A dynamic ring monitoring method and system

Abstract

The application discloses a kind of dynamic ring monitoring method and system, it is related to power plant detection technical field, including using sensor to collect the operation data of power plant key equipment, establish equipment dynamic ring model;Real-time state data of equipment is obtained, compared with dynamic ring model in real time, judge whether it is beyond normal working range;Intelligent diagnosis and prediction are carried out, according to the diagnosis result, corresponding fault alarm is generated, and is actively pushed using multiple channels;Collect operation data and fault feedback, continuously optimize equipment dynamic ring model and intelligent diagnosis algorithm, realize self-learning self-adaptation.The application reduces the number of artificial inspection, improves fault detection and positioning efficiency, saves operation manpower cost.System automatically pushes alarm also reduces communication coordination cost.Improves the equipment operation maintenance level of power plant, so that power plant develops in the direction of intelligentization, informatization and automation in equipment management.

Description

Technical Field

[0001] This invention relates to the field of power plant testing technology, and in particular to a dynamic environment monitoring method and system. Background Technology

[0002] Currently, power plant equipment condition monitoring mainly relies on regular manual inspections and simple over-limit alarms. This monitoring method suffers from detection lag and cannot achieve real-time status tracking. Furthermore, equipment faults cannot be automatically and intelligently diagnosed, relying primarily on the experience of maintenance personnel, resulting in a high rate of misdiagnosis. From a user perspective, it cannot monitor the dynamic status of equipment in real time, nor can it issue timely fault warnings and provide intelligent diagnosis. From a technical implementation perspective, it is impossible to build accurate dynamic models of the equipment, and reliable fault detection algorithms are lacking. From a product promotion perspective, it cannot provide intelligent equipment health management solutions. Therefore, there is an urgent need for a comprehensive intelligent monitoring and fault diagnosis technology solution for equipment operating status. This solution needs to utilize various sensors to collect equipment operating data, train a dynamic loop model of the equipment using the data, and compare and determine whether the status is abnormal in real time. Simultaneously, it needs to build a fault knowledge base, use intelligent algorithms to diagnose fault types and causes, and issue alarms through multiple channels. Continuous optimization of the model and algorithm is also required to achieve adaptive monitoring. Summary of the Invention

[0003] In view of the problems existing in the condition monitoring of power plant equipment, this invention is proposed.

[0004] Therefore, the problem that this invention aims to solve is that equipment malfunctions cannot be automatically and intelligently diagnosed, and mainly rely on the experience of maintenance personnel, which results in a high rate of misdiagnosis.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] In a first aspect, embodiments of the present invention provide an environmental monitoring method, which includes,

[0007] By using sensors to collect operating data of key equipment in power plants, a dynamic loop model of the equipment is established.

[0008] Acquire real-time status data of the equipment and compare it with the dynamic loop model in real time to determine whether it exceeds the normal operating range;

[0009] Perform intelligent diagnosis and prediction, generate corresponding fault alarms based on the diagnosis results, and actively push them through multiple channels;

[0010] Collect operational data and fault feedback, continuously optimize the equipment dynamic loop model and intelligent diagnostic algorithm, and achieve self-learning and self-adaptation.

[0011] As a preferred embodiment of the dynamic ring monitoring method of the present invention, the construction of the dynamic ring model includes the following steps:

[0012] Establish a physical model;

[0013] The theoretical calculation range of state parameters is determined based on the physical model;

[0014] The dynamic distribution of state parameters is obtained by training an LSTM neural network algorithm.

[0015] The dynamic distribution of the state parameters is modeled as a Gaussian distribution to obtain a dynamic loop model;

[0016] The physical model includes a physical model of a photovoltaic power generation unit;

[0017] The physical model of photovoltaic power generation resistor is shown in the following equation:

[0018]

[0019] In the formula, Iph is the photocurrent, Io is the saturation current, q is the electron charge, Vpv is the module output voltage, Rs is the module equivalent series resistance, n is the basic electron chemical potential factor, k is the Boltzmann constant, Tpv is the module temperature, Tr is the reference temperature, α is the temperature coefficient, Ppv is the module output power, η is the module conversion efficiency, and Ee is the incident light intensity.

[0020] Considering the heat loss mechanism of the components, a heat loss model is established:

[0021] Ploss = Ppv - ηEeS

[0022] In the formula, S is the light-receiving area of ​​the component, and Ploss is the power loss of the component.

[0023] As a preferred embodiment of the environmental monitoring method of the present invention, the LSTM neural network algorithm includes an input gate, a forget gate, a hidden layer, an output layer, candidate units, unit states, and an output.

[0024] The input gate is shown in the following formula:

[0025] it=σ(WxiXt+Whiht-1+WciCt-1+bi)

[0026] In the formula, it- represents the input gate state, Wxi- represents the weight matrix from input to the input gate, Xt represents the current input, Whi represents the weight matrix from hidden state to input gate, ht-1 represents the hidden state from the previous time step; Wci represents the weight matrix from cell state to input gate, Ct-1 represents the cell state from the previous time step, and bi represents the input gate bias.

[0027] The forgetting gate is shown in the following formula:

[0028] ft=σ(WxfXt+Whfht-1+WcfCt-1+bf)

[0029] In the formula, ft is the forget gate state, Wxf is the weight matrix, which is input to the forget gate, Whf is the weight matrix, which is input to the forget gate from the hidden state, Wcf is the weight matrix, which is input to the forget gate from the cell state, and bf is the forget gate bias.

[0030] The output layer is shown in the following formula:

[0031] ot=σ(WxoXt+Whoht-1+WcoCt-1+bo)

[0032] In the formula, ot is the output gate state, Wxo is the weight matrix from input to output gate, Who is the weight matrix from hidden state to output gate, Wco is the weight matrix from cell state to output gate, and bo is the output gate bias.

[0033] The candidate unit is shown in the following formula:

[0034] Ct~=tanh(WxcXt+Whcht-1+bc)

[0035] In the formula, Ct is the candidate unit state, Wxc is the weight matrix, Whc is the weight matrix input to the candidate unit, the hidden state to the candidate unit, and bc is the candidate unit bias.

[0036] The unit state is shown in the following formula:

[0037] ht = ot * tanh(Ct).

[0038] As a preferred embodiment of the environmental monitoring method of the present invention, the intelligent diagnosis includes calculating the distance between two vectors using Euclidean distance:

[0039]

[0040] In the formula, xi is the i-th input feature vector, and yi is the i-th sample feature vector;

[0041] The intelligent diagnosis also includes setting a threshold δ as follows:

[0042] δ=Dp+β*standard_deviation(D1,D2,...DN)

[0043] Where β is the empirical coefficient, standard_deviation is the standard deviation of the sample distance, and Dp is the P-th distance value.

[0044] As a preferred embodiment of the environmental monitoring method of the present invention, the fault alarm is divided into general level fault, medium level fault, severe level fault, and extremely severe level fault, and the general level fault < medium level fault < severe level fault < extremely severe level fault.

[0045] If the fault level is a general fault, the alarm system will emit a blue light, without making a sound or requiring external message push. If the same fault occurs repeatedly, the notification frequency will be controlled to avoid message fatigue.

[0046] If the fault level is medium, the alarm system will emit a yellow light, emit a long alarm sound with the same intermittent and relatively calm sound, and send an email to the person in charge's mobile terminal requesting confirmation and feedback on the handling measures. If no feedback is received within 15 minutes, it will automatically be upgraded to a serious level.

[0047] If the fault level is severe, the alarm system will emit an orange light, a short and sharp alarm sound, and send a text message to the person in charge's mobile terminal requesting confirmation and feedback on the handling measures. If no feedback is received within 5 minutes, the person in charge will be contacted using instant messaging tools to ensure immediate handling.

[0048] If the fault level is a critical fault, the alarm system will emit a red light, a rapid and sharp alarm sound, and notify the person in charge via instant communication tools.

[0049] In a preferred embodiment of the environmental monitoring method of the present invention, the average value and standard deviation of the state parameters are calculated as follows:

[0050]

[0051]

[0052] In the formula, μ is the mean of the state parameters, n is the length of the time series data, i.e. the number of samples, xi is the i-th sample value in the time series, σ is the standard deviation of the state parameters, ∑xi-x is the sum of all samples in the time series, and ∑(xi-μ)2 is the sum of squared deviations of each sample from the mean.

[0053] As a preferred embodiment of the environmental monitoring method of the present invention, the operating data includes temperature, humidity, voltage, current, light intensity, and battery status.

[0054] Secondly, embodiments of the present invention provide an environmental monitoring system, which includes: a data acquisition module, used to set up sensors, acquire equipment operating status data, and preprocess the data;

[0055] The dynamic loop model building module is used to build a physical model of the equipment, describe the working principle of the equipment, use historical data of the equipment to train an LSTM neural network, obtain a dynamic model of state parameters, and build a dynamic loop model based on the dynamic model.

[0056] The intelligent diagnostic module is used to acquire equipment status data in real time, extract and match features from the real-time data, calculate similarity, determine the fault type and cause, and generate fault alarms of different levels.

[0057] The model optimization module is used to retrain the model, adjust parameters, and improve the algorithm based on collected fault feedback and new data, so as to achieve continuous optimization and evolution of the model.

[0058] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of the above-described environmental monitoring method.

[0059] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the above-described environmental monitoring method.

[0060] The beneficial effects of this invention are as follows: By collecting equipment operation data in real time through various sensors and comparing it with a dynamic loop model, real-time tracking of equipment status can be achieved. When status parameters exceed the normal range, early warnings can be issued quickly to prevent further deterioration of the fault. By constructing a fault knowledge base and using intelligent algorithms to extract features and match them with samples, the specific patterns and causes of equipment faults can be automatically diagnosed, avoiding misjudgments by maintenance personnel and improving maintenance efficiency. Depending on the severity of the fault, the system can proactively push alarms of different colors, sounds, and frequencies, and provide reminders through multiple channels to ensure that the fault is taken seriously and addressed. This reduces the number of manual inspections while improving fault detection and location efficiency, saving on maintenance manpower costs. The system's automatic alarm push also reduces communication and coordination costs. Ultimately, this improves the equipment operation and maintenance level of power plants, enabling power plants to develop towards intelligent, information-based, and automated equipment management. Attached Figure Description

[0061] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0062] Figure 1 This is a flowchart of the environmental monitoring method. Detailed Implementation

[0063] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0064] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0065] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0066] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0067] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0068] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0069] Example 1

[0070] Reference Figure 1 This is the first embodiment of the present invention, which provides an environmental monitoring method, including:

[0071] S1: Use sensors to collect operating data of key equipment in the power plant and establish a dynamic loop model of the equipment.

[0072] Based on the operating principles of the power plant equipment, the key status parameters that need to be monitored are determined, including temperature, humidity, voltage, current, light intensity, and battery status.

[0073] Sensors are connected to the data acquisition unit via hardwired or wireless networks, and the acquisition unit collects sensor data. The acquisition unit connects to the industrial cloud platform of the data center via an industrial Ethernet switch, and different data acquisition frequencies are set for different equipment types, such as up to 1000Hz for high-speed rotating equipment and 10Hz for low-speed equipment. The acquired data is buffered and time-synchronized, and timestamped. The acquired data is then transmitted to the power plant's local data center via ultra-high voltage transmission through the cloud platform.

[0074] Physical models are constructed using collected operational data, such as physical models of photovoltaic power generation units.

[0075] The physical model of photovoltaic power generation resistor is shown in the following equation:

[0076]

[0077] In the formula, Iph is the photocurrent, Io is the saturation current, q is the electron charge, Vpv is the module output voltage, Rs is the module equivalent series resistance, n is the basic electron chemical potential factor, k is the Boltzmann constant, Tpv is the module temperature, Tr is the reference temperature, α is the temperature coefficient, Ppv is the module output power, η is the module conversion efficiency, and Ee is the incident light intensity.

[0078] Considering the heat loss mechanism of the components, a heat loss model is established:

[0079] Ploss = Ppv - ηEeS

[0080] In the formula, S is the light-receiving area of ​​the component, and Ploss is the power loss of the component.

[0081] Furthermore, the theoretical operating range of state parameters is determined based on the physical model. For example, the main parameters affecting the array operating voltage Vpv of a photovoltaic power generation unit are: photocurrent Iph, temperature Tpv, and illuminance Ee. Substituting these parameters into the electrical model, the formula for calculating Vpv can be obtained:

[0082]

[0083] Based on the parameter ranges in the component manual: Iph∈[8A,10A], Tpv∈[25℃,60℃], Ee∈[200W / m2,1000W / m2], leaving room for calculation error, we substitute the upper and lower limits to calculate the theoretical working range of Vpv: Vpv∈[360V,540V]. The calculation of other working ranges is similar.

[0084] Furthermore, collect all operational data of the equipment over the past 2-3 years, including status parameters, alarms, logs, etc., label the data, label fault-related data based on maintenance records and other information, label normal operation data based on no alarms and maintenance records, preprocess the data to remove errors and duplicate data, smooth the data to reduce noise, and interpolate to fill in missing values.

[0085] Furthermore, statistical features such as mean and standard deviation are extracted, frequency domain features such as FFT are extracted, and time domain features such as RMS and peak value are extracted. Oversampling enhancement is performed on a small amount of fault data to generate more fault state data and improve the robustness of the model. Finally, 80% of the data is randomly divided into a training set and 20% into a test set.

[0086] Furthermore, the dynamic distribution of state parameters is obtained by training with an LSTM neural network algorithm. The LSTM network structure includes an input layer: device state parameter vector Xt, a hidden layer: LSTM units, and an output layer: predicted state parameter values, candidate units, unit states, and outputs.

[0087] The input gate is shown in the following formula:

[0088] it=σ(WxiXt+Whiht-1+WciCt-1+bi)

[0089] In the formula, it- represents the input gate state, Wxi- represents the weight matrix from input to the input gate, Xt represents the current input, Whi represents the weight matrix from hidden state to input gate, ht-1 represents the hidden state from the previous time step; Wci represents the weight matrix from cell state to input gate, Ct-1 represents the cell state from the previous time step, and bi represents the input gate bias.

[0090] The forgetting gate is shown in the following formula:

[0091] ft=σ(WxfXt+Whfht-1+WcfCt-1+bf)

[0092] In the formula, ft is the forget gate state, Wxf is the weight matrix (input to the forget gate), Whf is the weight matrix (input to the forget gate), Wcf is the weight matrix (input to the forget gate), and bf is the forget gate bias.

[0093] The output layer is shown in the following formula:

[0094] ot=σ(WxoXt+Whoht-1+WcoCt-1+bo)

[0095] In the formula, ot is the output gate state, Wxo is the weight matrix from input to output gate, Who is the weight matrix from hidden state to output gate, Wco is the weight matrix from cell state to output gate, and bo is the output gate bias.

[0096] The candidate unit is shown in the following formula:

[0097] Ct~=tanh(WxcXt+Whcht-1+bc)

[0098] In the formula, Ct is the candidate unit state, Wxc is the weight matrix, Whc is the weight matrix input to the candidate unit, the hidden state is input to the candidate unit, and bc is the candidate unit bias.

[0099] The unit state is shown in the following formula:

[0100] ht = ot * tanh(Ct)

[0101] Furthermore, by minimizing the loss function, we obtain an LSTM model of all state parameters. For new input data, we use the LSTM model to make predictions, which is the dynamic distribution of the parameters.

[0102] Furthermore, the dynamic distribution of the state parameters is modeled as a Gaussian distribution, yielding the mean μ and standard deviation σ for each state parameter. The collected historical data is then processed according to a time-series normal distribution. For the smoothed time series, the mean μ and standard deviation σ of the state parameter are calculated.

[0103]

[0104]

[0105] In the formula, μ is the mean of the state parameters, n is the length of the time series data, i.e. the number of samples, xi is the i-th sample value in the time series, σ is the standard deviation of the state parameters, ∑xi-x is the sum of all samples in the time series, and ∑(xi-μ)2 is the sum of squared deviations of each sample from the mean.

[0106] Repeating the above steps, we finally obtain the Gaussian distribution model N(μ,σ²) for all state parameters. According to the 3σ criterion, the normal operating range of the state parameters is: [μ-3σ,μ+3σ]. Here, μ is the average value of the state parameters, reflecting the median level of normal operation; σ is the standard deviation, reflecting the range of fluctuation of the state parameters. Considering actual operation and maintenance, we appropriately narrow the range of the operating range to: [μ-2.5σ,μ+2.5σ]. We plot the normal operating range of the state parameters on a coordinate system, with the upper and lower limits corresponding to the upper and lower thresholds of the parameters, respectively. Connecting the upper and lower limit curves of each state parameter forms a closed region, which is the dynamic loop model of the equipment. The dynamic loop model intuitively reflects the allowable normal fluctuation range and operating range of different state parameters.

[0107] S2: Obtain the real-time status data of the device and compare it with the dynamic loop model in real time to determine whether it exceeds the normal operating range.

[0108] Real-time operating status data of the acquired equipment is collected, including parameters such as temperature, pressure, speed, and flow rate. Data preprocessing is performed, including removing outliers, smoothing filtering, filling in missing values, and normalization. The preprocessed data is then compared with the normal ranges of each parameter in the constructed dynamic loop model to determine whether each status parameter exceeds the normal threshold range of the dynamic loop model.

[0109] Based on the normal operating range of the status parameters obtained in step S1, for example: Assume the normal operating range of the photovoltaic module temperature (Tpv) is 25-60 degrees Celsius. If the detected Tpv is 40 degrees Celsius, it is within the allowable range and is considered normal. If Tpv rises above 65 degrees Celsius, it is considered abnormal, possibly due to insufficient module heat dissipation. If Tpv drops below 20 degrees Celsius, it is considered abnormal, possibly due to excessively low ambient temperature.

[0110] For wind turbine generator sets, the normal operating range of the generator speed n is assumed to be 500-1500 rpm. If the detected n is 1200 rpm, it is within the allowable range and is considered normal. If n drops below 400 rpm, it is considered abnormal, possibly due to a transmission box malfunction. If n increases to above 1600 rpm, it is considered abnormal, possibly due to excessive air retraction by the blades.

[0111] S3: Performs intelligent diagnosis and prediction, generates corresponding fault alarms based on the diagnosis results, and actively pushes them through multiple channels.

[0112] Based on the collected historical data, it is processed into historical fault data, fault modes and causes are labeled, and a fault knowledge base is constructed. Feature extraction is performed on the real-time input status parameter data to obtain key features reflecting the health status of the equipment. The extracted features are matched with sample features in the fault knowledge base, and feature similarity is calculated. Euclidean distance is used to calculate the distance between two vectors.

[0113]

[0114] In the formula, xi is the i-th input feature vector, and yi is the i-th sample feature vector;

[0115] From the training sample set, randomly select N pairs of samples (X1,Y1), (X2,Y2)...(XN,YN), and calculate the Euclidean distance between each pair of samples: D1=D(X1,Y1)D2=D(X2,Y2)...DN=D(XN,YN). Sort the above N distance values, let the Pth distance value be Dp, and take the threshold δ as:

[0116] δ=Dp+β*standard_deviation(D1,D2,...DN)

[0117] Where β is the empirical coefficient and standard_deviation is the standard deviation of the sample distance.

[0118] Set a distance threshold δ. If D(X,Y)<δ, then the input X is determined to have a high similarity with the sample Y. Select the k samples with the smallest distance D from X from all samples as the samples most similar to X. Return these k similar samples and their corresponding similarities: {Y1:0.85,Y2:0.82,...Yk:0.79}. The similarity calculation formula is: S(X,Y)=1-(D(X,Y) / δ). Finally, the top k similar samples Y and their similarities S of the input feature vector X are obtained. Based on the fault mode corresponding to sample Y, the fault type of X is predicted.

[0119] Define the severity level standards for faults, such as general, moderate, severe, and critical levels. For each fault mode, determine the corresponding standard fault level and establish a fault knowledge base. When a fault is detected, query the knowledge base to obtain the severity level of the fault mode. If the real-time status parameters exceed the judgment threshold of the standard level, the fault level can be upgraded.

[0120] If the fault level is a general fault, the alarm system will emit a blue light, without making a sound or requiring external message push. If the same fault occurs repeatedly, the notification frequency will be controlled to avoid message fatigue.

[0121] If the fault level is a major fault, the alarm system will emit a yellow light, emit a long, relatively calm alarm sound with the same intermittent intervals, and send an email to the person in charge's mobile terminal requesting confirmation and feedback on the handling measures. If no feedback is received within 15 minutes, the level will be automatically escalated to a serious fault.

[0122] If the fault level is severe, the alarm system will emit an orange light, a short, sharp alarm sound, and send a text message to the person in charge's mobile terminal requesting confirmation and feedback on handling measures. If no feedback is received within 5 minutes, the person in charge will be contacted via instant messaging to ensure immediate handling.

[0123] If the fault level is extremely high, the alarm system will emit a red light, a rapid and sharp alarm sound, and notify the person in charge via instant communication tools.

[0124] The following are examples of fault alerts and proactive push notifications:

[0125] Temperature Alarm (Severity Level): "Overheating alarm detected for #1 photovoltaic array. Temperature exceeds the red line of 65 degrees Celsius. Please immediately check the array cooling system and repair related cooling devices. Overheating of the modules can lead to reduced efficiency or even damage."

[0126] Voltage Alarm (Major Level): "PV grid-connected inverter voltage too low alarm, voltage has dropped to 320V. Please check the grid status immediately and report the handling situation. Low voltage will cause grid connection failure." Pressure Alarm (Moderate Level): "Gas compressor outlet pressure slightly low alarm, please check if there is a slow leak in the pipeline system. Low pressure will cause long-term unstable operation of the system."

[0127] Speed ​​alarm (severity level): "Overspeed alarm detected for generator #2. The speed has exceeded the red line limit of 1700 rpm. Please take immediate measures to reduce the load and check the blade adjustment system. Overspeed will cause unit power overload."

[0128] Temperature Alarm (General Level): "Wind turbine nacelle temperature slightly high alarm. Please check the nacelle ventilation and air conditioning system to ensure normal power output of the equipment."

[0129] S4: Collect operational data and fault feedback, continuously optimize the equipment dynamic loop model and intelligent diagnostic algorithm, and achieve self-learning and self-adaptation.

[0130] The system collects real-time equipment status data, records equipment operation logs, summarizes and stores various statistical data such as running time and load parameters, collects fault feedback, records all equipment fault alarm information, classifies and statistically analyzes the number and frequency of different types of faults, analyzes the root causes of faults, and records the maintenance process and result feedback.

[0131] Furthermore, the dynamic loop model is retrained and optimized using the newly acquired data, the parameter boundaries in the model are adjusted according to statistical data, the model topology and equations are modified through data-driven methods, the deviation between the model output and the actual operation is evaluated, and the optimization training is repeated until the model accuracy meets the requirements.

[0132] Furthermore, based on the new feedback, we increased the number of fault mode samples for the algorithm, adjusted the model structure, added feature extraction and judgment for newly discovered faults, improved the model parameters, improved prediction accuracy, adjusted and optimized the algorithm to address missed and false alarms, and evaluated the effect multiple times on the test set to ensure the algorithm is stable and efficient.

[0133] Furthermore, a closed-loop process for data feedback and model optimization is established, with large-scale full-data retraining conducted every 6 months, and key parameters such as model structure and training algorithms evaluated and optimized iteratively every 3 months. Small-scale incremental training is conducted monthly, using newly added data to fine-tune the model. If significant changes in operational status occur or new equipment failure modes are identified, rapid incremental training is performed immediately.

[0134] Furthermore, this embodiment also provides an environmental monitoring system, including: a data acquisition module, used to set up sensors, collect equipment operating status data, and preprocess the data;

[0135] The dynamic loop model building module is used to build a physical model of the equipment, describe the working principle of the equipment, use historical data of the equipment to train an LSTM neural network, obtain a dynamic model of state parameters, and build a dynamic loop model based on the dynamic model.

[0136] The intelligent diagnostic module is used to acquire equipment status data in real time, extract and match features from the real-time data, calculate similarity, determine the fault type and cause, and generate fault alarms of different levels.

[0137] The model optimization module is used to retrain the model, adjust parameters, and improve the algorithm based on collected fault feedback and new data, so as to achieve continuous optimization and evolution of the model.

[0138] This embodiment also provides a computer device applicable to the environmental monitoring method, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the environmental monitoring method proposed in the above embodiment.

[0139] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0140] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the environmental monitoring method proposed in the above embodiments.

[0141] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0142] Example 2

[0143] Reference Figure 1 This is the second embodiment of the present invention, which provides a dynamic environment monitoring method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0144] This invention was piloted in a power plant. The power plant, completed last year, has a total installed capacity of 100MW and is divided into multiple photovoltaic array areas. The #1 photovoltaic array area contains 20 single-axis solar tracking devices, under which are 60 sets of parallel double-glass photovoltaic strings. Each string contains 1500 photovoltaic modules, which are high-efficiency monocrystalline silicon modules with a power of 450W per module.

[0145] The operating status of photovoltaic (PV) modules directly affects power generation efficiency and power plant revenue. To monitor the status of array #1, a temperature sensor was installed at the input of one PV string to monitor the module temperature (Tpv) in real time. A voltage sensor and current transformer were installed at the output of the string to monitor the output voltage (Vpv) and output current (Ipv) in real time. These sensors are connected to a field data terminal via acquisition equipment, collecting data at a frequency of 10Hz. The data is then transmitted in real time via a dedicated power line to the power plant's meteorological service system and SCADA monitoring system, and stored in a historical database. These sensors are also connected to a fieldbus via acquisition equipment, using the OPC protocol to send data to the station-level monitoring system. The monitoring system has a data acquisition module in its industrial control computer, capable of collecting this sensor data at a frequency of 10Hz. After data buffering, it is sent to the plant-level data center and stored in a time-series database.

[0146] We extracted temperature, voltage, and current data from the historical database of this series of components over the past two years, and performed annotation and preprocessing to generate training and test sets. We then trained an LSTM network to obtain a dynamic model of these state parameters. By comparing real-time detection data with this dynamic model, we can accurately determine the operating status of the components, promptly detect anomalies, and perform fault reasoning. (See Table 1.)

[0147] Table 1 Training Number Table

[0148]

[0149]

[0150] The dynamic model with the above parameters is obtained by training an LSTM network, fitted with a Gaussian distribution, and the normal operating range is obtained. See Table 2 for details.

[0151] Table 2 Parameter Working Range Table

[0152] Tpv 25℃～60℃ Vpv 360V~540V Ipv 8A~12A

[0153] The real-time data is compared with the model to determine the running status.

[0154] Table 3. Operating Status Judgment Table

[0155] 1:00 35℃ 380V 10A normal 1:01 70℃ 380V 10A Tpv abnormal

[0156] The intelligent diagnostic module analyzes the collected real-time data, and the specific fault judgments for the day are shown in Table 4.

[0157] Table 4 Fault Diagnosis Table

[0158] 1:01 Component overheating serious

[0159] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring environmental conditions, characterized in that: include, By using sensors to collect operating data of key equipment in power plants, a dynamic loop model of the equipment is established. Acquire real-time status data of the equipment and compare it with the dynamic loop model in real time to determine whether it exceeds the normal operating range; Perform intelligent diagnosis and prediction, generate corresponding fault alarms based on the diagnosis results, and actively push them through multiple channels; Collect operational data and fault feedback, continuously optimize the equipment dynamic loop model and intelligent diagnostic algorithm, and achieve self-learning and self-adaptation; The construction of the dynamic ring model includes the following steps: Establish a physical model; The theoretical calculation range of state parameters is determined based on the physical model; The dynamic distribution of state parameters is obtained by training an LSTM neural network algorithm. The dynamic distribution of the state parameters is modeled as a Gaussian distribution to obtain a dynamic loop model; The physical model includes a physical model of a photovoltaic power generation unit; The physical model of photovoltaic power generation resistor is shown in the following equation: In the formula, Iph is the photocurrent, Io is the saturation current, q is the electron charge, Vpv is the module output voltage, Rs is the module equivalent series resistance, n is the basic electron chemical potential factor, k is the Boltzmann constant, Tpv is the module temperature, Tr is the reference temperature, and α is the temperature coefficient. Considering the heat loss mechanism of the components, a heat loss model is established: In the formula, Ppv is the output power of the module, η is the conversion efficiency of the module, Ee is the incident light intensity, S is the light-receiving area of ​​the module, and Ploss is the power loss of the module.

2. The environmental monitoring method as described in claim 1, characterized in that: The LSTM neural network algorithm includes an input gate, a forget gate, a hidden layer, an output layer, candidate cells, cell states, and an output. The input gate is shown in the following formula: In the formula, it is the input gate state, Wxi is the weight matrix from input to input gate, Xt is the current input, Whi is the weight matrix from hidden state to input gate, ht-1 is the hidden state from the previous time step; Wci is the weight matrix from cell state to input gate, Ct-1 is the cell state from the previous time step, and bi is the input gate bias. The forgetting gate is shown in the following formula: In the formula, ft is the forget gate state, Wxf is the weight matrix, which is input to the forget gate, Whf is the weight matrix, which is input to the forget gate from the hidden state, Wcf is the weight matrix, which is input to the forget gate from the cell state, and bf is the forget gate bias. The output layer is shown in the following formula: In the formula, ot is the output gate state, Wxo is the weight matrix from input to output gate, Who is the weight matrix from hidden state to output gate, Wco is the weight matrix from cell state to output gate, and bo is the output gate bias. The candidate unit is shown in the following formula: In the formula, Ct~ represents the candidate cell state, Wxc is the weight matrix input to the candidate cell, Whc is the weight matrix from the hidden state to the candidate cell, and bc is the candidate cell bias. The unit state is shown in the following formula: 。 3. The environmental monitoring method as described in claim 2, characterized in that: The intelligent diagnosis includes using Euclidean distance to calculate the distance between two vectors: In the formula, X i Let Y be the i-th input feature vector. i Let i be the feature vector of the i-th sample; The intelligent diagnosis also includes setting a threshold δ as follows: δ = Dp + β * standard_deviation, p=1, 2,...,N Where β is the empirical coefficient, standard_deviation is the standard deviation of the sample distance, and Dp is the p-th distance value.

4. The environmental monitoring method as described in claim 3, characterized in that: The fault alarms are classified into general level faults, medium level faults, severe level faults, and extremely severe level faults, with general level faults < medium level faults < severe level faults < extremely severe level faults. If the fault level is a general fault, the alarm system will emit a blue light, without making a sound or requiring external message push. If the same fault occurs repeatedly, the notification frequency will be controlled to avoid message fatigue. If the fault level is medium, the alarm system will emit a yellow light, emit a long alarm sound with the same intermittent and relatively calm sound, and send an email to the person in charge's mobile terminal requesting confirmation and feedback on the handling measures. If no feedback is received within 15 minutes, it will automatically be upgraded to a severe level. If the fault level is severe, the alarm system will emit an orange light, a short and sharp alarm sound, and send a text message to the person in charge's mobile terminal requesting confirmation and feedback on the handling measures. If no feedback is received within 5 minutes, the person in charge will be contacted using instant messaging tools to ensure immediate handling. If the fault level is a critical fault, the alarm system will emit a red light, a rapid and sharp alarm sound, and notify the person in charge via instant communication tools.

5. The environmental monitoring method as described in claim 4, characterized in that: The average value and standard deviation of the state parameters are calculated as follows: In the formula, μ is the mean of the state parameters, n is the length of the time series data, i.e. the number of samples, xi is the i-th sample value in the time series, and σ is the standard deviation of the state parameters.

6. The environmental monitoring method as described in claim 5, characterized in that: The operational data includes temperature, humidity, voltage, current, light intensity, and battery status.

7. An environmental monitoring system, based on the environmental monitoring method according to any one of claims 1 to 6, characterized in that: include, The data acquisition module is used to set up sensors, collect equipment operating status data, and preprocess the data; The dynamic loop model building module is used to build a physical model of the equipment, describe the working principle of the equipment, use historical data of the equipment to train an LSTM neural network, obtain a dynamic model of state parameters, and build a dynamic loop model based on the dynamic model. The intelligent diagnostic module is used to acquire equipment status data in real time, extract and match features from the real-time data, calculate similarity, determine the fault type and cause, and generate fault alarms of different levels. The model optimization module is used to retrain the model, adjust parameters, and improve the algorithm based on collected fault feedback and new data, so as to achieve continuous optimization and evolution of the model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the environmental monitoring method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the environmental monitoring method according to any one of claims 1 to 7.

Images