Production process intelligent monitoring method and system based on intelligent mine

Abstract

The invention provides a production process intelligent monitoring method and system based on an intelligent mine, and the method comprises the steps: collecting a target monitoring data set in the real-time production process of mine equipment, covering equipment vibration time sequence signals, environment temperature and humidity distribution data and an equipment energy consumption fluctuation curve, and carrying out the dynamic standardization processing, the method comprises the following steps: obtaining a standardized monitoring data set matched with an equipment type and a production process stage, calling a pre-trained multi-dimensional feature extraction model to carry out joint feature mapping, generating an equipment operation state, environment association and data abnormal fluctuation features, and carrying out dynamic fusion analysis on the features based on a preset abnormal mode identification model, and finally, according to the risk level, an alarm instruction is triggered, an optimization strategy is fed back to a production control terminal to adjust equipment operation parameters, and intelligent monitoring and optimization of the intelligent mine production process are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent mines, and more specifically, to an intelligent monitoring method and system for the production process based on intelligent mines. Background Art

[0002] In the field of mine production, with the continuous expansion of production scale and the increasing complexity of production processes, traditional production process monitoring methods are difficult to meet the requirements of efficient and safe production in modern mines. There are many limitations in existing mine production monitoring technologies, which seriously restrict the intelligent development of mine production.

[0003] Existing methods usually adopt fixed data processing methods and cannot be dynamically adjusted according to different equipment types and production process stages. As a result, the processed data is difficult to accurately reflect the characteristics of different equipment and process stages, reducing the availability and analysis value of the data and unable to provide strong support for subsequent risk assessment and optimization strategy formulation. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide an intelligent monitoring method for the production process based on intelligent mines, and the method includes:

[0005] Collect a set of target monitoring data of mine equipment in the real-time production process, and the set of target monitoring data includes equipment vibration time series signals, environmental temperature and humidity distribution data, and equipment energy consumption fluctuation curves;

[0006] Perform dynamic standardization processing on the set of target monitoring data to generate a set of standardized monitoring data that matches the equipment type and production process stage;

[0007] Call a pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the set of standardized monitoring data to generate equipment operation state features, environment-related features, and data abnormal fluctuation features;

[0008] Based on a preset abnormal pattern recognition model, perform dynamic fusion analysis on the equipment operation state features, environment-related features, and data abnormal fluctuation features to generate an equipment operation risk level and a set of production process optimization strategies;

[0009] Trigger corresponding alarm instructions according to the equipment operation risk level, and feedback the set of production process optimization strategies to the production control terminal to adjust the equipment operation parameters.

[0010] On the other hand, an embodiment of the present invention also provides an intelligent monitoring system for production processes based on a smart mine, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present invention collects the target monitoring data set of mining equipment in the real-time production process, comprehensively obtains key information such as equipment vibration, environmental temperature and humidity, and equipment energy consumption, performs dynamic standardization processing on the target monitoring data set, generates a standardized monitoring data set that matches the equipment type and production process stage, effectively eliminates the differences between data of different equipment and process stages, calls the pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the standardized monitoring data set, extracts key features such as equipment operation status, environmental association, and abnormal data fluctuations, and performs dynamic fusion analysis on the extracted features based on the preset abnormal pattern recognition model to generate equipment operation risk level and production process optimization strategy set, which can provide real-time insight into the equipment operation status, discover potential risks in advance, and formulate optimization strategies in a targeted manner. According to the equipment operation risk level, the corresponding alarm instruction is triggered, and the production process optimization strategy set is fed back to the production control terminal to adjust the equipment operation parameters, so as to realize real-time and precise control of the mine production process, effectively solve the problems of insufficient data utilization, untimely risk warning, and inaccurate optimization strategy in traditional mine production monitoring, improve the automation and intelligence of mine production, and reduce the incidence of production accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the execution flow of the intelligent monitoring method for production processes based on smart mines provided in an embodiment of the present invention.

[0013] Figure 2 It is a schematic diagram of exemplary hardware and software components of a production process intelligent monitoring system based on a smart mine provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 It is a flow chart of an intelligent monitoring method for a production process based on a smart mine provided by an embodiment of the present invention. The intelligent monitoring method for a production process based on a smart mine is introduced in detail below.

[0015] Step S110, collecting a target monitoring data set of mining equipment in a real-time production process, wherein the target monitoring data set includes equipment vibration time series signals, environmental temperature and humidity distribution data, and equipment energy consumption fluctuation curves.

[0016] In this embodiment, during the real-time production process of the mine, comprehensive monitoring and optimization work can be carried out for various types of mining equipment. Specifically, the mine has various types of equipment, covering multiple production process links such as mining, transportation, and crushing. For example, in the mining process, for a large mining machine with the model ABC-100, high-precision vibration sensors are installed at its key components such as the motor, transmission gears, and digging arm to collect the vibration time-series signal of the equipment in real time. For example, the above sensors can collect 100 data points per second, continuously record the vibration situation of the equipment during mining, and form a continuous vibration time-series signal data stream.

[0017] At the same time, multiple temperature and humidity sensors are arranged in the operation area of the mining machine to obtain the environmental temperature and humidity distribution data. For example, the above temperature and humidity sensors can be evenly distributed within a range with a radius of 10 meters, and record the temperature and humidity data every 5 minutes, so as to depict the environmental temperature and humidity distribution of this area. For example, at a certain moment, the sensor records that the temperature is 28 degrees Celsius and the humidity is 60%.

[0018] In addition, through the energy consumption monitoring system connected to the mining machine, the energy consumption fluctuation curve of the equipment can be monitored in real time, which can accurately record the electric energy consumption of the mining machine in different operating states, generate energy consumption data in minutes, and form the energy consumption fluctuation curve of the equipment. During a certain period of mining operation time, it is recorded that the energy consumption gradually rises from the initial 50 kilowatts to 80 kilowatts, and then stabilizes at about 70 kilowatts.

[0019] In the transportation process, for a transport truck with the model DEF-20, similar data collection is also carried out. For example, vibration sensors are installed on the engine, tires, and suspension system of the truck to collect the vibration time-series signal of the equipment. At the same time, temperature and humidity sensors are installed inside and outside the carriage to obtain the environmental temperature and humidity distribution data during transportation. And through the on-vehicle energy consumption monitoring device, the energy consumption fluctuation curve of the equipment during transportation is recorded. For example, during a transportation process, the energy consumption of the truck is 10 kilowatts when starting, the average energy consumption during driving is maintained at 30 kilowatts, and the energy consumption can reach 50 kilowatts when accelerating and climbing slopes.

[0020] In the crushing process, for a GHI-300 type crusher, vibration sensors are installed at its motor, crushing chamber, and transmission belt to collect the vibration time-series signal. For example, temperature and humidity sensors are arranged around the crusher to obtain the environmental temperature and humidity distribution data. Through the energy consumption monitoring device connected to the crusher, the energy consumption fluctuation curve of the equipment is recorded. For example, when the crusher starts, the energy consumption rapidly rises to 100 kilowatts, and after stable operation, the energy consumption remains at about 80 kilowatts.

[0021] Thus, the above-mentioned equipment vibration time series signals, environmental temperature and humidity distribution data, and equipment energy consumption fluctuation curves from different processes and different equipment can be integrated together to form a target monitoring data set.

[0022] Step S120, perform dynamic standardization processing on the target monitoring data set to generate a standardized monitoring data set that matches the equipment type and production process stage.

[0023] For example, for the ABC-100 type mining machine in the mining process, first obtain its historical operation parameter set, where the rated vibration threshold range of the equipment is 0 - 50 mm / s², the standard temperature and humidity adaptation interval is 20 - 30 °C for temperature and 40% - 70% for humidity, and the maximum energy consumption critical value is 100 kW.

[0024] According to the rated vibration threshold range of the equipment, perform amplitude normalization processing on the collected equipment vibration time series signal. For example, assume that the amplitude range of the collected vibration signal is 10 - 60 mm / s², and the following processing is carried out: subtract the minimum value 10 from each data point, and then divide by the difference between the maximum value and the minimum value 50 (60 - 10) to obtain a standardized vibration signal sequence. For example, the amplitude of the vibration signal collected at a certain moment is 30 mm / s², and after processing, it is (30 - 10) ÷ 50 = 0.4.

[0025] Then, determine the current environmental temperature and humidity reference values according to the current mining process stage. Assume that the current reference temperature is 25 °C and the reference humidity is 50%. Perform deviation correction processing on the environmental temperature and humidity distribution data based on the standard temperature and humidity adaptation interval. For the collected temperature data, if it is 28 °C, the deviation is 28 - 25 = 3 °C, and it is corrected within the standard temperature and humidity adaptation interval. The corrected temperature is (28 - 20) ÷ (30 - 20) = 0.8. For the humidity data, if the collected humidity is 60%, the deviation is 60 - 50 = 10%, and the corrected humidity is (60 - 40) ÷ (70 - 40) = 0.67, generating corrected temperature and humidity distribution data.

[0026] Next, perform dynamic scaling processing on the equipment energy consumption fluctuation curve according to the maximum energy consumption critical value. Assume that in the collected energy consumption fluctuation curve, the energy consumption value is between 50 - 80 kW, and the following scaling processing is carried out: subtract the minimum value 50 from each energy consumption data point, and then divide by the difference between the maximum value and the minimum value 30 (80 - 50) to generate a standardized energy consumption curve. For example, the energy consumption value at a certain moment is 70 kW, and after processing, it is (70 - 50) ÷ 30 = 0.67.

[0027] Thus, the standardized vibration signal sequence, the corrected temperature and humidity distribution data, and the standardized energy consumption curve can be integrated into a standardized monitoring data set for the ABC-100 type mining machine in the mining process stage.

[0028] For the DEF-20 type transport truck in the transportation process, obtain its historical operation parameter set. The rated vibration threshold range of the equipment is 0 - 30 mm / s², the standard temperature and humidity adaptation range is temperature 15 - 25 °C and humidity 30% - 60%, and the maximum energy consumption critical value is 60 kW. Process the collected target monitoring data in the same way as above to generate a standardized monitoring data set for the DEF-20 type transport truck in the transportation process stage.

[0029] For the GHI-300 type crusher in the crushing process, obtain its historical operation parameter set. The rated vibration threshold range of the equipment is 0 - 80 mm / s², the standard temperature and humidity adaptation range is temperature 22 - 32 °C and humidity 45% - 75%, and the maximum energy consumption critical value is 120 kW. Process the collected data accordingly to generate a standardized monitoring data set for the GHI-300 type crusher in the crushing process stage.

[0030] Step S130, call the pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the standardized monitoring data set to generate equipment operation state features, environment-related features, and data abnormal fluctuation features.

[0031] For example, for the standardized vibration signal sequence of the ABC-100 type mining machine in the mining process, time-frequency conversion processing can be performed. For example, first perform a windowed Fourier transform on the standardized vibration signal sequence. Assume the window function length is 100 data points and the overlap rate is 50% to generate a vibration spectrogram. Then calculate the energy integral value of each frequency interval in the vibration spectrogram. For example, divide the frequency range into multiple intervals such as 0 - 100 Hz, 100 - 200 Hz, etc., calculate the energy integral value within each interval, and generate a frequency-domain energy distribution histogram. Detect the main resonance peak frequency in the vibration spectrogram. Assume the standard resonance frequency is 150 Hz and the detected main resonance peak frequency is 160 Hz, calculate the difference 160 - 150 = 10 Hz to generate the resonance peak offset. Perform a sliding window statistics on the frequency-domain energy distribution histogram with a window size of 10 intervals to extract the low-frequency energy ratio feature and the high-frequency mutation count feature. Assume the low-frequency energy ratio is 30% and the high-frequency mutation count is 5 times. After logarithmic transformation of these two features, they are combined into the vibration frequency-domain energy distribution feature. For example, after logarithmic transformation, the low-frequency energy ratio feature is ln(0.3), the high-frequency mutation count feature is ln(5), and after combination, the vibration frequency-domain energy distribution feature is formed.

[0032] Then, perform spatial interpolation on the corrected temperature and humidity distribution data. Assume there are 5 temperature and humidity sensors in a 10×10 square meter working area, and generate the temperature and humidity gradient change characteristics of the entire area through an interpolation algorithm. For example, calculate the temperature change rate and humidity change rate between adjacent sensors to generate the temperature and humidity gradient change characteristics. At the same time, mark the local abnormal temperature areas. For example, if the temperature in a certain area is significantly higher than the surrounding areas, mark it as a local abnormal temperature area and generate the local abnormal temperature area mark.

[0033] Conduct slope change analysis on the standardized energy consumption curve. Perform first-order difference calculation on the standardized energy consumption curve to generate the energy consumption change rate sequence. Assume that within a certain period of time, the energy consumption data points are 0.5, 0.6, 0.7, etc. After the first-order difference calculation, the energy consumption change rate sequence is 0.1, 0.1, etc. Detect the time intervals in the energy consumption change rate sequence that continuously exceed the rated slope threshold (assumed to be 0.15), and record their start timestamps and end timestamps. Calculate the average slope value within this time interval as the energy consumption increase rate characteristic. Assume that the slope values within the time interval are 0.2, 0.2, 0.18, and the average slope value is (0.2 + 0.2 + 0.18) ÷ 3 = 0.193, which is used as the energy consumption increase rate characteristic. Determine the peak duration characteristic based on the difference between the start timestamp and the end timestamp. Assume that the start timestamp is 10 minutes and the end timestamp is 15 minutes, and the peak duration characteristic is 5 minutes. Detect the extreme points of the energy consumption change rate sequence to generate the energy consumption mutation event mark and the corresponding event duration.

[0034] Input the vibration frequency domain energy distribution characteristics, resonance peak offset, temperature and humidity gradient change characteristics, local abnormal temperature area mark, energy consumption increase rate characteristic, and peak duration characteristic into the multi-dimensional feature extraction model. Through the cross-modal attention mechanism, perform correlation weighting on the vibration frequency domain energy distribution characteristics and the temperature and humidity gradient change characteristics. Map the vibration frequency domain energy distribution characteristics to a set of query vectors, and map the temperature and humidity gradient change characteristics to a set of key vectors. Then, calculate the cosine similarity between each query vector and the key vector to generate the initial attention weight matrix. Assume the query vector is [0.5, 0.3] and the key vector is [0.4, 0.6], and the cosine similarity calculation is (0.5 × 0.4 + 0.3 × 0.6) ÷ (sqrt(0.52 + 0.3 2 ) × sqrt(0.4 2+0.62))=0.78, generate the initial attention weight matrix. Perform temperature parameter adjustment (assuming the temperature parameter is 0.1) and Softmax normalization on the initial attention weight matrix to generate standardized attention weights. Perform dynamic weighted summation of the temperature and humidity gradient change features according to the standardized attention weights to generate environmentally sensitive vibration features. Multiply the environmentally sensitive vibration features and the temperature and humidity gradient change features element by element to generate environmentally associated features.

[0035] The resonance peak offset and energy consumption increase rate characteristics are jointly encoded through the time series convolutional network to generate the equipment operation status characteristics. The abnormal detection algorithm is used to perform pattern matching on the local abnormal temperature zone mark and peak duration characteristics to generate data abnormal fluctuation characteristics.

[0036] The equipment in the transportation process and the crushing process are processed in the same manner as above to generate the equipment operation status characteristics, environment correlation characteristics and data abnormal fluctuation characteristics of the corresponding equipment respectively.

[0037] Step S140, based on a preset abnormal pattern recognition model, a dynamic fusion analysis is performed on the equipment operation status characteristics, environment association characteristics and data abnormal fluctuation characteristics to generate an equipment operation risk level and a production process optimization strategy set.

[0038] In this embodiment, the equipment operation status characteristics of the ABC-100 mining machine in the mining process can be spliced with the environmental correlation characteristics to generate equipment environment coupling characteristics. Then, the risk level prediction branch in the abnormal pattern recognition model is called to perform convolution processing on the equipment environment coupling characteristics, and the equipment failure probability value and the environmental interference coefficient are output. Assume that the output equipment failure probability value is 0.3 and the environmental interference coefficient is 0.2.

[0039] Next, the strategy generation branch in the abnormal pattern recognition model is called to analyze the abnormal fluctuation characteristics of the data and extract the abnormal trigger conditions and fluctuation propagation paths. Assume that the abnormal trigger condition is that the number of high-frequency mutations in the vibration frequency domain energy distribution characteristics exceeds 8 times, and the fluctuation propagation path is from the mining machine motor to the transmission gear and then to the excavator arm.

[0040] The equipment operation risk level is determined based on the weighted sum of the equipment failure probability value and the environmental interference coefficient. Assuming that the weight of the equipment failure probability value is 0.6 and the weight of the environmental interference coefficient is 0.4, the calculated weighted sum is 0.3×0.6+0.2×0.4=0.26. According to the preset risk level classification standard, this value is in the medium risk range, and the equipment operation risk level is medium risk.

[0041] Subsequently, based on the anomaly trigger condition, the preset optimization operation template library is matched to generate equipment parameter adjustment instructions and environmental control plans. For example, for an anomaly trigger condition where the high-frequency mutation count exceeds 8 times, the generated equipment parameter adjustment instructions are to reduce the operating speed of the mining arm of the mining machine by 20% and lower the cooling system startup threshold by 5 degrees Celsius; the environmental control plan is to increase the ventilation volume in the operation area by 30%.

[0042] Next, based on the wave propagation path, the list of affected equipment is determined, and a collaborative optimization strategy for the production process is generated. By traversing the production process topology diagram based on the wave propagation path, the affected equipment is determined to be the transmission gear and the equipment related to the mining arm. The equipment load rate and task priority of each process section are reallocated. For example, the load rate of the equipment related to the transmission gear is reduced by 15%, and the task priority of the mining arm equipment is increased. The above-mentioned equipment parameter adjustment instructions, environmental control plan, and collaborative optimization strategy for the production process are integrated into a set of production process optimization strategies.

[0043] For the equipment in the transportation process and the crushing process, the above dynamic fusion analysis is also carried out to generate their respective equipment operation risk levels and sets of production process optimization strategies.

[0044] Step S150, trigger the corresponding alarm instruction according to the equipment operation risk level, and feedback the set of production process optimization strategies to the production control terminal to adjust the equipment operation parameters.

[0045] In this embodiment, when the equipment operation risk level of the ABC-100 type mining machine in the mining process is medium risk, an equipment maintenance warning instruction is generated and the spare parts inventory verification process is started. At the same time, the equipment parameter adjustment instructions in the set of production process optimization strategies are parsed into the rotation speed adjustment value of the target equipment, the cooling system startup threshold, and the power supply frequency correction parameter. Query the reference rotation speed range of the target equipment in the current production process from the equipment historical operation database, which is 1000 - 1500 revolutions per minute. Calculate the actual value of the target rotation speed based on the percentage coefficient corresponding to the rotation speed adjustment value (assumed to be 20%). For example, if the reference rotation speed is 1200 revolutions per minute, the adjusted rotation speed is 1200×(1 - 0.2) = 960 revolutions per minute, and a rotation speed control instruction including an acceleration time curve is generated. Determine the compensation amount of the cooling system startup threshold according to the equipment heat dissipation performance parameters and the current environmental temperature. Assume that the equipment heat dissipation performance parameters indicate that the cooling system startup threshold should be reduced by 5 degrees Celsius at the current environmental temperature, and a dynamically changing cooling trigger condition is generated. Convert the power supply frequency correction parameter into the phase adjustment amount of the inverter modulation waveform through the power supply network impedance characteristic model.

[0046] Convert the environmental control plan into the operation mode of the ventilation equipment and the work schedule of the dust removal system. For example, adjust the operation mode of the ventilation equipment to increase the ventilation volume by 30%, and adjust the work schedule of the dust removal system to increase the dust removal operation once per hour.

[0047] According to the collaborative optimization strategy of production process, the equipment load rate and task priority of each process segment are reallocated. For example, the load rate of transmission gear-related equipment is reduced by 15%, and the task priority of excavator arm equipment is increased.

[0048] Through the production control terminal, the speed adjustment value, cooling system start-up threshold, power supply frequency correction parameters, ventilation equipment operation mode, dust removal system work schedule, equipment load rate and task priority are sent to the corresponding actuator to adjust the equipment operation parameters and ensure the stable and efficient operation of the mine production process.

[0049] For the equipment in the transportation process and crushing process, the corresponding alarm instructions are triggered according to the respective equipment operation risk levels, and the corresponding production process optimization strategy set is fed back to the production control terminal to adjust the equipment operation parameters. For example, when the equipment operation risk level in the transportation process is low risk, the equipment status monitoring prompt information is generated and sent to the inspection terminal; when the equipment operation risk level in the crushing process is high risk, the equipment emergency shutdown instruction and personnel evacuation alarm signal are generated, and the corresponding parameter adjustment and strategy implementation are carried out.

[0050] Based on the above steps, the embodiment of the present invention collects the target monitoring data set of mining equipment in the real-time production process, comprehensively obtains key information such as equipment vibration, environmental temperature and humidity, and equipment energy consumption, performs dynamic standardization processing on the target monitoring data set, generates a standardized monitoring data set that matches the equipment type and production process stage, effectively eliminates the differences between data of different equipment and process stages, calls the pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the standardized monitoring data set, extracts key features such as equipment operation status, environmental association, and abnormal data fluctuations, and performs dynamic fusion analysis on the extracted features based on the preset abnormal pattern recognition model to generate equipment operation risk level and production process optimization strategy set, which can provide real-time insight into the equipment operation status, discover potential risks in advance, and formulate optimization strategies in a targeted manner. According to the equipment operation risk level, the corresponding alarm instruction is triggered, and the production process optimization strategy set is fed back to the production control terminal to adjust the equipment operation parameters, realizing real-time and precise control of the mine production process, effectively solving the problems of insufficient data utilization, untimely risk warning, and inaccurate optimization strategy in traditional mine production monitoring, improving the automation and intelligence of mine production, and reducing the incidence of production accidents.

[0051] In a possible implementation, step S120 includes:

[0052] Step S121, obtaining a set of historical operating parameters of mining equipment, wherein the set of historical operating parameters includes a rated vibration threshold range of the equipment, a standard temperature and humidity adaptation range, and a maximum energy consumption critical value.

[0053] In this embodiment, in the above-mentioned mine, various mining equipment operates continuously, ensuring the orderly progress of multiple production processes such as mining, transportation, and crushing. For the set of target monitoring data collected for each piece of equipment, dynamic standardization processing and joint feature mapping processing need to be carried out to obtain key features such as equipment operating status, environmental correlation, and abnormal data fluctuations.

[0054] For the ABC-100 type mining machine in the mining process, first, its historical operating parameter set needs to be obtained. This set covers the rated vibration threshold range of the equipment, the standard temperature and humidity adaptation interval, and the maximum energy consumption critical value. Through long-term data accumulation and analysis, it is known that the rated vibration threshold range of the ABC-100 type mining machine is from 10 mm / s² to 50 mm / s², the standard temperature and humidity adaptation interval is a temperature of 20 °C to 30 °C and a humidity of 40% to 70%, and the maximum energy consumption critical value is 100 kW.

[0055] Step S122: Match the corresponding rated vibration threshold range of the equipment according to the equipment type, and perform amplitude normalization processing on the equipment vibration time series signal to generate a standardized vibration signal sequence.

[0056] In this embodiment, during a certain mining time period, the amplitude of the equipment vibration time series signal collected fluctuates between 15 mm / s² and 60 mm / s². The process of amplitude normalization is to first determine the minimum value of 15 mm / s² and the maximum value of 60 mm / s² in this set of data, and then for each collected vibration signal amplitude data point, subtract the minimum value of 15 from this data point, and then divide by the difference between the maximum value and the minimum value (60 - 15 = 45). For example, if one of the collected vibration signal amplitudes is 30 mm / s², then after amplitude normalization, it is (30 - 15) ÷ 45 = 15 ÷ 45 = 1 / 3. In this way, a standardized vibration signal sequence is generated.

[0057] Step S123: Determine the current environmental temperature and humidity reference value according to the production process stage, and perform deviation correction processing on the environmental temperature and humidity distribution data based on the standard temperature and humidity adaptation interval to generate corrected temperature and humidity distribution data.

[0058] In this embodiment, it is assumed that the current mining operation is in an area at a depth of 200 meters underground. After long-term environmental monitoring and analysis of this area, the current environmental temperature and humidity reference values are determined to be a temperature of 25 degrees Celsius and a humidity of 50%. Deviation correction processing is performed on the environmental temperature and humidity distribution data based on the standard temperature and humidity adaptation range. A plurality of temperature and humidity sensors are arranged in this operation area. At a certain moment, the collected temperature data is 28 degrees Celsius and the humidity data is 60%. For the temperature data, the deviation from the reference temperature is 28 - 25 = 3 degrees Celsius. Correction is performed within the standard temperature and humidity adaptation range, and the correction process is (28 - 20) ÷ (30 - 20) = 8 ÷ 10 = 0.8. For the humidity data, the deviation is 60 - 50 = 10%, and the correction process is (60 - 40) ÷ (70 - 40) = 20 ÷ 30 = 2 / 3, thereby generating the corrected temperature and humidity distribution data.

[0059] Step S124, perform dynamic scaling processing on the equipment energy consumption fluctuation curve according to the maximum energy consumption critical value to generate a standardized energy consumption curve.

[0060] Step S125, integrate the standardized vibration signal sequence, the corrected temperature and humidity distribution data, and the standardized energy consumption curve into the standardized monitoring data set.

[0061] For example, in this mining operation, the collected equipment energy consumption fluctuation curve shows that the energy consumption value varies between 30 kilowatts and 80 kilowatts. During dynamic scaling processing, first find the minimum value of 30 kilowatts and the maximum value of 80 kilowatts. For each energy consumption data point, subtract the minimum value of 30 from this data point, and then divide by the difference between the maximum value and the minimum value (80 - 30 = 50). For example, at a certain moment, the energy consumption value is 60 kilowatts. After processing, it is (60 - 30) ÷ 50 = 30 ÷ 50 = 0.6, thereby generating a standardized energy consumption curve. Finally, integrate the standardized vibration signal sequence, the corrected temperature and humidity distribution data, and the standardized energy consumption curve into the standardized monitoring data set for the ABC-100 type mining machine in the mining process stage.

[0062] In the transportation process, a similar data processing is also carried out on the DEF-20 type transport truck. For example, its historical operation parameter set can be obtained, and it is known that the rated vibration threshold range of the equipment is 5 mm / s² to 30 mm / s², the standard temperature and humidity adaptation range is a temperature of 15 °C to 25 °C and a humidity of 30% to 60%, and the maximum energy consumption critical value is 60 kW. The amplitude of the collected equipment vibration time series signal is between 8 mm / s² and 25 mm / s². Perform amplitude normalization on it. Taking an amplitude of 15 mm / s² as an example, (15 - 8) ÷ (25 - 8) = 7 ÷ 17 ≈ 0.41, and generate a standardized vibration signal sequence. The current transportation route is in an open-pit mining area, and the current ambient temperature and humidity reference values are determined to be a temperature of 20 °C and a humidity of 40%. At a certain moment, the temperature of 22 °C and the humidity of 50% are collected. The temperature correction is (22 - 15) ÷ (25 - 15) = 7 ÷ 10 = 0.7, and the humidity correction is (50 - 30) ÷ (60 - 30) = 20 ÷ 30 = 2 / 3, obtaining the corrected temperature and humidity distribution data. The energy consumption values in the collected equipment energy consumption fluctuation curve are between 10 kW and 40 kW. Perform dynamic scaling on the energy consumption value of 30 kW, (30 - 10) ÷ (40 - 10) = 20 ÷ 30 = 2 / 3, generate a standardized energy consumption curve, and after integration, form a standardized monitoring data set for the DEF-20 type transport truck in the transportation process stage.

[0063] The same is true for the GHI-300 type crusher in the crushing process. Its historical operation parameter set shows that the rated vibration threshold range of the equipment is 20 mm / s² to 80 mm / s², the standard temperature and humidity adaptation range is a temperature of 22 °C to 32 °C and a humidity of 45% to 75%, and the maximum energy consumption critical value is 120 kW. The amplitude of the collected vibration signal is between 30 mm / s² and 70 mm / s². Perform normalization on the amplitude of 40 mm / s², (40 - 30) ÷ (70 - 30) = 10 ÷ 40 = 0.25, and generate a standardized vibration signal sequence. The ambient temperature and humidity reference values are determined to be a temperature of 25 °C and a humidity of 55% in the current crushing workshop. At a certain moment, the temperature of 27 °C and the humidity of 65% are collected. The temperature correction is (27 - 22) ÷ (32 - 22) = 5 ÷ 10 = 0.5, and the humidity correction is (65 - 45) ÷ (75 - 45) = 20 ÷ 30 = 2 / 3, obtaining the corrected temperature and humidity distribution data. The energy consumption values in the collected energy consumption fluctuation curve are between 40 kW and 100 kW. Perform dynamic scaling on the energy consumption value of 70 kW, (70 - 40) ÷ (100 - 40) = 30 ÷ 60 = 0.5, generate a standardized energy consumption curve, and then integrate it into a standardized monitoring data set for the GHI-300 type crusher in the crushing process stage.

[0064] In a possible implementation manner, step S130 includes:

[0065] Step S131: Perform time-frequency conversion processing on the standardized vibration signal sequence to extract the vibration frequency-domain energy distribution characteristics and the resonance peak offset.

[0066] For example, for the standardized vibration signal sequence of the ABC-100 type mining machine in the mining process, perform time-frequency conversion processing to extract the vibration frequency-domain energy distribution characteristics and the resonance peak offset. First, perform windowed Fourier transform on the standardized vibration signal sequence. The window function selects a length of 128 data points, and the overlap rate is set to 50%, thereby generating a vibration spectrogram. In the vibration spectrogram, divide the frequency range into multiple intervals, such as 0 - 50Hz, 50 - 100Hz, 100 - 150Hz, etc. Calculate the energy integral value in each interval, and generate a frequency-domain energy distribution histogram through the cumulative calculation of the energies in each interval. After detection, the main resonance peak frequency in the vibration spectrogram is 160Hz, while the standard resonance frequency of the equipment is 150Hz. Calculate the difference between the two, 160 - 150 = 10Hz, to obtain the resonance peak offset. Perform sliding window statistics on the frequency-domain energy distribution histogram. The window size is set to 10 intervals. During the statistical process, it is found that the low-frequency energy accounts for 35%, and the number of high-frequency mutations is 6 times. Perform logarithmic transformation on these two characteristics. The low-frequency energy proportion characteristic is transformed into ln(0.35), and the high-frequency mutation number characteristic is transformed into ln(6). Combine the transformed characteristics into the vibration frequency-domain energy distribution characteristics.

[0067] Step S132: Perform spatial interpolation processing on the corrected temperature and humidity distribution data to generate the temperature and humidity gradient change characteristics and the local abnormal temperature zone markers.

[0068] For example, in the working area of the ABC-100 type mining machine, 9 temperature and humidity sensors are arranged within a range of 15×15 square meters. Through the spatial interpolation algorithm, calculate the temperature change rate and humidity change rate between adjacent sensors according to the data collected by each sensor. For example, the temperatures of two adjacent sensors are 27 degrees Celsius and 28 degrees Celsius respectively, and the distance is 3 meters. Then the temperature change rate is (28 - 27)÷3 = 1÷3 ≈ 0.33 degrees Celsius / meter. Calculate the temperature and humidity gradient change characteristics of the entire area in this way. At the same time, through the analysis of the data at each point, mark the local abnormal temperature zones. For example, the temperature in a certain area is significantly higher than the surrounding area, reaching 32 degrees Celsius, and it is marked as a local abnormal temperature zone, generating local abnormal temperature zone markers.

[0069] Step S133: Perform slope change analysis on the standardized energy consumption curve to extract the energy consumption increase rate characteristics and the peak duration characteristics.

[0070] In this embodiment, a first-order difference calculation is performed on the standardized energy consumption curve, that is, the difference between adjacent data points is calculated. Suppose the data points of the standardized energy consumption curve within a certain period of time are 0.4, 0.5, 0.6, 0.7, etc. After the first-order difference calculation, the energy consumption change rate sequence obtained is 0.1, 0.1, 0.1, etc. Set the rated slope threshold to 0.15, and detect the time interval in which the energy consumption change rate sequence continuously exceeds this threshold. After investigation, it is found that within a certain time period, the energy consumption change rate from the 10th minute to the 15th minute continuously exceeds 0.15. Record its start timestamp of 10 minutes and end timestamp of 15 minutes. Within this time interval, the energy consumption change rates are 0.16, 0.18, 0.2, 0.17 respectively. Calculate the average slope value, (0.16 + 0.18 + 0.2 + 0.17)÷4 = 0.71÷4 = 0.1775, and take it as the energy consumption rising rate feature. According to the difference between the start timestamp and the end timestamp, 15 - 10 = 5 minutes, determine that the peak duration feature is 5 minutes. Detect the extreme points of the energy consumption change rate sequence, find that there are energy consumption mutation events, record the corresponding event duration, and generate energy consumption mutation event marks and the corresponding event durations.

[0071] Step S134, input the vibration frequency domain energy distribution feature, resonance peak offset, temperature and humidity gradient change feature, local abnormal temperature zone mark, energy consumption rising rate feature, and peak duration feature into the multi-dimensional feature extraction model, and perform correlation weighting on the vibration frequency domain energy distribution feature and the temperature and humidity gradient change feature through a cross-modal attention mechanism to generate an environment correlation feature.

[0072] In this embodiment, the vibration frequency domain energy distribution feature can be mapped to a set of query vectors, assumed to be [ln(0.35), ln(6)], and the temperature and humidity gradient change feature can be mapped to a set of key vectors, assumed to be [0.33, 0.25]. Calculate the cosine similarity between each query vector and the key vector. Taking the first query vector and the first key vector as an example, the cosine similarity calculation is (ln(0.35)×0.33)÷(sqrt(ln2(0.35)+ln2(6))×sqrt(0.33 2 +0.252)), and a specific value is obtained through calculation. And so on to generate an initial attention weight matrix. Adjust the temperature parameter of the initial attention weight matrix, set the temperature parameter to 0.1, and then perform Softmax normalization processing to generate a standardized attention weight. Dynamically weight and sum the temperature and humidity gradient change features according to the standardized attention weight to generate an environment-sensitive vibration feature. Multiply the environment-sensitive vibration feature and the temperature and humidity gradient change feature element by element to generate an environment correlation feature.

[0073] Step S135, jointly encode the formant offset and the energy consumption increase rate feature through a temporal convolutional network to generate a device operating state feature.

[0074] Step S136, perform pattern matching on the local abnormal temperature zone mark and the peak duration feature through an anomaly detection algorithm to generate a data abnormal fluctuation feature.

[0075] For example, the formant offset of 10 Hz and the energy consumption increase rate feature of 0.1775 can be used as inputs, and after a series of processes such as the convolutional layer and pooling layer in the network, a device operating state feature is generated. Perform pattern matching on the local abnormal temperature zone mark and the peak duration feature through an anomaly detection algorithm. For example, in the preset abnormal pattern, it is specified that when the temperature in the local abnormal temperature zone exceeds 30 degrees Celsius and the peak duration exceeds 4 minutes, it is an abnormal situation. The current temperature in the local abnormal temperature zone is 32 degrees Celsius, and the peak duration is 5 minutes, meeting the abnormal pattern condition, and a data abnormal fluctuation feature is generated.

[0076] For the DEF-20 type transport truck in the transportation process and the GHI-300 type crusher in the crushing process, the combined feature mapping process is also carried out according to the above detailed process, and the device operating state feature, the environment correlation feature and the data abnormal fluctuation feature are generated respectively.

[0077] In a possible implementation manner, step S131 includes:

[0078] Step S1311, perform windowed Fourier transform on the standardized vibration signal sequence to generate a vibration spectrogram.

[0079] In this embodiment, for the ABC-100 type mining machine in the mining process, perform time-frequency conversion processing on its standardized vibration signal sequence to extract key features, and at the same time perform slope change analysis on the standardized energy consumption curve, and generate an environment correlation feature through a cross-modal attention mechanism.

[0080] For the standardized vibration signal sequence of the ABC-100 type mining machine, first perform windowed Fourier transform to generate a vibration spectrogram. In actual operation, a Hanning window with a length of 256 data points is selected as the window function. The reason for choosing the Hanning window is that it can perform well in reducing spectral leakage. At the same time, set the overlap rate to 75%, which means that 75% of the data points between adjacent windows overlap. In the collected standardized vibration signal sequence, taking a sequence containing 1024 data points as an example, perform windowed Fourier transform according to the set parameters. Starting from the first data point, take 256 data points as a group in turn and apply the Hanning window function for processing. For each group of data, convert the time-domain signal to the frequency domain through Fourier transform to obtain the corresponding spectral information. In this way, after processing multiple groups of data, a vibration spectrogram reflecting the frequency characteristics of the standardized vibration signal sequence is generated.

[0081] Step S1312, calculate the energy integral value of each frequency interval in the vibration spectrogram to generate a frequency-domain energy distribution histogram.

[0082] In this embodiment, the frequency range of the vibration spectrogram can be divided into multiple intervals. For example, from 0 Hz to 500 Hz, it is divided into 20 equal-width intervals, and the width of each interval is 25 Hz. For each interval, calculate the sum of the squares of the spectral values in this interval to approximately represent the energy of this interval. Taking the first interval from 0 Hz to 25 Hz as an example, find all the spectral values in this interval in the vibration spectrogram, square the above spectral values respectively and then add them up. Suppose there are 10 spectral values in this interval, which are 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0 respectively. Then the energy integral value of this interval is the square of 0.1 plus the square of 0.2 plus the square of 0.3 plus the square of 0.4 plus the square of 0.5 plus the square of 0.6 plus the square of 0.7 plus the square of 0.8 plus the square of 0.9 plus the square of 1.0, that is, 0.01 + 0.04 + 0.09 + 0.16 + 0.25 + 0.36 + 0.49 + 0.64 + 0.81 + 1.0 = 3.85. Calculate the energy integral value of each frequency interval in the same way, and then use the frequency interval as the abscissa and the energy integral value as the ordinate to draw a frequency-domain energy distribution histogram.

[0083] Step S1313, detect the main resonance peak frequency in the vibration spectrogram, and calculate the difference from the equipment standard resonance frequency to generate a resonance peak offset.

[0084] In this embodiment, by analyzing the vibration spectrogram, the frequency point with the largest spectral value is found, which is the main resonance peak frequency. After detection, it is found that the main resonance peak frequency is 180 Hz. The standard resonance frequency of the ABC-100 type mining machine is 150 Hz. The difference between the two is calculated as 180 - 150 = 30 Hz, and this 30 Hz is the resonance peak offset. The resonance peak offset can reflect the difference between the current vibration state of the device and the standard state.

[0085] Step S1314, perform a sliding window statistics on the frequency domain energy distribution histogram, and extract the low-frequency energy ratio feature and the high-frequency mutation count feature.

[0086] In this embodiment, the size of the sliding window is set to 5 intervals, and it slides with a step size of 1 interval. For each window position, calculate the ratio of the total energy of the low-frequency interval (assumed to be from 0 Hz to 100 Hz, a total of 4 intervals) in the total energy of the entire window as the low-frequency energy ratio feature. For example, at a certain window position, the energy integral values of the 5 intervals within the window are 2.0, 3.0, 4.0, 5.0, and 6.0 respectively. The total energy of the low-frequency interval (the first 4 intervals) is 2.0 + 3.0 + 4.0 + 5.0 = 14.0, and the total energy of the entire window is 2.0 + 3.0 + 4.0 + 5.0 + 6.0 = 20.0. Then the low-frequency energy ratio is 14.0 divided by 20.0, which is 0.7. At the same time, during the sliding process, detect the change of the energy integral value in the high-frequency interval (assumed to be from 300 Hz to 500 Hz, a total of 8 intervals). If the energy integral value of a certain interval changes by more than a certain threshold (assumed to be 2 times) compared with the adjacent interval, it is determined as one high-frequency mutation. During the entire sliding window statistics process, record the number of high-frequency mutations. Assume that during the statistics process, it is found that there are 3 high-frequency mutations, then the high-frequency mutation count feature is 3.

[0087] Step S1315, perform a logarithmic transformation on the low-frequency energy ratio feature and the high-frequency mutation count feature and combine them into the vibration frequency domain energy distribution feature.

[0088] In this embodiment, for example, perform a logarithmic transformation on the low-frequency energy ratio feature 0.7 to obtain ln(0.7). Perform a logarithmic transformation on the high-frequency mutation count feature 3 to obtain ln(3). Combine these two features after logarithmic transformation to form the vibration frequency domain energy distribution feature. This vibration frequency domain energy distribution feature combines the low-frequency energy distribution and the high-frequency mutation situation, and more comprehensively reflects the characteristics of the device vibration signal in the frequency domain.

[0089] In a possible implementation manner, step S133 includes:

[0090] Step S1331, perform a first-order difference calculation on the standardized energy consumption curve to generate an energy consumption change rate sequence.

[0091] For example, for the standardized energy consumption curve of the ABC-100 type mining machine, first perform a first-order difference calculation to generate an energy consumption change rate sequence. The standardized energy consumption curve is a curve composed of a series of energy consumption data points that change over time. Taking a section of standardized energy consumption curve data containing 100 time points as an example, the data points are 0.4, 0.45, 0.5, 0.55, 0.6, etc. in sequence. The first-order difference calculation is to calculate the difference between adjacent two data points. For the first data point 0.4 and the second data point 0.45, the difference is 0.45 - 0.4 = 0.05; for the second data point 0.45 and the third data point 0.5, the difference is 0.5 - 0.45 = 0.05. And so on. After calculating these 100 data points, the obtained energy consumption change rate sequence is 0.05, 0.05, 0.05, 0.05, etc.

[0092] Step S1332, detect the time intervals in the energy consumption change rate sequence that continuously exceed the rated slope threshold, and record their start time stamps and end time stamps.

[0093] For example, set the rated slope threshold to 0.08. In the energy consumption change rate sequence, start checking from the first data point. When it is found that a certain data point and its subsequent consecutive data points all exceed 0.08, start recording the start time stamp. Suppose at the 10th time point, the energy consumption change rate is 0.09, which exceeds the rated slope threshold. Continue to check the subsequent data points and find that from the 10th time point to the 20th time point, the energy consumption change rate all exceeds 0.08. Then the start time stamp is the 10th time point and the end time stamp is the 20th time point.

[0094] Step S1333, calculate the average slope value within the time interval as the energy consumption increase rate feature.

[0095] For example, within the above-determined time interval (from the 10th time point to the 20th time point), the energy consumption change rate sequence is 0.09, 0.1, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19, 0.2. Calculate the average value of the above data, that is, (0.09 + 0.1 + 0.11 + 0.12 + 0.13 + 0.14 + 0.15 + 0.16 + 0.17 + 0.18 + 0.19 + 0.2) divided by 11 (the number of data points), and get (1.74) divided by 11 ≈ 0.158. This 0.158 is the energy consumption increase rate feature.

[0096] Step S1334, determine the peak duration feature according to the difference between the start timestamp and the end timestamp.

[0097] For example, if the start timestamp is the 10th time point and the end timestamp is the 20th time point, the difference between them is 20 - 10 = 10 time units (assuming each time point interval is 1 minute, then the peak duration is 10 minutes), and this is the peak duration feature.

[0098] Step S1335, perform extreme point detection on the energy consumption change rate sequence to generate energy consumption mutation event markers and corresponding event durations.

[0099] For example, in the energy consumption change rate sequence, find the points that are larger or smaller than adjacent data points. These points are extreme points. For example, in the sequence 0.05, 0.05, 0.05, 0.06, 0.05, 0.04, 0.06 is a maximum point and 0.04 is a minimum point. When an extreme point is detected, record the time point when the extreme point appears and the duration of the extreme point. Assume that the maximum point 0.06 starts at the 25th time point and lasts until the 27th time point, then mark this as an energy consumption mutation event, and the event duration is 3 time units (3 minutes), thus generating energy consumption mutation event markers and corresponding event durations.

[0100] In a possible implementation, step S134 includes:

[0101] Step S1341, map the vibration frequency domain energy distribution feature to a set of query vectors, and map the temperature and humidity gradient change feature to a set of key vectors.

[0102] For example, for the ABC - 100 type mining machine, assume that the vibration frequency domain energy distribution feature after processing is [ln(0.7), ln(3)], and map it to the set of query vectors Q = [[ln(0.7), ln(3)]]. Assume that the temperature and humidity gradient change feature after processing is [0.2, 0.3], and map it to the set of key vectors K = [[0.2, 0.3]].

[0103] Step S1342, calculate the cosine similarity between each query vector and the key vector to generate an initial attention weight matrix.

[0104] For example, for the query vector [ln(0.7), ln(3)] and the key vector [0.2, 0.3], the formula for calculating the cosine similarity is the dot product of the two vectors divided by the product of their magnitudes. First, calculate the dot product, ln(0.7)×0.2 + ln(3)×0.3. ln(0.7) is approximately -0.357, and ln(3) is approximately 1.099. Then the dot product is -0.357×0.2 + 1.099×0.3 = -0.0714 + 0.3297 = 0.2583. Next, calculate the magnitude of the query vector, sqrt(ln²(0.7) + ln²(3)) = sqrt((-0.357)² + (1.099)²) = sqrt(0.127449 + 1.207801) = sqrt(1.33525) ≈ 1.155. The magnitude of the key vector is sqrt(0.2² + 0.3 2 ) = sqrt(0.04 + 0.09) = sqrt(0.13) ≈ 0.361. Then the cosine similarity is 0.2583 divided by (1.155×0.361), 0.2583÷(1.155×0.361) = 0.2583÷0.417955 ≈ 0.618. From this, an initial attention weight matrix is generated. Assume that this initial attention weight matrix has only one element 0.618.

[0105] Step S1343, perform temperature parameter adjustment and Softmax normalization on the initial attention weight matrix to generate a normalized attention weight.

[0106] For example, set the temperature parameter τ = 0.1. Adjust each element in the initial attention weight matrix by the temperature parameter, divide the element value by the temperature parameter, and get 0.618÷0.1 = 6.18. The Softmax normalization process is to perform an exponential operation on the adjusted element value and then divide it by the sum of all element values after the exponential operation. Here there is only one element, and after the exponential operation, it is e to the power of 6.18, approximately 485.5. The normalized attention weight is this value divided by the sum of all element values after the exponential operation (here there is only one element, so it is itself), that is, the normalized attention weight is 485.5÷485.5 = 1.

[0107] Step S1344, perform dynamic weighted summation on the temperature and humidity gradient change features according to the normalized attention weight to generate environment - sensitive vibration features.

[0108] For example, the temperature and humidity gradient change features are [0.2, 0.3], and the normalized attention weight is 1. The dynamic weighted summation is to multiply the normalized attention weight by each element of the temperature and humidity gradient change features and then add them up, that is, 1×0.2 + 1×0.3 = 0.5, generating the environment - sensitive vibration feature 0.5.

[0109] Step S1345, multiply the environment-sensitive vibration feature and the temperature-humidity gradient change feature element by element to generate the environment-related feature.

[0110] For example, if the environment-sensitive vibration feature is 0.5 and the temperature-humidity gradient change feature is [0.2, 0.3], the element-by-element multiplication gives [0.5×0.2, 0.5×0.3] = [0.1, 0.15], which is the generated environment-related feature.

[0111] For the DEF-20 type transport truck in the transportation process and the GHI-300 type crusher in the crushing process, they are also processed according to the above detailed process. When performing time-frequency conversion processing on their standardized vibration signal sequences, slope change analysis on the standardized energy consumption curve, and generating environment-related features through the cross-modal attention mechanism, corresponding calculations and processing are carried out according to the characteristics of their respective equipment and the collected data to obtain key features reflecting the equipment operation status, energy consumption situation, and the relationship with the environment.

[0112] In a possible implementation manner, step S140 includes:

[0113] Step S141, splice the equipment operation status feature and the environment-related feature to generate an equipment-environment coupling feature.

[0114] In this embodiment, the equipment operation status feature of the ABC-100 type mining machine is obtained by jointly encoding the resonance peak offset and the energy consumption increase rate feature. Suppose its value is [0.6, 0.8], and these two values respectively reflect the operation status of the equipment in terms of vibration and energy consumption. The environment-related feature is generated by correlation weighting of the vibration frequency domain energy distribution feature and the temperature-humidity gradient change feature through the cross-modal attention mechanism. Suppose it is [0.4, 0.5], which reflects the relationship between the equipment operation and environmental factors. Splice these two sets of features, that is, combine them in order to form the equipment-environment coupling feature [0.6, 0.8, 0.4, 0.5], which integrates the information of the equipment's own operation status and environmental factors.

[0115] Step S142, call the risk level prediction branch in the abnormal mode recognition model to perform convolution processing on the equipment-environment coupling feature, and output the equipment failure probability value and the environmental interference coefficient.

[0116] In this embodiment, the risk level prediction branch in the abnormal pattern recognition model includes a series of convolutional layers, activation functions, and fully connected layers. When the device environment coupling feature [0.6, 0.8, 0.4, 0.5] is convolved, it first passes through the first convolution layer, which has multiple convolution kernels, and each convolution kernel performs a convolution operation with the device environment coupling feature. For example, one of the convolution kernels is [0.2, 0.3, 0.1, 0.4], which is convolved with the device environment coupling feature, and the corresponding elements are multiplied and summed: 0.6×0.2+0.8×0.3+0.4×0.1+0.5×0.4=0.12+0.24+0.04+0.2=0.6. After the operation of multiple convolution kernels, a new set of data is obtained. Then, the above data is processed by an activation function, such as a ReLU function, and values less than 0 are changed to 0, and values greater than 0 remain unchanged. After multiple layers of convolution and activation function processing, it enters the fully connected layer. The fully connected layer performs comprehensive calculations on the previously processed data and finally outputs the device failure probability value and the environmental interference coefficient. Assume that after this series of processing, the output device failure probability value is 0.3 and the environmental interference coefficient is 0.2. The device failure probability value reflects the possibility of device failure, and the environmental interference coefficient reflects the degree of influence of environmental factors on device operation.

[0117] Step S143, calling the strategy generation branch in the abnormal pattern recognition model to perform pattern analysis on the abnormal data fluctuation characteristics, and extracting the abnormal triggering conditions and fluctuation propagation paths.

[0118] In this embodiment, the abnormal data fluctuation characteristics are obtained by pattern matching the local abnormal temperature zone mark and the peak duration characteristics. Assume that the abnormal data fluctuation characteristics show that in a certain time period, the temperature of the local abnormal temperature zone reaches 35 degrees Celsius (higher than the normal range), and the peak duration is 8 minutes (exceeding the normal standard). The strategy generation branch can conduct an in-depth analysis of the above-mentioned abnormal data fluctuation characteristics, and by comparing with the preset abnormal pattern library, it is identified that this is an abnormal pattern in which the energy consumption of the equipment is abnormally increased due to excessively high ambient temperature. Then, the abnormal trigger condition is extracted as the temperature of the local abnormal temperature zone exceeds 32 degrees Celsius and the peak duration exceeds 6 minutes. The fluctuation propagation path affects the cooling system of the equipment from the local high temperature area, and then affects the energy consumption of the equipment, and then affects other related equipment through the power transmission system between the equipment. Clarifying the abnormal trigger conditions and fluctuation propagation paths is crucial for the subsequent formulation of targeted optimization strategies.

[0119] Step S144, determining the equipment operation risk level according to the weighted sum of the equipment failure probability value and the environmental interference coefficient, wherein the equipment operation risk level includes low risk, medium risk and high risk.

[0120] In this embodiment, it is assumed that the weight of the preset equipment failure probability value is 0.6 and the weight of the environmental interference coefficient is 0.4. The weighted sum is calculated as: 0.3×0.6 + 0.2×0.4 = 0.18 + 0.08 = 0.26. According to the pre-established risk level classification standard, when the weighted sum is less than 0.3, it is a low risk; between 0.3 and 0.6 is a medium risk; and greater than 0.6 is a high risk. Since the calculated weighted sum is 0.26, the current equipment operation risk level of the ABC-100 type mining machine is a low risk. Accurately determining the risk level can enable corresponding measures to be taken in a timely manner to ensure the stable operation of the equipment.

[0121] Step S145: Based on the abnormal trigger condition, match the preset optimized operation template library to generate an equipment parameter adjustment instruction and an environmental regulation plan.

[0122] For example, for the abnormal trigger condition extracted previously, that is, the temperature in the local abnormal temperature area exceeds 32 degrees Celsius and the peak duration exceeds 6 minutes, match it in the preset optimized operation template library. The template library sets corresponding optimization measures for this situation. The generated equipment parameter adjustment instruction is to reduce the start threshold of the cooling system of the mining machine by 3 degrees Celsius, so that the cooling system can start earlier to cope with the high-temperature environment. At the same time, appropriately reduce the excavation speed of the equipment to reduce the load of the equipment, thereby reducing energy consumption. Specifically, adjust the excavation speed from the original 10 cubic meters per minute to 8 cubic meters per minute. The environmental regulation plan is to increase the ventilation volume of the operation area, increase the operating power of the ventilation equipment by 20%, to reduce the temperature in the local abnormal temperature area. And adjust the working schedule of the dust removal system to increase the dust removal frequency during high-temperature periods, from once per hour originally to once every half hour, to improve the operation environment and reduce the impact of dust on the equipment. The above instructions and plans are aimed at eliminating or reducing the impact of the abnormal trigger condition on the equipment operation and improving the stability and reliability of the equipment.

[0123] Step S146: Determine the list of affected equipment according to the wave propagation path and generate a collaborative optimization strategy for the production process.

[0124] In this embodiment, according to the previously determined wave propagation path, that is, from the local high-temperature area to the heat dissipation system of the equipment, which in turn affects the energy consumption of the equipment, and then affects other associated equipment through the power transmission system between the equipment. After analysis, it is determined that the affected equipment includes the transportation equipment sharing the power transmission system with the ABC-100 type mining machine, and other auxiliary equipment located in the same working area. In order to ensure the coordinated operation of the entire production process, a coordinated optimization strategy for the production process is generated. For the affected transportation equipment, appropriately reduce its transportation speed and load to relieve the pressure on the power transmission system. For example, reduce the transportation speed of the transportation equipment from 30 kilometers per hour to 25 kilometers per hour, and reduce the amount of ore transported each time from 20 tons to 18 tons. For other auxiliary equipment, adjust its working hours to avoid full-load operation at the same time during the high-temperature operation period of the mining machine. Reallocate the equipment load rate and task priority of each process section, appropriately reduce the task priority of the mining machine, and at the same time increase the task priority of the transportation equipment and auxiliary equipment in ensuring the normal operation of the equipment. Through the above-mentioned coordinated optimization strategy for the production process, the coordinated work between various equipment is realized, and the impact of abnormal fluctuations on the entire production process is reduced.

[0125] Step S147, integrate the equipment parameter adjustment instruction, the environmental control plan and the coordinated optimization strategy for the production process into the production process optimization strategy set.

[0126] In this embodiment, the production process optimization strategy set includes comprehensive optimization measures for the ABC-100 type mining machine and its related production links, aiming to improve the equipment operation efficiency, reduce risks, and ensure the stability and high efficiency of the entire production process.

[0127] In a possible implementation manner, step S150 includes:

[0128] Step S151, when the equipment operation risk level is low risk, generate equipment status monitoring prompt information and send it to the patrol terminal.

[0129] For example, when the equipment operation risk level of the ABC-100 type mining machine is low risk, generate equipment status monitoring prompt information and send it to the patrol terminal. The content of the prompt information is "The current operation risk level of the ABC-100 type mining machine is low risk, but the local abnormal temperature area and energy consumption changes still need to be concerned. It is recommended that the patrol personnel strengthen the inspection of relevant areas." After receiving this information, the patrol personnel will focus on the above aspects during the daily patrol and discover potential problems in a timely manner.

[0130] Step S152, when the equipment operation risk level is medium risk, generate an equipment maintenance warning instruction and start the spare parts inventory verification process.

[0131] For example, the equipment maintenance warning instruction notifies the maintenance team to conduct a comprehensive inspection and maintenance of the ABC-100 type mining machine to prevent possible failures in advance. At the same time, start the spare parts inventory verification process to check the inventory of spare parts related to this mining machine, such as the components of the cooling system, transmission parts, etc. Ensure that there are sufficient spare parts in the inventory so that they can be replaced in time when needed, reducing the equipment downtime. For example, after verification, it is found that the inventory of a key valve spare part in the cooling system is insufficient, and procurement is promptly arranged for replenishment.

[0132] Step S153, when the equipment operation risk level is high risk, generate an equipment emergency shutdown instruction and a personnel evacuation warning signal.

[0133] For example, an emergency shutdown instruction can be immediately sent to the ABC-100 type mining machine to stop its operation, avoiding greater losses to the equipment and personnel caused by possible serious failures. At the same time, start the personnel evacuation warning signal to notify the personnel in the nearby operation area to quickly evacuate to a safe area. Through methods such as the broadcast system and warning lights, ensure that the personnel can receive the evacuation information in time and evacuate safely.

[0134] Step S154, parse the equipment parameter adjustment instruction in the production process optimization strategy set into the rotational speed adjustment value of the target equipment, the cooling system startup threshold, and the power supply frequency correction parameter.

[0135] For example, the adjustment of the mining speed of the mining machine in the equipment parameter adjustment instruction is converted into the rotational speed adjustment value. Assuming that the mining speed of the mining machine has a certain proportional relationship with the motor rotational speed, after calculation, the motor rotational speed is adjusted from the original 1500 revolutions per minute to 1200 revolutions per minute. The cooling system startup threshold is reduced from the original 38 degrees Celsius to 35 degrees Celsius. For the power supply frequency correction parameter, according to the electrical characteristics and operation requirements of the equipment, the power supply frequency is fine-tuned from 50Hz to 49.5Hz to meet the requirements of the equipment in the new operation state.

[0136] Step S155, convert the environmental control plan into the operation mode of the ventilation equipment and the working schedule of the dust removal system.

[0137] For example, the operation mode of the ventilation equipment is adjusted to increase the operation power of the ventilation equipment by 20%. The original operation power of the ventilation equipment is 50 kilowatts, and after the increase, it becomes 50×(1 + 20%) = 50×1.2 = 60 kilowatts, and the operation time of the ventilation equipment is adjusted so that it operates continuously during high-temperature periods. The working schedule of the dust removal system is increased from once per hour to once every half hour, and the specific working time is reasonably arranged according to the production operation time to ensure that the dust concentration can be effectively reduced during the equipment operation.

[0138] Step S156: Reallocate the equipment load rates and task priorities of each process segment according to the collaborative optimization strategy for the production process.

[0139] For example, for the mining process, reduce the load rate of the ABC-100 type mining machine from the original 80% to 60%, and reduce its excavation volume to lower the working intensity of the equipment. For the transportation process, reduce the load rate of the transportation equipment from the original 70% to 60%, and at the same time adjust the transportation task priority to give priority to ensuring the ore transportation needs of the mining machine. For other auxiliary processes, reasonably adjust the equipment load rate and task priority according to their correlation with the mining machine to ensure the coordinated operation of the entire production process.

[0140] Step S157: Send the rotation speed adjustment value, cooling system start threshold, power supply frequency correction parameter, ventilation equipment operation mode, dust removal system work schedule, equipment load rate, and task priority to the corresponding actuators through the production control terminal.

[0141] For example, the production control terminal sends the above adjustment parameters and instructions accurately to the controllers and actuators of each device through a wired or wireless communication network. For example, send the rotation speed adjustment value to the motor controller of the mining machine, and the motor controller adjusts the rotation speed of the motor according to the received instruction; send the cooling system start threshold to the control module of the cooling system, and when the temperature reaches the new threshold, the cooling system starts automatically; send the power supply frequency correction parameter to the inverter of the power supply system, and the inverter adjusts the output power supply frequency. The actuators of the ventilation equipment and the dust removal system make corresponding adjustments according to the received operation mode and work schedule. The equipment in each process segment adjusts its own operating state according to the reallocated load rate and task priority to achieve the optimization and adjustment of the entire production process, and ensure the safe, stable, and efficient operation of the mine production.

[0142] For the DEF-20 type transport truck in the transportation process and the GHI-300 type crusher in the crushing process, the same detailed process of dynamic fusion analysis, risk level assessment, strategy generation, and instruction execution is carried out. According to their respective equipment operating state characteristics, environmental correlation characteristics, and data abnormal fluctuation characteristics, combined with the preset abnormal mode recognition model, determine the corresponding equipment operating risk level, generate a targeted set of production process optimization strategies, and adjust the equipment operating parameters through the production control terminal to ensure the stable operation of the entire mine production system.

[0143] In a possible implementation manner, step S146 includes:

[0144] Step S1461: Traverse the production process topology diagram based on the fluctuation propagation path, and extract the set of adjacent process nodes that have a signal transmission path with the current data abnormal fluctuation characteristics.

[0145] For example, the production process topology diagram details the connection relationships and signal transmission paths between various processes in the mine production process. Taking the ABC-100 type mining machine as an example, assuming that the abnormal data fluctuations are caused by the excessive temperature in a local abnormal temperature area affecting the equipment energy consumption, through traversing the topology diagram, it is found that the adjacent process nodes directly associated with the mining machine include the ore transportation link and the pre-treatment link before crushing. The ore transportation link is responsible for transporting the mined ore to the designated location, while the pre-treatment link before crushing conducts preliminary screening and processing of the ore. There are material transmission and signal interaction between the above links and the mining machine, constituting the possible propagation path of the abnormal data fluctuations. Therefore, the extracted set of adjacent process nodes covers the node where the DEF-20 type transport truck in the transportation process is located, and some screening equipment nodes in the pre-treatment process.

[0146] Step S1462: Query the equipment identifiers of the equipment sharing the power supply system or connected by conveyor belts in the set of adjacent process nodes according to the equipment dependency graph, and generate a target equipment set.

[0147] In this embodiment, the equipment dependency graph records the power supply and mechanical connection relationships between various types of equipment in the mine. In the above set of adjacent process nodes, by querying the graph, it can be seen that the DEF-20 type transport truck shares some lines of the same power supply system with the ABC-100 type mining machine, which means that the abnormal fluctuations of the mining machine may affect the transport truck through the power system. At the same time, the screening equipment in the pre-treatment process is connected to the mining machine by a conveyor belt, and during the process of the ore being transported from the mining machine to the screening equipment, the abnormal fluctuations may also be transmitted accordingly. Based on the above relationships, it is determined that the target equipment set includes the DEF-20 type transport truck, the GHI-10 type screening equipment in the pre-treatment process, and other auxiliary equipment closely related to them. The above equipment is interrelated in the production process, and the abnormality of one equipment may trigger a chain reaction, so they are all included in the target equipment set for further analysis.

[0148] Step S1463: Obtain the historical operation parameters of each equipment in the target equipment set, and calculate the signal attenuation coefficient of the abnormal data fluctuation characteristics on the fluctuation propagation path.

[0149] For example, for the DEF-20 type transport truck, its historical operation parameters include data such as speed and energy consumption under different loads, as well as records of operation stability under various environmental conditions. For the GHI-10 type screening equipment, the historical operation parameters cover information such as the vibration frequency of the equipment, the screening efficiency, and the wear conditions of different components. By analyzing the above historical data and combining with the specific situation of the abnormal fluctuation of the current ABC-100 type mining machine data, the signal attenuation coefficient is calculated. Assuming that the abnormal fluctuation of the mining machine propagates in the form of energy and is transmitted to other equipment through the power supply system and the conveyor belt. Taking the power supply system as an example, when calculating the signal attenuation coefficient from the mining machine to the transport truck, factors such as the resistance and capacitance of the power line on the energy transmission are considered. Assuming that the resistance of the power line will cause the transmitted energy to decay by a certain proportion, after a series of analyses and data comparisons, the calculated signal attenuation coefficient from the mining machine to the transport truck is 0.6. This means that when the abnormal fluctuation of the mining machine is transmitted to the transport truck, its intensity will be weakened to 60% of the original. For the screening equipment connected by the conveyor belt, considering the influence of factors such as the material, length of the conveyor belt, and the conveying volume of the ore on the signal transmission, the calculated signal attenuation coefficient is 0.7. The above signal attenuation coefficients reflect the differences in the influence degrees of the abnormal fluctuation on each equipment in different propagation paths.

[0150] Step S1464, according to the signal attenuation coefficient, screen the list of affected equipment identifiers in the target equipment set whose influence degree exceeds a preset threshold.

[0151] In this embodiment, the preset threshold is determined based on long-term production experience and equipment performance research and is used to judge whether the equipment is significantly affected. Assuming that the preset threshold is 0.5, that is, when the signal attenuation coefficient is greater than 0.5, it is considered that the equipment is significantly affected. Based on this standard, the signal attenuation coefficient of the DEF-20 type transport truck is 0.6, exceeding the preset threshold; the signal attenuation coefficient of the GHI-10 type screening equipment is 0.7, also exceeding the preset threshold. Therefore, the screened list of affected equipment identifiers includes the DEF-20 type transport truck and the GHI-10 type screening equipment. The above equipment may be greatly affected when the mining machine has abnormal data fluctuations and needs to be optimized and adjusted specifically.

[0152] Step S1465, based on the list of affected equipment identifiers, analyze the process time sequence dependency and rearrange the start order of each equipment in the list of affected equipment identifiers.

[0153] In this embodiment, the process timing dependency describes the sequence of each process in time and the mutual restrictive relationship. In mine production, the mining work of the mining machine is the starting point of the entire process. Only after the ore is mined will it enter the transportation and pretreatment links. Therefore, the startup of the DEF-20 type transport truck and the GHI-10 type screening equipment depends on the normal operation of the mining machine. However, due to abnormal fluctuations in the data of the mining machine, in order to ensure the continuity and stability of production, it is necessary to readjust the startup sequence of the above-mentioned equipment. After detailed analysis, the new startup sequence is determined as follows: After the mining machine starts and runs for a period of time to ensure its relatively stable operating state, first start the GHI-10 type screening equipment, because the screening equipment can preliminarily process the ore mined by the mining machine and reduce the pressure in the subsequent transportation link. After the screening equipment runs stably, then start the DEF-20 type transport truck to transport the screened ore to the designated location. Such an adjustment of the startup sequence can avoid production chaos caused by the subsequent equipment starting blindly when the mining machine has abnormal fluctuations.

[0154] Step S1466, generate a dynamic buffer time window configuration according to the signal attenuation coefficient and the preset buffer time window parameter mapping table.

[0155] In this embodiment, the preset buffer time window parameter mapping table is formulated based on a large number of production experiments and data analyses, which stipulates the buffer time windows corresponding to different signal attenuation coefficients. For the DEF-20 type transport truck, its signal attenuation coefficient is 0.6, and the corresponding buffer time window is found in the mapping table. Assuming that the mapping table stipulates that when the signal attenuation coefficient is between 0.5 and 0.7, the buffer time window is 5 - 10 minutes. Considering the actual operating conditions of the transport truck and the possible degree of influence, it is determined to configure an 8-minute buffer time window for it. For the GHI-10 type screening equipment, the signal attenuation coefficient is 0.7. According to the mapping table, the corresponding buffer time window is 8 - 12 minutes, and finally it is determined to configure a 10-minute buffer time window for it. The setting of the above buffer time windows is to give enough time for the equipment to adapt to the possible influence of abnormal fluctuations after the equipment starts, and to avoid the equipment running in an unstable state.

[0156] Step S1467, match the startup sequence with the dynamic buffer time window configuration to generate a production process collaborative optimization strategy including equipment execution timing and buffer intervals.

[0157] For example, a specific collaborative optimization strategy for production processes can be formulated according to the newly determined startup sequence and buffer time window configuration. First, start the ABC-100 type mining machine. After it has been running for 10 minutes (which is used for the initialization and stable operation of the mining machine itself), start the GHI-10 type screening equipment, and at the same time set a 10-minute buffer time window for the GHI-10 type screening equipment. Within these 10 minutes, the screening equipment can gradually adapt to the possible abnormal fluctuation effects transmitted from the mining machine and perform its own parameter adjustment and operation state optimization. After the GHI-10 type screening equipment has been running for 8 minutes (i.e., within its buffer time window), start the DEF-20 type transport truck and set an 8-minute buffer time window for the transport truck. Within the buffer time window of the transport truck, it can adjust parameters such as transport speed and load according to the abnormal fluctuation signals received by itself to ensure the smoothness of the transport process. In this way, by combining the startup sequence of the equipment with the buffer time window, a complete set of collaborative optimization strategies for production processes is formed, effectively reducing the impact of data abnormal fluctuations on the entire production process and improving the collaboration and stability of production.

[0158] For example, in a possible implementation manner, step S154 includes:

[0159] Step S1541, query the reference speed range of the target equipment in the current production process according to the equipment historical operation database.

[0160] In this embodiment, the equipment historical operation database stores the operation data of the transport truck under different production conditions. For example, after querying, it is known that in the current production process of transporting ore, the reference speed range of the DEF-20 type transport truck is 1200 - 1800 revolutions per minute.

[0161] Step S1542, calculate the actual value of the target speed based on the percentage coefficient corresponding to the speed adjustment value, and generate a speed control instruction including an acceleration time curve.

[0162] For example, assume that according to the optimization strategy, the rotational speed of the transport truck needs to be reduced by 20%. Then, within the benchmark rotational speed range, the intermediate value of 1500 revolutions per minute is taken as the calculation basis. The actual value of the target rotational speed is calculated as 1500×(1 - 20%) = 1500×0.8 = 1200 revolutions per minute. At the same time, in order to ensure that the transport truck smoothly adjusts from the current rotational speed to the target rotational speed, a rotational speed control instruction including an acceleration time curve needs to be generated. Considering the inertia and mechanical properties of the transport truck, the acceleration time is set to 5 minutes. Within these 5 minutes, the rotational speed gradually decreases from the current rotational speed to 1200 revolutions per minute according to a certain curve. For example, within the first 2 minutes, the rotational speed decreases slowly, by 50 revolutions per minute; within the next 2 minutes, the rate of decrease in rotational speed increases, by 100 revolutions per minute; in the last 1 minute, the rotational speed is finely adjusted to 1200 revolutions per minute. Such an acceleration time curve can avoid damage to the equipment caused by too rapid adjustment of the rotational speed and ensure the safety and stability of the transportation process.

[0163] Step S1543: Determine the compensation amount of the cooling system startup threshold according to the equipment heat dissipation performance parameters and the current ambient temperature, and generate a dynamically changing cooling trigger condition.

[0164] For example, the heat dissipation performance parameters of the DEF-20 type transport truck record its heat dissipation capacity under different temperature and load conditions. Assume the current ambient temperature is 30 degrees Celsius. According to the heat dissipation performance parameters, at this temperature, in order to ensure the normal operation of key components such as the engine of the transport truck, the cooling system needs to be started in advance. After analysis and calculation, the compensation amount of the cooling system startup threshold is determined to be 5 degrees Celsius. That is, the original cooling system startup threshold was 40 degrees Celsius, and now it is adjusted to 40 - 5 = 35 degrees Celsius. The generated dynamically changing cooling trigger condition is: when the engine temperature of the transport truck reaches 35 degrees Celsius, the cooling system automatically starts to ensure that the engine operates within an appropriate temperature range and prevent equipment failures caused by excessive temperature.

[0165] Step S1544: Convert the power supply frequency correction parameter into a phase adjustment amount of the inverter modulation waveform through the power supply network impedance characteristic model.

[0166] In this embodiment, the impedance characteristic model of the power supply network describes the relationship between impedance and parameters such as frequency and voltage in the power supply network. For example, assuming that according to the optimization strategy, the power supply frequency correction parameter needs to be adjusted to decrease by 0.5 Hz, that is, from the original 50 Hz to 49.5 Hz. Through calculation and analysis using the impedance characteristic model of the power supply network, considering the influence of factors such as the resistance, inductance, and capacitance of the power supply line on power transmission, as well as the working principle and performance requirements of the inverter, the phase adjustment amount of the inverter modulation waveform is calculated. The specific calculation process involves the comprehensive application of multiple electrical parameters and formulas. For example, according to Ohm's law and the principle of electromagnetic induction, the current and voltage changes in the power supply line are analyzed to determine the influence on the phase of the inverter modulation waveform. After a series of calculations, it is finally determined that the phase of the inverter modulation waveform needs to be adjusted by 15 degrees to ensure that the transport truck can operate normally at the new power supply frequency and guarantee the stable and reliable power supply of the equipment.

[0167] For the GHI-10 type screening equipment, the parsing of the equipment parameter adjustment instructions is also carried out according to the above detailed process. Query the reference speed range from its equipment historical operation database, calculate the actual value of the target speed, and generate a speed control instruction; combine the equipment heat dissipation performance parameters and the current ambient temperature to determine the compensation amount of the cooling system start threshold, and generate a dynamically changing cooling trigger condition; convert the power supply frequency correction parameter into the phase adjustment amount of the inverter modulation waveform through the impedance characteristic model of the power supply network. Through the precise adjustment and optimization of the above equipment parameters, the synergy and stability of the entire production process are further improved, the equipment operation risk is reduced, and the efficient and safe operation of the mine production is guaranteed.

[0168] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of a production process intelligent monitoring system 100 based on a smart mine that can implement the inventive concept provided in some embodiments of the present invention. For example, the processor 120 can be used on the production process intelligent monitoring system 100 based on a smart mine and is used to execute the functions in the present invention.

[0169] The production process intelligent monitoring system 100 based on a smart mine can be a general-purpose server or a special-purpose server, both of which can be used to implement the production process intelligent monitoring method based on a smart mine of the present invention. Although only one server is shown in the present invention, for convenience, the functions described in the present invention can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0170] For example, the intelligent monitoring system 100 for the production process of an intelligent mine may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as disks, ROMs, or RAMs, or any combination thereof. Exemplarily, the intelligent monitoring system 100 for the production process of an intelligent mine may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present invention can be implemented according to these program instructions. The intelligent monitoring system 100 for the production process of an intelligent mine further includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0171] For ease of explanation, only one processor is described in the intelligent monitoring system 100 for the production process of an intelligent mine. However, it should be noted that the intelligent monitoring system 100 for the production process of an intelligent mine in the present invention may also include multiple processors. Therefore, the steps performed by one processor described in the present invention may also be jointly performed or separately performed by multiple processors. For example, if the processor of the intelligent monitoring system 100 for the production process of an intelligent mine performs steps A and B, it should be understood that steps A and B may also be jointly performed by two different processors or separately performed in one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0172] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned intelligent monitoring method for the production process of an intelligent mine is implemented.

[0173] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the previous description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. An intelligent monitoring method for the production process based on an intelligent mine, characterized in that The method includes: Collecting a set of target monitoring data of mining equipment in the real-time production process, where the set of target monitoring data includes equipment vibration time series signals, environmental temperature and humidity distribution data, and equipment energy consumption fluctuation curves; Performing dynamic standardization processing on the set of target monitoring data to generate a set of standardized monitoring data that matches the equipment type and production process stage; Invoking a pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the set of standardized monitoring data to generate equipment operation status features, environmental correlation features, and data abnormal fluctuation features; Based on a preset abnormal pattern recognition model, performing dynamic fusion analysis on the equipment operation status features, environmental correlation features, and data abnormal fluctuation features to generate an equipment operation risk level and a set of production process optimization strategies; Triggering a corresponding alarm instruction according to the equipment operation risk level, and feeding back the set of production process optimization strategies to the production control terminal to adjust the equipment operation parameters.

2. The intelligent monitoring method for the production process based on the intelligent mine according to claim 1, characterized in that, The performing dynamic standardization processing on the set of target monitoring data to generate a set of standardized monitoring data that matches the equipment type and production process stage includes: Obtaining a set of historical operation parameters of the mining equipment, where the set of historical operation parameters includes the rated vibration threshold range of the equipment, the standard temperature and humidity adaptation interval, and the maximum energy consumption critical value; Matching the corresponding rated vibration threshold range of the equipment according to the equipment type, and performing amplitude normalization processing on the equipment vibration time series signals to generate a standardized vibration signal sequence; Determining the current environmental temperature and humidity reference value according to the production process stage, and performing deviation correction processing on the environmental temperature and humidity distribution data based on the standard temperature and humidity adaptation interval to generate corrected temperature and humidity distribution data; Performing dynamic scaling processing on the equipment energy consumption fluctuation curve according to the maximum energy consumption critical value to generate a standardized energy consumption curve; Integrating the standardized vibration signal sequence, the corrected temperature and humidity distribution data, and the standardized energy consumption curve into the set of standardized monitoring data.

3. The intelligent monitoring method for the production process based on the intelligent mine according to claim 2, wherein The invoking a pre-trained multi-dimensional feature extraction model to perform joint feature mapping processing on the set of standardized monitoring data to generate equipment operation status features, environmental correlation features, and data abnormal fluctuation features includes: Performing time-frequency conversion processing on the standardized vibration signal sequence to extract vibration frequency domain energy distribution features and resonance peak offset amounts; Performing spatial interpolation processing on the corrected temperature and humidity distribution data to generate temperature and humidity gradient change features and local abnormal temperature zone markers; Performing slope change analysis on the standardized energy consumption curve to extract energy consumption rising rate features and peak duration features; Inputting the vibration frequency domain energy distribution features, resonance peak offset amounts, temperature and humidity gradient change features, local abnormal temperature zone markers, energy consumption rising rate features, and peak duration features into the multi-dimensional feature extraction model, and performing correlation weighting on the vibration frequency domain energy distribution features and the temperature and humidity gradient change features through a cross-modal attention mechanism to generate environmental correlation features; Jointly encoding the resonance peak offset amount and the energy consumption rising rate features through a temporal convolutional network to generate equipment operation status features; Perform pattern matching on the marked local abnormal temperature region and peak duration characteristics through an anomaly detection algorithm to generate data abnormal fluctuation characteristics.

4. The intelligent monitoring method for the production process based on the intelligent mine according to claim 3, wherein, The time-frequency conversion process of the standardized vibration signal sequence and the extraction of vibration frequency-domain energy distribution characteristics and resonance peak offset include: Perform windowed Fourier transform on the standardized vibration signal sequence to generate a vibration spectrogram; Calculate the energy integral value of each frequency interval in the vibration spectrogram to generate a frequency-domain energy distribution histogram; Detect the main resonance peak frequency in the vibration spectrogram and calculate the difference from the standard resonance frequency of the device to generate a resonance peak offset; Perform sliding window statistics on the frequency-domain energy distribution histogram to extract the low-frequency energy ratio characteristic and high-frequency mutation count characteristic; Perform logarithmic transformation on the low-frequency energy ratio characteristic and high-frequency mutation count characteristic and combine them into the vibration frequency-domain energy distribution characteristic.

5. The intelligent monitoring method for the production process based on the intelligent mine according to claim 3, characterized in that The slope change analysis of the standardized energy consumption curve and the extraction of energy consumption increase rate characteristics and peak duration characteristics include: Perform first-order difference calculation on the standardized energy consumption curve to generate an energy consumption change rate sequence; Detect the time intervals in the energy consumption change rate sequence that continuously exceed the rated slope threshold, and record their start time stamps and end time stamps; Calculate the average slope value within the time interval as the energy consumption increase rate characteristic; Determine the peak duration characteristic based on the difference between the start time stamp and the end time stamp; Perform extreme point detection on the energy consumption change rate sequence to generate energy consumption mutation event marks and corresponding event durations.

6. The intelligent monitoring method for the production process based on the intelligent mine according to claim 3, characterized in that The correlation weighting of the vibration frequency-domain energy distribution characteristic and the temperature and humidity gradient change characteristic through a cross-modal attention mechanism to generate an environment correlation characteristic includes: Map the vibration frequency-domain energy distribution characteristic to a query vector set, and map the temperature and humidity gradient change characteristic to a key vector set; Calculate the cosine similarity between each query vector and key vector to generate an initial attention weight matrix; Perform temperature parameter adjustment and Softmax normalization processing on the initial attention weight matrix to generate a standardized attention weight; Perform dynamic weighted summation on the temperature and humidity gradient change characteristic according to the standardized attention weight to generate an environment-sensitive vibration characteristic; Multiply the environment-sensitive vibration characteristic and the temperature and humidity gradient change characteristic element by element to generate the environment correlation characteristic.

7. The intelligent monitoring method for the production process based on an intelligent mine according to claim 1, wherein The dynamic fusion analysis of the device operation state characteristic, environment correlation characteristic, and data abnormal fluctuation characteristic based on a preset abnormal pattern recognition model to generate a device operation risk level and a set of production process optimization strategies includes: Perform feature splicing on the device operation state characteristic and the environment correlation characteristic to generate a device-environment coupling characteristic; Call the risk level prediction branch in the abnormal pattern recognition model to perform convolution processing on the device-environment coupling characteristic, and output a device failure probability value and an environment interference coefficient; Call the strategy generation branch in the abnormal pattern recognition model to perform pattern analysis on the data abnormal fluctuation characteristic, and extract abnormal trigger conditions and fluctuation propagation paths; Determine the device operation risk level according to the weighted sum of the device failure probability value and the environmental interference coefficient, where the device operation risk level includes low risk, medium risk, and high risk; Match the preset optimization operation template library based on the abnormal trigger condition, and generate device parameter adjustment instructions and environmental regulation plans; Determine the list of affected devices according to the fluctuation propagation path, and generate a collaborative optimization strategy for the production process; Integrate the device parameter adjustment instructions, environmental regulation plans, and collaborative optimization strategies for the production process into the production process optimization strategy set.

8. The intelligent monitoring method for the production process based on the intelligent mine according to claim 7, wherein, Trigger corresponding alarm instructions according to the device operation risk level, and feedback the production process optimization strategy set to the production control terminal to adjust the device operation parameters, including: When the device operation risk level is low risk, generate device status monitoring prompt information and send it to the patrol terminal; When the device operation risk level is medium risk, generate a device maintenance warning instruction and start the spare parts inventory verification process; When the device operation risk level is high risk, generate a device emergency shutdown instruction and a personnel evacuation alarm signal; Parse the device parameter adjustment instructions in the production process optimization strategy set into the rotational speed adjustment value, cooling system start threshold, and power supply frequency correction parameters of the target device; Convert the environmental regulation plan into the operation mode of the ventilation device and the working schedule of the dust removal system; Reallocate the device load rate and task priority of each process segment according to the collaborative optimization strategy of the production process; Send the rotational speed adjustment value, cooling system start threshold, power supply frequency correction parameters, ventilation device operation mode, dust removal system working schedule, device load rate, and task priority to the corresponding execution mechanism through the production control terminal.

9. The intelligent monitoring method for the production process based on the intelligent mine according to claim 7, characterized in that The determining the list of affected devices according to the fluctuation propagation path and generating a collaborative optimization strategy for the production process includes: Traverse the production process topology diagram based on the fluctuation propagation path, and extract the set of adjacent process nodes that have a signal transmission path with the current data abnormal fluctuation characteristics; Query the device identifiers of the devices sharing the power supply system or connected by the conveyor belt in the set of adjacent process nodes according to the device dependency graph, and generate a target device set; Obtain the historical operation parameters of each device in the target device set, and calculate the signal attenuation coefficient of the data abnormal fluctuation characteristics on the fluctuation propagation path; Screen the list of affected device identifiers in the target device set whose degree of influence exceeds a preset threshold according to the signal attenuation coefficient; Analyze the process time sequence dependency based on the list of affected device identifiers, and rearrange the start order of each device in the list of affected device identifiers; Generate a dynamic buffer time window configuration according to the signal attenuation coefficient and the preset buffer time window parameter mapping table; Match the start order with the dynamic buffer time window configuration, and generate a collaborative optimization strategy for the production process including device execution time sequence and buffer interval.

10. An intelligent monitoring system for the production process based on an intelligent mine, characterized in that, The intelligent monitoring system for the production process based on the intelligent mine includes a processor and a memory. The memory is connected to the processor. The memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the intelligent monitoring method for the production process based on the intelligent mine according to any one of claims 1-9 above.

Images