in a fire scene, and it can prevent toxic and harmful gases from flowing to the operating site or the location of personnel, and is an effective means of mine disaster emergency rescue.

Take the working surface of the ventilation system of the left wing as an example, as shown in Figure 8. If a fire breaks out at the flame mark, observe the direction of wind flow. The toxic and harmful gas formed by the fire will be transported to the working face, and the workers on the working face will be harmed by the toxic and harmful gas. At this time, the damper near the working face must be regulated to change the direction of the wind flow from upwind to downwind, so as to effectively ensure the fresh air flow into the work site, avoid poisonous and harmful gases from pouring into the work face, causing casualties, and provide emergency rescue for mines. Precious time. The mine intelligent ventilation management and control platform can simulate the process of fire disaster and ventilation emergency rescue.

Step 1: The precise sensing module acquires real-time wind speed, temperature, humidity, and time-series data of each gas concentration by deploying wind speed, temperature, humidity, and gas sensors near the working face and along the trough of the working face, and monitors the working status of the working face in real time.

Step 2: Upload the cleaned data to the mine ventilation brain system, and call the ventilation pattern recognition algorithm to identify the status of the ventilation system. When the wind speed, temperature and humidity, and gas sensor monitoring values are input into the fire detector, the following formula can be used to determine whether a fire has occurred, as well as the location and level of the catastrophe.

In the formula, χ ₂ is the fire detector; V is the monitoring value of the wind speed sensor, m ³ /s; T is the monitoring value of the temperature sensor, ℃; C is the monitoring value of the gas sensor, %.

Step 3: When χ ₂ = "catastrophe", input the state recognition result into the knowledge database, and trigger the emergency plan event of local backwind on the working face through retrieval, matching or reasoning rules.

Step 4: Distribute the action command to the relevant dampers near the working face through the feedback adjustment module. After the execution is completed, confirm whether the location of the personnel on the working face is safe, and generate the next emergency strategy, such as calculating the disaster avoidance route. Feedback the execution results and the executed ventilation system data to the mine ventilation brain system, and store them in the knowledge database.

As shown in Figure 9, it is only necessary to change the opening and closing conditions of the four dampers, and the working face can be changed from upward ventilation and local reverse wind to downward ventilation. At this time, the toxic and harmful gas generated by the fire will not be transported to the work site, which can ensure the safety of the workers.

Example 3: Scheme optimization decision-making during normal ventilation period

The technical route of Embodiment 3 is: Accurate data sensing and collection→ventilation pattern recognition→abnormality of ventilation power or ventilation supply-demand ratio detector→knowledge database matching→scheme optimization decision-making

The plan optimization decision-making during the normal ventilation period is to optimize and control the ventilation system on a daily basis, so that the ventilation system is always in the state of optimal economy and highest safety. The decision-making process of scheme optimization is illustrated by an example.

Step 1: The precise sensing module monitors the operating conditions, operating power consumption and efficiency of the main fans in real time through the wind speed, wind pressure, and motor parameter sensors installed at the main fans, and the wind speed, wind speed, and Gas, temperature and humidity, and positioning sensors monitor the required air volume and actual air supply volume of the working face in real time.

Step 2: Upload the working conditions, power consumption, efficiency and other data of the main ventilators and the wind demand and supply data of each working face to the mine ventilation brain system, and call the ventilation power detector and the supply-demand ratio detector to identify the ventilation mode.

In the formula, χ ₃ is a ventilation power detector; χ ₄ is a ventilation supply-demand ratio detector; W is the power of the main fan, kW; η is the efficiency of the main fan, %; L is the number of people working on the working face;

Step 3: When χ ₃ or χ ₄ = "failure", invoke the intelligent decision-making algorithm of the scheme. Initialize the objective function, the global optimal value and the optimal position, and z particles, each particle is composed of s wind resistance of adjustable dampers. A collection of particles looks like this:

Step 4: Allocate z particles to the multi-core CPU to generate a parallel computing pool of p processes, and each process allocates z / p particles for calculation.

，个体最优位置R _i ^*，粒子群的全局最优值

与全局最优位置R _g ^*。根据更新规则更新个体当前位置，同时更新个体与种群最优值与最优位置。第t+1次参数更新计算过程如下：Step 5: Calculate the individual optimal values of z particles respectively

, the individual optimal position R _i ^* , the global optimal value of the particle swarm

and the global optimal position R _g ^* . Update the current position of the individual according to the update rule, and update the optimal value and optimal position of the individual and the population at the same time. The calculation process of the t +1th parameter update is as follows:

达到预先设定的要求或超过一定迭代步数max t后，停止计算，记录此时的全局最优位置：Step 6: Repeat step 5 continuously, when the global optimal value

After reaching the pre-set requirements or exceeding a certain number of iterations max t , stop the calculation and record the global optimal position at this time:

Step 7: Convert the windage resistance of each damper to be adjusted in R _g ^* into a specific damper adjustment opening, and distribute each damper adjustment command through the feedback adjustment module. After the execution is completed, the execution result and the executed ventilation system data are fed back to the mine ventilation brain system, and stored in the knowledge database.

Claims (10)

1. A self-perception, self-decision and self-execution mine intelligent ventilation control system, characterized in that it includes: The precise sensing module is used to collect various ventilation parameters of the ventilation pipe network in real time, and clean the collected parameters to form basic data, which is uploaded to the mine ventilation brain system; The mine ventilation brain system is used to analyze the basic data provided by the precise perception module in real time, identify the operation mode of the ventilation system and generate a decision-making plan, and transmit the decision-making plan to the feedback adjustment module, The mine ventilation brain system includes a ventilation pattern recognition module and a decision-making module, The ventilation pattern recognition module is composed of multi-channel parallel fuzzy detectors, and the mathematical model (1) of the ventilation pattern recognition module is expressed as follows:

(1) In the formula, χ _det is the ventilation mode output by the detector det , including normal, early warning, catastrophe and failure modes; ∨ is the maximum membership degree mode; σ _det ^{[ mod ]} is the membership degree function of the detector det mode mod ; X is the input feature, different detectors read different sensor data as input values; The feedback adjustment module is used to receive decision-making schemes, control the ventilation facilities to adjust the ventilation system, The knowledge database module is used to store the structured and unstructured data of the ventilation system, receive and store the real-time data collected by the accurate perception module in real time, and the decision-making plan data of the decision-making module. 2. The self-sensing, self-decision-making and self-executing mine intelligent ventilation management and control system according to claim 1, wherein the precise sensing module includes a ventilation parameter monitoring module and a ventilation power monitoring module, and the ventilation parameter monitoring module includes The wind speed sensor, wind pressure sensor, gas sensor and gas temperature and humidity sensor installed in the mine ventilation pipe network; the ventilation power monitoring module includes the wind speed sensor, wind pressure sensor, motor parameter sensor, speed sensor, and switch stop sensor. 3. The self-sensing, self-decision-making and self-executing mine intelligent ventilation management and control system according to claim 1 or 2, wherein said multi-channel parallel fuzzy detector comprises a ventilation power detector, a ventilation resistance detector, a ventilation structure detection Detector, Air Volume Supply and Demand Ratio Detector, Roadway Deformation Detector, Fire Detector, Dust Damage Detector and Gas Detector. 4. The self-aware, self-decision-making and self-executing mine intelligent ventilation management and control system according to claim 3 is characterized in that said decision-making module comprises a scheme optimization decision-making module and an emergency plan decision-making module, The scheme optimization decision-making module is used for the regulation of the ventilation mode when the ventilation is abnormal with the adjustment of the air volume as the solution. By initializing the objective function and calculating the ventilation system adjustment scheme through the backbone particle swarm algorithm of parallel computing, the ventilation system is realized. Find the optimal solution between system ventilation power consumption and required air volume; The emergency plan decision-making module is used to adjust the air volume as a solution for ventilation anomalies or disasters that are difficult to solve. 5. The self-perception, self-decision and self-execution mine intelligent ventilation management and control system according to claim 4, characterized in that said scheme optimization decision-making module includes the following calculation process:

In the formula,

The expression of the objective function for the optimal decision-making of the scheme, which is composed of weighted economic items and security items, by solving min

The optimal solution is obtained; ω is the weight of the ventilation power consumption item, 0< ω <1; ζ is the expression of the economic item of the objective function; ψ is the expression of the security item of the objective function; k is the number of fans ; The number of wind locations; q _af is the air volume of fan a , m ³ /s;

is the wind pressure characteristic function of fan a ;

is the power characteristic function of the fan a ; ε is a minimum value that avoids the denominator being 0; q _bs is the air supply volume of the wind-using site b , m ³ /s; q _bl is the lower limit of the air volume of the wind-using site b , m ³ /s; q _bu is the upper limit of wind volume at wind site b , m ³ /s; q _br is the required air volume at wind site b , m ³ /s; Constantly update the position of particle i and calculate the objective function of the corresponding position

, all particles are optimized by continuously gathering to the optimal position, the number of lanes is set to m , the number of dampers to be adjusted is set to s , the number of particles is set to z , The position of particle i is composed of the equivalent wind resistance of s dampers to be adjusted:

In the formula, R _i ( t ) is the position of particle i when it is updated for the tth time; r ₁ , r ₂ ,…, r _s are the equivalent wind resistances of all dampers to be adjusted, The t +1th update equation of a single particle is as follows:

In the formula, R _i ( t + 1) is the sampling position of particle i at the t + 1th update; N is a normal distribution; μ _i ( t ) is the mean value of the normal distribution after the tth update; σ _i ( t ) is the variance of the normal distribution after the tth update; Among them, the mean and variance of the normal distribution are obtained by the following formula:

In the formula, R _i ^* ( t ) is the optimal position of particle i at the t -th update; R _g ^* ( t ) is the optimal position of the population at the t -th update, Use the ventilation network calculation algorithm to solve the air volume of all roadways of particle i at the t +1th update:

In the formula, Q _i ( t+1 ) is the air volume of all tunnels at the t + 1st calculation of particle i , composed of q ₁ , q ₂ , … , q _m , m ³ /s;

Solving algorithms for ventilation networks; Calculate the global optimal position of the population and the optimal position of particle i after the t +1th update:

In the formula: R _g ^* ( t + 1) is the global optimal position of the population after the t + 1th update, R _i ^* ( t + 1) is the optimal position of particle i after the t + 1th update, R is particle position, When the calculation reaches the maximum number of iterations max t or the value of the objective function

After reaching a certain accuracy, the calculation is stopped, and the equivalent wind resistance R _g ^* corresponding to the optimal position of the population at this time is converted into the opening of the damper to form an optimization scheme. 6. The self-perception, self-decision and self-execution mine intelligent ventilation management and control system according to claim 5 is characterized in that the calculation process of the plan optimization decision is parallelized by using a parallel computing architecture, and the parallel computing architecture includes a particle layer and a computing layer and the shared layer, The particle layer is composed of different processes. Each process contains several particles as the basic unit. There is no direct communication between the particle group processes. This layer stores the current position of all particles after the t -th update

and optimal location

, i =1, 2,…, z ; The calculation layer connects the particle layer and the sharing layer, collects information from the sharing layer and the particle layer respectively, realizes data exchange and high-speed parallel computing between the two layers, and feeds back better calculation results to the particle layer and the sharing layer; The specific process of t + 1 update is: (1) information collection, each process in the calculation layer collects the global optimal position from the shared layer

and the global optimum

, read the optimal position of each particle from the particle layer

and particle optimum

; (2) Calculation of particle position and objective function value, using the optimal position of the particle

and the global optimal position

Calculate the position of the particle at time t +1

, and then calculate the objective function value of the position

; (3) Update the optimal position and optimal value, if the objective function value

less than the global optimum

, then the t +1th global optimal position

Set as

, the global optimum is set to

; if the objective function value

less than particle optimal value

, then the optimal particle position of the t +1th time

Set as

, the optimal value of the particle

Set as

; The shared layer is responsible for saving the global optimal position and global optimal value. It is a shared memory between multiple processes, and a process lock ensures the security of inter-process communication. Only one process can operate the shared memory at a time. 7. The self-perception, self-decision and self-execution mine intelligent ventilation management and control system according to claim 6, characterized in that the mine intelligent ventilation management and control system also includes a ventilation pipe network model, and the ventilation pipe network model is based on real A mock-up of the scaled construction of the mine. 8. A self-perception, self-decision and self-execution mine intelligent ventilation control method is completed based on the self-perception, self-decision and self-execution mine intelligent ventilation control system as described in one of claims 1-7, characterized in that it comprises the following steps: S1. The precise sensing module collects various ventilation parameters of the ventilation pipe network in real time, cleans the collected parameters, and uploads them to the mine ventilation brain system; S2. The mine ventilation brain system analyzes and identifies the ventilation mode in real time, generates a decision-making plan, and transmits the decision-making plan to the feedback adjustment module; S3. The feedback adjustment module receives the decision-making scheme, and controls the ventilation facilities to adjust the ventilation system. 9. The self-perception, self-decision and self-execution mine intelligent ventilation control method according to claim 8, characterized in that in the step S2, the ventilation pattern recognition obtains the ventilation pattern output by each detector through the mathematical model (1), including Normal, early warning, failure and catastrophe. 10. The self-sensing, self-decision-making and self-executing mine intelligent ventilation control method according to claim 9, characterized in that the operating mode of the ventilation system in the step S2 is identified as an abnormal ventilation that can be solved by adjusting the air volume , to generate a decision-making scheme for scheme optimization; when the operating mode of the ventilation system is identified as a ventilation anomaly or disaster that is difficult to solve by adjusting the air volume as a solution, a decision-making scheme for an emergency plan is generated.

Images