A sensorless intelligent control method and system for electromechanical equipment

Abstract

The present invention relates to the field of electromechanical control technology, and in particular to a non-sensing intelligent control method and system for electromechanical equipment. An ultrasonic flow meter, a digital pressure gauge and a frequency converter are used to obtain basic data of a water pump, which is stored in a cloud storage system. An on-site terminal monitors the current, power and rotation speed of the electromechanical equipment in real time, obtains the flow and energy consumption through machine learning analysis and forms a report; provides suggestions for optimizing the operating efficiency of the electromechanical equipment; analyzes abnormal fluctuations in current, power and rotation speed to issue an alarm in time, and displays the alarm on a terminal after analysis, thereby realizing a non-sensing intelligent control system for electromechanical equipment that integrates monitoring, analysis, early warning, control and visualization.

Description

Technical Field

[0001] The present invention relates to the field of electromechanical control technology, and in particular to an intelligent system for monitoring, analyzing, warning and controlling a sensorless water pump. Background Art

[0002] Water pumps are widely used, including urban water supply, sewage systems, civil and construction systems, agricultural water conservancy systems, power station systems, chemical systems, etc. With the continuous development of urbanization, the demand for water pumps is greater. The user's water demand changes over time, which does not match the rated flow of the water pump unit. The long-term inefficient operation of the water pump consumes a lot of electricity. Users can only see the rated parameters from the motor nameplate of the water pump, and cannot see the flow-head curve, flow-power curve and flow-efficiency curve, let alone set the operating point in the efficient operation range.

[0003] The rated power operating efficiency of electromechanical equipment is the highest. For building water use, users' water use is cyclical, so operating at industrial frequency will cause huge energy waste, and will cause system fatigue and increased maintenance costs. Summary of the invention

[0004] In order to overcome the shortcomings of the above existing technologies, a sensorless and intelligent electromechanical equipment non-sensing intelligent control method is provided by using a data acquisition monitoring system and a mathematical model established thereby. The method is as follows:

[0005] Collecting flow and pressure data of a primary electromechanical device, and collecting device information of the electromechanical device in real time, the device information including current, power data, operating frequency data, speed data, temperature data, and vibration data; preprocessing the device information to generate a first data set, the preprocessing including: converting pressure data into head, using vibration data to reduce noise and filter environmental noise, and cutting and compressing historical data generated by the collected device information over time;

[0006] Using the deep learning GRU algorithm, based on the first data set as historical data, the historical data is divided into fixed-length data as labels using a smoothing window, The historical data are combined after being lagged by 1 to 5 digits. Historical data is divided into fixed-length data using a smoothing window as input The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain a reset gate. The state information at the previous moment contains the previous features. The function of the reset gate is to control the state information at the previous moment to be written into the current state candidate set. The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain an update gate. The update gate controls the ratio of the state at the previous moment to the state candidate set at the current moment to obtain the current state information. The current moment output is obtained from the current moment state. The absolute error between the current moment label and the current moment output is used as back propagation for training. , , and Matrix, error convergence to get the model, take the lagged data as input, and get the predicted data;

[0007] The predicted data is combined with the historical data, and the combined data is used as the input of the matrix model, and the cycle is repeated until the preset length ends; the reset gate:

[0008]

[0009] The update gate:

[0010]

[0011] The current state candidate set:

[0012]

[0013] Current status:

[0014]

[0015] Output at the current moment:

[0016]

[0017] in, Input for the current moment, is the state at the previous moment, , , and is a trainable matrix and is an activation function.

[0018] Furthermore, the flow value of the pre-set electromechanical equipment is read, the operating frequency pressure data of the electromechanical equipment is collected, the pressure data is converted into a head, and the maximum operating frequency performance curve of the electromechanical equipment is obtained by using orthogonal polynomial least squares fitting. The median of the maximum operating frequency and the minimum operating frequency is taken, and the frequency conversion curve of the median frequency can be obtained according to the water pump similarity law. The maximum value point of the frequency conversion curve is taken. If the horizontal coordinate of the maximum value point is less than the flow value, the median frequency is set to the minimum operating frequency, and the above operation is repeated. If the horizontal coordinate of the maximum value point is greater than the flow value, the median frequency is set to the maximum operating frequency, and the above operation is repeated. If the absolute difference between the horizontal coordinate of the maximum value point and the flow rate is less than the threshold value, the median frequency is obtained as the highest efficiency operating frequency of the electromechanical equipment.

[0019] Preferably, the electromechanical equipment automatically sets the operating point manually or using predicted data, and the operating point includes flow data and head data. After the system obtains the operating point, the maximum operating frequency performance curve of the electromechanical equipment is obtained by using orthogonal polynomial least squares fitting, and the median of the maximum operating frequency and the minimum operating frequency is taken. The frequency conversion curve of the median frequency can be obtained according to the water pump similarity law, and the horizontal coordinate of the operating point is substituted into the frequency conversion curve to obtain the vertical coordinate of the frequency conversion curve. If the vertical coordinate of the frequency conversion curve is less than the vertical coordinate of the operating point, the median frequency is set to the minimum operating frequency, and the above operation is repeated. If the vertical coordinate of the frequency conversion curve is greater than the vertical coordinate of the operating point, the median frequency is set to the maximum operating frequency, and the above operation is repeated. If the absolute difference between the vertical coordinate of the frequency conversion curve and the vertical coordinate of the operating point is less than a set value, the median frequency is obtained as the highest efficiency operating frequency of the electromechanical equipment.

[0020] A non-sensing intelligent control system for electromechanical equipment, characterized in that the non-sensing intelligent control method for electromechanical equipment is executed and applied, including: a data acquisition and monitoring system, a cloud storage system, a field terminal, and a remote terminal, wherein the data acquisition and monitoring system includes an acquisition system and a monitoring system, wherein the acquisition system is composed of an ultrasonic flow meter and a digital pressure gauge, wherein the ultrasonic flow meter is used to read the flow data of the electromechanical equipment, and the digital pressure gauge is used to read the pressure data of the electromechanical equipment;

[0021] The monitoring system is composed of a frequency converter, a temperature sensor, and a vibration measuring instrument. The frequency converter is connected to the electromechanical equipment and is used to control the electromechanical equipment and monitor the current, power, operating frequency, and speed of the electromechanical equipment in real time. The temperature sensor is installed at the head and tail of the electromechanical equipment. The vibration measuring instrument is used to monitor the vibration of the electromechanical equipment in real time;

[0022] The cloud storage system runs in the cloud and is used to store data;

[0023] The field terminal is connected to the data acquisition and monitoring system, the frequency converter communicates with the field terminal through the Modbus_TCP protocol, the temperature sensor communicates with the field terminal through the Modbus_RTU protocol, the field terminal is provided with a memory for storing data and a processor for performing calculations, the memory can upload data to the cloud storage system, the data preprocessing system converts pressure data into lift, uses vibration data to reduce noise and filter environmental noise, and cuts and compresses historical data generated by the collected equipment information over time;

[0024] According to the non-sensing intelligent control method for electromechanical equipment according to claim 1, the data analysis system includes an automatic optimization module, a prediction module, and an early warning module;

[0025] The prediction module is used to obtain predicted limit state data from the historical data of electromechanical equipment collected by the on-site terminal through the deep learning GRU algorithm;

[0026] The automatic optimization module is used to calculate the predicted limit state data to obtain the limit performance curve, and obtain the optimal operating frequency of the electromechanical equipment by reading the preset flow value of the electromechanical equipment and using the iterative approximation algorithm.

[0027] The early warning module takes the data of the prediction module and the vibration frequency of the electromechanical equipment as the decision basis, controls the on-site terminal to issue a braking signal in time, and sends abnormal information to the remote terminal.

[0028] The remote terminal can download data from the on-site terminal or the cloud storage system. The remote terminal is provided with a memory for storing data and a processor for performing calculations. The remote terminal can send operating point information to the on-site terminal. After receiving the operating point information, the on-site terminal adjusts the parameters of the inverter.

[0029] Preferably, the electromechanical equipment includes but is not limited to water pumps and central air conditioners.

[0030] Preferably, the remote terminal includes but is not limited to a computer, a mobile phone, and a tablet.

[0031] Beneficial effects of the present invention: The present invention uses an ultrasonic flow meter, a digital pressure gauge and a frequency converter to obtain basic data of the water pump, and stores it in a cloud storage system. The on-site terminal monitors the current, power and speed of the electromechanical equipment in real time, obtains the flow and energy consumption through machine learning analysis and forms a report; provides suggestions for optimizing the operating efficiency of the electromechanical equipment; analyzes abnormal fluctuations in current, power and speed to issue an alarm in time, and displays it on the terminal after analysis, thereby realizing a non-sensing intelligent control system for electromechanical equipment that integrates monitoring, analysis, early warning, control and visualization. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] For ease of explanation, the present invention is described in detail with reference to the following preferred embodiments and the accompanying drawings.

[0033] Figure 1 A schematic diagram of the system structure of a non-sensing intelligent control method and system for electromechanical equipment according to the present invention;

[0034] Figure 2 A system execution flow chart of a non-sensing intelligent control method and system for electromechanical equipment of the present invention;

[0035] Figure 3 The present invention is a flowchart for executing a method for non-sensing intelligent control of electromechanical equipment. DETAILED DESCRIPTION

[0036] In order to make the implementation purpose, technical scheme and features of the present invention application clearer, the technical scheme implemented by the present invention application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the examples of the present invention application, but not all implementation cases. The embodiments of the present invention application described and shown in the drawings here can be arranged and designed in different configurations.

[0037] Therefore, the detailed description of the embodiments provided in the invention application is not intended to limit the scope of the invention claimed in the present application, but merely represents a specific implementation case of the present application. Based on the implementation case in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of the invention application.

[0038] This implementation case provides a sensorless, intelligent electromechanical equipment sensorless intelligent control method. The solution provided by the embodiment of the present invention is described in detail below:

[0039] The collection system collects flow and pressure data of electromechanical equipment only once on site before the equipment is put into operation, and collects equipment information of electromechanical equipment in real time, the equipment information includes current, power data, operating frequency data, speed data, temperature data, and vibration data; the equipment information is preprocessed to generate a first data set, the preprocessing includes: converting pressure data into head, using vibration data to reduce noise and filter environmental noise, and cutting and compressing historical data generated by the collected equipment information over time;

[0040] Using the deep learning GRU algorithm, based on the first data set as historical data, the historical data is divided into fixed-length data as labels using a smoothing window, The historical data are combined after being lagged by 1 to 5 digits. Historical data is divided into fixed-length data using a smoothing window as input The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain a reset gate. The state information at the previous moment contains the previous features. The function of the reset gate is to control the state information at the previous moment to be written into the current state candidate set. The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain an update gate. The update gate controls the ratio of the state at the previous moment to the state candidate set at the current moment to obtain the current state information. The current moment output is obtained from the current moment state. The absolute error between the current moment label and the current moment output is used as back propagation for training. , , and Matrix, error convergence to get the model, take the lagged data as input, and get the predicted data;

[0041] The predicted data is combined with the historical data, and the combined data is used as the input of the matrix model, and the cycle is repeated until the preset length ends; the reset gate:

[0042]

[0043] The update gate:

[0044]

[0045] The current state candidate set:

[0046]

[0047] Current status:

[0048]

[0049] Output at the current moment:

[0050]

[0051] in, Input for the current moment, is the state of the previous moment, , and is a trainable matrix, is the activation function.

[0052] The method reads the flow value of the pre-set electromechanical equipment, collects the working frequency pressure data of the electromechanical equipment, converts the pressure data into the head, uses the orthogonal polynomial least squares fitting to obtain the maximum working frequency performance curve of the electromechanical equipment, takes the median of the maximum working frequency and the minimum working frequency, and obtains the frequency conversion curve of the median frequency according to the water pump similarity law, takes the maximum value point of the frequency conversion curve, if the horizontal coordinate of the maximum value point is less than the flow value, then the median frequency is set to the minimum working frequency, and the above operation is repeated. If the horizontal coordinate of the maximum value point is greater than the flow value, then the median frequency is set to the maximum working frequency, and the above operation is repeated. If the absolute difference between the horizontal coordinate of the maximum value point and the flow is less than the threshold value, then the median frequency is obtained as the working frequency with the highest efficiency of the electromechanical equipment.

[0053] The electromechanical equipment automatically sets the operating point manually or using predicted data, and the operating point includes flow data and head data. After the system obtains the operating point, the maximum operating frequency performance curve of the electromechanical equipment is obtained by using orthogonal polynomial least squares fitting, and the median of the maximum operating frequency and the minimum operating frequency is taken. According to the water pump similarity law, the frequency conversion curve of the median frequency can be obtained, and the horizontal coordinate of the operating point is substituted into the frequency conversion curve to obtain the vertical coordinate of the frequency conversion curve. If the vertical coordinate of the frequency conversion curve is less than the vertical coordinate of the operating point, the median frequency is set to the minimum operating frequency, and the above operation is repeated. If the vertical coordinate of the frequency conversion curve is greater than the vertical coordinate of the operating point, the median frequency is set to the maximum operating frequency, and the above operation is repeated. If the absolute difference between the vertical coordinate of the frequency conversion curve and the vertical coordinate of the operating point is less than a set value, the median frequency is obtained as the highest efficiency operating frequency of the electromechanical equipment.

[0054] A non-sensing intelligent control system for electromechanical equipment, characterized in that the non-sensing intelligent control method for electromechanical equipment is executed and applied, including: a data acquisition and monitoring system, a cloud storage system, a field terminal, and a remote terminal, wherein the data acquisition and monitoring system includes an acquisition system and a monitoring system, wherein the acquisition system is composed of an ultrasonic flow meter and a digital pressure gauge, wherein the ultrasonic flow meter is used to read the flow data of the electromechanical equipment, and the digital pressure gauge is used to read the pressure data of the electromechanical equipment;

[0055] The monitoring system is composed of a frequency converter, a temperature sensor, and a vibration measuring instrument. The frequency converter is connected to the electromechanical equipment and is used to control the electromechanical equipment and monitor the current, power, operating frequency, and speed of the electromechanical equipment in real time. The temperature sensor is installed at the head and tail of the electromechanical equipment. The vibration measuring instrument is used to monitor the vibration of the electromechanical equipment in real time;

[0056] The cloud storage system runs in the cloud and is used to store data;

[0057] The field terminal is connected to the data acquisition and monitoring system, the frequency converter communicates with the field terminal through the Modbus_TCP protocol, the temperature sensor communicates with the field terminal through the Modbus_RTU protocol, the field terminal is provided with a memory for storing data and a processor for performing calculations, the memory can upload data to the cloud storage system, the data preprocessing system converts pressure data into lift, uses vibration data to reduce noise and filter environmental noise, and cuts and compresses historical data generated by the collected equipment information over time;

[0058] According to the method for non-sensing intelligent control of electromechanical equipment according to claim 1, the data analysis system includes an automatic optimization module, a prediction module, and an early warning module.

[0059] The prediction module is used to obtain predicted limit state data from the historical data of electromechanical equipment collected by the on-site terminal through the deep learning GRU algorithm;

[0060] The automatic optimization module is used to calculate the predicted limit state data to obtain the limit performance curve, and obtain the optimal operating frequency of the electromechanical equipment by reading the preset flow value of the electromechanical equipment and using the iterative approximation algorithm.

[0061] The early warning module takes the data of the prediction module and the vibration frequency of the electromechanical equipment as the decision basis, controls the on-site terminal to issue a braking signal in time, and sends abnormal information to the remote terminal.

[0062] The data visualization system is used to display the analysis results of the prediction module.

[0063] The remote terminal can download data from the on-site terminal or the cloud storage system. The remote terminal is provided with a memory for storing data and a processor for performing calculations. The remote terminal can send operating point information to the on-site terminal. After receiving the operating point information, the on-site terminal adjusts the parameters of the inverter.

[0064] The electromechanical equipment includes but is not limited to water pumps and central air conditioners.

[0065] The remote terminal includes but is not limited to a computer, a mobile phone, and a tablet.

[0066] An embodiment of the present invention takes a water pump as an example and discloses a method and system for non-sensing intelligent control of electromechanical equipment. Figure 1 As shown, it includes data acquisition and monitoring system, cloud storage system, on-site terminal and remote terminal; among them,

[0067] The data acquisition monitoring system is connected to the water pump to monitor the status of the water pump in real time and transmit the status data to the data transmission system;

[0068] The data acquisition and monitoring system sends the received data to the cloud storage system and the on-site terminal;

[0069] The remote terminal downloads data from the cloud storage system, the data is pre-processed, analyzed, and finally visualized.

[0070] It is particularly noted that the sensorless intelligent control method and system for electromechanical equipment in the embodiments of the present invention are applicable to all water supply and drainage systems.

[0071] The data acquisition and monitoring system includes an acquisition system and a monitoring system.

[0072] The data acquisition system includes an ultrasonic flow meter and a digital pressure gauge. When the water pump motor is in rated operation, the ultrasonic flow meter and the digital pressure gauge are used to measure 20 sets of different flow values ​​and corresponding pressure data (pressures P1, P2, and P3 at the water pump inlet and outlet and user end). The monitoring system consists of a frequency converter, a temperature sensor, and a vibration measuring instrument. The frequency converter monitors the current, operating frequency, power, and speed in real time. The default monitoring time interval is 5 seconds each time, and the user can modify the monitoring time interval by himself.

[0073] It should be noted that in order to ensure the accuracy of the measurement, the data acquisition system can use 3 sets of ultrasonic flow meters and digital pressure gauges for measurement.

[0074] The data acquisition system and field terminals are installed at the pumping station. The field terminal supports wired or wireless data transmission. The Ethernet interface is used to implement the Modbus TCP communication protocol to transmit current, operating frequency, power and speed, temperature, and vibration data to the gateway. The data is then transmitted to the cloud storage system via wireless methods such as 4G, 5G, and WIFI. The staff downloads the data to the remote terminal, which displays the data on the terminal after the remote terminal completes the calculation.

[0075] It should be noted that users can build their own cloud storage system or rent a third-party cloud server. For security reasons, users can also choose to transfer data directly to a remote terminal without uploading the data to a cloud storage system.

[0076] On-site terminals and remote terminals include data preprocessing system, data analysis system, and data visualization system;

[0077] The data preprocessing system will segment and extract the received data to obtain the target data, which is text data. The target data will be converted into integers. Some data also need to be converted. For example, the data collected by the pressure gauge is converted into the head of the water pump, and the collected ultrasonic waves need to remove environmental noise; convert the pressure data into head, use the vibration measuring instrument to monitor the vibration data of the water pump motor to reduce the noise and filter the impact of the environment, and cut and compress the historical data generated by the collected equipment information over time;

[0078] The data analysis system includes automatic optimization module, prediction module and early warning module;

[0079] The prediction module uses ultrasonic flowmeter and digital pressure gauge to collect and convert flow data and head data. Since it is impossible to operate the pump at the limit state and collect the data of the pump at the limit state, it is impossible to fit the real pump performance curve. Therefore, the prediction module is very necessary.

[0080] Collecting flow and pressure data of a water pump and collecting equipment information of the water pump in real time, the equipment information including current, power data, operating frequency data, speed data, temperature data, and vibration data; preprocessing the equipment information to generate a first data set, the preprocessing including: converting pressure data into head, using vibration data to reduce noise and filter environmental noise, and cutting and compressing historical data generated by the collected equipment information over time;

[0081] Using the deep learning GRU algorithm, based on the first data set as historical data, the historical data is divided into fixed-length data as labels using a smoothing window, The historical data are combined after being lagged by 1 to 5 digits. Historical data is divided into fixed-length data using a smoothing window as input The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain a reset gate. The state information at the previous moment contains the previous features. The function of the reset gate is to control the state information at the previous moment to be written into the current state candidate set. The input at the current moment is combined with the state at the previous moment and then undergoes linear and nonlinear operations to obtain an update gate. The update gate controls the ratio of the state at the previous moment to the state candidate set at the current moment to obtain the current state information. The current moment output is obtained from the current moment state. The absolute error between the current moment label and the current moment output is used as back propagation for training. , , and Matrix, error convergence to get the model, take the lagged data as input, and get the predicted data;

[0082] The automatic optimization module automatically finds the best operating frequency according to the analysis results of the prediction module and transmits the operating frequency to the inverter, thereby achieving high efficiency and energy saving;

[0083] The data visualization system is used to display the analysis results of the prediction module.

[0084] The predicted data is combined with the historical data, and the combined data is used as the input of the matrix model, and the cycle is repeated until the preset length ends; the reset gate:

[0085]

[0086] The update gate:

[0087]

[0088] The current state candidate set:

[0089]

[0090] Current status:

[0091]

[0092] Output at the current moment:

[0093]

[0094] in, Input for the current moment, is the state at the previous moment, , , and is a trainable matrix, is the activation function.

[0095] After the flow data and pressure data of the water pump are converted into the limit performance curve of the water pump through the pressure and head conversion formula,

[0096] Read the preset flow value of the water pump, use the iterative approximation algorithm, set two frequencies, variable 1 of the operating frequency, and variable 2 of 0 Hz, obtain the performance curve of variable 2 according to the similarity law, calculate the intersection of the performance curve and the limit performance curve, if the preset flow value is greater than the horizontal coordinate of the intersection, calculate the frequency at the intersection and assign it to variable 2, otherwise calculate the frequency at the intersection and assign it to variable 1, and continue to iterate until the difference between the absolute value of the flow value and the horizontal coordinate of the intersection is less than the set threshold.

[0097] After the flow data and pressure data of the water pump are converted into the limit performance curve of the water pump through the pressure and head conversion formula,

[0098] The pump can be set to the operating point manually or automatically using the predicted data. After the system obtains the operating point, it is based on the pump performance data read on site. Use iterative approximation algorithm to obtain the operating point performance curve within a certain error. Get the operating frequency of this operating point.

[0099] The early warning module uses the analysis results of the prediction module and the abnormal vibration of the water pump as the basis for decision-making. The controller promptly sends a braking signal and sends abnormal information to the staff for immediate maintenance. The life of the water pump is determined based on the analysis results, and preparations for replacement are made in advance.

[0100] The system has fault warning and has a certain fault tolerance for faults, and will not issue an immediate stop command to the water pump. For example, if a sensor in the system is damaged, the on-site terminal will only send a warning message to inform the staff to replace the sensor. A single sensor monitors a time series. If a single time series anomaly is used as a control strategy, it will affect the normal operation of the system, so multivariate time series must be considered. The existing technology does not consider the relationship between multivariate time series, such as wavelet analysis, ARIMA, etc. The present invention uses a graph attention network to implement it, which combines graph neural networks and attention mechanisms. The multivariate time series is sliced ​​using a sliding window to obtain a matrix. The elements of the first column of the matrix are used as nodes, and the correlation (matrix) between each node is calculated through self-attention, and the same operation is performed on the other columns of the matrix. The correlation matrix is ​​a symmetric matrix. If there are more than two values ​​in the correlation matrix that are lower than the set threshold (the label is a binary sequence, normal is 0, abnormal is 1, the label length is consistent with the number of time series slices, and the threshold is obtained by back propagation), then it means that more than one sensor data is abnormal, and the water pump is abnormal.

[0101] Compute Node and nodes Relevance:

[0102]

[0103] in, , Node and nodes , is a trainable matrix, is the activation function, is the total number of nodes.

[0104] The data visualization system dynamically displays the collected data and statically displays the analysis results, making the status of the water pump clear at a glance.

Claims

1. A non-sensing intelligent control method for electromechanical equipment, characterized in that: Collect the flow and pressure data of primary electromechanical equipment, and collect the equipment information of electromechanical equipment in real time, the equipment information includes current, power data, operating frequency data, speed data, temperature data, and vibration data; Preprocessing the equipment information to generate a first data set, the preprocessing comprising: converting the pressure data into lift, using vibration data to reduce noise and filter environmental noise, and cutting and compressing historical data generated by the collected equipment information over time; Using the deep learning GRU algorithm, the first data set is used as historical data, the historical data is divided into fixed-length data using a smoothing window as a label, the historical data is lagged by 1 to 5 bits and then merged, the lagged historical data is divided into fixed-length data using a smoothing window as input, the input at the current moment is merged with the state at the previous moment, and then linear and nonlinear operations are performed to obtain a reset gate. The state information at the previous moment contains the previous features, and the function of the reset gate is to control the state information at the previous moment to be written into the current state candidate set; The input at the current moment is combined with the state at the previous moment, and then undergoes linear and nonlinear operations to obtain the update gate. The update gate controls the ratio of the state at the previous moment to the candidate set of the state at the current moment to obtain the current state information; The current state is used to get the current output. The absolute error between the current label and the current output is used as the back propagation for training. , , and Matrix, error convergence to get the model, take the lagged data as input, and get the predicted data; The predicted data is combined with the historical data, and the combined data is used as the input of the matrix model, and the cycle is repeated until the preset length ends; the reset gate: , The update gate: , The current state candidate set: , Current status: , Output at the current moment: , in, Input for the current moment, is the state at the previous moment, , , and is a trainable matrix, is an activation function; the flow value of the electromechanical equipment set in advance is read, the working frequency pressure data of the electromechanical equipment is collected, the pressure data is converted into the head, the maximum working frequency performance curve of the electromechanical equipment is obtained by using the orthogonal polynomial least squares fitting, the median of the maximum working frequency and the minimum working frequency is taken, and the frequency conversion curve of the median frequency can be obtained according to the water pump similarity law, the maximum value point of the frequency conversion curve is taken, if the horizontal coordinate of the maximum value point is less than the flow value, the median frequency is set to the minimum working frequency, and the above operation is repeated. If the horizontal coordinate of the maximum value point is greater than the flow value, the median frequency is set to the maximum working frequency, and the above operation is repeated. If the absolute difference between the horizontal coordinate of the maximum value point and the flow is less than the threshold value, the median frequency is obtained as the working frequency with the highest efficiency of the electromechanical equipment.

2. According to the method for non-sensing intelligent control of electromechanical equipment described in claim 1, the electromechanical equipment automatically sets the operating point manually or using predicted data, and the operating point includes flow data and head data. After the system obtains the operating point, the maximum operating frequency performance curve of the electromechanical equipment is obtained by using orthogonal polynomial least squares fitting, and the median of the maximum operating frequency and the minimum operating frequency is taken. According to the water pump similarity law, the frequency conversion curve of the median frequency can be obtained, and the horizontal coordinate of the operating point is substituted into the frequency conversion curve to obtain the vertical coordinate of the frequency conversion curve. If the vertical coordinate of the frequency conversion curve is less than the vertical coordinate of the operating point, the median frequency is set to the minimum operating frequency, and the above operation is repeated. If the vertical coordinate of the frequency conversion curve is greater than the vertical coordinate of the operating point, the median frequency is set to the maximum operating frequency, and the above operation is repeated. If the absolute difference between the vertical coordinate of the frequency conversion curve and the vertical coordinate of the operating point is less than a set value, the median frequency is obtained as the highest efficiency operating frequency of the electromechanical equipment.

3. A non-sensing intelligent control system for electromechanical equipment, characterized in that: Execution and application of the method as claimed in any one of claims 1 to 2, comprising: a data acquisition and monitoring system, a cloud storage system, a field terminal, and a remote terminal, wherein the data acquisition and monitoring system comprises an acquisition system and a monitoring system, wherein the acquisition system is composed of an ultrasonic flow meter and a digital pressure gauge, wherein the ultrasonic flow meter is used to read the flow data of the electromechanical equipment, and the digital pressure gauge is used to read the pressure data of the electromechanical equipment; The monitoring system is composed of a frequency converter, a temperature sensor, and a vibration measuring instrument. The frequency converter is connected to the electromechanical equipment and is used to control the electromechanical equipment and monitor the current, power, operating frequency, and speed of the electromechanical equipment in real time. The temperature sensor is installed at the head and tail of the electromechanical equipment. Vibration measuring instruments are used to monitor the vibration of electromechanical equipment in real time; The cloud storage system runs in the cloud and is used to store data; The field terminal includes a data preprocessing system, a data analysis system, and a data visualization system. The data preprocessing system is connected to the data acquisition and monitoring system. The frequency converter communicates with the field terminal through the Modbus_TCP protocol. The temperature sensor communicates with the field terminal through the Modbus_RTU protocol. The field terminal is provided with a memory for storing data and a processor for performing calculations. The memory can upload data to a cloud storage system. The data preprocessing system converts pressure data into lift, uses vibration data to reduce noise and filter environmental noise, and cuts and compresses historical data generated by the collected equipment information over time; The data analysis system includes an automatic optimization module, a prediction module, and an early warning module; The prediction module is used to obtain predicted limit state data from the historical data of electromechanical equipment collected by the on-site terminal through the deep learning GRU algorithm; The automatic optimization module is used to calculate the predicted limit state data to obtain the limit performance curve, and obtain the optimal operating frequency of the electromechanical equipment by reading the preset flow value of the electromechanical equipment and using the iterative approximation algorithm; The early warning module uses the data of the prediction module and the vibration frequency of the electromechanical equipment as the basis for decision-making, controls the on-site terminal to issue a braking signal in time, and sends abnormal information to the remote terminal; The data visualization system is used to display the analysis results of the prediction module; The remote terminal can download data from the on-site terminal or the cloud storage system. The remote terminal is provided with a memory for storing data and a processor for performing calculations. The remote terminal can send operating point information to the on-site terminal. After receiving the operating point information, the on-site terminal adjusts the parameters of the inverter.

4. A non-sensing intelligent control system for electromechanical equipment according to claim 3, wherein the electromechanical equipment includes but is not limited to water pumps and central air conditioners.

5. According to the non-sensing intelligent control system for electromechanical equipment as described in claim 3, the remote terminal includes but is not limited to a computer, a mobile phone, and a tablet.

Images