Intelligent water station sampling signal circulation transmission method and system

Abstract

The invention discloses a circulation transmission method and system for sampling signals of an intelligent water station, and relates to the technical field of artificial intelligence, and the method comprises the steps: dividing the sampling signals into slow variables and fast variables according to the change rate of the sampling signals, and dividing the slow variables and the fast variables into signal subsets with different priorities according to the importance indexes of the sampling signals; for the signal subset of each variable category, defining a slow variable depth Q network state space, an action space and a reward function and a fast variable depth Q network state space, an action space and a reward function, and constructing a slow variable depth Q network and a fast variable depth Q network; the edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable depth Q network and the fast variable depth Q network, and executes the global sampling decision and collects sampling signals; and the edge device divides the sampling signals into data packets according to the priority, and performs grading processing on the data packets according to the variable category and the priority sequence.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and system for transmitting sampling signals of a smart water station. Background Art

[0002] Smart water stations are comprehensive solutions that use new-generation information technologies such as the Internet of Things, automation, and big data to achieve intelligent upgrades and renovations of water stations, thereby improving water station operation and management efficiency, optimizing scheduling, reducing costs, and improving services. It deploys various sensors and smart devices at various links in the water station to collect data such as water level, flow, pressure, and water quality, and transmits these sampled data to the cloud management platform in real time for storage, analysis, and visualization, providing data support for the intelligent operation, early warning, and optimized scheduling of the water station.

[0003] The Chinese patent with the publication number CN111365186A discloses a water level signal measurement and transmission system for a hydropower station surge tank, which solves the problem in the prior art that the signal cannot be collected and transmitted to the computer monitoring system and the speed regulator system due to the remote location of the surge tank. The system includes a local signal acquisition box arranged next to the surge tank, a water level signal acquisition module arranged in the surge tank, a computer monitoring system LCU, a host computer, a unit LCU and a speed regulator control cabinet; the water level signal acquisition module is used to collect the water level signal of the surge tank in real time and transmit it to the local signal acquisition box; the local signal acquisition box processes the water level signal and transmits the water level signal remotely to the computer monitoring system LCU; the computer monitoring system LCU uploads the water level signal to the host computer, and at the same time transmits the water level signal to each unit LCU, and each unit LCU transmits the water level signal to the corresponding speed regulator control cabinet to participate in the PID adjustment of the speed regulator.

[0004] There are many types of signals collected by water stations, including slow variables such as water level and pressure that are sampled once a minute, and fast variables such as water hammer and pipe burst that are sampled in milliseconds. However, most systems currently lack a mechanism to distinguish data of different priorities, and all use a simple first-in-first-out mode, which may cause some important emergency data to be buried, affecting fault diagnosis. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems in the related art to a certain extent. To this end, one purpose of the present application is to propose a method and system for transmitting sampling signals of a smart water station, so as to realize hierarchical transmission of sampling signals.

[0006] One aspect of the present application provides a method for transmitting a sampling signal of a smart water station, including:

[0007] Step S100: dividing the sampled signal into different variable categories according to the sampled signal change rate, wherein the variable category includes slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets of different priorities according to the importance index of the sampled signal;

[0008] Step S200: for each variable category signal subset, define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function, build and train the slow variable deep Q network and the fast variable deep Q network;

[0009] Step S300: The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal;

[0010] Step S400: The edge device divides the sampled signal into data packets according to the priority, and performs hierarchical processing on the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets.

[0011] The specific method of dividing the sampled signals into different variable categories according to the sampled signal change rate, wherein the variable categories include slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets of different priorities according to the importance index of the sampled signals is as follows:

[0012] Step S110: Obtain the i-th sampling signal X at the current time t i (t) and the i-th sampling signal X at the previous time t-1 i (t-1), calculate the rate of change of the i-th sampling signal;

[0013] Step S120: Preset the upper limit of the slow variable change rate and the upper limit of the fast variable change rate. i (t), if its changing rate is less than or equal to the upper limit of the changing rate of the slow variable, the sampled signal is classified as a slow variable; if its changing rate is greater than or equal to the upper limit of the changing rate of the fast variable, the sampled signal is classified as a fast variable; if its changing rate is greater than the upper limit of the changing rate of the slow variable and less than the upper limit of the changing rate of the fast variable, the sampled signal is classified as an undetermined variable;

[0014] Step S130: for the sampled signal divided into the undetermined variable, extract its mean, variance and deviation, and calculate the comprehensive index of the sampled signal;

[0015] Step S140: Setting the segmentation threshold Cthreshold , if the comprehensive index is less than or equal to the division threshold, the corresponding undetermined variable is divided into a slow variable; if the comprehensive index is greater than the division threshold, the corresponding undetermined variable is divided into a fast variable;

[0016] Step S150: defining an importance index, and calculating the importance index for each sampled signal;

[0017] Step S160: For each type of variable, all sampling signals under the variable category are sorted from large to small according to the value of the importance index to obtain an ordered signal set, the priority number P of the slow variables and the priority number K of the fast variables are set, and the ordered signal set of each type of variable is divided into signal subsets according to the priority number of the corresponding variable category, and the signal subsets include P slow variable subsets and K fast variable subsets.

[0018] The calculation method of the importance index is:

[0019] Step S151: Obtain the number of abnormalities of the i-th sampling signal in the historical time as the fault frequency N faulti , obtain the severity L of the fault caused by the abnormality of the i-th sampling signal faulti , calculate the fault information score of the i-th sampling signal;

[0020] Step S152: Calculate the statistical feature score of the i-th sampling signal according to the mean, standard deviation and deviation of the i-th sampling signal;

[0021] Step S153: calculating the interaction term between the fault information score and the statistical feature score according to the fault information score and the statistical feature score of the i-th sampling signal;

[0022] Step S154: Based on the fault information score, the statistical feature score, and the interaction term between the fault information score and the statistical feature score of the i-th sampling signal, calculate the importance index of the i-th sampling signal.

[0023] The specific method of defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and constructing and training the slow variable deep Q network and the fast variable deep Q network is:

[0024] Step S210: For all slow variable subsets, define the state space of the slow variable deep Q network, the slow variable state s includes the water quality state WQ1(t), environmental parameter ENV1(t) and time feature TIME(t) at the current time t, then s = [WQ1(t), ENV1(t), TIME(t)];

[0025] Step S220: For all slow variable subsets, define the action space of the slow variable deep Q network, the slow variable action a includes the sampling frequency f and the sampling point distribution p, then a = [f, p];

[0026] Step S230: For all slow variable subsets, define the reward function r of the slow variable deep Q network slow , the reward function is calculated by weighted summation based on the sampling frequency score and the sampling point coverage score;

[0027] The calculation method of the reward function of the slow variable deep Q network is:

[0028] The sampling frequency penalty coefficient is preset, and the exponential of the opposite number of the multiplication result of the sampling frequency and the sampling frequency penalty coefficient is used as the sampling frequency score r freq ;

[0029] Calculate the sampling point coverage score r based on the sampling point distribution cov ;

[0030] The reward function r of the slow variable deep Q network is calculated by weighted summing the sampling frequency score and the sampling point coverage score. slow ;

[0031] Step S240: Use the slow variable subset to train a slow variable deep Q network, the input of the slow variable deep Q network is the slow variable state s, and the output is the Q value of each slow variable action a taken under the slow variable state s. According to the current slow variable state s(t) and the slow variable action a(t), the environment will transfer to the next slow variable state s(t+1), and return the reward function r slow , output the Q value of the current slow variable action. When the reward function value of the slow variable deep Q network reaches convergence, it means that the training of the slow variable deep Q network is completed;

[0032] Step S250: For all fast variable subsets, define the state space of the fast variable deep Q network, the fast variable state S includes the instantaneous event type EVENT(t), water quality state WQ2(t), and environmental parameter ENV2(t) at the current time t, then the fast variable state S = [EVENT(t), WQ2(t), ENV2(t)];

[0033] Step S260: for all fast variable subsets, define the action space of the fast variable deep Q network, the fast variable action A includes the fast variable sampling frequency f2 and the trigger threshold th, then A = [f2, th];

[0034] Step S270: For all fast variable subsets, define the reward function r of the fast variable deep Q network fast , the reward function of the fast variable deep Q network is based on the event detection rate score revt , false negative rate score r miss and false alarm rate score r false Perform weighted sum calculation to obtain;

[0035] The reward function r of the fast variable deep Q network fast The calculation method is:

[0036] Step S271: Use the current fast variable sampling decision to count the number of real events N detected in the historical time det and the total number of events that actually occurred N evt , calculate the ratio of the number of real events to the total number of events that actually occurred to get the event detection rate score r evt ;

[0037] Step S272: Count the number of missed events in the historical time and calculate the number of missed events N miss With the total number of events N evt The ratio of the false negative rate score r miss ;

[0038] Step S273: Count the number of false alarm events in the historical time and calculate the number of false alarm events N false With the total number of events N evt The ratio of the false alarm rate score r false ;

[0039] Step S274: Score the event detection rate r evt , false negative rate score r miss and false alarm rate score r false Perform weighted summation to obtain the reward function r of the fast variable deep Q network fast ;

[0040] Step S280: Use the fast variable subset to train the fast variable deep Q network, the input of the fast variable deep Q network is the fast variable state S, and the output is the Q value of each fast variable action A taken under the fast variable state S. According to the current fast variable state S(t) and the fast variable action S(t), the environment will transfer to the next fast variable state S(t+1), and return the reward function at the same time, and output the Q value of the current fast variable action. When the reward function value of the fast variable deep Q network reaches convergence, it indicates that the training of the fast variable deep Q network is completed.

[0041] The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the specific method for the edge device to execute the global sampling decision and collect the sampled signal is:

[0042] Step S310: The edge device receives the slow variable states {s1(t), s2(t), ..., s P (t)} and the fast variable states of K priority fast variable subsets {S1(t),S2(t),...,S K (t)};

[0043] Step S320: For the slow variable state s of the mth priority m (t), and input it into the slow variable deep Q network to obtain the optimal slow variable action a* m (t), as the optimal slow variable sampling decision for the mth priority; for the fast variable state S of the nth priority n (t), and input it into the fast variable deep Q network to obtain the optimal fast variable action A* n (t), as the optimal fast variable sampling decision of the nth priority; summarizing the slow variable sampling decisions of all priorities of the slow variables and the fast variable sampling decisions of all priorities of the fast variables to obtain the global sampling decision;

[0044] Step S330: The edge device executes a global sampling decision, samples the sensor corresponding to the slow variable of the mth priority level at the sampling frequency f and the sampling point distribution p, collects the slow variable sampling signal and marks the slow variable priority label [m]; samples the sensor corresponding to the fast variable of the nth priority level at the fast variable sampling frequency f2 and the trigger threshold th, collects the fast variable sampling signal and marks the fast variable priority label [n].

[0045] The edge device divides the sampled signal into data packets according to the priority, and the specific method of hierarchically processing the data packets according to the variable category and priority order is:

[0046] Step S410: Divide the collected slow variable sampling signals and fast variable sampling signals into P slow variable data packets and K fast variable data packets according to the priority, the first P1 priorities of the slow variables are used as the high priorities of the slow variables, the last P2 priorities are used as the low priorities of the slow variables, the first K1 priorities of the fast variables are used as the high priorities of the fast variables, and the last K2 priorities are used as the low priorities of the fast variables;

[0047] Step S420: applying different data compression, encryption and transmission strategies to the slow variable data packets and the fast variable data packets;

[0048] Step S430: When the network is congested, the edge device and the network node first transmit a high-priority fast variable data packet, then a high-priority slow variable data packet, then a low-priority fast variable data packet, and finally a low-priority slow variable data packet;

[0049] Step S440: After the cloud receives the data packet, the high-priority fast variable data packet directly enters the memory-level cache, the high-priority slow variable data packet enters the SSD storage, and the low-priority fast variable data packet and slow variable data packet enter the data warehouse.

[0050] One aspect of the present application provides a flow transmission system for a smart water station sampling signal, comprising:

[0051] A sampling signal classification module is used to classify the sampling signals into different variable categories according to the sampling signal change rate, wherein the variable categories include slow variables and fast variables, and classify the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampling signals;

[0052] A deep Q network building module is used to define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and to build and train the slow variable deep Q network and the fast variable deep Q network;

[0053] A sampling decision output module is used for the variable state of the edge device receiving the signal subset, and outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network. The edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal;

[0054] The sampling signal processing module is used for the edge device to divide the sampling signal into data packets according to the priority, and to perform hierarchical processing on the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets.

[0055] One aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a method for transmitting a sampling signal of a smart water station are implemented.

[0056] One aspect of the present application provides a readable storage medium, which stores a computer program, and the computer program is suitable for loading by a processor to execute steps in a method for streaming transmission of sampling signals of a smart water station.

[0057] Compared with the prior art, the method and system for transmitting the sampling signal of a smart water station proposed in this application have the following advantages:

[0058] This application uses a deep reinforcement learning method to adaptively adjust the sampling strategy. The sampled signals are divided into slow variables and fast variables according to the rate of change, and the priorities are divided according to the importance index. The deep Q network of slow variables and fast variables is constructed to select the optimal sampling action. This adaptive hierarchical sampling method can reduce data redundancy and improve sampling efficiency while ensuring the quality of important signals.

[0059] This application defines the state space, action space and reward function for variable characteristics. Slow variable sampling focuses on long-term stability, takes sampling frequency and distribution as actions, minimizes frequency and maximizes coverage. Fast variable sampling focuses on real-time detection of emergencies, takes sampling frequency and trigger threshold as actions, improves detection rate, and reduces missed detection and false alarms.

[0060] The edge device in this application outputs the globally optimal hierarchical sampling decision based on the deep Q network, comprehensively considers multiple priorities of slow variables and fast variables, and intelligently generates a global sampling strategy at the edge, which can overcome the limitations of centralized systems, reduce communication and computing loads, and improve system real-time and robustness.

[0061] This application processes and transmits sampled data in a hierarchical manner according to variable categories and priorities. Edge devices group and package sampled signals according to priority, and adopt differentiated compression, encryption, and transmission strategies for different data packets to ensure the quality and security of key data. When the network is congested, fast variable high-priority data is transmitted first to avoid important events being overwhelmed. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A method flow chart of a method for transferring and transmitting sampling signals of a smart water station provided in this application;

[0063] Figure 2 Flowchart of the importance index calculation method provided for this application;

[0064] Figure 3 A flowchart of the global sampling decision execution process provided for this application;

[0065] Figure 4 A functional module diagram of a flow transmission system for a smart water station sampling signal provided in this application;

[0066] Figure 5 A schematic diagram of the structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0067] In order to better understand the present application, various aspects of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application and do not limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0068] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0069] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0070] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0071] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0072] Example 1

[0073] like Figure 1 As shown, a method for transmitting a sampling signal of a smart water station provided by the present application includes:

[0074] Step S100: dividing the sampled signal into different variable categories according to the sampled signal change rate, wherein the variable category includes slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampled signal.

[0075] The specific method of dividing the sampled signals into different variable categories according to the sampled signal change rate, wherein the variable categories include slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets of different priorities according to the importance index of the sampled signals is as follows:

[0076] Step S110: Obtain the i-th sampling signal X at the current time t i (t) and the i-th sampling signal X at the previous time t-1 i (t-1), calculate the rate of change of the i-th sampling signal;

[0077] The calculation formula for the change rate of the i-th sampling signal is: Wherein, Δt is the sampling time interval between the current time t and the previous time t-1;

[0078] Step S120: Preset the upper limit of the slow variable change rate and the upper limit of the fast variable change rate. i (t), if its changing rate is less than or equal to the upper limit of the changing rate of the slow variable, the sampled signal is classified as a slow variable; if its changing rate is greater than or equal to the upper limit of the changing rate of the fast variable, the sampled signal is classified as a fast variable; if its changing rate is greater than the upper limit of the changing rate of the slow variable and less than the upper limit of the changing rate of the fast variable, the sampled signal is classified as an undetermined variable;

[0079] Step S130: for the sampled signal divided into the undetermined variable, extract its mean, variance and deviation, and calculate the comprehensive index of the sampled signal;

[0080] The calculation formula of the mean is: i =E[X i (t)'], where E[·] represents the mathematical expectation, X i (t)' is the sampled signal divided into undetermined variables;

[0081] The calculation formula of the variance is: in, is the variance of the sampled signal;

[0082] The calculation formula of the deviation is:

[0083] The comprehensive index C i The calculation formula is: C i =α1×δ i +α2×Si , where δ i is the standard deviation of the sampled signal, S i is the deviation of the sampling signal, α1 and α2 are the weight coefficients of variance and deviation respectively;

[0084] The weight coefficients of the variance and the deviation are set by those skilled in the art according to actual needs.

[0085] Step S140: Setting the segmentation threshold C threshold , if the comprehensive index is less than or equal to the division threshold, the corresponding undetermined variable is divided into a slow variable; if the comprehensive index is greater than the division threshold, the corresponding undetermined variable is divided into a fast variable;

[0086] Step S150: defining an importance index, and calculating the importance index for each sampled signal;

[0087] Figure 2 Flowchart of the importance index calculation method provided for this application;

[0088] The calculation method of the importance index is:

[0089] Step S151: Obtain the number of abnormalities of the i-th sampling signal in the historical time as the fault frequency Get the severity of the fault caused by the abnormality of the i-th sampling signal Calculate the fault information score of the i-th sampling signal;

[0090] The calculation formula of the fault information score of the i-th sampling signal is: in, is the fault frequency of the i-th sampling signal, is the severity of the fault caused by the abnormality of the i-th sampling signal, λ N , L are the weight coefficients of fault frequency and severity, respectively. are the maximum values ​​of the fault frequency and severity of all sampled signals respectively;

[0091] The severity of the fault is evaluated by those skilled in the art from the perspectives of safety, economy, and environment in combination with their historical experience, and the weight coefficients of the fault frequency and severity are determined by those skilled in the art based on experience.

[0092] Step S152: Calculate the statistical feature score of the i-th sampling signal according to the mean, standard deviation and deviation of the i-th sampling signal;

[0093] The calculation formula of the statistical feature score is: Among them, μ max , δmax , S max are the mean μ of the i-th sampling signal respectively. i , standard deviation δ i and deviation S i The maximum value of

[0094] Step S153: calculating the interaction term between the fault information score and the statistical feature score according to the fault information score and the statistical feature score of the i-th sampling signal;

[0095] The calculation formula of the interaction term between the fault information score and the statistical feature score is:

[0096]

[0097] Step S154: calculating the importance index of the i-th sampling signal based on the fault information score, the statistical feature score, and the interaction term between the fault information score and the statistical feature score of the i-th sampling signal;

[0098] The calculation formula of the importance index is: i =[α×f(stats i )+β×g(fault i )×h(stats i ,fault i )], where α and β are the balance factors controlling the fault information score and the statistical feature score respectively;

[0099] The sum of the balancing factors of the control fault information score and the statistical feature score is 1, which is determined by those skilled in the art based on experience.

[0100] Step S160: for each type of variable, all sampled signals under the variable category are sorted from large to small according to the value of the importance index to obtain an ordered signal set, the priority number P of the slow variable and the priority number K of the fast variable are set, and the ordered signal set of each type of variable is divided into signal subsets according to the priority number of the corresponding variable category, and the signal subsets include P slow variable subsets and K fast variable subsets;

[0101] The ordered signal set of each type of variable is divided into signal subsets at equal intervals according to the priority number of the corresponding variable category, each signal subset contains the same number of sampling signals, and the boundary of the subset corresponds to the P-quantile or K-quantile of the importance index.

[0102] Step S100 classifies and prioritizes the sampled signals, laying the foundation for subsequent hierarchical sampling and processing. This step avoids information redundancy and resource waste, and provides a guarantee for improving the perception of key events and the real-time performance of the system.

[0103] Step S200: for each variable category signal subset, define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function, build and train the slow variable deep Q network and the fast variable deep Q network;

[0104] The slow variable deep Q network and the fast variable deep Q network are used to learn the optimal sampling strategy under different priorities. The input is the state of the slow variable and the fast variable of each priority, and the output is the Q value of each action taken in this state.

[0105] The specific method of defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and constructing and training the slow variable deep Q network and the fast variable deep Q network is:

[0106] Step S210: For all slow variable subsets, define the state space of the slow variable deep Q network, the slow variable state s includes the water quality state WQ1(t), environmental parameter ENV1(t) and time feature TIME(t) at the current time t, then s = [WQ1(t), ENV1(t), TIME(t)];

[0107] Step S220: For all slow variable subsets, define the action space of the slow variable deep Q network, the slow variable action a includes the sampling frequency f and the sampling point distribution p, then a = [f, p];

[0108] The sampling frequency is [f min , f max ], where f min is the minimum value that the sampling frequency can take, f max is the maximum value that the sampling frequency can take;

[0109] The sampling point distribution represents the weight distribution of the slow variables of the i-th priority at different locations, represented by a k-dimensional vector, satisfying ∑ k p k =1, where k represents the dimension of the sampling point distribution, p k Represents the sampling point distribution of the kth dimension;

[0110] Step S230: For all slow variable subsets, define the reward function r of the slow variable deep Q network slow , the reward function is calculated by weighted summation based on the sampling frequency score and the sampling point coverage score;

[0111] The calculation method of the reward function of the slow variable deep Q network is:

[0112] The sampling frequency penalty coefficient is preset, and the exponential of the opposite number of the multiplication result of the sampling frequency and the sampling frequency penalty coefficient is used as the sampling frequency score r freq ;

[0113] The calculation formula for the sampling frequency score is: freq =exp(-λ f ×f), where λ f is the sampling frequency penalty coefficient, exp() is the exponential function;

[0114] Calculate the sampling point coverage score r based on the sampling point distribution cov ;

[0115] The calculation formula of the sampling point coverage score is: cov =-∑ k p k ×log(p k ), where log() is a logarithmic function;

[0116] The reward function r of the slow variable deep Q network is calculated by weighted summing the sampling frequency score and the sampling point coverage score. slow ;

[0117] The calculation formula of the reward function of the slow variable deep Q network is: slow =w f × freq +w c × cov , where w f 、w c They are the weight coefficients of sampling frequency score and sampling point coverage score respectively;

[0118] The weight coefficients of the sampling frequency score and the sampling point coverage score are determined by those skilled in the art according to actual needs.

[0119] Step S240: Use the slow variable subset to train a slow variable deep Q network, the input of the slow variable deep Q network is the slow variable state s, and the output is the Q value of each slow variable action a taken under the slow variable state s. According to the current slow variable state s(t) and the slow variable action a(t), the environment will transfer to the next slow variable state s(t+1), and return the reward function r slow , output the Q value of the current slow variable action. When the reward function value of the slow variable deep Q network reaches convergence, it means that the training of the slow variable deep Q network is completed;

[0120] Step S250: For all fast variable subsets, define the state space of the fast variable deep Q network, the fast variable state S includes the instantaneous event type EVENT(t), water quality state WQ2(t), and environmental parameter ENV2(t) at the current time t, then the fast variable state S = [EVENT(t), WQ2(t), ENV2(t)];

[0121] The instantaneous event type EVENT(t) includes but is not limited to water hammer and pipe burst;

[0122] Step S260: for all fast variable subsets, define the action space of the fast variable deep Q network, the fast variable action A includes the fast variable sampling frequency f2 and the trigger threshold th, then A = [f2, th];

[0123] The fast variable sampling frequency represents the time interval for sampling the fast variable, and the value can be [f2 min , f2 max ], where f2 min and f2 max are the minimum and maximum values ​​that the fast variable sampling frequency can take, respectively;

[0124] Step S270: For all fast variable subsets, define the reward function r of the fast variable deep Q network fast , the reward function of the fast variable deep Q network is based on the event detection rate score r evt , false negative rate score r miss and false alarm rate score r false Perform weighted sum calculation to obtain;

[0125] The reward function r of the fast variable deep Q network fast The calculation method is:

[0126] Step S271: Use the current fast variable sampling decision to count the number of real events N detected in the historical time det and the total number of events that actually occurred N evt , calculate the ratio of the number of real events to the total number of events that actually occurred to get the event detection rate score r evt ;

[0127] The calculation formula of the event detection rate score is:

[0128] Step S272: Count the number of missed events in the historical time and calculate the number of missed events N miss With the total number of events N evt The ratio of the false negative rate score r miss ;

[0129] The calculation formula of the false negative rate score is:

[0130] Step S273: Count the number of false alarm events in the historical time and calculate the number of false alarm events N false With the total number of events N evt The ratio of the false alarm rate score r false ;

[0131] The calculation formula for the false alarm rate score is:

[0132] Step S274: Score the event detection rate r evt , false negative rate score r miss and false alarm rate score r false Perform weighted summation to obtain the reward function r of the fast variable deep Q network fast ;

[0133] The reward function r of the fast variable deep Q network fast The function expression is: fast =w evt × evt -w miss × miss -w false × false , where w evt 、w miss 、w false They are the weight coefficients of event detection rate score, missed alarm rate score, and false alarm rate score respectively;

[0134] The weight coefficients of the event detection rate score, missed alarm rate score, and false alarm rate score are set by those skilled in the art according to actual needs.

[0135] Step S280: Use the fast variable subset to train the fast variable deep Q network, the input of the fast variable deep Q network is the fast variable state S, and the output is the Q value of each fast variable action A taken under the fast variable state S. According to the current fast variable state S(t) and the fast variable action S(t), the environment will transfer to the next fast variable state S(t+1), and return the reward function at the same time, and output the Q value of the current fast variable action. When the reward function value of the fast variable deep Q network reaches convergence, it indicates that the training of the fast variable deep Q network is completed.

[0136] Step S200 uses deep reinforcement learning to build an intelligent sampling model for slow variables and fast variables. By learning the mapping of variable states to optimal sampling strategies, adaptive optimization of the sampling process is achieved. Compared with a fixed sampling scheme, this step can dynamically adjust the sampling behavior and improve resource utilization efficiency.

[0137] Step S300: The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal;

[0138] Figure 3 A flowchart of the global sampling decision execution process provided for this application;

[0139] The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the specific method for the edge device to execute the global sampling decision and collect the sampled signal is:

[0140] Step S310: The edge device receives the slow variable states {s1(t), s2(t), ..., s P (t)} and the fast variable states of K priority fast variable subsets {S1(t),S2(t),...,S K (t)};

[0141] Step S320: For the slow variable state s of the mth priority m (t), and input it into the slow variable deep Q network to obtain the optimal slow variable action a* m (t), as the optimal slow variable sampling decision for the mth priority; for the fast variable state S of the nth priority n (t), and input it into the fast variable deep Q network to obtain the optimal fast variable action A* n (t), as the optimal fast variable sampling decision of the nth priority; summarizing the slow variable sampling decisions of all priorities of the slow variables and the fast variable sampling decisions of all priorities of the fast variables to obtain the global sampling decision;

[0142] Step S330: The edge device executes a global sampling decision, samples the sensor corresponding to the slow variable of the mth priority level at the sampling frequency f and the sampling point distribution p, collects the slow variable sampling signal and marks the slow variable priority label [m]; samples the sensor corresponding to the fast variable of the nth priority level at the fast variable sampling frequency f2 and the trigger threshold th, collects the fast variable sampling signal and marks the fast variable priority label [n].

[0143] Step S300 deploys the trained deep Q network to the edge device to realize distributed intelligent sampling decision.

[0144] Step S400: The edge device divides the sampled signal into data packets according to the priority, and hierarchically processes the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets;

[0145] The edge device divides the sampled signal into data packets according to the priority, and the specific method of hierarchically processing the data packets according to the variable category and priority order is:

[0146] Step S410: Divide the collected slow variable sampling signals and fast variable sampling signals into P slow variable data packets and K fast variable data packets according to the priority, the first P1 priorities of the slow variables are used as the high priorities of the slow variables, the last P2 priorities are used as the low priorities of the slow variables, the first K1 priorities of the fast variables are used as the high priorities of the fast variables, and the last K2 priorities are used as the low priorities of the fast variables;

[0147] Step S420: applying different data compression, encryption and transmission strategies to the slow variable data packets and the fast variable data packets;

[0148] For slow variable data packets, low compression rate Cm, high encryption strength Em and high service quality transmission Qm are used in descending order of priority. The smaller m is, the higher the priority, the lower the compression rate, the higher the encryption strength and the higher the service quality.

[0149] For fast variable data packets, the lowest compression rate Cn, the highest encryption strength En and the highest service quality Qn are used for transmission in descending order of priority. The smaller n is, the higher the priority is, the lower the compression rate is, the higher the encryption strength and the higher the service quality is.

[0150] Step S430: When the network is congested, the edge device and the network node first transmit a high-priority fast variable data packet, then a high-priority slow variable data packet, then a low-priority fast variable data packet, and finally a low-priority slow variable data packet;

[0151] Step S440: After the cloud receives the data packet, the high-priority fast variable data packet directly enters the memory-level cache, the high-priority slow variable data packet enters the SSD storage, and the low-priority fast variable data packet and slow variable data packet enter the data warehouse.

[0152] Step S400 performs hierarchical coding, transmission and storage on the collected data. Through differentiated processing strategies, priority acquisition and timely reporting of important data are guaranteed, avoiding the loss and delay of key information. The multi-level scheduling mechanism during network congestion ensures the smooth flow of emergency data and improves the fault tolerance and robustness of the system.

[0153] Example 2

[0154] like Figure 4 As shown, a flow transmission system of a smart water station sampling signal provided by the present application includes:

[0155] A sampling signal classification module is used to classify the sampling signals into different variable categories according to the sampling signal change rate, wherein the variable categories include slow variables and fast variables, and classify the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampling signals;

[0156] A deep Q network building module is used to define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and to build and train the slow variable deep Q network and the fast variable deep Q network;

[0157] A sampling decision output module is used for the variable state of the edge device receiving the signal subset, and outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network. The edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal;

[0158] The sampling signal processing module is used for the edge device to divide the sampling signal into data packets according to the priority, and to perform hierarchical processing on the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets.

[0159] Example 3

[0160] Figure 5 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 5 As shown, according to another aspect of the present application, an electronic device is also provided. The electronic device may include one or more processors and one or more memories. The memory stores a computer readable code, and when the computer readable code is executed by one or more processors, it can execute the above-mentioned method for transmitting the sampling signal of a smart water station.

[0161] The method or system according to the embodiment of the present application can also be used by Figure 5 The electronic device architecture shown in FIG. Figure 5As shown, the electronic device may include a bus, one or more CPUs, a read-only memory (ROM), a random access memory (RAM), a communication port connected to a network, an input / output component, a hard disk, etc. A storage device in the electronic device, such as a ROM or a hard disk, may store a method for transferring and transmitting a sampling signal of a smart water station provided in the present application. A method for circulating and transmitting sampling signals of a smart water station may, for example, include: dividing the sampling signals into different variable categories according to the sampling signal change rate, the variable categories including slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets of different priorities according to the importance index of the sampling signals; for each signal subset of the variable category, defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function, constructing and training the slow variable deep Q network and the fast variable deep Q network; the edge device receives the variable state of the signal subset, outputs the global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, the edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal; the edge device divides the sampling signal into data packets according to the priority, and hierarchically processes the data packets according to the variable category and priority order; the data packets include the slow variable data packets and the fast variable data packets. Furthermore, the electronic device may also include a user interface. Of course, Figure 5 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 5 One or more components of an electronic device are shown.

[0162] Example 4

[0163] Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, a method for transferring and transmitting a sampling signal of a smart water station according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache). Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0164] In addition, according to the implementation mode of the present application, the process described in the above reference flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: dividing the sampled signal into different variable categories according to the sampling signal change rate, the variable category includes slow variables and fast variables, dividing the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampled signal; for each signal subset of each variable category, defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function , construct and train a slow variable deep Q network and a fast variable deep Q network; the edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal; the edge device divides the sampling signal into data packets according to the priority, and hierarchically processes the data packets according to the variable category and the priority order; the data packets include the slow variable data packets and the fast variable data packets. When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0165] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0166] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0167] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for transmitting sampling signals of a smart water station, characterized in that: include: Dividing the sampled signals into different variable categories according to the sampling signal change rate, wherein the variable categories include slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampled signals; For each subset of signals in each variable category, define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function, build and train the slow variable deep Q network and the fast variable deep Q network; The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal; The edge device divides the sampled signal into data packets according to the priority, and performs hierarchical processing on the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets.

2. A method for transmitting a sampling signal of a smart water station according to claim 1, characterized in that: The specific method of dividing the sampled signals into different variable categories according to the sampled signal change rate, wherein the variable categories include slow variables and fast variables, and dividing the slow variables and fast variables into signal subsets of different priorities according to the importance index of the sampled signals is as follows: Get the i-th sampling signal X at the current time t i (t) and the i-th sampling signal X at the previous time t-1 i (t-1), calculate the rate of change of the i-th sampling signal; Preset the upper limit of the slow variable change rate and the upper limit of the fast variable change rate. For each sampling signal X i (t), if its changing rate is less than or equal to the upper limit of the changing rate of the slow variable, the sampled signal is classified as a slow variable; if its changing rate is greater than or equal to the upper limit of the changing rate of the fast variable, the sampled signal is classified as a fast variable; If its changing rate is greater than the upper limit of the changing rate of the slow variable and less than the upper limit of the changing rate of the fast variable, the sampled signal is classified as an undetermined variable; For the sampling signal divided into the undetermined variable, the mean, variance and deviation are extracted, and the comprehensive index of the sampling signal is calculated; Set the partition threshold C threshold , if the comprehensive index is less than or equal to the division threshold, the corresponding undetermined variable is divided into a slow variable; if the comprehensive index is greater than the division threshold, the corresponding undetermined variable is divided into a fast variable; Define an importance index and calculate its importance index for each sampled signal; For each type of variable, all sampling signals under the variable category are sorted from large to small according to the value of the importance index to obtain an ordered signal set, the priority number P of the slow variables and the priority number K of the fast variables are set, and the ordered signal set of each type of variable is divided into signal subsets according to the priority number of the corresponding variable category, and the signal subsets include P slow variable subsets and K fast variable subsets.

3. A method for transmitting a sampling signal of a smart water station as claimed in claim 2, characterized in that: The calculation method of the importance index is: Get the number of abnormalities of the i-th sampling signal in the historical time as the fault frequency N faulti , obtain the severity L of the fault caused by the abnormality of the i-th sampling signal faulti , calculate the fault information score of the i-th sampling signal; Calculate the statistical feature score of the i-th sampling signal according to the mean, standard deviation and deviation of the i-th sampling signal; According to the fault information score and the statistical feature score of the i-th sampling signal, an interaction term between the fault information score and the statistical feature score is calculated; Based on the fault information score, the statistical feature score, and the interaction term between the fault information score and the statistical feature score of the ith sampling signal, the importance index of the ith sampling signal is calculated.

4. A method for transmitting a sampling signal of a smart water station as claimed in claim 3, characterized in that: The specific method of defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and constructing and training the slow variable deep Q network and the fast variable deep Q network includes: For all slow variable subsets, the state space of the slow variable deep Q network is defined. The slow variable state s includes the water quality state WQ1(t), environmental parameters ENV1(t) and time characteristics TIME(t) at the current time t, then s = [WQ1(t), ENV1(t), TIME(t)]; For all slow variable subsets, define the action space of the slow variable deep Q network, the slow variable action a includes the sampling frequency f and the sampling point distribution p, then a = [f, p]; For all slow variable subsets, define the reward function r of the slow variable deep Q network slow , the reward function is calculated by weighted summation based on the sampling frequency score and the sampling point coverage score; The calculation method of the reward function of the slow variable deep Q network is: The sampling frequency penalty coefficient is preset, and the exponential of the opposite number of the multiplication result of the sampling frequency and the sampling frequency penalty coefficient is used as the sampling frequency score r freq ; Calculate the sampling point coverage score r based on the sampling point distribution cov ; The reward function r of the slow variable deep Q network is calculated by weighted summing the sampling frequency score and the sampling point coverage score. slow ; The slow variable deep Q network is trained using the slow variable subset. The input of the slow variable deep Q network is the slow variable state s, and the output is the Q value of each slow variable action a taken under the slow variable state s. According to the current slow variable state s(t) and slow variable action a(t), the environment will transfer to the next slow variable state s(t+1), and return the reward function r slow , output the Q value of the current slow variable action. When the reward function value of the slow variable deep Q network reaches convergence, it means that the training of the slow variable deep Q network is completed.

5. A method for transmitting a sampling signal of a smart water station as claimed in claim 4, characterized in that: The method of defining the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and constructing and training the slow variable deep Q network and the fast variable deep Q network also includes: For all fast variable subsets, the state space of the fast variable deep Q network is defined. The fast variable state S includes the instantaneous event type EVENT(t), water quality state WQ2(t), and environmental parameter ENV2(t) at the current time t. Then the fast variable state S = [EVENT(t), WQ2(t), ENV2(t)]; For all fast variable subsets, define the action space of the fast variable deep Q network, the fast variable action A includes the fast variable sampling frequency f2 and the trigger threshold th, then A = [f2, th]; For all fast variable subsets, define the reward function r of the fast variable deep Q network fast , the reward function of the fast variable deep Q network is based on the event detection rate score r evt , false negative rate score r miss and false alarm rate score r false Perform weighted sum calculation to obtain; The reward function r of the fast variable deep Q network fast The calculation method is: Count the number of real events N detected in historical time using the current fast variable sampling decision det and the total number of events that actually occurred N evt , calculate the ratio of the number of real events to the total number of events that actually occurred to get the event detection rate score r evt ; Count the number of missed events in the historical time and calculate the number of missed events N miss With the total number of events N evt The ratio of the false negative rate score r miss ; Count the number of false positives in the historical period and calculate the number of false positives N false With the total number of events N evt The ratio of the false alarm rate score r false ; The event detection rate is scored as r evt , false negative rate score r miss and false alarm rate score r false Perform weighted summation to obtain the reward function r of the fast variable deep Q network fast ; The fast variable deep Q network is trained using the fast variable subset. The input of the fast variable deep Q network is the fast variable state S, and the output is the Q value of each fast variable action A taken under the fast variable state S. According to the current fast variable state S(t) and the fast variable action S(t), the environment will transfer to the next fast variable state S(t+1), and return the reward function at the same time, and output the Q value of the current fast variable action. When the reward function value of the fast variable deep Q network reaches convergence, it indicates that the training of the fast variable deep Q network is completed.

6. A method for transmitting a sampling signal of a smart water station as claimed in claim 5, characterized in that: The edge device receives the variable state of the signal subset, outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network, and the specific method for the edge device to execute the global sampling decision and collect the sampled signal is: The edge device receives the slow variable states {s1(t),s2(t),...,s P (t)} and the fast variable states of K priority fast variable subsets {S1(t),S2(t),...,S K (t)}; For the mth priority slow variable state s m (t), and input it into the slow variable deep Q network to obtain the optimal slow variable action a* m (t), as the optimal slow variable sampling decision of the mth priority; For the nth priority fast variable state S n (t), and input it into the fast variable deep Q network to obtain the optimal fast variable action A* n (t), as the optimal fast variable sampling decision of nth priority; Summarize the slow variable sampling decisions of all priorities of the slow variables and the fast variable sampling decisions of all priorities of the fast variables to obtain the global sampling decision; The edge device executes the global sampling decision, samples the sensor corresponding to the slow variable of the mth priority at the sampling frequency f and the sampling point distribution p, collects the slow variable sampling signal and marks the slow variable priority label [m]; The sensor corresponding to the fast variable of the nth priority is sampled at the fast variable sampling frequency f2 and the trigger threshold th, the fast variable sampling signal is collected and marked with the fast variable priority label [n].

7. A method for transmitting a sampling signal of a smart water station as claimed in claim 6, characterized in that: The edge device divides the sampled signal into data packets according to the priority, and the specific method of hierarchically processing the data packets according to the variable category and priority order is: The collected slow variable sampling signals and fast variable sampling signals are divided into P slow variable data packets and K fast variable data packets according to the priority, the first P1 priorities of the slow variables are used as the high priority of the slow variables, and the last P2 priorities are used as the low priority of the slow variables, the first K1 priorities of the fast variables are used as the high priority of the fast variables, and the last K2 priorities are used as the low priority of the fast variables; Apply different data compression, encryption and transmission strategies to slow variable data packets and fast variable data packets; When the network is congested, the edge devices and network nodes transmit high-priority fast variable packets first, then high-priority slow variable packets, then low-priority fast variable packets, and finally low-priority slow variable packets; After the cloud receives the data packet, the high-priority fast variable data packet directly enters the memory-level cache, the high-priority slow variable data packet enters the SSD storage, and the low-priority fast variable data packet and slow variable data packet enter the data warehouse.

8. A flow transmission system for a sampling signal of a smart water station, which is used to implement a flow transmission method for a sampling signal of a smart water station according to any one of claims 1 to 7, characterized in that: include: A sampling signal classification module is used to classify the sampling signals into different variable categories according to the sampling signal change rate, wherein the variable categories include slow variables and fast variables, and classify the slow variables and fast variables into signal subsets with different priorities according to the importance index of the sampling signals; A deep Q network building module is used to define the slow variable deep Q network state space, action space, reward function and the fast variable deep Q network state space, action space, reward function for each variable category signal subset, and to build and train the slow variable deep Q network and the fast variable deep Q network; A sampling decision output module is used for the variable state of the edge device receiving the signal subset, and outputs a global sampling decision according to the slow variable deep Q network and the fast variable deep Q network. The edge device executes the global sampling decision and collects the sampling signal; the variable state includes the slow variable state and the fast variable state, the global sampling decision includes the slow variable sampling decision and the fast variable sampling decision, and the sampling signal includes the slow variable sampling signal and the fast variable sampling signal; The sampling signal processing module is used for the edge device to divide the sampling signal into data packets according to the priority, and to perform hierarchical processing on the data packets according to the variable category and priority order; the data packets include slow variable data packets and fast variable data packets.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for transferring and transmitting sampling signals of a smart water station as described in any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in the method for streaming and transmitting sampling signals of a smart water station as described in any one of claims 1-7.

Images