Intelligent deep metal mine stope temperature control strategy

Abstract

The invention discloses an intelligent deep metal mine stope temperature control strategy, and belongs to the technical field of ore stope temperature control, and the strategy comprises the steps: reading the flow, temperature, pressure head and other actual system information in a PLC data block in real time; identifying characteristic parameters of each power component and the pipe network through experimental measurement and numerical model fitting; predicting the heat conductance of each heat exchanger in real time through an artificial neural network model; the minimum total energy consumption of the system is used as an optimization target, the overall heat transfer constraint and the flow constraint of the thermal management system are combined, a Lagrange optimization equation is established, the optimal frequency of each power component is solved, the frequency is input into a PLC in real time, the ventilation condition of each mine field is changed, and the temperature of each mine field is controlled. The mine field temperature control system integrates data reading, characteristic parameter identification, neural network model prediction, optimization solution and real-time feedback regulation, can effectively control the mine field temperature under the condition of low power consumption, and has the characteristics of high adaptability, high response speed, good control effect, high system robustness and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control in ore stope, and in particular to an intelligent temperature control strategy for deep metal ore stope. Background Art

[0002] Nowadays, the world economic life is inseparable from metal mineral resources. However, there are many problems in the process of mining, and the treatment of heat damage is one of the main difficulties restricting the mining depth of metal minerals. The problem of high-temperature heat damage will affect the normal work of workers and equipment, and reduce the service life of mechanical equipment. The stope ventilation system is the most commonly used method to control the stope temperature and solve the problem of high-temperature heat damage. However, its power consumption is generally large, accounting for 25% of the total power consumption of the whole mine. Traditional PID control can meet the ventilation requirements of the stope, but it is difficult to control the stope temperature under the condition of optimal power consumption.

[0003] Adaptive control is one of the mainstream trends in the development research of modern control systems. Compared with PID control, it has the advantages of fast response speed, high stability, and easy addition of constraints. However, the control accuracy is closely related to the accuracy of the model. As a typical complex thermal system, the ventilation and cooling system of deep metal ore stope has many heat transfer and flow constraint conditions, and a method that can effectively control the stope temperature needs to be studied. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent temperature control strategy for deep metal ore stope, which can effectively control the ore stope temperature under low power consumption, and has good temperature control effect and environmental adaptability.

[0005] To achieve the above object, the present invention provides an intelligent temperature control strategy for deep metal ore stope, including the following steps:

[0006] S1. Obtain system temperature, flow rate and pressure difference data;

[0007] S2. Judge whether the system needs to be optimized. If it needs to be optimized, execute S3; if not, execute S11;

[0008] S3. Initialize the heat conductance kA of the heat exchanger i and the parameters of the refrigerator, and give the system boundary conditions;

[0009] S4. Solve the Lagrangian equation to obtain the working frequency and the outlet temperatures T chwr and T cw,i ;

[0010] S5. Based on the flow constraints of the return air cooling circuit RACC, the intermediate cooling circuit ICC, the stope cooling circuit SCL, and the air pipeline CAL, calculate the flow rates of the cooling air and water in each circuit;

[0011] S6. Input these flow rates into the heat exchanger neural network model to predict the thermal conductivity kA of each heat exchanger i , and calculate the error between the thermal conductivity predicted by the neural network model and the initial value;

[0012] S7, determine whether the error is less than 0.01, if the error is less than 0.01, execute S8, if the error is greater than 0.01, return to S4;

[0013] S8, outputting the frequency of each power component after optimization;

[0014] S9, judging whether the system is to be regulated, if so, executing S10, if not, ending;

[0015] S10, writing the optimized frequency into the PLC to control the operation of each power component;

[0016] S11. The system records data.

[0017] Preferably, in S1, the specific steps are as follows:

[0018] S1.1. Start the system, set the initial frequency and run it normally for a period of time until the system is stable, and record the temperature, flow, pressure and frequency information of each part of the system within a period of time after stabilization;

[0019] S1.2. Record the information after median filtering into a file. During operation, the system records the file at regular intervals. During initial operation, this file is used as the data source for model establishment.

[0020] Preferably, in S3, the specific steps are as follows:

[0021] S3.1. Initialize heat exchanger thermal conductivity kA i and refrigerator parameters, if the refrigerator power P re and the outlet temperature T of the water from the evaporator and condenser chws and T cw,o ;

[0022] S3.2, given the system boundary conditions, if the outlet temperature of the stope simulation box is T out,i , return air cooler temperature T ct,i , air cooler inlet temperature T aaj,i , given the heat transfer capacity Q of each heat exchanger i .

[0023] Preferably, in S4, the specific steps are as follows:

[0024] S4.1. The model between the flow rate, frequency and pressure head of each variable frequency pump and variable frequency fan is expressed as:

[0025] H = a 0 ω 2 + a 1 ωm + a 2 m 2 ;

[0026] Wherein, H is the indenter m, ω is the frequency Hz, m is the flow rate kg / s, a 0 , a 1 , a 2 are characteristic parameters;

[0027] S4.2. For a loop with a fixed pipe network structure, without considering the static head, the relationship between the head and the flow rate is expressed as:

[0028] H = d 0 m 2 ;

[0029] Wherein, d 0 is the dynamic head coefficient of the pipe network;

[0030] S4.3. Ignoring motor losses and mechanical losses, the power consumed by the power components is calculated using their flow rate and head:

[0031] P aj = m aaj gH aj , j = 1, 2,..., 6;

[0032] P ch1 = m 01 gH ch1 ;

[0033] P ch3 = m 04 gH ch3 ;

[0034] P ct = m ct gH ct ;

[0035] P m = m m gH ct ;

[0036] Wherein, P aj (j = 1, 2,..., 10) is the power of the jth variable-frequency fan, H aj is the head of the jth fan; P ch1 is the power of the No. 1 chilled water loop pump, H ch1 is the head of the No. 1 chilled water loop pump; P ch3 is the power of the No. 3 chilled water loop pump, H ch3 is the head of the No. 3 chilled water loop pump; P ctis the power of the return air cooling circuit pump, H ct For the return air cooling circuit water pump; P m is the power of the intermediate cooling circuit pump, H m is the pressure head of the intermediate cooling circuit water pump; g is the acceleration due to gravity;

[0037] S4.4. Taking the minimum total energy consumption of the system as the optimization goal, combined with the overall heat transfer constraints and flow constraints of the thermal management system, the following Lagrangian function is constructed:

[0038]

[0039] Among them, lamda k (k=1, 2, ..., 10) is the Lagrange multiplier;

[0040] S4.5. Let the equation be about ω aaj ,ω m ,ω ct 、m awj and lamda k The partial derivative of is zero, and the following optimization equation is obtained:

[0041]

[0042] Solve the equations to obtain the operating frequency of each power component with the lowest total power consumption and the outlet temperature T of the evaporator and condenser chwr and T cw,i .

[0043] Preferably, S4.1 and S4.2 both adopt fitting regression methods, in which S4.1 fits the characteristic parameters of the cooling tower loop circulation pump, the intermediate loop circulation pump, the mine cooling loop circulation pump and the air pipeline fan according to the experimental measurement data; in S4.2, the characteristic parameters of the return air cooling loop network, the intermediate cooling loop network, the chilled water loop network and the mine cooling air loop network are fitted according to the experimental measurement data; in S4.4, the overall heat transfer constraint is based on the heat flow method.

[0044] Preferably, in S5, the specific steps are as follows:

[0045] S5.1. When the system is running stably, the driving force provided by the power components should be consistent with the flow resistance in the pipe network. The power component constraint equation and the pipe network resistance constraint equation are combined to establish the overall flow constraint model of the system:

[0046] a 0,aj ω aaj 2 +a 1,aj ω aaj m aaj +a 2,aj m aaj2 = d aj m aaj 2 , j = 1, 2, …, 6;

[0047] a 0,ch1 ω ch1 2 + a 1,ch1 ω ch1 m 01 + a 2,ch1 m 01 2 = d 0 m 0 2 + d 01 m 01 2 + d 1 m aw1 2 + d 03 m 03 2 ;

[0048] a 0,ch3 ω ch3 2 + a 1,ch3 ω ch3 m 04 + a 2,ch3 m 04 2 = d 0 m 0 2 + d 04 m 04 2 + d 2 m aw2 2 + d 06 m 06 2 ;

[0049] a 0,ch3 ω ch3 2 + a 1,ch3 ω ch3 m 04 + a 2,ch3 m 04 2 = d 0 m 0 2 + d 04 m 04 2 + d 2 m aw2 2 + d 06 m06 2 ;

[0050] a 0,ct ω ct 2 +a 1,ct ω ct m ct +a 2,ct m ct 2 =d ct m ct 2 ;

[0051] a 0,m ω m 2 +a 1,m ω m m m +a 2,m m m 2 =d m m m 2 ;

[0052] wherein, a 0,aj , a 1,aj and a 2,aj are characteristic parameters of the j-th variable-frequency fan, ω aaj is the operating frequency of the j-th fan, d aj is the characteristic parameter of the flow resistance of the j-th air cooling pipeline; a 0,ch1 , a 1,ch1 and a 2,ch1 are characteristic parameters of the No. 1 chilled water circuit pump, ω ch1 is the frequency of the No. 1 chilled water circuit pump; a 0,ch3 , a 1,ch3 and a 2,ch3 are characteristic parameters of the No. 3 chilled water circuit pump, ω ch3 is the frequency of the No. 3 chilled water circuit pump; a 0,ct , a 1,ct and a 2,ct are characteristic parameters of the return air cooling circuit pump, ω ct is the frequency of the return air cooling circuit pump, d ct is the characteristic parameter of the flow resistance of the return air cooling circuit; a 0,m , a 1,m and a 2,m are characteristic parameters of the intermediate cooling circuit pump, ω m is the frequency of the intermediate cooling circuit pump, d m is the characteristic parameter of the intermediate cooling circuit pump;

[0053] S5.2. Calculate the flow rates of the cooling air and water in each loop using the flow constraint model.

[0054] Preferably, in S6, the specific steps are as follows:

[0055] S6.1. Input the flow rates in S5 into the heat exchanger neural network model to predict the thermal conductance kA of each heat exchanger i ;

[0056] S6.2. Input the mass flow rates m ch and m m of the water in the evaporator and condenser, and the inlet temperatures T chwr and T cw,i into the refrigerator neural network model to predict the new power P re of the refrigerator and the outlet temperatures T chws and T cw,o of the water in the evaporator and condenser;

[0057] S6.3. Calculate the error between the thermal conductance predicted by the neural network and the initial value.

[0058] Preferably, in S6.1 and S6.2, the layers of the established neural network model are connected by activation functions. Using the mean square error as the criterion, the data within the set time recorded by the system are used for repeated iterative training, and the parameters in the neural network are updated by the gradient method to finally obtain a multi-layer neural network model with the characteristics of the training data.

[0059] Preferably, in S8, display the frequencies of each power component, the flow rates of each pipeline, the thermal conductance of each heat exchanger, and the heat transfer amount after the optimized calculation; in S10, if intelligent regulation is selected, write the optimized frequencies into the PLC using the OPC protocol to control the operation of each power component.

[0060] Preferably, in S11, use the pandas library to write the recorded system operation parameters into an excel table for storage when the system runs to the set time, which is convenient for subsequent analysis and calculation.

[0061] Therefore, by adopting the above intelligent deep metal mine stope temperature control strategy, the present invention can effectively solve the heat damage problem in the deep mine stope and effectively reduce energy consumption, and has the characteristics of strong adaptability, fast response speed, good control effect, and high system robustness.

[0062] The technical solution of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0063] Figure 1 is the intelligent temperature control flow chart of the stope in the embodiment of the intelligent deep metal mine stope temperature control strategy of the present invention;

[0064] Figure 2 is the schematic diagram of the actual six-stope system structure and the optimized control interface of the embodiment of the present invention;

[0065] Figure 3 is the schematic diagram of the equivalent thermal resistance network of the actual six-stope system of the embodiment of the present invention;

[0066] Figure 4 is the schematic diagram of the flow resistance network of the actual six-stope system of the embodiment of the present invention. Specific Embodiments

[0067] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0068] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. The terms such as "comprising" or "including" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0069] Embodiment 1

[0070] As Figure 1 shown, the present invention provides an intelligent temperature control strategy for deep metal mine stopes, including the following steps:

[0071] S1. The system adopted in this embodiment is as Figure 2 shown, and the system temperature, flow rate and pressure difference data are obtained. The specific steps are as follows:

[0072] S1.1. Turn on the system, give an initial frequency and run it normally for a period of time until the system is stable, and record the temperature, flow rate, pressure and frequency information at various parts of the system for a period of time after stability.

[0073] S1.2. Record the information after median filtering into a file, and the system records the file every once in a while during operation. This file is used as the data source for model establishment during the initial operation.

[0074] S2. Determine whether the system needs to be optimized. If it needs to be optimized, execute S3; if not, execute S11.

[0075] S3. Initialize the heat transfer coefficient kA of the heat exchanger i and the parameters of the refrigerator, and specify the system boundary conditions. The specific steps are as follows:

[0076] S3.1. Initialize the heat transfer coefficient kA of the heat exchanger i and the parameters of the refrigerator. If the power P of the refrigerator re and the outlet temperatures T of the water in the evaporator and condenser chws and T cw,o .

[0077] S3.2. Specify the system boundary conditions. If the outlet temperature T of the stope simulation box out,i , the temperature T of the return air air cooler ct,i , the inlet temperature T of the air cooler aaj,i , specify the heat transfer quantity Q of each heat exchanger i .

[0078] S4. Solve the Lagrangian equation to obtain the operating frequency and the outlet temperatures T of the evaporator and condenser chwr and T cw,i , and the specific steps are as follows:

[0079] S4.1. The model representing the relationship between the flow rate, frequency, and head of each variable-frequency pump and variable-frequency fan is known as:

[0080] H = a 0 ω 2 + a 1 ωm + a 2 m 2

[0081] where H is the head in m, ω is the frequency in Hz, m is the flow rate in kg / s, and a 0 , a 1 , a 2 are characteristic parameters.

[0082] S4.2. For a loop with a fixed pipe network structure, without considering the static head, the relationship between the head and the flow rate is expressed as:

[0083] H = d 0 m 2

[0084] where d 0 is the dynamic head coefficient of the pipe network.

[0085] S4.3. Ignoring the motor loss and mechanical loss, the power consumed by the power component is calculated using its flow rate and head:

[0086] Paj = m aaj gH aj , j = 1, 2, …, 6

[0087] P ch1 = m 01 gH ch1

[0088] P ch3 = m 04 gH ch3

[0089] P ct = m ct gH ct

[0090] P m = m m gH ct

[0091] Among them, P aj (j = 1, 2, …, 10) is the power of the jth variable-frequency fan, and H aj is the head of the jth fan; P ch1 is the power of the No. 1 chilled water circuit pump, and H ch1 is the head of the No. 1 chilled water circuit pump; P ch3 is the power of the No. 3 chilled water circuit pump, and H ch3 is the head of the No. 3 chilled water circuit pump; P ct is the power of the return air cooling circuit pump, and H ct is the return air cooling circuit water pump; P m is the power of the intermediate cooling circuit pump, and H m is the head of the intermediate cooling circuit water pump; g is the acceleration due to gravity.

[0092] S4.4. With the lowest total energy consumption of the system as the optimization goal, combined with the overall heat transfer constraints and flow constraints of the thermal management system, construct the following Lagrangian function:

[0093]

[0094] Among them, lamda k (k = 1, 2, …, 10) is the Lagrange multiplier.

[0095] S4.5. Let the partial derivatives of the equation with respect to ω aaj , ω m , ω ct , m awj and lamda k be zero to obtain the following optimization equations:

[0096]

[0097] Solve the system of equations to obtain the operating frequencies of each power component with the lowest total system power consumption and the outlet temperatures T of the evaporator and condenser chwr and T cw,i .

[0098] Among them, both S4.1 and S4.2 adopt the fitting regression method. In S4.1, the characteristic parameters of the cooling tower loop (RACC) circulation pump, intermediate loop (ICC) circulation pump, stope cooling loop (SCL) circulation pump, and air pipeline (CAL) fan are fitted according to the experimental measurement data. In S4.2, the characteristic parameters of the return air cooling loop network, intermediate cooling loop network, chilled water loop network, and stope cooling air loop network are fitted according to the experimental measurement data. In S4.4, the overall heat transfer constraint is based on the heat flow method, as Figure 3 shown.

[0099] S5. Based on the flow constraints of the return air cooling loop RACC, intermediate cooling loop ICC, stope cooling loop SCL, and air pipeline CAL, calculate the flow rates of the cooling air and water in each loop. The specific steps are as follows:

[0100] S5.1. When the system is operating stably, the driving force provided by the power components should be consistent with the flow resistance in the pipe network. Combine the power component constraint equation and the pipe network resistance constraint equation to establish an overall system flow constraint model:

[0101] a 0,aj ω aaj 2 +a 1,aj ω aaj m aaj +a 2,aj m aaj 2 =d aj m aaj 2 , j = 1, 2, …, 6

[0102] a 0,ch1 ω ch1 2 +a 1,ch1 ω ch1 m 01 +a 2,ch1 m 01 2 =d 0 m 0 2 +d 01 m 01 2 +d 1 m aw1 2 +d 03 m03 2

[0103] a 0,ch3 ω ch3 2 +a 1,ch3 ω ch3 m 04 +a 2,ch3 m 04 2 =d 0 m 0 2 +d 04 m 04 2 +d 2 m aw2 2 +d 06 m 06 2

[0104] a 0,ch3 ω ch3 2 +a 1,ch3 ω ch3 m 04 +a 2,ch3 m 04 2 =d 0 m 0 2 +d 04 m 04 2 +d 2 m aw2 2 +d 06 m 06 2

[0105] a 0,ct ω ct 2 +a 1,ct ω ct m ct +a 2,ct m ct 2 =d ct m ct 2

[0106] a 0,m ω m 2 +a 1,m ω m m m +a 2,m m m2 = d m m m 2

[0107] wherein, a 0,aj 、a 1,aj and a 2,aj are characteristic parameters of the j-th variable-frequency fan, ω aaj is the operating frequency of the j-th fan, d aj is the characteristic parameter of the flow resistance of the j-th air cooling pipeline; a 0,ch1 、a 1,ch1 and a 2,ch1 are characteristic parameters of the No. 1 chilled water loop pump, ω ch1 is the frequency of the No. 1 chilled water loop pump; a 0,ch3 、a 1,ch3 and a 2,ch3 are characteristic parameters of the No. 3 chilled water loop pump, ω ch3 is the frequency of the No. 3 chilled water loop pump; a 0,ct 、a 1,ct and a 2,ct are characteristic parameters of the return air cooling loop pump, ω ct is the frequency of the return air cooling loop pump, d ct is the characteristic parameter of the flow resistance of the return air cooling loop; a 0,m 、a 1,m and a 2,m are characteristic parameters of the intermediate cooling loop pump, ω m is the frequency of the intermediate cooling loop pump, d m is the characteristic parameter of the intermediate cooling loop pump.

[0108] S5.2. Calculate the flow rates of the cooling air and water in each loop using the flow constraint model.

[0109] S6. Input these flow rates into the heat exchanger neural network model to predict the thermal conductance kA i of each heat exchanger, and calculate the error between the thermal conductance predicted by the neural network model and the initial value. The specific steps are as follows:

[0110] S6.1. Input the flow rates in S5 into the heat exchanger neural network model to predict the thermal conductance kA i of each heat exchanger.

[0111] S6.2. Input the mass flow rates m ch and m m of the water in the evaporator and condenser, and the inlet temperatures T chwr and T cw,i of the evaporator and condenser into the refrigerator neural network model to predict the new power P re of the refrigerator and the outlet temperatures T of the water in the evaporator and condenserchws and T cw,o 。

[0112] S6.3. Calculate the error between the thermal conductivity predicted by the neural network and the initial value.

[0113] Among them, the layers of the neural network model established in S6.1 and S6.2 are connected by activation functions. Using the Mean-Square Error (MSE) as the criterion, the data recorded by the system for a period of time is used for repeated iterative training. The parameters in the neural network are updated by the gradient method, and finally a multi-layer neural network model with the characteristics of the training data is obtained.

[0114] S7. Determine whether the error is less than 0.01. If the error is less than 0.01, execute S8; if the error is greater than 0.01, return to S4.

[0115] S8. Output the optimized frequencies of each power component, and display the frequencies of each power component, the flow rates of each pipeline, the thermal conductivities of each heat exchanger, and the heat transfer amounts after the optimized calculation.

[0116] S9. Determine whether the system is regulated. If it is regulated, execute S10; if it is not regulated, end.

[0117] S10. Write the optimized frequencies into the PLC to control the operation of each power component. In this implementation, the OPC protocol is used to write the optimized frequencies into the PLC to control the operation of each power component.

[0118] S11. The system records data. Using the pandas library, when the system runs to the set time, the recorded system operation parameters are written into an excel table for storage, which is convenient for subsequent analysis and calculation.

[0119] Therefore, the present invention adopts the above-mentioned intelligent deep metal mine stope temperature control strategy, and adaptively controls the system based on the neural network model, ensuring the effectiveness of the control, being able to effectively control the mine temperature under the condition of low power consumption, and having good temperature control effect and environmental adaptability.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent deep metal mine temperature control strategy, characterized in that: The following steps are involved: S1. Obtain system temperature, flow and pressure difference data; S2, determine whether the system needs to be optimized, if it is optimized, execute S3, if not, execute S11; S3. Initialize the heat exchanger thermal conductivity kA i and refrigerator parameters, and given system boundary conditions; S4. Solve the Lagrange equation to obtain the operating frequency and the outlet temperature T of the evaporator and condenser chwr and T cw,i ; S5. Based on the flow constraints of the return air cooling circuit RACC, the intermediate cooling circuit ICC, the stope cooling circuit SCL, and the air pipeline CAL, calculate the flow rates of cooling air and water in each circuit; S6. Input these flow rates into the heat exchanger neural network model to predict the thermal conductivity kA of each heat exchanger i , and calculate the error between the thermal conductivity predicted by the neural network model and the initial value; S7, determine whether the error is less than 0.01, if the error is less than 0.01, execute S8, if the error is greater than 0.01, return to S4; S8, outputting the frequency of each power component after optimization; S9, judging whether the system is to be regulated, if so, executing S10, if not, ending; S10, writing the optimized frequency into the PLC to control the operation of each power component; S11. The system records data.

2. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized by: In S1, the specific steps are as follows: S1.

1. Start the system, set the initial frequency and run it normally for a period of time until the system is stable, and record the temperature, flow, pressure and frequency information of each part of the system within a period of time after stabilization; S1.

2. Record the information after median filtering into a file. During operation, the system records the file at regular intervals. During initial operation, this file is used as the data source for model establishment.

3. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized by: In S3, the specific steps are as follows: S3.

1. Initialize heat exchanger thermal conductivity kA i and refrigerator parameters, if the refrigerator power P re and the outlet temperature T of the water from the evaporator and condenser chws and T cw,o ; S3.2, given the system boundary conditions, if the outlet temperature of the stope simulation box is T out,i , return air cooler temperature T ct,i , air cooler inlet temperature T aaj,i , given the heat transfer capacity Q of each heat exchanger i .

4. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized in that: In S4, the specific steps are as follows: S4.

1. The model between the flow rate, frequency and pressure head of each variable frequency pump and variable frequency fan is expressed as: H=a0ω 2 +a1ωm+a2m 2 ; Among them, H is the pressure head m, ω is the frequency Hz, m is the flow rate kg / s, a0, a1, a2 are characteristic parameters; S4.

2. For a circuit with a fixed pipe network structure, ignoring the static pressure head, the relationship between pressure head and flow is expressed as: H=d0m 2 ; Among them, d0 is the dynamic pressure head coefficient of the pipe network; S4.

3. Ignoring motor losses and mechanical losses, the power consumed by the power components is calculated using their flow and pressure head: P aj =m aaj gH aj ,j=1,2,…,6; P ch1 =m 01 gH ch1 ; P ch3 =m 04 gH ch3 ; P ct =m ct gH ct ; P m =m m gH ct ; Among them, P aj (j=1, 2, ..., 10) is the power of the jth variable frequency fan, H aj is the pressure head of the jth fan; P ch1 is the power of No. 1 chilled water circuit pump, H ch1 is the pressure head of No. 1 chilled water circuit pump; P ch3 is the power of No. 3 chilled water circuit pump, H ch3 is the pressure head of No. 3 chilled water circuit pump; P ct is the power of the return air cooling circuit pump, H ct For the return air cooling circuit water pump; P m is the power of the intermediate cooling circuit pump, H m is the pressure head of the intermediate cooling circuit water pump; g is the acceleration due to gravity; S4.

4. Taking the minimum total energy consumption of the system as the optimization goal, combined with the overall heat transfer constraints and flow constraints of the thermal management system, the following Lagrangian function is constructed: Among them, lamda k (k=1, 2, ..., 10) is the Lagrange multiplier; S4.

5. Let the equation be about ω aaj ,ω m ,ω ct 、m awj and lamda k The partial derivative of is zero, and the following optimization equation is obtained: Solve the equations to obtain the operating frequency of each power component with the lowest total power consumption and the outlet temperature T of the evaporator and condenser chwr and T cw,i .

5. The intelligent deep metal mine temperature control strategy according to claim 4 is characterized by: Both S4.1 and S4.2 use fitting regression methods. In S4.1, the characteristic parameters of the cooling tower loop circulation pump, the intermediate loop circulation pump, the mine cooling loop circulation pump and the air pipeline fan are fitted according to the experimental measurement data; in S4.2, the characteristic parameters of the return air cooling loop network, the intermediate cooling loop network, the chilled water loop network and the mine cooling air loop network are fitted according to the experimental measurement data; in S4.4, the overall heat transfer constraint is based on the heat flow method.

6. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized by: In S5, the specific steps are as follows: S5.

1. When the system is running stably, the driving force provided by the power components should be consistent with the flow resistance in the pipe network. The power component constraint equation and the pipe network resistance constraint equation are combined to establish the overall flow constraint model of the system: a 0,aj ω aaj 2 +a 1,aj ω aaj m aaj +a 2,aj m aaj 2 =d aj m aaj 2 ,j=1,2,…,6; a 0,ch1 ω ch1 2 +a 1,ch1 ω ch1 m 01 +a 2,ch1 m 01 2 =d0m0 2 +d 01 m 01 2 +d1m aw1 2 +d 03 m 03 2 ; a 0,ch3 ω ch3 2 +a 1,ch3 ω ch3 m 04 +a 2,ch3 m 04 2 =d0m0 2 +d 04 m 04 2 +d2m aw2 2 +d 06 m 06 2 ; a 0,ch3 ω ch3 2 +a 1,ch3 ω ch3 m 04 +a 2,ch3 m 04 2 =d0m0 2 +d 04 m 04 2 +d2m aw2 2 +d 06 m 06 2 ; a 0,ct ω ct 2 +a 1,ct ω ct m ct +a 2,ct m ct 2 =d ct m ct 2 ; a 0,m ω m 2 +a 1,m ω m m m +a 2,m m m 2 =d m m m 2 ; Among them, a 0,aj 、a 1,aj and a 2,aj is the characteristic parameter of the jth variable frequency fan, ω aaj is the operating frequency of the jth fan, d aj is the characteristic parameter of the flow resistance of the jth air cooling pipeline; a 0,ch1 、a 1,ch1 and a 2,ch1 is the characteristic parameter of the No. 1 chilled water circuit pump, ω ch1 is the frequency of the No. 1 chilled water circuit pump; a 0,ch3 、a 1,ch3 and a 2,ch3 is the characteristic parameter of No. 3 chilled water circuit pump, ω ch3 is the frequency of the No. 3 chilled water circuit pump; a 0,ct 、a 1,ct and a 2,ct is the characteristic parameter of the return air cooling circuit pump, ω ct is the frequency of the return air cooling circuit pump, d ct is the characteristic parameter of the return air cooling circuit flow resistance; a 0,m 、a 1,m and a 2,m is the characteristic parameter of the intermediate cooling circuit pump, ω m is the frequency of the intercooling circuit pump, d m is the characteristic parameter of the intermediate cooling circuit pump; S5.

2. Use the flow constraint model to calculate the flow rates of cooling air and water in each circuit.

7. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized by: In S6, the specific steps are as follows: S6.

1. Input the flow rate in S5 into the heat exchanger neural network model to predict the thermal conductivity kA of each heat exchanger i ; S6.

2. The mass flow rate of water in the evaporator and condenser m ch and m m , the inlet temperature T of the evaporator and condenser chwr and T cw,i Input into the refrigerator neural network model to predict the new power P of the refrigerator re and the outlet temperature T of the water from the evaporator and condenser chws and T cw,o ; S6.

3. Calculate the error between the thermal conductivity predicted by the neural network and the initial value.

8. The intelligent deep metal mine temperature control strategy according to claim 7 is characterized by: Each layer of the neural network model established in S6.1 and S6.2 is connected by an activation function. The mean square error is used as a standard, and the data recorded by the system within a set time is used for repeated iterative training. The parameters in the neural network are updated by the gradient method, and finally a multi-layer neural network model with the characteristics of the training data is obtained.

9. The intelligent deep metal mine temperature control strategy according to claim 1 is characterized by: In S8, the frequency of each power component, the flow rate of each pipeline, the thermal conductivity of each heat exchanger and the heat transfer capacity after optimization calculation are displayed; in S10, if intelligent control is selected, the OPC protocol is used to write the optimized frequency into the PLC to control the operation of each power component.

10. The intelligent deep metal mine temperature control strategy according to claim 1, characterized in that: In S11, the pandas library is used to write the recorded system operation parameters into an Excel table for storage when the system runs to the set time, which is convenient for subsequent analysis and calculation.

Images