Performance detection method and system for hydrogen compressor of hydrogen refueling station

Abstract

The invention relates to the field of compressor performance detection, and particularly discloses a hydrogen refueling station hydrogen compressor performance detection method and system, and the method employs an artificial intelligence technology based on deep learning to carry out the continuous monitoring and data analysis of the outlet pressure, inlet pressure and inlet temperature of a hydrogen compressor in a working state. The method comprises the following steps: respectively mining time sequence change modes of various working state parameters, then performing time sequence collaborative analysis on inlet and outlet pressure data to reveal the compression degree of hydrogen, and further combining temperature change characteristics at an inlet of a compressor on the basis to realize intelligent evaluation on an exhaust temperature reference value of the compressor, so as to realize the intelligent evaluation of the exhaust temperature reference value of the compressor. And whether the exhaust temperature performance of the compressor is qualified or not can be judged according to the deviation between the actual exhaust temperature and the reference value. Through the mode, real-time monitoring and accurate evaluation of the performance of the hydrogen compressor can be realized, so that timely maintenance or adjustment is facilitated, and stable operation of the hydrogen compressor and safe supply of hydrogen are ensured.

Description

Technical Field

[0001] The present application relates to the field of compressor performance detection, and more specifically, to a method and system for detecting the performance of a hydrogen compressor at a hydrogen refueling station. Background Art

[0002] With the transformation of the global energy structure and the enhancement of environmental awareness, hydrogen energy, as a clean and efficient energy carrier, is gradually becoming an important part of the future energy system. As a key link in the hydrogen energy industry chain, hydrogen refueling stations are responsible for providing hydrogen refueling services for hydrogen fuel cell vehicles and other hydrogen energy application equipment.

[0003] During the operation of a hydrogen refueling station, the hydrogen compressor is one of the core equipment to ensure efficient and safe storage and transportation of hydrogen. The hydrogen compressor compresses hydrogen through components such as the compression chamber, the air inlet valve, the air outlet valve and the compressor drive. When the air inlet valve is opened, hydrogen enters the compression chamber through the air intake pipe. Then, the compressor drive starts to work and drives the compressor to achieve mechanical compression. As the drive works, the volume of the compression chamber gradually decreases and the hydrogen is gradually compressed. When the hydrogen is compressed to a certain pressure, the air inlet valve closes to prevent the hydrogen from flowing back. At the same time, the air outlet valve opens to push the compressed hydrogen into the hydrogen storage container or the delivery pipeline.

[0004] The stability and efficiency of hydrogen compressor performance are directly related to the operating efficiency, energy consumption level and safety of hydrogen refueling stations. Traditionally, the detection of hydrogen compressor performance mainly relies on regular manual inspection and maintenance, by measuring the compressor's inlet and outlet pressures, inlet and outlet temperatures and other key parameters, and evaluating performance according to preset standards. However, this method has many limitations: first, the detection cycle is long, and it is difficult to detect potential performance degradation or failures in time; second, manual detection is easily affected by human factors, making it difficult to ensure the accuracy and consistency of the test results; third, this static detection method is easily affected by the transient behavior and short-term anomalies of the compressor, and cannot fully reflect the performance of the compressor in actual work, resulting in an incomplete and inaccurate evaluation of the compressor performance.

[0005] Therefore, a method and system for detecting the performance of a hydrogen compressor at a hydrogen refueling station 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 a hydrogen refueling station hydrogen compressor performance detection method and system, which uses artificial intelligence technology based on deep learning to continuously monitor and analyze the outlet pressure, inlet pressure and inlet temperature of the hydrogen compressor in the working state, and respectively mine the time series change pattern of each working state parameter, and then, through the time series collaborative analysis of the inlet and outlet pressure data, to reveal the compression degree of hydrogen, and on this basis, further combined with the temperature change characteristics at the compressor inlet, to achieve intelligent evaluation of the compressor exhaust temperature reference value, and then, the exhaust temperature performance of the compressor can be judged according to the deviation between the actual exhaust temperature and the reference value. In this way, real-time monitoring and accurate evaluation of the performance of the hydrogen compressor can be achieved, so as to facilitate timely maintenance or adjustment to ensure the stable operation of the hydrogen compressor and the safe supply of hydrogen.

[0007] According to one aspect of the present application, a method for detecting performance of a hydrogen compressor at a hydrogen refueling station is provided, comprising:

[0008] Collecting the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, wherein the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value;

[0009] Performing a time series analysis based on parameter sample dimensions on the operating value of the working state parameter to obtain a compressor inlet pressure time series associated implicit feature vector, a compressor outlet pressure time series associated implicit feature vector, and a compressor inlet temperature time series associated implicit feature vector;

[0010] Performing enhanced interactive fusion based on common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain a compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector;

[0011] Performing a time series interaction analysis on the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector to obtain a compressor pressure-compressor inlet temperature time series interaction feature vector;

[0012] Based on the compressor pressure-compressor inlet temperature time series interaction characteristic vector, it is determined whether the exhaust temperature performance of the hydrogen compressor is qualified.

[0013] According to another aspect of the present application, a hydrogen refueling station hydrogen compressor performance detection system is provided, which includes:

[0014] A state parameter acquisition module, used to acquire the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, wherein the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value;

[0015] A timing analysis module, used for performing timing analysis on the operating value of the working state parameter based on the parameter sample dimension to obtain a compressor inlet pressure timing-related implicit feature vector, a compressor outlet pressure timing-related implicit feature vector, and a compressor inlet temperature timing-related implicit feature vector;

[0016] A feature enhancement interactive fusion module is used to perform enhanced interactive fusion based on common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain a compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector;

[0017] A timing interaction analysis module, used for performing timing interaction analysis on the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector to obtain a compressor pressure-compressor inlet temperature timing interaction feature vector;

[0018] The exhaust temperature performance evaluation module is used to determine whether the exhaust temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interaction characteristic vector.

[0019] Compared with the prior art, the hydrogen refueling station hydrogen compressor performance detection method and system provided in this application uses artificial intelligence technology based on deep learning to continuously monitor and analyze the outlet pressure, inlet pressure and inlet temperature of the hydrogen compressor in the working state, and respectively mines out the time series change pattern of each working state parameter. Then, by performing time series collaborative analysis on the inlet and outlet pressure data, the compression degree of hydrogen is revealed, and on this basis, the temperature change characteristics at the compressor inlet are further combined to realize the intelligent evaluation of the compressor exhaust temperature reference value, and then, the exhaust temperature performance of the compressor can be judged whether it is qualified according to the deviation between the actual exhaust temperature and the reference value. In this way, real-time monitoring and accurate evaluation of the performance of the hydrogen compressor can be achieved, so as to facilitate timely maintenance or adjustment to ensure the stable operation of the hydrogen compressor and the safe supply of hydrogen. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] 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.

[0021] Figure 1 The present invention is a flowchart of a method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to an embodiment of the present application.

[0022] Figure 2 Schematic diagram of data flow of a method for detecting performance of a hydrogen compressor at a hydrogen refueling station according to an embodiment of the present application.

[0023] Figure 3 This is a flowchart of sub-step S3 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application.

[0024] Figure 4 This is a flowchart of sub-step S4 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application.

[0025] Figure 5 This is a flowchart of sub-step S5 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application.

[0026] Figure 6 4 is a block diagram of a hydrogen compressor performance detection system for a hydrogen refueling station according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] In response to the technical problems described in the above background technology, this application proposes an optimized hydrogen refueling station hydrogen compressor performance detection method, which uses artificial intelligence technology based on deep learning to continuously monitor and analyze the outlet pressure, inlet pressure and inlet temperature of the hydrogen compressor in the working state, and respectively mine the time series change pattern of each working state parameter. Then, through the time series collaborative analysis of the inlet and outlet pressure data, the compression degree of hydrogen is revealed, and on this basis, the temperature change characteristics at the compressor inlet are further combined to realize the intelligent evaluation of the compressor exhaust temperature reference value. Then, the exhaust temperature performance of the compressor can be judged whether it is qualified based on the deviation between the actual exhaust temperature and the reference value. In this way, real-time monitoring and accurate evaluation of the performance of the hydrogen compressor can be achieved, so as to facilitate timely maintenance or adjustment to ensure the stable operation of the hydrogen compressor and the safe supply of hydrogen.

[0032] Figure 1 The present invention is a flowchart of a method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the method for detecting the performance of hydrogen compressors in hydrogen refueling stations according to an embodiment of the present application. Figure 1 and Figure 2As shown, the hydrogen station hydrogen compressor performance detection method includes the following steps: S1, collecting the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, wherein the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value; S2, performing a time series analysis on the operating values ​​of the working state parameters based on the parameter sample dimension to obtain the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet temperature time series associated implicit feature vector; S3, performing an enhanced interactive fusion based on the common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector; S4, performing a time series interactive analysis on the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector and the compressor inlet temperature time series associated implicit feature vector to obtain the compressor pressure-compressor inlet temperature time series interactive feature vector; S5, determining whether the exhaust temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interactive feature vector.

[0033] In the above-mentioned hydrogen station hydrogen compressor performance detection method, the step S1 collects the operating values ​​of the working state parameters of the hydrogen compressor under working conditions at a predetermined sampling frequency, and the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value. It should be understood that based on the laws of physics and engineering design principles, there is a close correlation between the inlet pressure, outlet pressure and inlet temperature of the hydrogen compressor under normal working conditions. Specifically, the difference between the inlet pressure and outlet pressure of the hydrogen compressor represents the pressure increase value of the hydrogen compressed in the compressor, reflecting the degree of compression of the hydrogen. At the same time, based on the physical law ideal gas state equation: PV = nRT, where P represents pressure, V represents volume, n represents the number of gas moles, R represents the ideal gas constant, and T represents temperature. Under ideal conditions, the gas fully exchanges heat with the outside world during the compression process, so that the heat generated by factors such as friction when the gas flows inside the compressor can be transferred out in time, and the gas temperature remains unchanged throughout the compression process. At this time, since the temperature remains unchanged, the density of the gas will increase with the increase of pressure, and no additional energy is needed to overcome the adverse effects of the temperature increase, which helps to reduce energy losses in the compression process and improve the efficiency of the compressor. However, it is difficult for an actual hydrogen compressor to achieve complete isothermal compression. In the actual compression process, there will always be some heat that cannot be completely transferred out, resulting in an increase in temperature. Therefore, based on the physical principles of the hydrogen compression process, this application analyzes the degree of compression of hydrogen by collecting the inlet and outlet pressure data of the hydrogen compressor under working conditions, and estimates the exhaust temperature of the compressor in combination with the degree of compression of hydrogen and the temperature data of the compressor inlet, thereby evaluating the performance of the hydrogen compressor based on the deviation between the actual exhaust temperature and the estimated value.

[0034] In the operation of hydrogen refueling stations, the performance of hydrogen compressors, as one of the core equipment, directly affects the quality and efficiency of hydrogen supply. In order to ensure the efficient and safe operation of hydrogen compressors, real-time monitoring of their working state parameters is essential. Specifically, collecting the operating values ​​of the hydrogen compressor in the working state at a predetermined sampling frequency, especially the compressor inlet pressure value, compressor outlet pressure value, and compressor inlet temperature value, is a key step to achieve this goal.

[0035] First of all, in order to accurately collect the working state parameters of the compressor, corresponding sensors need to be installed at key positions of the compressor. The selection and installation of these sensors need to consider multiple factors to ensure the accuracy and reliability of the data. The accuracy of the sensor directly affects the reliability of the data, while the response speed determines whether the instantaneous changes can be captured in time. For hydrogen compressors, high-precision and fast-response sensors should be selected. For example, the accuracy of the pressure sensor should reach ±0.1% FS, and the accuracy of the temperature sensor should reach ±0.5℃. The response time should also be as short as possible, usually required to be in the millisecond level. The working environment of the hydrogen compressor usually has the characteristics of high temperature, high pressure, flammability and explosion, so the sensor must have good temperature resistance, pressure resistance and explosion-proof performance. When selecting a sensor, you should pay attention to its protection level and explosion-proof certification to ensure that it can work stably in harsh environments. For example, the pressure sensor should choose IP67 protection level, and the temperature sensor should choose explosion-proof certified products. The installation position of the sensor has a great influence on the accuracy of the data. The inlet pressure sensor and temperature sensor should be installed on the straight pipe section of the compressor inlet pipe, away from elbows, valves and other places that may cause turbulence to reduce measurement errors. The outlet pressure sensor should be installed at an appropriate location on the compressor outlet pipe, also to avoid interference from other factors. Specifically, the installation of the sensor:

[0036] Inlet pressure sensor: Select a straight pipe section with a larger diameter and use a dedicated installation tool to fix the sensor on the pipe. Ensure that the connection between the sensor and the pipe is well sealed to prevent hydrogen leakage. After installation, use the calibration equipment to perform a preliminary calibration on the sensor to ensure that its measurement value is accurate.

[0037] Inlet temperature sensor: Also select a straight pipe section with a larger diameter, insert the temperature sensor into the pipe, and ensure that its probe part is completely immersed in hydrogen. Use a special fixture to fix the sensor on the pipe to ensure that it is stable and not loose. After installation, use the calibration equipment to perform preliminary calibration on the sensor to ensure that its measurement value is accurate.

[0038] Outlet pressure sensor: Select a straight pipe section with a larger diameter and use a special installation tool to fix the sensor on the pipe. Make sure the connection between the sensor and the pipe is well sealed to prevent hydrogen leakage. After installation, use the calibration equipment to perform a preliminary calibration on the sensor to ensure that its measurement value is accurate.

[0039] The selection of sampling frequency is an important part in the design of real-time monitoring systems. Too high a sampling frequency will increase the burden of data processing, while too low a sampling frequency may result in missing key information. For hydrogen compressors, it is recommended to adopt a higher sampling frequency, such as once per second or higher, to ensure that any abnormal changes that may affect the safe operation of the equipment can be captured in time. Different models of hydrogen compressors have different operating characteristics, and the sampling frequency should be determined based on the dynamic response time and working cycle of the equipment. For example, for high-speed compressors, the sampling frequency should be appropriately increased to capture rapidly changing parameters. Although high-frequency sampling can provide more detailed data, it will also increase the cost of hardware and software. Therefore, on the premise of meeting the monitoring needs, the economic cost should be comprehensively considered to select the appropriate sampling frequency.

[0040] In addition, the data collected by the sensor needs to be transmitted to the central control system through appropriate communication means. The efficiency and accuracy of data collection also depend on the integration and compatibility of the system. In actual deployment, wired or wireless methods are usually used to achieve remote data transmission. For example, RS485 is a commonly used wired transmission protocol, which is suitable for scenarios with short distances and relatively stable environments. The advantages of this transmission method are stable transmission and strong anti-interference ability, but the wiring is relatively complex and the cost is high. During installation, it is necessary to ensure that the shielding layer of the cable is well grounded to reduce electromagnetic interference. Modbus is a universal communication protocol that supports a variety of physical layer interfaces, such as RS232, RS485, etc. The Modbus protocol is simple and easy to use and is suitable for communication with a variety of industrial equipment. During installation, it is necessary to ensure that the communication addresses between devices do not conflict and configure the correct baud rate and data format. Wi-Fi is a common wireless transmission method that is suitable for applications with difficult wiring or strong mobility. The advantages of wireless transmission are easy installation and high flexibility, but it is greatly affected by environmental interference and has a limited transmission distance. During installation, it is necessary to select a suitable wireless module to ensure its signal coverage and transmission stability. Zigbee is a low-power, short-range wireless communication technology suitable for the establishment of sensor networks. The Zigbee protocol supports self-organizing networks and multi-hop transmission, and is suitable for large-scale sensor network applications. During installation, it is necessary to ensure that the communication paths between nodes are unobstructed to avoid signal blind spots. LoRa is a long-range, low-power wireless communication technology suitable for long-distance data transmission. LoRa technology has strong penetration and anti-interference capabilities, and is suitable for data transmission outdoors and in complex environments. During installation, it is necessary to select a suitable antenna and frequency band to ensure its transmission distance and stability.

[0041] In summary, collecting the operating values ​​of the working state parameters of the hydrogen compressor under working conditions at a predetermined sampling frequency is a complex and delicate task, involving multiple links such as sensor selection and installation, sampling frequency setting, data collection and transmission. Through this series of steps, it can be ensured that the collected data is accurate and reliable, providing a solid foundation for subsequent performance evaluation. In practical applications, it is also necessary to flexibly adjust the data collection strategy according to the specific compressor type and application scenario to achieve the best effect.

[0042] In the above-mentioned hydrogen station hydrogen compressor performance detection method, the step S2 performs a time series analysis based on the parameter sample dimension on the operating value of the working state parameter to obtain the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet temperature time series associated implicit feature vector. In a specific example of the present application, the step S2 further includes: data sorting of the operating value of the working state parameter according to the parameter sample dimension to obtain the time queue of the compressor inlet pressure value, the time queue of the compressor outlet pressure value and the time queue of the compressor inlet temperature value. It should be understood that, considering that different working state parameters usually have different data dimensions and time series change patterns, therefore, in order to improve the accuracy of data analysis, the present application further sorts the operating value of the working state parameter according to the parameter sample dimension to separate the time queue of each working state parameter, and maintain the time series correspondence between the data of each parameter, forming the time queue of the compressor inlet pressure value, the time queue of the compressor outlet pressure value and the time queue of the compressor inlet temperature value. In this way, the independence and accuracy of each parameter in the subsequent data analysis process can be effectively ensured, and information interference between different parameters can be avoided.

[0043] In a specific example of the present application, the step S2 also includes: inputting the time queue of the compressor inlet pressure value, the time queue of the compressor outlet pressure value and the time queue of the compressor inlet temperature value into the timing encoder based on the LSTM-RNN hybrid model to obtain the compressor inlet pressure timing-related implicit feature vector, the compressor outlet pressure timing-related implicit feature vector and the compressor inlet temperature timing-related implicit feature vector. Specifically, in order to more accurately capture the timing change characteristics of various working state parameters of the hydrogen compressor, the present application adopts the LSTM-RNN hybrid model as the timing encoder to extract the timing features of the time queue data of each working state parameter respectively, so as to obtain the compressor inlet pressure timing-related implicit feature vector, the compressor outlet pressure timing-related implicit feature vector and the compressor inlet temperature timing-related implicit feature vector. In the LSTM-RNN hybrid model, RNN (recurrent neural network) is good at processing sequence data and capturing short-term dependencies, while the LSTM unit can effectively capture long-distance timing dependencies through its unique gating mechanism and has long-term memory capabilities. Therefore, by combining the two, the long-term and short-term time series variation characteristics of various working state parameters of the hydrogen compressor can be captured simultaneously, thereby gaining a more comprehensive understanding of the working mode and performance changes of the compressor.

[0044] In the above-mentioned hydrogen station hydrogen compressor performance detection method, the step S3 performs enhanced interactive fusion based on common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector. It should be understood that since the inlet and outlet pressures of the hydrogen compressor comprehensively reflect the degree of compression of hydrogen, in order to more accurately characterize the working state of the compressor, the present application further performs collaborative analysis on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector. In particular, considering that traditional interactive fusion methods such as feature splicing cannot effectively utilize the complementary information between data, resulting in poor feature fusion effect, based on this, the present application proposes an enhanced interactive fusion method based on common feature guidance, by mining the common features between the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector, to identify and compensate for the complementary information between the two, thereby achieving a deeper level of feature fusion.

[0045] Figure 3 Flow chart of sub-step S3 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application. Figure 3As shown, the step S3 includes the following steps: S31, inputting the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector into a common feature extraction network to obtain a compressor inlet and outlet pressure time series common feature representation vector; S32, inputting the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet and outlet pressure time series common feature representation vector into a probabilistic unit based on a Sigmoid function to obtain a probabilistic compressor inlet pressure time series associated implicit feature vector, a probabilistic compressor outlet pressure time series associated implicit feature vector and a probabilistic compressor inlet and outlet pressure time series common feature representation vector; S33 , respectively extract the difference features of the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor outlet pressure time series associated implicit feature vector relative to the probabilistic compressor inlet and outlet pressure time series common feature representation vector to obtain the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector; S34, based on the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector, perform complementary reinforcement interaction fusion on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector.

[0046] Specifically, in step S31, the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are input into a common feature extraction network to obtain a compressor inlet and outlet pressure time series common feature representation vector, which is expressed as:

[0047] X J =tanh(W VT (X1⊕X2)+b VT )

[0048] Wherein, X1 represents the implicit characteristic vector of the compressor inlet pressure time series association, X2 represents the implicit characteristic vector of the compressor outlet pressure time series association, X J Represents the common characteristics of the compressor inlet and outlet pressure time series representation vector, W VT and b VT They represent the weight parameter matrix and bias term respectively, tanh represents the hyperbolic tangent function, and ⊕ represents point-by-point addition.

[0049] That is, a common feature extraction network is constructed using neural network units to learn the common timing patterns between the compressor inlet pressure timing-related implicit feature vector and the compressor outlet pressure timing-related implicit feature vector, and generate a compressor inlet and outlet pressure timing-related common feature representation vector.

[0050] Specifically, in step S32, the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet and outlet pressure time series common feature representation vector are input into a probabilistic unit based on a Sigmoid function to obtain a probabilistic compressor inlet pressure time series associated implicit feature vector, a probabilistic compressor outlet pressure time series associated implicit feature vector and a probabilistic compressor inlet and outlet pressure time series common feature representation vector, which are expressed as follows:

[0051] X a =sigmoid(X1)

[0052] X b =sigmoid(X2)

[0053] X c =sigmoid(X J )

[0054] Among them, sigmoid represents the sigmoid function, X a represents the implicit eigenvector of the probabilistic compressor inlet pressure time series association, X b represents the implicit eigenvector of the probabilistic compressor outlet pressure time series association, X c Represents the common feature representation vector of the probabilistic compressor inlet and outlet pressure time series.

[0055] That is, the sigmoid function is used to probabilistically process the implicit feature vector associated with the compressor inlet pressure time series, the implicit feature vector associated with the compressor outlet pressure time series, and the generated common feature representation vector of the compressor inlet and outlet pressure time series, respectively, to ensure that the value range of the feature vector is between 0 and 1, which is convenient for subsequent data processing.

[0056] Specifically, the step S33 further includes: calculating the dot division vector between the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor inlet and outlet pressure time series common feature representation vector, and calculating the base-two logarithmic value of the absolute value of each eigenvalue in the dot division vector to obtain the compressor inlet pressure time series difference feature standardized representation vector; calculating the dot product vector between the compressor inlet pressure time series difference feature standardized representation vector and the probabilistic compressor inlet pressure time series associated implicit feature vector, and calculating the exponential function value with base e and each eigenvalue of the dot product vector as the exponent to obtain the indexed compressor inlet pressure time series difference feature representation vector; inputting the indexed compressor inlet pressure time series difference feature representation vector into the softmax function for normalization processing to obtain the compressor inlet pressure time series difference feature representation vector, which is expressed as follows:

[0057]

[0058] Among them, X ai , X bi and X ci denote the ith eigenvalue of the probabilistic compressor inlet pressure time series associated implicit eigenvalue, the ith eigenvalue of the probabilistic compressor outlet pressure time series associated implicit eigenvalue, and the ith eigenvalue of the probabilistic compressor inlet and outlet pressure time series common feature representation vector, log(·) denotes a logarithmic function with base 2, exp(·) denotes a natural exponential function, softmax denotes a normalized exponential function, and x a ′ and x b ′ respectively represent the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector.

[0059] That is, the feature differences of the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor outlet pressure time series associated implicit feature vector relative to the probabilistic compressor inlet and outlet pressure time series common feature representation vector are calculated respectively, so as to mine the unique feature information of the compressor inlet pressure feature and the outlet pressure feature in addition to the common features, thereby generating the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector, and using the two as weights, the original compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are compensated and strengthened by performing point multiplication by position, so as to ensure that during the fusion process, as much unique information of the two is retained as much as possible.

[0060] Specifically, the step S34 further includes: calculating the compressor inlet pressure timing difference feature representation vector and the compressor inlet pressure timing associated implicit feature vector by position point multiplication to obtain the compressor inlet pressure timing feature enhancement compensation vector; multiplying the compressor outlet pressure timing difference feature representation vector and the compressor outlet pressure timing associated implicit feature vector by position point multiplication to obtain the compressor outlet pressure timing feature enhancement compensation vector; cascading the compressor inlet pressure timing feature enhancement compensation vector, the compressor outlet pressure timing feature enhancement compensation vector and the compressor inlet and outlet pressure timing common feature representation vector to obtain the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector, which is expressed by the formula:

[0061] x 1a =X1⊙x a '

[0062] x 2b =X2⊙x b '

[0063] v1 = concat[X J ;x 1a ;x 2b ]

[0064] Among them, ⊙ represents the dot product, x 1a and x 2b They respectively represent the compressor inlet pressure timing feature enhancement compensation vector and the compressor outlet pressure timing feature enhancement compensation vector, concat[·;·;·] represents the cascade operation, and v1 represents the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector.

[0065] That is, the original compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are compensated and strengthened by means of position point multiplication to ensure that the unique information of the two is retained as much as possible during the fusion process, and the compensated and strengthened compressor inlet pressure time series associated implicit feature vector, compressor outlet pressure time series associated implicit feature vector and compressor inlet and outlet pressure time series common feature representation vector are fused to comprehensively consider the common information between the two and their respective unique information, and generate the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector, thereby effectively improving the quality of feature interaction fusion and achieving accurate description of the working state of the hydrogen compressor.

[0066] In the above-mentioned hydrogen refueling station hydrogen compressor performance detection method, the step S4 performs a timing interaction analysis on the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector to obtain the compressor pressure-compressor inlet temperature timing interaction feature vector. It should be understood that under the same intake temperature and pressure, the higher the degree of hydrogen compression, the more heat will be generated, and thus the exhaust temperature of the compressor will increase accordingly. Therefore, in order to comprehensively consider the compressor inlet temperature and compression degree to estimate the exhaust temperature of the compressor, the present application further performs a timing interaction analysis on the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector. Among them, Figure 4 FIG. 4 is a flow chart of sub-step S4 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application. Figure 4As shown, the step S4 includes the steps of: S41, using a co-space mapper based on a fully connected layer to co-space map and modulate the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector to obtain a modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and a modulated compressor inlet temperature timing associated implicit feature vector, wherein the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector have the same characteristic scale; S42, based on the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector The feature mapping deviation of the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector is measured relative to the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector, and the modulated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector are compensated to obtain the compensated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector; S43, input the compensated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector into the multidimensional significant interaction network between features to obtain the compressor pressure-compressor inlet temperature timing series interaction feature vector.

[0067] Specifically, the step S41 is expressed by the formula:

[0068] v t1 =Relu(W cm v1+b cm )

[0069] v t2 =Relu(W cm v2+b cm )

[0070] Wherein, v1 represents the compressor inlet pressure-outlet pressure time series cooperative compensation interaction representation vector, v2 represents the compressor inlet temperature time series associated implicit feature vector, W cm and b cm Represent the weight matrix and bias term respectively, Relu represents the activation function, v t1 and v t2 They respectively represent the interactive representation vector of the coordinated compensation of the compressor inlet pressure-outlet pressure time series after modulation and the implicit eigenvector associated with the compressor inlet temperature time series after modulation.

[0071] That is, since the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector come from different data sources and are in different feature spaces, cross-space semantic interaction cannot be achieved. Therefore, this application first uses a fully connected layer to construct a common space mapper to perform feature modulation on the two, and map the two to the same feature space to ensure that the two have the same feature scale to facilitate subsequent interactive processing.

[0072] It should be understood that the present application takes into account that part of the original feature information may be lost during the feature space mapping process, resulting in the destruction of the intrinsic connection or covariance relationship between the compressor pressure feature and the inlet temperature feature, thereby increasing the difficulty of subsequent time series interaction information mining. Therefore, in the technical solution of the present application, after the co-space mapping is performed, the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector after the co-space mapping is modulated are further compensated for feature mapping deviation.

[0073] In a specific example of the present application, the step S42 further includes: interpolating the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector so that they have the same characteristic scale as the modulated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector to obtain the interpolated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the interpolated compressor inlet temperature timing associated implicit feature vector, which is expressed by the formula:

[0074] v c1 =L(v1)

[0075] v c2 =L(v2)

[0076] Among them, L(·) represents the interpolation process, v c1 and v c2 They respectively represent the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector after interpolation processing and the compressor inlet temperature time series associated implicit feature vector after interpolation processing.

[0077] That is, the original compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector are interpolated so that the interpolated features have the same feature scale as the features after co-space modulation, and while ensuring scale alignment, more important information of the original features is retained, thereby improving the model's perception of detail information.

[0078] In a specific example of the present application, the step S42 further includes: calculating the compensation factor between the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector after the interpolation processing and the compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector after the modulation to obtain the compressor pressure feature map deviation compensation factor; calculating the compensation factor between the compressor inlet temperature timing associated implicit feature vector after the interpolation processing and the compressor inlet temperature timing associated implicit feature vector after the modulation to obtain the compressor inlet temperature feature map deviation compensation factor, which is expressed by the formula:

[0079]

[0080] in, and They respectively represent the compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector after modulation and the compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector after interpolation processing, log(·) represents a logarithmic function with base 2, exp(·) represents a natural exponential function, l represents the characteristic scale of the compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector after modulation, s1 represents the compressor pressure characteristic mapping deviation compensation factor, and They respectively represent the eigenvalue of the i-th position in the compressor inlet temperature time series associated implicit feature vector after modulation and the compressor inlet temperature time series associated implicit feature vector after interpolation processing, and s2 represents the compressor inlet temperature characteristic mapping deviation compensation factor.

[0081] That is, by comparing and analyzing the pressure characteristics and inlet temperature characteristics after interpolation processing with the pressure characteristics and inlet temperature characteristics after co-space modulation, the deviation generated in the co-space mapping process can be identified, and the corresponding compensation factor can be calculated accordingly to quantify the degree of feature deviation introduced in the transformation process.

[0082] In a specific example of the present application, the step S42 also includes: based on the compressor pressure feature map deviation compensation factor and the compressor inlet temperature feature map deviation compensation factor, the modulated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector are compensated to obtain the compensated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector. More specifically, the square of the compressor pressure feature map deviation compensation factor and its sum are calculated as correction factors, and each eigenvalue of the modulated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector is divided by the correction factor to obtain the compensated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector, which is expressed by the formula:

[0083]

[0084] Among them, v b1 and v b2 They respectively represent the interactive representation vector of the synergistic compensation of the compressor inlet pressure-outlet pressure time series after compensation and the implicit eigenvector associated with the compressor inlet temperature time series after compensation.

[0085] That is, based on the calculated compensation factor, compensation correction is performed on the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector after the above-mentioned co-space modulation, so as to eliminate the information distortion caused by the feature space transformation and improve the model's sensitivity to detail information.

[0086] Specifically, in step S43, the compensated compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compensated compressor inlet temperature time series associated implicit feature vector are input into the multi-dimensional significant interaction network between features to obtain the compressor pressure-compressor inlet temperature time series interaction feature vector, which is expressed as follows:

[0087]

[0088] Among them, ⊙ represents the dot product, Indicates point addition, Indicates point reduction, represents point division, conv1D represents a one-dimensional convolution operation, MaxPool represents a maximum pooling operation, and v in Represents the compressor pressure-compressor inlet temperature time series interaction feature vector.

[0089] That is, a multi-dimensional interactive analysis is performed on the compensated compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector. Through multi-level feature interactive fusion, the correlation patterns of pressure features and inlet temperature features at different levels are learned, and the correlation response laws between the two are fully excavated, thereby generating a compressor pressure-compressor inlet temperature time series interaction feature vector.

[0090] In the above hydrogen station hydrogen compressor performance detection method, the step S5 determines whether the exhaust temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interaction feature vector. Figure 5 FIG. 5 is a flow chart of sub-step S5 of the method for detecting the performance of the hydrogen compressor of a hydrogen refueling station according to an embodiment of the present application. Figure 5 As shown, the step S5 includes the steps of: S51, inputting the compressor pressure-compressor inlet temperature time series interaction feature vector into the compressor exhaust temperature reference value estimation module based on the decoder to obtain the compressor exhaust temperature reference decoding value; S52, collecting the actual exhaust temperature value of the hydrogen compressor; S53, in response to the deviation value between the actual exhaust temperature value and the compressor exhaust temperature reference decoding value being greater than a preset threshold, determining that the exhaust temperature performance of the hydrogen compressor is unqualified.

[0091] Specifically, in step S51, the compressor pressure-compressor inlet temperature time series interaction feature vector is input into a compressor exhaust temperature reference value estimation module based on a decoder to obtain a compressor exhaust temperature reference decoded value. Specifically, the decoder adopts a multi-layer perceptron structure, and processes the compressor pressure-compressor inlet temperature time series interaction feature vector layer by layer to gradually learn the physical state change law of hydrogen during the compression process, thereby generating a compressor exhaust temperature reference value under specific temperature conditions and compression requirements, thereby indicating the expected level of the compressor exhaust temperature, and providing an important reference for the operation monitoring and maintenance of the compressor.

[0092] Specifically, the step S52 and the step S53 collect the actual value of the exhaust temperature of the hydrogen compressor, and in response to the deviation between the actual value of the exhaust temperature and the reference decoded value of the compressor exhaust temperature being greater than a preset threshold, determine that the exhaust temperature performance of the hydrogen compressor is unqualified. Specifically, the actual value of the exhaust temperature of the hydrogen compressor is collected and compared with the reference decoded value of the compressor exhaust temperature to evaluate the deviation between the actual operating state of the compressor and the expected state, and determine that the exhaust temperature performance of the hydrogen compressor is unqualified. If the deviation between the actual value of the exhaust temperature and the reference decoded value of the compressor exhaust temperature exceeds the preset threshold, it indicates that the exhaust temperature performance of the compressor does not meet the standard requirements. At this time, the compressor needs to be further inspected and maintained to ensure its normal operation. For example, check whether the cooling system of the compressor is working properly, check whether there is wear or blockage inside the compression chamber, and check the operating status of the compressor drive device. Accordingly, if the deviation value is within an acceptable range, it means that the exhaust temperature performance of the compressor meets the standard and can continue to operate.

[0093] In summary, a method for detecting the performance of a hydrogen compressor at a hydrogen refueling station based on an embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to continuously monitor and analyze the outlet pressure, inlet pressure, and inlet temperature of the hydrogen compressor in a working state, and respectively mines out the time series change patterns of various working state parameters, and then, through time series collaborative analysis of the inlet and outlet pressure data, to reveal the degree of compression of hydrogen, and on this basis, further combined with the temperature change characteristics at the compressor inlet, to achieve intelligent evaluation of the compressor exhaust temperature reference value, and then, the exhaust temperature performance of the compressor can be judged based on the deviation between the actual exhaust temperature and the reference value. In this way, real-time monitoring and accurate evaluation of the performance of the hydrogen compressor can be achieved, so as to facilitate timely maintenance or adjustment, and ensure the stable operation of the hydrogen compressor and the safe supply of hydrogen.

[0094] Furthermore, a hydrogen refueling station hydrogen compressor performance detection system is also provided.

[0095] Figure 6 FIG. 1 is a block diagram of a hydrogen compressor performance detection system for a hydrogen refueling station according to an embodiment of the present application. Figure 6As shown, according to the hydrogen station hydrogen compressor performance detection system 100 of the embodiment of the present application, it includes: a state parameter acquisition module 110, which is used to collect the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, and the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value; a timing analysis module 120, which is used to perform a timing analysis on the operating values ​​of the working state parameters based on the parameter sample dimension to obtain the compressor inlet pressure timing associated implicit feature vector, the compressor outlet pressure timing associated implicit feature vector and the compressor inlet temperature timing associated implicit feature vector; a feature enhancement interaction fusion module 130, which is used to analyze the compressor The inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are subjected to enhanced interactive fusion guided by common features to obtain the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector; the time series interaction analysis module 140 is used to perform time series interactive analysis on the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector and the compressor inlet temperature time series associated implicit feature vector to obtain the compressor pressure-compressor inlet temperature time series interaction feature vector; the exhaust temperature performance evaluation module 150 is used to determine whether the exhaust temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interaction feature vector.

[0096] Here, those skilled in the art will appreciate that the specific operations of each module in the above-mentioned hydrogen station hydrogen compressor performance detection system have been referenced above. Figures 1 to 5 The performance detection method of the hydrogen compressor at the hydrogen refueling station has been introduced in detail, and therefore, its repeated description will be omitted.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

[0098] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for detecting the performance of a hydrogen compressor at a hydrogen refueling station, characterized in that: include: Collecting the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, wherein the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value; Performing a time series analysis based on parameter sample dimensions on the operating value of the working state parameter to obtain a compressor inlet pressure time series associated implicit feature vector, a compressor outlet pressure time series associated implicit feature vector, and a compressor inlet temperature time series associated implicit feature vector; Performing enhanced interactive fusion based on common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain a compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector; Performing a time series interaction analysis on the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector to obtain a compressor pressure-compressor inlet temperature time series interaction feature vector; Based on the compressor pressure-compressor inlet temperature time series interaction characteristic vector, it is determined whether the exhaust temperature performance of the hydrogen compressor is qualified.

2. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 1, characterized in that: The operating value of the working state parameter is subjected to a time series analysis based on the parameter sample dimension to obtain a compressor inlet pressure time series associated implicit feature vector, a compressor outlet pressure time series associated implicit feature vector, and a compressor inlet temperature time series associated implicit feature vector, including: Arrange the operating values ​​of the working state parameters according to the parameter sample dimension to obtain a time queue of compressor inlet pressure values, a time queue of compressor outlet pressure values, and a time queue of compressor inlet temperature values; The time queue of the compressor inlet pressure value, the time queue of the compressor outlet pressure value and the time queue of the compressor inlet temperature value are respectively input into the timing encoder based on the LSTM-RNN hybrid model to obtain the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet temperature time series associated implicit feature vector.

3. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 2, characterized in that: The compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are subjected to enhanced interactive fusion guided by common features to obtain a compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector, including: Inputting the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector into a common feature extraction network to obtain a compressor inlet and outlet pressure time series common feature representation vector; Inputting the compressor inlet pressure time series associated implicit feature vector, the compressor outlet pressure time series associated implicit feature vector and the compressor inlet and outlet pressure time series common feature representation vector into a probabilistic unit based on a Sigmoid function to obtain a probabilistic compressor inlet pressure time series associated implicit feature vector, a probabilistic compressor outlet pressure time series associated implicit feature vector and a probabilistic compressor inlet and outlet pressure time series common feature representation vector; Respectively extracting the difference features of the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor outlet pressure time series associated implicit feature vector relative to the probabilistic compressor inlet and outlet pressure time series common feature representation vector to obtain a compressor inlet pressure time series difference feature representation vector and a compressor outlet pressure time series difference feature representation vector; Based on the compressor inlet pressure timing difference feature representation vector and the compressor outlet pressure timing difference feature representation vector, the compressor inlet pressure timing associated implicit feature vector and the compressor outlet pressure timing associated implicit feature vector are complementary enhanced interactively fused to obtain the compressor inlet pressure-outlet pressure timing collaborative compensation interactive representation vector.

4. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 3, characterized in that: Respectively extracting the difference features of the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor outlet pressure time series associated implicit feature vector relative to the probabilistic compressor inlet and outlet pressure time series common feature representation vector to obtain the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector, including: Calculate the dotted division vector between the probabilistic compressor inlet pressure time series associated implicit feature vector and the probabilistic compressor inlet and outlet pressure time series common feature representation vector, and calculate the base-two logarithm of the absolute value of each eigenvalue in the dotted division vector to obtain the compressor inlet pressure time series difference feature standardized representation vector; Calculate the dot product vector between the normalized representation vector of the compressor inlet pressure time series difference feature and the probabilistic compressor inlet pressure time series associated implicit feature vector, and calculate the exponential function value with e as the base and each eigenvalue of the dot product vector as the exponent to obtain the indexed compressor inlet pressure time series difference feature representation vector; The indexed compressor inlet pressure time series difference feature representation vector is input into a softmax function for normalization processing to obtain the compressor inlet pressure time series difference feature representation vector.

5. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 4, characterized in that: Based on the compressor inlet pressure time series difference feature representation vector and the compressor outlet pressure time series difference feature representation vector, the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector are complementary reinforced interactively fused to obtain the compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector, including: Calculate the point-by-point multiplication between the compressor inlet pressure time series difference feature representation vector and the compressor inlet pressure time series associated implicit feature vector to obtain the compressor inlet pressure time series feature enhancement compensation vector; Multiplying the compressor outlet pressure time series difference feature representation vector and the compressor outlet pressure time series associated implicit feature vector by the position point to obtain a compressor outlet pressure time series feature enhancement compensation vector; The compressor inlet pressure timing feature enhancement compensation vector, the compressor outlet pressure timing feature enhancement compensation vector and the compressor inlet and outlet pressure timing common feature representation vector are cascaded to obtain the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector.

6. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 5, characterized in that: Performing a time series interaction analysis on the compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compressor inlet temperature time series associated implicit feature vector to obtain a compressor pressure-compressor inlet temperature time series interaction feature vector, including: Using a fully connected layer-based co-space mapper, the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector are co-space mapped and modulated to obtain a modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and a modulated compressor inlet temperature timing associated implicit feature vector, wherein the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector have the same characteristic scale; Based on the feature mapping deviation of the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector relative to the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector, the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector are compensated to obtain the compensated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector; The compensated compressor inlet pressure-outlet pressure time series collaborative compensation interaction representation vector and the compensated compressor inlet temperature time series associated implicit feature vector are input into the multi-dimensional significant interaction network between features to obtain the compressor pressure-compressor inlet temperature time series interaction feature vector.

7. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 6, characterized in that: The modulated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector are compensated to obtain the compensated compressor inlet pressure-outlet pressure timing synergistic compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector, including: Interpolating the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector so that they have the same characteristic scale as the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector after modulation and the compressor inlet temperature timing associated implicit feature vector after modulation to obtain the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector after interpolation and the compressor inlet temperature timing associated implicit feature vector after interpolation; Calculating a compensation factor between the compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector after the interpolation process and the compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector after the modulation process to obtain a compressor pressure characteristic map deviation compensation factor; Calculating a compensation factor between the interpolated compressor inlet temperature time series associated implicit feature vector and the modulated compressor inlet temperature time series associated implicit feature vector to obtain a compressor inlet temperature characteristic map deviation compensation factor; Based on the compressor pressure characteristic mapping deviation compensation factor and the compressor inlet temperature characteristic mapping deviation compensation factor, the modulated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the modulated compressor inlet temperature timing associated implicit feature vector are compensated to obtain the compensated compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compensated compressor inlet temperature timing associated implicit feature vector.

8. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 7, characterized in that: The method of compensating the modulated compressor inlet pressure-outlet pressure timing cooperative compensation interaction representation vector based on the compressor pressure characteristic map deviation compensation factor to obtain the compensated compressor inlet pressure-outlet pressure timing cooperative compensation interaction representation vector comprises: The square of the compressor pressure characteristic mapping deviation compensation factor and the sum of the factors are calculated as correction factors, and each eigenvalue of the modulated compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector is divided by the correction factor to obtain the compensated compressor inlet pressure-outlet pressure timing coordinated compensation interaction representation vector.

9. The method for detecting the performance of a hydrogen compressor at a hydrogen refueling station according to claim 8, characterized in that: Determining whether the exhaust gas temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interaction characteristic vector includes: Inputting the compressor pressure-compressor inlet temperature time series interaction feature vector into a compressor exhaust temperature reference value estimation module based on a decoder to obtain a compressor exhaust temperature reference decoded value; Collecting the actual value of the exhaust temperature of the hydrogen compressor; In response to a deviation value between the actual exhaust temperature value and the compressor exhaust temperature reference decoded value being greater than a preset threshold, it is determined that the exhaust temperature performance of the hydrogen compressor is unqualified.

10. A hydrogen compressor performance detection system for a hydrogen refueling station, characterized in that: include: A state parameter acquisition module, used to acquire the operating values ​​of the working state parameters of the hydrogen compressor in the working state at a predetermined sampling frequency, wherein the working state parameters include the compressor inlet pressure value, the compressor outlet pressure value and the compressor inlet temperature value; A timing analysis module, used for performing timing analysis on the operating value of the working state parameter based on the parameter sample dimension to obtain a compressor inlet pressure timing-related implicit feature vector, a compressor outlet pressure timing-related implicit feature vector, and a compressor inlet temperature timing-related implicit feature vector; A feature enhancement interactive fusion module is used to perform enhanced interactive fusion based on common feature guidance on the compressor inlet pressure time series associated implicit feature vector and the compressor outlet pressure time series associated implicit feature vector to obtain a compressor inlet pressure-outlet pressure time series collaborative compensation interactive representation vector; A timing interaction analysis module, used for performing timing interaction analysis on the compressor inlet pressure-outlet pressure timing collaborative compensation interaction representation vector and the compressor inlet temperature timing associated implicit feature vector to obtain a compressor pressure-compressor inlet temperature timing interaction feature vector; The exhaust temperature performance evaluation module is used to determine whether the exhaust temperature performance of the hydrogen compressor is qualified based on the compressor pressure-compressor inlet temperature time series interaction characteristic vector.

Images