Intelligent refrigeration equipment control system and method for high temperature environment

Abstract

The present application discloses an intelligent refrigeration equipment control system and method for high-temperature environments, which uses data analysis and processing technology based on deep learning to perform data regularization and time-series association on working state parameters, and extracts local time-series features and significantly aggregates the temperature data of high-temperature natural gas in the entire time domain, so as to adaptively control the compressor speed value of the monitored refrigeration equipment at the next time point according to the principal component significant matching fusion feature between the time-series features of the working state of the refrigeration equipment and the aggregated features of the temperature data of the high-temperature natural gas, and make corresponding adjustments. In this way, the system can collect multiple parameters and provide a more comprehensive view of the system status. At the same time, it can respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to an intelligent refrigeration equipment control system and method for high temperature environments. Background Art

[0002] The MRC (Mixed Refrigerant Cycle) refrigeration process is a technology that uses the principle of molecular sieve adsorption to reduce the gas temperature. It is often used in the natural gas liquefaction process and can effectively reduce the temperature of associated gas, thereby separating light hydrocarbon components and improving the recovery rate of light hydrocarbons. Specifically, the existing associated gas light hydrocarbon recovery device based on the MRC refrigeration process includes: an inlet separation skid, a compressor, a molecular sieve skid, a cold box, a low-temperature separator, a reabsorption tower, a deethanizer, a liquefied gas tower and a central control system; wherein the inlet separation skid is used to deliquidize the associated gas to obtain deliquidized associated gas; the compressor is used to increase the deliquidized associated gas to obtain pressurized associated gas; the molecular sieve skid is used to dehydrate the pressurized associated gas to obtain dehydrated associated gas; the dehydrated associated gas is pre-cooled by the cold box and then separated into gas and liquid by the low-temperature separator and the reabsorption tower to obtain natural gas and hydrocarbon liquid parts; the deethanizer is used to distill the hydrocarbon liquid part, wherein the liquid obtained from the bottom of the deethanizer enters the liquefied gas tower for light hydrocarbon stabilization to obtain light hydrocarbons and liquefied petroleum gas.

[0003] It should be understood that natural gas is usually at a high temperature during processing, and its core lies in recovering high-value liquefied petroleum gas (LPG) and light hydrocarbons in it to improve the comprehensive utilization efficiency and economic value of resources and reduce environmental pollution. The main goal of the light hydrocarbon recovery unit is to increase the recovery rate of light hydrocarbons, reduce oil and gas losses, and ensure that natural gas meets the transportation standards and quality requirements of liquefied petroleum gas. In order to achieve this goal, a mixed refrigerant refrigeration equipment is usually used (that is, the cold box is a mixed refrigerant refrigeration equipment), which can optimize the operating parameters according to the specific conditions of the raw gas, thereby effectively improving the recovery rate of C3 (propane) and C4 (butane). Therefore, as a key equipment, the mixed refrigerant refrigeration equipment plays a vital role in improving the overall recovery rate.

[0004] However, traditional technology usually uses preset fixed parameters when controlling refrigeration equipment. Although this method is simple and easy to use, it has obvious defects. Specifically, due to the lack of real-time monitoring and dynamic adjustment mechanisms, when external conditions change, such as pressure or temperature fluctuations of raw gas, the system cannot respond in time, resulting in reduced refrigeration efficiency and may even affect product quality. The fixed parameter control method ignores the dynamic changes under actual working conditions, causing the equipment operation to often deviate from the optimal state, which not only wastes energy, but also may increase equipment wear. In addition, traditional control usually only focuses on a single or a few parameters, and often ignores the interdependence and overall dynamic behavior of the parameters in the system, which limits the ability to fully understand the system, thereby affecting the optimization of the control strategy.

[0005] Therefore, an optimized refrigeration equipment control system for high temperature environments is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent refrigeration equipment control system and method for high temperature environment, which collects the time series of the working state parameters (inlet temperature, inlet pressure, outlet temperature and outlet pressure) of the monitored refrigeration equipment by a sensor group, and obtains the time series of the temperature data of high temperature natural gas, and adopts data analysis and processing technology based on deep learning to perform data regularization and time series association on the working state parameters, extract local time series features and significantly aggregate the temperature data of the high temperature natural gas in the whole time domain, so as to adaptively control the compressor speed value of the monitored refrigeration equipment at the next time point according to the main component significant matching fusion feature between the working state time series feature of the refrigeration equipment and the aggregated feature of the high temperature natural gas temperature data, and adjust it accordingly. In this way, the system can collect multiple parameters and provide a more comprehensive system status view. At the same time, it can respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment.

[0007] According to one aspect of the present application, there is provided an intelligent refrigeration equipment control system for a high temperature environment, which includes: a state parameter acquisition module, which is used to acquire a time series of working state parameters of a monitored refrigeration equipment collected by a sensor group, wherein the working state parameters include inlet temperature, inlet pressure, outlet temperature and outlet pressure; a temperature data acquisition module, which is used to acquire a time series of temperature data of high-temperature natural gas; a temperature local time series feature extraction module, which is used to pass the time series of the temperature data of the high-temperature natural gas through a sequence encoder based on 1D-CNN to obtain a time series of local time series feature vectors of the temperature of the high-temperature natural gas; a high-temperature natural gas temperature time series aggregation module, which is used to pass the time series of the local time series feature vectors of the temperature of the high-temperature natural gas through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significant aggregation representation vector; a data regularization module, which is used to convert the time series of the working state parameters into a time series of local time series feature vectors of the temperature of the high-temperature natural gas; The columns are regularized according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix; a state parameter time series association feature extraction module is used to obtain the working state time series association feature vector by a working state parameter time series association feature extractor including a void convolutional neural network and a recursive neural network; a temperature time series state-working state significant fusion module is used to obtain the temperature time series state-working state sparse significant matching fusion representation vector by a significant fusion network based on feature principal component fine-grained optimization matching through the semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series association feature vector; a control result generation module is used to obtain a control result based on the temperature time series state-working state sparse significant matching fusion representation vector; a speed adjustment module is used to adjust the compressor speed of the monitored refrigeration equipment through a frequency converter based on the control result.

[0008] According to another aspect of the present application, a method for controlling an intelligent refrigeration device for a high-temperature environment is provided, which comprises: obtaining a time series of working state parameters of a monitored refrigeration device collected by a sensor group, wherein the working state parameters include an inlet temperature, an inlet pressure, an outlet temperature and an outlet pressure; obtaining a time series of temperature data of a high-temperature natural gas; passing the time series of the temperature data of the high-temperature natural gas through a sequence encoder based on 1D-CNN to obtain a time series of a local time series feature vector of the temperature of the high-temperature natural gas; passing the time series of the local time series feature vector of the temperature of the high-temperature natural gas through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significance aggregation representation vector; and dividing the time series of the working state parameters into Data is regularized according to the time dimension and the working state parameter sample dimension to obtain a time series of the working state parameter matrix; the time series of the working state parameter matrix is ​​passed through a working state parameter time series association feature extractor including a void convolutional neural network and a recurrent neural network to obtain a working state time series association feature vector; the high-temperature natural gas temperature time series node semantic significant aggregation representation vector and the working state time series association feature vector are passed through a significant fusion network based on feature principal component fine-grained optimization matching to obtain a temperature time series state-working state sparse significant matching fusion representation vector; based on the temperature time series state-working state sparse significant matching fusion representation vector, a control result is obtained; based on the control result, the compressor speed of the monitored refrigeration equipment is adjusted through a frequency converter.

[0009] Compared with the prior art, the present application provides an intelligent refrigeration equipment control system and method for high temperature environment, which collects the time series of the working state parameters (inlet temperature, inlet pressure, outlet temperature and outlet pressure) of the monitored refrigeration equipment by a sensor group, and obtains the time series of the temperature data of high temperature natural gas, and uses data analysis and processing technology based on deep learning to perform data regularization and time series association on the working state parameters, extract local time series features and significantly aggregate the temperature data of the high temperature natural gas in the whole time domain, so as to adaptively control the compressor speed value of the monitored refrigeration equipment at the next time point according to the main component significant matching fusion feature between the time series feature of the working state of the refrigeration equipment and the aggregated feature of the temperature data of the high temperature natural gas, and make corresponding adjustments. In this way, the system can collect multiple parameters and provide a more comprehensive system status view. At the same time, it can respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 The block diagram is a control system of an intelligent refrigeration equipment for a high temperature environment according to an embodiment of the present application.

[0012] Figure 2 Schematic diagram of data flow of a smart refrigeration equipment control system for a high temperature environment according to an embodiment of the present application.

[0013] Figure 3 The present invention is a flowchart of a method for controlling an intelligent refrigeration device for a high temperature environment according to an embodiment of the present application. DETAILED DESCRIPTION

[0014] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0015] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0016] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0017] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0018] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0019] Traditional technology usually uses preset fixed parameters when controlling refrigeration equipment. Although this method is simple and easy, it has obvious defects. Specifically, due to the lack of real-time monitoring and dynamic adjustment mechanism, when external conditions change, such as pressure or temperature fluctuations of raw gas, the system cannot respond in time, resulting in reduced refrigeration efficiency and may even affect product quality. The fixed parameter control method ignores the dynamic changes under actual working conditions, causing the equipment operation to often deviate from the optimal state, which not only wastes energy but also may increase equipment wear. In addition, traditional control usually only focuses on a single or a few parameters, and often ignores the interdependence and overall dynamic behavior of the parameters in the system, which limits the ability to fully understand the system, thereby affecting the optimization of the control strategy. Therefore, an optimized refrigeration equipment control system for high temperature environments is desired.

[0020] In the technical solution of the present application, an intelligent refrigeration equipment control system for high temperature environment is proposed. Figure 1 The block diagram is a control system of an intelligent refrigeration equipment for a high temperature environment according to an embodiment of the present application. Figure 2 FIG. 1 is a data flow diagram of a smart refrigeration equipment control system for a high temperature environment according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to an embodiment of the present application, an intelligent refrigeration equipment control system 300 for a high temperature environment includes: a state parameter acquisition module 310, which is used to acquire a time series of working state parameters of a monitored refrigeration equipment collected by a sensor group, wherein the working state parameters include inlet temperature, inlet pressure, outlet temperature and outlet pressure; a temperature data acquisition module 320, which is used to acquire a time series of temperature data of high-temperature natural gas; a temperature local time series feature extraction module 330, which is used to pass the time series of the temperature data of the high-temperature natural gas through a sequence encoder based on 1D-CNN to obtain a time series of a local time series feature vector of the temperature of the high-temperature natural gas; a high-temperature natural gas temperature time series aggregation module 340, which is used to pass the time series of the local time series feature vector of the temperature of the high-temperature natural gas through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significant aggregation representation vector; a data regularization module 350, which is used to convert the time series of the working state parameters into a local time series feature vector of the temperature of the high-temperature natural gas; The sequence is data regularized according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix; the state parameter time series association feature extraction module 360 ​​is used to obtain the working state time series association feature vector by a working state parameter time series association feature extractor including a void convolutional neural network and a recursive neural network; the temperature time series state-working state significant fusion module 370 is used to obtain the temperature time series state-working state sparse significant matching fusion representation vector by a significant fusion network based on feature principal component fine-grained optimization matching; the control result generation module 380 is used to obtain the control result based on the temperature time series state-working state sparse significant matching fusion representation vector; the speed adjustment module 390 is used to adjust the compressor speed of the monitored refrigeration equipment through the inverter based on the control result.

[0021] In particular, the state parameter acquisition module 310 and the temperature data acquisition module 320 are used to acquire the time series of the working state parameters of the monitored refrigeration equipment collected by the sensor group, the working state parameters including inlet temperature, inlet pressure, outlet temperature and outlet pressure; and to acquire the time series of the temperature data of the high-temperature natural gas. It should be understood that through the principal component significant matching fusion feature between the time series feature of the working state of the refrigeration equipment and the aggregation feature of the temperature data of the high-temperature natural gas, the compressor speed value of the monitored refrigeration equipment at the next time point can be adaptively controlled and adjusted accordingly. In this way, it is possible to respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of the refrigeration equipment. Therefore, first, the time series of the working state parameters of the monitored refrigeration equipment collected by the sensor group is acquired, the working state parameters including inlet temperature, inlet pressure, outlet temperature and outlet pressure; and the time series of the temperature data of the high-temperature natural gas is acquired.

[0022] In particular, the temperature local time series feature extraction module 330 is used to pass the time series of the temperature data of the high-temperature natural gas through a sequence encoder based on 1D-CNN to obtain a time series of local time series feature vectors of the temperature of the high-temperature natural gas. Considering that the temperature change trend and fluctuation of the high-temperature natural gas in different local times in the time series of the temperature data of the high-temperature natural gas are different, and 1D-CNN is good at capturing local patterns and trends in time series data, based on this, in the technical solution of the present application, the time series of the temperature data of the high-temperature natural gas is passed through a sequence encoder based on 1D-CNN to obtain a time series of local time series feature vectors of the temperature of the high-temperature natural gas, so that specific patterns of temperature changes, such as fluctuations, peaks or abnormal behaviors, can be identified. It is worth mentioning that the one-dimensional convolutional neural network (1D-CNN) is a deep learning model for processing one-dimensional data with time series correlation. It extracts features by applying convolution operations on the one-dimensional space of the input data.

[0023] In particular, the high-temperature natural gas temperature time series aggregation module 340 is used to pass the time series of the local time series feature vector of the high-temperature natural gas temperature through the temperature local time series semantic aggregation network based on node significance attenuation to obtain the high-temperature natural gas temperature time series node semantic significant aggregation representation vector. Considering that different time points in the time series of the local time series feature vector of the high-temperature natural gas temperature have different influences and correlations on the current time point, in order to identify and strengthen the key nodes in the temperature data, that is, those parts that are most important for understanding the entire time series changes, thereby highlighting the overall trend and pattern of natural gas temperature changes, in the technical solution of the present application, the time series of the local time series feature vector of the high-temperature natural gas temperature is passed through the temperature local time series semantic aggregation network based on node significance attenuation to obtain the high-temperature natural gas temperature time series node semantic significant aggregation representation vector. That is, the temperature local time series semantic aggregation network based on node significance attenuation is an innovative extraction and aggregation feature model based on significant feature analysis.

[0024] In an embodiment of the present application, the time series of the local time series feature vector of the high-temperature natural gas temperature is passed through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significance aggregation representation vector, including: first, calculating the feature significance description factor of each local time series feature vector of the high-temperature natural gas temperature in the time series of the local time series feature vector of the high-temperature natural gas temperature, wherein the feature significance description factor is related to the mean and variance of each local time series feature vector of the high-temperature natural gas temperature; in particular, the feature significance factor is calculated based on the variance and variance of each local time series feature vector of the high-temperature natural gas temperature, the mean provides the average state of the natural gas temperature within a period of time, and the variance represents the degree of discreteness between the temperatures, and by calculating the feature significance factor, more critical features can be highlighted. Then, based on the distance span between each local time series feature vector of the high-temperature natural gas temperature in the time queue of the local time series feature vector of the high-temperature natural gas temperature and the current local time series feature vector of the high-temperature natural gas temperature, the feature significance attenuation factor of each local time series feature vector of the high-temperature natural gas temperature is constructed; so that the model dynamically adjusts the weight of each feature according to the distance between the feature vectors, and gives a higher weight to the feature that is closer to the current node. Then, based on the feature significance attenuation factors of the feature significance descriptors of the local time series feature vectors of the temperature of the high-temperature natural gas, the feature significance descriptors of the local time series feature vectors of the temperature of the high-temperature natural gas are modulated to obtain a sequence of the feature significance attenuation descriptors of the local time series feature of the temperature of the high-temperature natural gas; in this way, the features consistent with the current features are reflected and highlighted, and the features with a long distance or large difference are reduced to obtain a sequence of the feature significance attenuation descriptors of the local time series feature of the temperature of the high-temperature natural gas. Furthermore, the sequence of the feature significance attenuation descriptors of the local time series feature of the temperature of the high-temperature natural gas is input into the gated mask module to obtain a sequence of the feature significance attenuation weight factors of the local time series feature of the temperature of the high-temperature natural gas; that is, through normalization and gated masking processing, it is ensured that each feature has equal importance in the model, and some features are dynamically selected or suppressed through the gating mechanism, so that the model can focus on the most important information to obtain a sequence of the feature significance attenuation weight factors of the local time series feature of the temperature of the high-temperature natural gas. Finally, the sequence of significance attenuation weight factors of the local time series characteristics of the high-temperature natural gas temperature is used as the sequence of weights, and the weighted sum of the time series of the local time series feature vectors of the high-temperature natural gas temperature is calculated to obtain the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node.

[0025] The process of calculating the characteristic significance description factor of each local time series characteristic vector of the high-temperature natural gas temperature in the time series of the local time series characteristic vector of the high-temperature natural gas temperature, wherein the characteristic significance description factor is related to the mean and variance of each local time series characteristic vector of the high-temperature natural gas temperature, comprises: respectively calculating the mean and variance of the local time series characteristic vector of the high-temperature natural gas temperature to obtain the local time series characteristic mean and the local time series characteristic variance of the high-temperature natural gas temperature; calculating the position difference between the local time series characteristic vector of the high-temperature natural gas temperature and the local time series characteristic mean of the high-temperature natural gas temperature to obtain the local time series difference vector of the high-temperature natural gas temperature; calculating the fourth power of each characteristic value in the local time series difference vector of the high-temperature natural gas temperature to obtain the local time series modulation difference vector of the high-temperature natural gas temperature; calculating the expected value of the local time series modulation difference vector of the high-temperature natural gas temperature to obtain the local time series expected value of the high-temperature natural gas temperature; dividing the local time series expected value of the high-temperature natural gas temperature by the square of the local time series characteristic variance of the high-temperature natural gas temperature to obtain the characteristic significance description factor corresponding to the local time series characteristic vector of the high-temperature natural gas temperature.

[0026] More specifically, based on the distance span between each local time series feature vector of the high-temperature natural gas temperature in the time queue of the local time series feature vector of the high-temperature natural gas temperature and the current local time series feature vector of the high-temperature natural gas temperature, the process of constructing the characteristic significance attenuation factor of each local time series feature vector of the high-temperature natural gas temperature includes: extracting the maximum value of each local time series feature vector of the high-temperature natural gas temperature to obtain a sequence of the maximum values ​​of the local time series feature of the high-temperature natural gas temperature; extracting the maximum value of the current local time series feature vector of the high-temperature natural gas temperature to obtain the current local feature maximum value of the high-temperature natural gas temperature; calculating the distance between the current local feature maximum value of the high-temperature natural gas temperature and the high-temperature natural gas temperature; The sequences of the maximum values ​​of the local time series characteristics of the temperature are subtracted by position to obtain a sequence of high-temperature natural gas temperature time series offset values; the distance span values ​​between the various high-temperature natural gas temperature local time series characteristic vectors and the current high-temperature natural gas temperature local time series characteristic vector are counted to obtain a sequence of high-temperature natural gas temperature distance span values, and the squares of each value in the sequence of the high-temperature natural gas temperature distance span values ​​are calculated to obtain a sequence of high-temperature natural gas temperature distance span modulation values; the sequence of the high-temperature natural gas temperature time series offset values ​​is divided by the sequence of the high-temperature natural gas temperature distance span modulation values ​​by position to obtain a characteristic significance attenuation factor of each high-temperature natural gas temperature local time series characteristic vector.

[0027] More specifically, the process of inputting the sequence of the high-temperature natural gas temperature local temporal feature significance attenuation description factors into the gated mask module to obtain the sequence of the high-temperature natural gas temperature local temporal feature significance attenuation weight factors includes: performing normalization processing based on the Sigmoid function on each high-temperature natural gas temperature local temporal feature significance attenuation description factor in the sequence of the high-temperature natural gas temperature local temporal feature significance attenuation description factors to obtain a set of normalized high-temperature natural gas temperature local temporal feature significance attenuation description factors; and inputting each normalized high-temperature natural gas temperature local temporal feature significance attenuation description factor in the set of normalized high-temperature natural gas temperature local temporal feature significance attenuation description factors into the gated function for masking processing to obtain the sequence of the high-temperature natural gas temperature local temporal feature significance attenuation weight factors.

[0028] In summary, in the above embodiment, the time series of the local time series feature vector of the high-temperature natural gas temperature is processed through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significant aggregation representation vector, including: processing the time series of the local time series feature vector of the high-temperature natural gas temperature through a temperature local time series semantic aggregation network based on node significance attenuation with the following semantic aggregation formula to obtain the high-temperature natural gas temperature time series node semantic significant aggregation representation vector; wherein the semantic aggregation formula is: ;in, represents the time series of the local time series characteristic vector of the high-temperature natural gas temperature, , and are the time series of the local time series characteristic vector of the high-temperature natural gas temperature. , and The local time series characteristic vector of high-temperature natural gas temperature, To extract the maximum value of a vector, express and The distance span between For the said The characteristic significance attenuation factor of the local time series characteristic vector of the high-temperature natural gas temperature, It is The local time series characteristic vector of the high-temperature natural gas temperature The eigenvalues ​​at the positions, and They are The mean and square of the variance of the local time series characteristic vector of the high-temperature natural gas temperature, It is The significant attenuation description factor of the local time series characteristic of the high-temperature natural gas temperature corresponding to the local time series characteristic vector of the high-temperature natural gas temperature, yes function, It is the first in the set of descriptive factors of the significant attenuation of the local time series characteristics of the normalized high-temperature natural gas temperature. A normalized high-temperature natural gas temperature local time series characteristic significance attenuation description factor, For masking, is the first in the sequence of significant attenuation weight factors of the local time series characteristics of the high-temperature natural gas temperature. The significant attenuation weight factor of the local time series characteristics of high-temperature natural gas temperature, is the predetermined threshold, is the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node.

[0029] In particular, the data regularization module 350 is used to regularize the time series of the working state parameters according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix. Considering that the time series of the working state parameters has a time series feature in the time dimension, and there is mutual influence and correlation between the various working parameters in terms of time, based on this, in the technical solution of the present application, the time series of the working state parameters is regularized according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix.

[0030] In particular, the state parameter time series correlation feature extraction module 360 ​​is used to pass the time series of the working state parameter matrix through a working state parameter time series correlation feature extractor including a hole convolutional neural network and a recursive neural network to obtain a working state time series correlation feature vector. Considering that the time series of the working state parameter matrix contains the correlation between each time point and the correlation between each parameter feature information, in order to more accurately capture and mine the characteristics of the working state parameters in the time dimension and the space dimension, so as to better characterize the characteristics of the working state, in the technical solution of the present application, the time series of the working state parameter matrix is ​​passed through a working state parameter time series correlation feature extractor including a hole convolutional neural network and a recursive neural network to obtain a working state time series correlation feature vector. It is worth mentioning that the hole convolutional neural network is a convolutional neural network (CNN) that uses a hole convolution operation to expand the receptive field while maintaining spatial resolution. Recursive neural network (RNN) is a neural network that processes sequential data, such as text, time series, and audio. RNN has feedback connections that allow information to propagate across time steps in the network.

[0031] In particular, the temperature time series state-working state significant fusion module 370 is used to obtain the temperature time series state-working state sparse significant matching fusion representation vector by combining the semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector through a significant fusion network based on fine-grained optimization matching of feature principal components. Considering that the semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node expresses the most critical and prominent temperature change characteristics of high-temperature natural gas in the entire time domain. The working state time series associated feature vector reveals the dynamic changes of the equipment working state parameters over time and their relationship with each other. In addition, the temperature change of high-temperature natural gas may affect the operating parameters of the refrigeration equipment, such as the load and efficiency of the compressor, and these changes may be reflected in the working state time series associated feature vector. By analyzing the main component association between the two vectors, the performance of the refrigeration equipment can be better understood and predicted, so as to more accurately control the compressor speed value. Therefore, in the technical solution of the present application, the semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are passed through a significant fusion network based on fine-grained optimization matching of feature principal components to obtain a temperature time series state-working state sparse significant matching fusion representation vector. In particular, the significant fusion network based on fine-grained optimization matching of feature principal components is a technology that integrates feature sparsification, principal component analysis (PCA) and significant feature fusion, and is used to construct a key and significant feature correlation representation between feature vectors.

[0032] In an embodiment of the present application, the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are subjected to a significant fusion network based on fine-grained optimization matching of feature principal components to obtain a temperature time series state-working state sparse significant matching fusion representation vector, including: first, the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are standardized to obtain a standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and a standardized working state time series associated feature vector; then, the sample covariance matrix of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and the standardized working state time series associated feature vector is calculated to obtain a standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and a standardized working state time series associated feature vector. The semantically significant temporal aggregation sample covariance matrix of the high-temperature natural gas temperature node and the working state temporal association covariance matrix are obtained; then, the semantically significant temporal aggregation sample covariance matrix of the high-temperature natural gas temperature node and the working state temporal association covariance matrix are subjected to feature vector extraction based on matrix decomposition to obtain a set of principal component feature vectors of the semantically significant temporal aggregation of the high-temperature natural gas temperature node and a set of principal component feature vectors of the working state temporal association; in particular, the feature vector extraction based on matrix decomposition can extract the most important features in the data, namely the principal components, which can capture the main change trends in the data, making the features sparse, namely removing unimportant features, so as to better retain and refine the potential key feature information in the original features. Furthermore, the set of semantically significant temporal aggregation principal component feature vectors of high-temperature natural gas temperature nodes and the set of working state temporal association principal component feature vectors are input into the maximum approximate query matching network to obtain the set of best matching pairs of semantically significant temporal aggregation principal component feature vectors of high-temperature natural gas temperature nodes and working state temporal association principal component feature vectors; further, the best matching pairs of each semantically significant temporal aggregation principal component feature vector of high-temperature natural gas temperature node and working state temporal association principal component feature vector in the set of best matching pairs of semantically significant temporal aggregation principal component feature vectors of high-temperature natural gas temperature nodes and working state temporal association principal component feature vectors are input into the semantic fine-grained gated joint module to obtain the set of high-temperature natural gas temperature-working state component fusion feature vectors; then the set of high-temperature natural gas temperature-working state component fusion feature vectors is cascaded to obtain the temperature temporal state-working state sparse significant matching fusion representation vector.

[0033] Among them, the process of normalizing the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector to obtain the standardized semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the standardized working state time series associated feature vector includes: respectively calculating the mean and standard deviation of the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node to obtain the mean of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node and the standard deviation of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node; comparing the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node with the mean of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node by position; After subtraction, the calculated high-temperature natural gas temperature time series offset vector and the standard deviation of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node are divided by position to obtain the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector; the mean and standard deviation of the working state time series association feature vector are calculated respectively to obtain the working state time series association feature mean and the working state time series association feature standard deviation; after subtracting the working state time series association feature vector from the working state time series association feature mean by position, the calculated working state time series offset vector and the standard deviation of the working state time series association feature are divided by position to obtain the standardized working state time series association feature vector.

[0034] More specifically, the process of calculating the sample covariance matrix of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and the standardized working state time series association feature vector to obtain the high-temperature natural gas temperature node semantically significant time series aggregation sample covariance matrix and the working state time series association covariance matrix includes: multiplying the transposed vector of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector by the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector, and then dividing the obtained standardized high-temperature natural gas temperature time series association matrix by the value obtained by subtracting one from the length of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector by position to obtain the high-temperature natural gas temperature node semantically significant time series aggregation sample covariance matrix; multiplying the transposed vector of the standardized working state time series association feature vector by the standardized working state time series association feature vector, and then dividing the obtained standardized working state time series association matrix by the value obtained by subtracting one from the length of the standardized working state time series association feature vector by position to obtain the working state time series association covariance matrix.

[0035] More specifically, the process of inputting the set of semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes and the set of working state temporal association principal component feature vectors into the maximum approximate query matching network to obtain a set of best matching pairs of semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes and working state temporal association principal component feature vectors includes: extracting predetermined semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes from the set of semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes; calculating the cosine similarity between the predetermined semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes and each working state temporal association principal component feature vector in the set of working state temporal association principal component feature vectors to obtain a set of matching query similarities; and taking the working state temporal association principal component feature vector corresponding to the largest matching query similarity in the set of matching query similarities and the predetermined semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature nodes as the best matching pair of the predetermined semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature nodes and the working state temporal association principal component feature vector.

[0036] More specifically, the process of inputting the best matching pairs of the semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature node and the principal component feature vector associated with the working state time series in the set of the best matching pairs of the semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature node and the principal component feature vector associated with the working state time series into the semantic fine-grained gated joint module to obtain the set of high-temperature natural gas temperature-working state component fusion feature vectors includes: respectively calculating the position difference, position dot multiplication and position addition between the best matching pairs of the semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature node and the principal component feature vector associated with the working state time series to obtain the high-temperature natural gas temperature-working state component fusion feature vector. state time series principal component difference vector, high-temperature natural gas temperature-working state time series principal component dot product vector and high-temperature natural gas temperature-working state time series principal component sum vector; cascade the high-temperature natural gas temperature-working state time series principal component difference vector, the high-temperature natural gas temperature-working state time series principal component dot product vector and the high-temperature natural gas temperature-working state time series principal component sum vector and perform one-dimensional convolution encoding to obtain a high-temperature natural gas temperature-working state time series principal component multi-dimensional fusion vector; perform local window-based maximum pooling processing on the high-temperature natural gas temperature-working state time series principal component multi-dimensional fusion vector to obtain the high-temperature natural gas temperature-working state component fusion feature vector.

[0037] In summary, in the above embodiment, the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are processed through a significant fusion network based on fine-grained optimization matching of feature principal components to obtain a temperature time series state-working state sparse significant matching fusion representation vector, including: the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are processed through a significant fusion network based on fine-grained optimization matching of feature principal components with the following principal component feature fusion formula to obtain the temperature time series state-working state sparse significant matching fusion representation vector; wherein the principal component feature fusion formula is: ;in, and respectively represent the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector, and are the means of the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the associated feature vector of the working state time series, and are the standard deviations of the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the associated feature vector of the working state time series, and are the semantically significant aggregation representation vector of the standardized high-temperature natural gas temperature time series node and the standardized working state time series associated feature vector, respectively. and They are and The transposed vector of and are the lengths of the semantically significant aggregation representation vector of the standardized high-temperature natural gas temperature time series node and the length of the standardized working status time series associated feature vector, and are respectively the semantically significant temporal aggregation sample covariance matrix of the high-temperature natural gas temperature node and the temporal association covariance matrix of the working state, and They are respectively the principal component orthogonal matrix of high-temperature natural gas temperature time series aggregation and the principal component orthogonal matrix of working state time series correlation. and They are respectively the high-temperature natural gas temperature time series aggregation diagonal matrix and the working state time series correlation diagonal matrix, The diagonal elements of the matrix are The high-temperature natural gas temperature time series aggregate diagonal matrix, are the weight values ​​of the principal component feature vectors of semantically significant temporal aggregation of each high-temperature natural gas temperature node, The diagonal elements of the matrix are The working state timing correlation diagonal matrix, are the weight values ​​of the principal component eigenvectors associated with each working state time series, and They are and The transposed matrix of is each semantically significant temporal aggregated principal component feature vector of a high-temperature natural gas temperature node in the set of semantically significant temporal aggregated principal component feature vectors of the high-temperature natural gas temperature node, is each working state time series associated principal component feature vector in the set of working state time series associated principal component feature vectors, To calculate the and The vector inner product between To calculate the one-norm of a vector, To return the maximum value value, is the maximum approximate matching value, , and They are position difference, position dot multiplication and position addition respectively. For cascade processing, is a one-dimensional convolutional coding operation, is the maximum pooling operation, It is the first in the set of high-temperature natural gas temperature-working state component fusion feature vectors. A high-temperature natural gas temperature-operating state component fusion feature vector, is the number of feature vectors in the set of high-temperature natural gas temperature-working state component fusion feature vectors, is the temperature time series state-operating state sparse significant matching fusion representation vector.

[0038] In particular, the control result generation module 380 and the speed adjustment module 390 are used to obtain a control result based on the temperature time series state-working state sparse significant matching fusion representation vector; and based on the control result, the compressor speed of the monitored refrigeration equipment is adjusted through the frequency converter. In a specific example of the present application, the temperature time series state-working state sparse significant matching fusion representation vector is passed through a classifier-based compressor speed controller to obtain the control result, and the control result is used to indicate whether the compressor speed value of the monitored refrigeration equipment at the next time point should be increased, decreased or unchanged. That is, the temperature time series state-working state sparse significant matching fusion representation vector obtained by sparsely fusion of the high-temperature natural gas temperature time series node semantic significant aggregation representation vector and the working state time series associated feature vector is classified and processed, so as to adaptively control the compressor speed value of the monitored refrigeration equipment at the next time point and make corresponding adjustments. In this way, the system can collect multiple parameters and provide a more comprehensive system status view. At the same time, it can respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment.

[0039] In a preferred example, the temperature time series state-working state sparse significant matching fusion representation vector is passed through a classifier-based compressor speed controller to obtain a control result, including: calculating the absolute value sum of each eigenvalue of the temperature time series state-working state sparse significant matching fusion representation vector to obtain a first temperature time series state-working state sparse significant matching fusion spatial structure value, and calculating the square root of the sum of the squares of each eigenvalue of the temperature time series state-working state sparse significant matching fusion representation vector to obtain a second temperature time series state-working state sparse significant matching fusion spatial structure value; the temperature time series state-working state sparse significant matching fusion is passed through a classifier-based compressor speed controller to obtain a control result, including: calculating the absolute value sum of each eigenvalue of the temperature time series state-working state sparse significant matching fusion representation vector to obtain a first temperature time series state-working state sparse significant matching fusion spatial structure value, and calculating the square root of the sum of the squares of each eigenvalue of the temperature time series state-working state sparse significant matching fusion representation vector to obtain a second temperature time series state-working state sparse significant matching fusion spatial structure value; Each eigenvalue of the representation vector is multiplied by the first temperature time series state-working state sparse significant matching fusion space structure value and the second temperature time series state-working state sparse significant matching fusion space structure value to obtain the first temperature time series state-working state sparse significant matching fusion structure reference value and the second temperature time series state-working state sparse significant matching fusion structure reference value corresponding to each eigenvalue; each eigenvalue of the temperature time series state-working state sparse significant matching fusion representation vector is multiplied by the length of the temperature time series state-working state sparse significant matching fusion representation vector and the square root of the length to obtain the first temperature time series state-working state sparse significant matching fusion structure reference value and the second temperature time series state-working state sparse significant matching fusion structure reference value corresponding to each eigenvalue. a temperature time series state-working state sparse significant matching fusion scale transformation value and a second temperature time series state-working state sparse significant matching fusion scale transformation value; dividing the first temperature time series state-working state sparse significant matching fusion structure reference value by the difference between the first temperature time series state-working state sparse significant matching fusion space structure value and the first temperature time series state-working state sparse significant matching fusion scale transformation value to obtain the first temperature time series state-working state sparse significant matching fusion transformation adjustment value; dividing the second temperature time series state-working state sparse significant matching fusion structure reference value by the second temperature time series state-working state sparse significant matching fusion scale transformation value; The difference between the fusion spatial structure value and the second temperature time series state-working state sparse significant matching fusion scale transformation value is matched to obtain the second temperature time series state-working state sparse significant matching fusion transformation adjustment value; the weighted sum of the first temperature time series state-working state sparse significant matching fusion transformation adjustment value and the second temperature time series state-working state sparse significant matching fusion transformation adjustment value is calculated to obtain each eigenvalue of the optimized temperature time series state-working state sparse significant matching fusion representation vector; the optimized temperature time series state-working state sparse significant matching fusion representation vector is passed through a classifier-based compressor speed controller to obtain a control result.

[0040] Here, the temperature time series state-operating state sparse significant matching fusion representation vector is recorded as The optimization expression is: ;in, is the temperature time series state-operating state sparse significant matching fusion representation vector, represents the set of real numbers, The first one represents the temperature time series state-operating state sparse significant matching fusion representation vector The eigenvalues ​​at the positions, Indicates the length of the temperature timing state-working state sparse significant matching fusion representation vector, Indicates the first temperature time series state-working state sparse significant matching fusion space structure value, Indicates the second temperature time series state-working state sparse significant matching fusion space structure value, It means point multiplication by position. It means adding by position. represents the second temperature time series state-working state sparse significant matching fusion transformation adjustment vector, represents the second temperature time series state-working state sparse significant matching fusion transformation adjustment value, represents the first temperature time series state-working state sparse significant matching fusion transformation adjustment vector, represents the first temperature time series state-working state sparse significant matching fusion transformation adjustment value, represents the weighted hyperparameter, Represents the optimized temperature timing state-operating state sparse significant matching fusion representation vector.

[0041] Here, in the preferred example, considering that the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series association feature vector respectively represent the time series node significance attenuation aggregation feature of the one-dimensional local time series association feature of the temperature data of the high-temperature natural gas and the sample-time series cross-dimension local association feature of the working state parameter, when performing significant fusion based on feature principal component query matching, the principal component feature differences on the time series dimension and the time series-sample cross-dimension will have different significance fusion weights based on query matching, so that the temperature time series state-working state sparse significant matching fusion representation vector will also have a diversified set expression distribution of fusion features. Therefore, it is expected to improve the balance between the regression mapping accuracy and completeness of the temperature time series state-working state sparse significant matching fusion representation vector when performing class regression through a classifier-based compressor speed controller, thereby improving the accuracy of the control result obtained.

[0042] In the preferred example, for the spatial structure information of the feature set of the temperature time series state-working state sparse significant matching fusion representation vector in the high-dimensional space, the scale-based frame transformation of each feature value of the temperature time series state-working state sparse significant matching fusion representation vector is performed by taking the class norm space structured representation of the temperature time series state-working state sparse significant matching fusion representation vector as the reference window, and the spatial structure-based frame attention weight adjustment of each feature value of the temperature time series state-working state sparse significant matching fusion representation vector is implemented to ensure the invariance of the spatial transformation (translation, scaling and rotation) of the temperature time series state-working state sparse significant matching fusion representation vector under the interaction of feature space, so as to achieve the balanced executableness between mapping accuracy and mapping completeness in the class regression process based on the discretized feature distribution of the temperature time series state-working state sparse significant matching fusion representation vector, and improve the accuracy of the control result obtained by the temperature time series state-working state sparse significant matching fusion representation vector through the classifier-based compressor speed controller. In this way, the system can collect multiple parameters and provide a more comprehensive system status view. At the same time, it can respond instantly to changes in external conditions and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment.

[0043] As described above, the intelligent refrigeration equipment control system 300 for high temperature environment according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with an intelligent refrigeration equipment control algorithm for high temperature environment. In a possible implementation, the intelligent refrigeration equipment control system 300 for high temperature environment according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent refrigeration equipment control system 300 for high temperature environment can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent refrigeration equipment control system 300 for high temperature environment can also be one of the many hardware modules of the wireless terminal.

[0044] Alternatively, in another example, the intelligent refrigeration equipment control system 300 for high temperature environment and the wireless terminal may also be separate devices, and the intelligent refrigeration equipment control system 300 for high temperature environment may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0045] Furthermore, a method for controlling an intelligent refrigeration device for a high temperature environment is also provided.

[0046] Figure 3FIG. 1 is a flow chart of a method for controlling an intelligent refrigeration device for a high temperature environment according to an embodiment of the present application. Figure 3 As shown, according to the embodiment of the present application, the intelligent refrigeration equipment control method for a high temperature environment includes the following steps: S1, obtaining a time series of working state parameters of the monitored refrigeration equipment collected by a sensor group, wherein the working state parameters include inlet temperature, inlet pressure, outlet temperature and outlet pressure; S2, obtaining a time series of temperature data of high-temperature natural gas; S3, passing the time series of temperature data of high-temperature natural gas through a sequence encoder based on 1D-CNN to obtain a time series of local time series feature vectors of high-temperature natural gas temperature; S4, passing the time series of local time series feature vectors of high-temperature natural gas temperature through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significant aggregation representation vector; S5, dividing the time series of working state parameters into Data is regularized according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix; S6, the time series of the working state parameter matrix is ​​passed through a working state parameter time series association feature extractor including a void convolutional neural network and a recurrent neural network to obtain a working state time series association feature vector; S7, the high-temperature natural gas temperature time series node semantic significant aggregation representation vector and the working state time series association feature vector are passed through a significant fusion network based on feature principal component fine-grained optimization matching to obtain a temperature time series state-working state sparse significant matching fusion representation vector; S8, based on the temperature time series state-working state sparse significant matching fusion representation vector, a control result is obtained; S9, based on the control result, the compressor speed of the monitored refrigeration equipment is adjusted through a frequency converter.

[0047] In summary, according to the embodiment of the present application, the intelligent refrigeration equipment control method for high temperature environment is explained, which collects the time series of the working state parameters (inlet temperature, inlet pressure, outlet temperature and outlet pressure) of the monitored refrigeration equipment by the sensor group, and obtains the time series of the temperature data of high-temperature natural gas, and adopts the data analysis and processing technology based on deep learning to perform data regularization and time series association on the working state parameters, extract local time series features and significantly aggregate the temperature data of the high-temperature natural gas in the whole time domain, so as to adaptively control the compressor speed value of the monitored refrigeration equipment at the next time point according to the main component significant matching fusion feature between the time series feature of the working state of the refrigeration equipment and the aggregated feature of the temperature data of the high-temperature natural gas, and make corresponding adjustments. In this way, the system can collect multiple parameters and provide a more comprehensive system status view. At the same time, it can respond to changes in external conditions in real time and dynamically adjust the compressor speed, so that the equipment can maintain the best operating state under different working conditions, improve the refrigeration efficiency, and realize the intelligent control of refrigeration equipment.

[0048] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An intelligent refrigeration equipment control system for high temperature environment, characterized in that: include: A state parameter acquisition module, used to acquire a time series of working state parameters of the monitored refrigeration equipment collected by the sensor group, wherein the working state parameters include inlet temperature, inlet pressure, outlet temperature and outlet pressure; A temperature data acquisition module, used to acquire the time series of temperature data of high-temperature natural gas; A temperature local time series feature extraction module, used for obtaining a time series of a local time series feature vector of the temperature of the high-temperature natural gas by passing the time series of the temperature data of the high-temperature natural gas through a sequence encoder based on 1D-CNN; A high-temperature natural gas temperature time series aggregation module is used to obtain a high-temperature natural gas temperature time series node semantic significant aggregation representation vector by passing the time series of the high-temperature natural gas temperature local time series feature vector through a temperature local time series semantic aggregation network based on node significance attenuation; A data regularization module, used for regularizing the time series of the working state parameters according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix; A state parameter time series correlation feature extraction module, used for obtaining a work state time series correlation feature vector by passing the time series of the work state parameter matrix through a work state parameter time series correlation feature extractor including a hole convolutional neural network and a recursive neural network; A temperature time series state-working state significant fusion module is used to obtain a temperature time series state-working state sparse significant matching fusion representation vector by combining the semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector through a significant fusion network based on fine-grained optimization matching of feature principal components; A control result generation module, used to obtain a control result based on the temperature time series state-operating state sparse significant matching fusion representation vector; A speed adjustment module, used for adjusting the speed of the compressor of the monitored refrigeration equipment through a frequency converter based on the control result; Wherein, the high-temperature natural gas temperature time series aggregation module includes: A high-temperature natural gas temperature local time series characteristic significance description factor calculation unit, used to calculate the characteristic significance description factor of each high-temperature natural gas temperature local time series characteristic vector in the time series of the high-temperature natural gas temperature local time series characteristic vector, wherein the characteristic significance description factor is related to the mean and variance of each high-temperature natural gas temperature local time series characteristic vector; A high-temperature natural gas temperature local time series feature significance attenuation factor construction unit is used to construct a feature significance attenuation factor of each high-temperature natural gas temperature local time series feature vector based on the distance span between each high-temperature natural gas temperature local time series feature vector in the time queue of the high-temperature natural gas temperature local time series feature vector and the current high-temperature natural gas temperature local time series feature vector; A high-temperature natural gas temperature local time series characteristic significance description factor modulation unit, used for modulating the characteristic significance description factor of each high-temperature natural gas temperature local time series characteristic vector based on the characteristic significance attenuation factor of each high-temperature natural gas temperature local time series characteristic vector to obtain a sequence of high-temperature natural gas temperature local time series characteristic significance attenuation description factors; A high-temperature natural gas temperature local time series characteristic significance attenuation weight factor calculation unit, used for inputting the sequence of high-temperature natural gas temperature local time series characteristic significance attenuation description factors into a gated mask module to obtain a sequence of high-temperature natural gas temperature local time series characteristic significance attenuation weight factors; A semantically significant aggregation unit for high-temperature natural gas temperature time series nodes, used to calculate the weighted sum of the time series of the high-temperature natural gas temperature local time series feature vectors using the sequence of the high-temperature natural gas temperature local time series feature significance attenuation weight factors as the sequence of weights to obtain the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node; The high-temperature natural gas temperature local time series characteristic significance attenuation weight factor calculation unit is used to: Performing normalization processing based on the Sigmoid function on each significant attenuation description factor of the local time series characteristic of the high-temperature natural gas temperature in the sequence of significant attenuation description factors of the local time series characteristic of the high-temperature natural gas temperature to obtain a set of normalized significant attenuation description factors of the local time series characteristic of the high-temperature natural gas temperature; Input each normalized high-temperature natural gas temperature local time series characteristic significance attenuation description factor in the set of normalized high-temperature natural gas temperature local time series characteristic significance attenuation description factors into a gating function for masking processing to obtain a sequence of high-temperature natural gas temperature local time series characteristic significance attenuation weight factors; The high-temperature natural gas temperature local time series characteristic significance description factor calculation unit is used to: Respectively calculating the mean and variance of the local time series characteristic vector of the high-temperature natural gas temperature to obtain the mean of the local time series characteristic of the high-temperature natural gas temperature and the variance of the local time series characteristic of the high-temperature natural gas temperature; Calculating the position difference between the local time series characteristic vector of the high-temperature natural gas temperature and the mean value of the local time series characteristic of the high-temperature natural gas temperature to obtain the local time series differential vector of the high-temperature natural gas temperature; Calculating the fourth power of each eigenvalue in the local time series difference vector of the high-temperature natural gas temperature to obtain a local time series modulation difference vector of the high-temperature natural gas temperature; Calculating the expected value of the local time series modulation difference vector of the high-temperature natural gas temperature to obtain the local time series expected value of the high-temperature natural gas temperature; Dividing the expected value of the local time series of the high-temperature natural gas temperature by the square of the characteristic variance of the local time series of the high-temperature natural gas temperature to obtain a characteristic significance description factor corresponding to the characteristic vector of the local time series of the high-temperature natural gas temperature; The high-temperature natural gas temperature local time series characteristic significance attenuation factor construction unit is used for: Extracting the maximum values ​​of the local time series characteristic vectors of the high-temperature natural gas temperature to obtain a sequence of maximum values ​​of the local time series characteristic vectors of the high-temperature natural gas temperature; Extracting the maximum value of the local time series characteristic vector of the current high-temperature natural gas temperature to obtain the current local characteristic maximum value of the high-temperature natural gas temperature; Calculate the positional subtraction between the current high-temperature natural gas temperature local characteristic maximum value and the sequence of the high-temperature natural gas temperature local time series characteristic maximum values ​​to obtain a sequence of high-temperature natural gas temperature time series offset values; Counting the distance span values ​​between each of the high-temperature natural gas temperature local time series feature vectors and the current high-temperature natural gas temperature local time series feature vector to obtain a sequence of high-temperature natural gas temperature distance span values, and calculating the squares of each value in the sequence of high-temperature natural gas temperature distance span values ​​to obtain a sequence of high-temperature natural gas temperature distance span modulation values; Dividing the sequence of high-temperature natural gas temperature time series offset values ​​by the sequence of high-temperature natural gas temperature distance span modulation values ​​by position to obtain characteristic significance attenuation factors of the local time series characteristic vectors of the high-temperature natural gas temperature; Wherein, the temperature time series state-working state significant fusion module includes: A high-temperature natural gas temperature working state feature standardization unit is used to standardize the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector to obtain a standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and a standardized working state time series associated feature vector; A high-temperature natural gas temperature working state sample covariance calculation unit is used to calculate the sample covariance matrix of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector and the standardized working state time series associated feature vector to obtain the high-temperature natural gas temperature node semantically significant time series aggregation sample covariance matrix and the working state time series associated covariance matrix; A high-temperature natural gas temperature working state principal component feature extraction unit is used to extract feature vectors based on matrix decomposition from the semantically significant temporal aggregation sample covariance matrix of the high-temperature natural gas temperature node and the working state temporal association covariance matrix to obtain a set of high-temperature natural gas temperature node semantically significant temporal aggregation principal component feature vectors and a set of working state temporal association principal component feature vectors; A high-temperature natural gas temperature working state feature query matching unit, used to input the set of semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes and the set of working state temporal association principal component feature vectors into a maximum approximate query matching network to obtain a set of optimal matching pairs of semantically significant temporal aggregation principal component feature vectors of the high-temperature natural gas temperature nodes and working state temporal association principal component feature vectors; A high-temperature natural gas temperature working state feature semantic association unit is used to input the best matching pairs of each high-temperature natural gas temperature node semantically significant temporal aggregation principal component feature vector and the working state temporal association principal component feature vector in the set of the best matching pairs of the high-temperature natural gas temperature node semantically significant temporal aggregation principal component feature vector and the working state temporal association principal component feature vector into a semantic fine-grained gated joint module to obtain a set of high-temperature natural gas temperature-working state component fusion feature vectors; A temperature time series state-working state sparse significant matching fusion unit, used for cascading the set of high-temperature natural gas temperature-working state component fusion feature vectors to obtain the temperature time series state-working state sparse significant matching fusion representation vector; Wherein, the high-temperature natural gas temperature working state feature query and matching unit is used to: Extracting predetermined high-temperature natural gas temperature node semantically significant temporal aggregation principal component feature vectors from the set of high-temperature natural gas temperature node semantically significant temporal aggregation principal component feature vectors; Calculating the cosine similarity between the predetermined high-temperature natural gas temperature node semantically significant time series aggregated principal component feature vector and each working state time series associated principal component feature vector in the set of working state time series associated principal component feature vectors to obtain a set of matching query similarities; The working state time series correlation principal component feature vector corresponding to the largest matching query similarity in the set of matching query similarities and the predetermined high-temperature natural gas temperature node semantically significant time series aggregation principal component feature vector are used as the best matching pair of the predetermined high-temperature natural gas temperature node semantically significant time series aggregation principal component feature vector and the working state time series correlation principal component feature vector; Wherein, the high-temperature natural gas temperature working state characteristic standardization unit is used for: Calculating the mean and standard deviation of the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node respectively to obtain the mean of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node and the standard deviation of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node; After positionally subtracting the semantically significant aggregation representation vector of the high-temperature natural gas temperature time series node from the mean value of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node, positionally dividing the calculated high-temperature natural gas temperature time series offset vector and the standard deviation of the semantically significant aggregation feature of the high-temperature natural gas temperature time series node to obtain the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector; Respectively calculating the mean and standard deviation of the working state time series correlation feature vector to obtain the working state time series correlation feature mean and the working state time series correlation feature standard deviation; After subtracting the working state time series correlation feature vector from the working state time series correlation feature mean by position, the calculated working state time series offset vector and the working state time series correlation feature standard deviation are divided by position to obtain the standardized working state time series correlation feature vector; Wherein, the high-temperature natural gas temperature working state sample covariance calculation unit is used to: After multiplying the transposed vector of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector by the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector, the obtained standardized high-temperature natural gas temperature time series association matrix is ​​divided by the value obtained by subtracting one from the length of the standardized high-temperature natural gas temperature time series node semantically significant aggregation representation vector to obtain the high-temperature natural gas temperature node semantically significant time series aggregation sample covariance matrix; After multiplying the transposed vector of the standardized working state time series association feature vector by the standardized working state time series association feature vector, dividing the obtained standardized working state time series association matrix by the value obtained by subtracting one from the length of the standardized working state time series association feature vector by position to obtain the working state time series association covariance matrix; Wherein, the high-temperature natural gas temperature working state feature semantic association unit is used for: Respectively calculating the position difference, position dot product and position addition between the best matching pairs of the semantically significant temporal aggregation principal component feature vector of the high-temperature natural gas temperature node and the working state temporal association principal component feature vector to obtain the high-temperature natural gas temperature-working state temporal principal component difference vector, the high-temperature natural gas temperature-working state temporal principal component dot product vector and the high-temperature natural gas temperature-working state temporal principal component sum vector; The high-temperature natural gas temperature-working state time series principal component difference vector, the high-temperature natural gas temperature-working state time series principal component dot product vector and the high-temperature natural gas temperature-working state time series principal component sum vector are cascaded and then one-dimensional convolutional encoded to obtain a high-temperature natural gas temperature-working state time series principal component multi-dimensional fusion vector; The multi-dimensional fusion vector of the high-temperature natural gas temperature-working state time series principal component is subjected to a local window-based maximum pooling process to obtain the high-temperature natural gas temperature-working state component fusion feature vector.

2. The intelligent refrigeration equipment control system for high temperature environment according to claim 1, characterized in that: The control result generation module is used to: pass the temperature time series state-working state sparse significant matching fusion representation vector through a classifier-based compressor speed controller to obtain the control result, and the control result is used to indicate whether the compressor speed value of the monitored refrigeration equipment at the next time point should increase, decrease or remain unchanged.

3. A method for controlling an intelligent refrigeration device for a high temperature environment, using the intelligent refrigeration device control system for a high temperature environment according to claim 1, characterized in that: include: Acquire a time series of working state parameters of the monitored refrigeration equipment collected by the sensor group, wherein the working state parameters include inlet temperature, inlet pressure, outlet temperature and outlet pressure; Get the time series of temperature data of high-temperature natural gas; The time series of the temperature data of the high-temperature natural gas is passed through a sequence encoder based on 1D-CNN to obtain a time series of local time series feature vectors of the temperature of the high-temperature natural gas; The time series of the local time series feature vector of the high-temperature natural gas temperature is passed through a temperature local time series semantic aggregation network based on node significance attenuation to obtain a high-temperature natural gas temperature time series node semantic significance aggregation representation vector; Regularizing the time series of the working state parameters according to the time dimension and the working state parameter sample dimension to obtain the time series of the working state parameter matrix; The time series of the working state parameter matrix is ​​passed through a working state parameter time series correlation feature extractor including a dilated convolutional neural network and a recurrent neural network to obtain a working state time series correlation feature vector; The semantic significant aggregation representation vector of the high-temperature natural gas temperature time series node and the working state time series associated feature vector are fused through a significant fusion network based on fine-grained optimization matching of feature principal components to obtain a temperature time series state-working state sparse significant matching fusion representation vector; Based on the temperature time series state-operating state sparse significant matching fusion representation vector, a control result is obtained; Based on the control result, the speed of the compressor of the monitored refrigeration equipment is adjusted by a frequency converter.

Images