Intelligent shower head system capable of saving water

Abstract

The invention relates to the field of shower head systems, and discloses a water-saving intelligent shower head system which comprises a shower head body and a control box, the shower head is provided with an electromagnetic valve, a temperature sensor and a flow sensor, the electromagnetic valve is used for controlling a water flow switch of the shower head, and the flow sensor is used for measuring the flow of water flowing to the shower head; a heater, a controller, a pump body and a Wi F i receiving antenna are arranged in the control box, the pump body is used for pumping water back into the water heater when the temperature of water transmitted to the shower head by the water heater is lower than a preset value, the heater is used for heating water flowing to the shower head from the water heater, and the Wi F i receiving antenna is used for receiving a signal of the Wi F i transmitting antenna; the controller recognizes the activity of the human body in the bathroom based on the Wi F i channel state information and the water flow received by the Wi F i receiving antenna in combination with the convolutional neural network. Cold water sprayed out by the shower head in advance is pumped back into the water heater, water waste is avoided, human body activity is recognized based on the Wi F i signal, and protection and early warning are conducted on the elderly living alone.

Description

Technical Field

[0001] The present invention relates to the field of shower systems, and particularly to an intelligent shower system for water conservation. Background Art

[0002] Existing intelligent shower systems usually control the shower water flow through water flow and temperature sensors to ensure the comfort of users during use. However, most of the existing technologies do not consider the hot water waiting time, and the waiting time for hot water depends on three factors: the distance from the water heater, the pipe diameter, and the flow rate. The influence of distance is very obvious. If the distance that the hot water needs to reach is longer, the required time will be longer. During the waiting process, users can only waste the cold water discharged first, which is not conducive to water conservation. Also, due to the housing cost limitation in big cities, the bathroom space is generally small. When users take a direct shower or cannot avoid it in time, the cold water sprayed by the shower head is likely to cause physical discomfort.

[0003] In addition, the bathroom is a high-risk area for the elderly living alone. Due to the slippery floor in the bathroom and the inconvenient movement of the elderly, they are prone to falling, or due to the stimulation of cold water and hot water, the elderly may suddenly get sick. In terms of home monitoring in the existing technologies, it mostly relies on cameras, but cameras are not suitable for bathrooms; some of the existing technologies use ultrasonic sensors, but ultrasonic transmission is easily interfered. That is, when the shower head is turned on, the water flow sprayed by the shower head and the water mist diffused in the bathroom will interfere with the ultrasonic sensor. Summary of the Invention

[0004] In view of the above technical problems, the present invention provides an intelligent shower system for water conservation.

[0005] Other features and advantages of the present disclosure will become apparent from the following detailed description, or will be partially learned through the practice of the present disclosure.

[0006] An intelligent shower system for water conservation includes a shower head body and a control box. The control box is connected between the shower head and the water heater, wherein:

[0007] The shower head is provided with a solenoid valve, a temperature sensor, and a flow sensor. The temperature sensor is used to sense the water temperature inside the shower head. The solenoid valve is used to control the water flow switch of the shower head. The flow sensor is used to measure the water flow rate flowing to the shower head;

[0008] A heater, a controller, a pump body, and a Wi-Fi receiving antenna are provided inside the control box. The pump body is connected to the water inlet pipe and the water outlet pipe of the water heater and the shower head. The heater is arranged between the connection pipes of the water heater and the shower head. The pump body is used to pump the water back to the water heater when the temperature of the water transmitted from the water heater to the shower head is lower than a preset value. The heater is used to heat the water flowing from the water heater to the shower head. The Wi-Fi receiving antenna is used to receive the signals of one or more Wi-Fi transmitting antennas at a fixed position. Based on the fuzzy control algorithm, the controller takes the data of the temperature sensor and the data of the flow sensor as input variables in real time, and takes the first control value of the heater as the output variable, so that the water outlet temperature of the shower head tends to the set value. The first control value is voltage or duty ratio. The controller is also used to capture the change in multipath signal propagation caused by human movement based on the Wi-Fi channel status information received by the Wi-Fi receiving antenna and the water flow rate, and combine it with a convolutional neural network to identify the activities of the human body in the bathroom.

[0009] Furthermore, the controller is also used to take the ambient temperature and ambient humidity measured by the ambient sensor as the input variables of the fuzzy control algorithm based on the ambient sensor arranged in the bathroom, cooperate with the data of the temperature sensor and the data of the flow sensor, and obtain the second control value of the solenoid valve and the first control value of the heater based on the preset fuzzy inference rules.

[0010] Furthermore, obtaining the first control value and the second control value based on the preset fuzzy inference rules includes:

[0011] Presetting the data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity into multiple fuzzy sets with different gradients;

[0012] Fuzzifying the data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity obtained in real time into the corresponding fuzzy sets, and performing logical reasoning and defuzzification on the fuzzified data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity based on the fuzzy inference rules to obtain the first control value and the second control value. The fuzzy inference rules at least include: when the ambient humidity is fuzzified into a high-gradient fuzzy set, the output second control value is a low-gradient value.

[0013] Furthermore, when the controller identifies the activities of the human body in the bathroom, it includes:

[0014] Preprocess the Wi-Fi channel state information, reconstruct it into the input matrix shape of the convolutional neural network, perform denoising processing on the reconstructed Wi-Fi channel state information based on the wavelet denoising method, and store it as a set of images, where each pixel point in the image represents the absolute value of a specified subcarrier at a specified antenna and time point;

[0015] Based on the convolutional neural network, classify the input images into human activity types, record the duration of the activity, and issue an alarm when the duration of a specified activity exceeds the threshold.

[0016] Furthermore, the convolutional neural network includes:

[0017] Three 2D convolutional layers;

[0018] There are Dropout layers after the first and third 2D convolutional layers;

[0019] After performing convolutional operations in each 2D convolutional layer, use max pooling, batch normalization, and the Leaky ReLU activation function;

[0020] The final output layer is a fully connected layer with 160 neurons, the activation function is Sigmoid, and the human activity type classification is performed by the Softmax classification layer.

[0021] Furthermore, when obtaining the classification of the human activity type, binarize the water flow rate, and based on the binarized water flow rate, weight the classification probabilities of each human activity type output by the convolutional neural network, and take the human activity type with the highest probability as the final human activity type.

[0022] Furthermore, the human activity types at least include:

[0023] Undressing;

[0024] Drying the body;

[0025] Static;

[0026] Washing the head and face in an excited state;

[0027] Washing the body and legs in an excited state;

[0028] Washing the head and face in a slow state;

[0029] Washing the body and legs in a slow state;

[0030] Falling down.

[0031] Further, when reconstructing the Wi-Fi channel state information into the shape of an input matrix, it includes:

[0032] Process the 4D Wi-Fi channel state information into a 3D input matrix. The Wi-Fi channel state information is expressed as n*m*f*t, where n is the Wi-Fi receiving antenna, m is the Wi-Fi transmitting antenna, f is the subcarrier of different frequencies for data transmission between the Wi-Fi receiving antenna and the Wi-Fi transmitting antenna, and t is the number of time steps. The 3D input matrix is expressed as H*W*D, where H and W are the height and width of the matrix respectively, and D is the depth. During reconstruction, set H = n*m*f, W = t, and D = 1, so that the subcarrier data of each pair of the Wi-Fi receiving antenna and the Wi-Fi transmitting antenna are stacked in sequence.

[0033] The technical solution of the present disclosure has the following beneficial effects:

[0034] Based on the settings of the pump body, heater, and temperature sensor, the cold water sprayed out by the shower head is pumped back into the water heater. On the one hand, it avoids wasting water, and on the other hand, it avoids discomfort caused by cold water to the human body; based on the fuzzy control algorithm to control the heating of the heater, the outlet water temperature of the shower head is more in line with the actual set temperature, providing a more comfortable experience;

[0035] Based on the Wi-Fi receiving antenna, Wi-Fi transmitting antenna, and convolutional neural network, using the strong penetration and anti-interference of Wi-Fi signals, classify the types of human activities to obtain the activity situation of the elderly in the bathroom and prevent risks; Wi-Fi devices already widely exist in daily life, and the detection can be completed using existing routers and receivers without the need for additional hardware deployment. Moreover, Wi-Fi device components can detect human activities and can also conduct data communication, enabling the controller to communicate with the outside world for alarms and other data analysis. Description of the Drawings

[0036] Figure 1 It is a structural block diagram of an intelligent shower system for saving water in this specification.

[0037] Among them, reference numerals: 100, shower head; 101, solenoid valve; 102, temperature sensor; 103, flow sensor; 200, control box; 201, heater; 202, controller; 203, pump body; 204, Wi-Fi receiving antenna; 205, Wi-Fi transmitting antenna; 206, environmental sensor. Detailed Embodiments

[0038] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. The features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present disclosure. However, those skilled in the art will realize that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be used. In other cases, well-known technical solutions are not shown or described in detail to avoid obscuring the various aspects of the present disclosure.

[0039] In addition, the accompanying drawings are only schematic illustrations of the present disclosure. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0040] As Figure 1 shown, the embodiment of the present specification provides an intelligent shower head system for saving water, including a shower head 100 body and a control box 200, where the control box 200 is connected between the shower head 100 and the water heater, and wherein:

[0041] The shower head 100 is provided with a solenoid valve 101, a temperature sensor 102, and a flow sensor 103. The temperature sensor 102 is used to sense the water temperature inside the shower head 100, the solenoid valve 101 is used to control the water flow switch of the shower head 100, and the flow sensor 103 is used to measure the water flow rate flowing into the shower head 100.

[0042] A heater 201, a controller 202, a pump body 203, and a Wi-Fi receiving antenna 204 are provided inside the control box 200. The pump body 203 is connected to the water inlet pipe and the water outlet pipe of the water heater and the shower head 100. The heater 201 is disposed between the connection pipe of the water heater and the shower head 100. The pump body 203 is configured to pump water back to the water heater when the temperature of the water transmitted from the water heater to the shower head 100 is lower than a preset value. The heater 201 is configured to heat the water flowing from the water heater to the shower head 100. The Wi-Fi receiving antenna 204 is configured to receive signals from one or more Wi-Fi transmitting antennas 205 at a fixed position. The controller 202, based on a fuzzy control algorithm, takes the data of the temperature sensor 102 and the data of the flow sensor 103 as input variables in real time, and takes the first control value of the heater 201 as an output variable, so that the water outlet temperature of the shower head 100 tends to a set value. The first control value is voltage or duty ratio. The controller 202 is further configured to capture the change in multipath signal propagation caused by human movement based on the Wi-Fi channel status information received by the Wi-Fi receiving antenna 204 and the water flow rate, and combine a convolutional neural network to identify the activities of a human body in the bathroom.

[0043] Among them, the temperature sensor 102 is used to measure the water temperature flowing to the shower head to help determine whether the water temperature is within a predetermined range; the flow sensor 103 is used to measure the water flow rate flowing to the shower head to provide input data of the water flow; the solenoid valve 101 is used to control the on-off of the water flow. When the water temperature does not reach the preset value, the solenoid valve 101 can close the water flow to prevent water waste and prevent cold water from directly spraying on people; important components such as a pump body 203, a heater 201, and a controller 202 are installed inside the control box 200. The pump body 203 ensures that when the water temperature from the central water heater is lower than the preset value, the water is pumped back to the water heater for heating. The heater 201 is installed between the water heater and the shower and is responsible for heating the water temperature to a suitable temperature and then sending it to the shower head 100; the controller 202 receives data from the temperature sensor 102 (detecting the water temperature) and the flow sensor 103 (measuring the water flow rate), and calculates a suitable control voltage signal using a fuzzy logic algorithm to adjust the working state of the heater 201. The goal is to ensure that the water temperature of the shower head outlet remains within the set temperature range, usually 38.5°C to 43.5°C. The output of the controller 202 is the voltage or duty cycle of the heater 201. During operation, when cold water flows into the system, the temperature sensor 102 sends a signal to the controller 202 to tell it whether the water temperature is too low. If the water temperature is low, the pump body 203 will return the water to the heater 201 for heating; if the water temperature is still lower than the set value, the fuzzy logic controller 202 will adjust the voltage of the heater 201 to heat the water temperature to the correct temperature. Once the water temperature reaches the set temperature, the water flow will continue to flow to the shower head 100; during this process, the fuzzy logic controller 202 will continuously adjust the working state of the heater 201 according to the real-time input data to ensure that the water temperature always remains within the set range. In some water heaters heated by natural gas, when there is water usage in other places in addition to the water usage of the shower head 100, the water flow rate will become smaller and the water temperature flowing out of the water heater will become higher. At this time, the solenoid valve 101 should be closed to avoid the user's physical discomfort caused by the water with too high a temperature.

[0044] In Wi-Fi-based human activity detection, the Wi-Fi receiving antenna 204 is installed at a fixed position in the bathroom and is responsible for receiving signals from multiple Wi-Fi transmitting antennas 205, such as those installed inside or outside the control box 200. The signals transmitted by the Wi-Fi antennas in the environment change with the presence, position, and movement of different objects; the Wi-Fi transmitting antenna 205 is responsible for transmitting Wi-Fi signals and can be installed in locations such as the living room and guest rooms and can exist in the form of a router. The Wi-Fi signals are affected by surrounding objects (such as walls, furniture, or the human body) during propagation, resulting in a multipath propagation phenomenon. This means that the Wi-Fi signals will be transmitted through multiple paths, possibly directly, or they may be reflected, refracted, or bypass objects, thereby causing changes in the signal strength, phase, and amplitude; the Wi-Fi channel state information refers to the information about signal strength, phase, frequency, etc. obtained through the propagation process of the Wi-Fi signals, and these information reflect the changes in the surrounding environment, including the movement of the human body; in addition to the Wi-Fi channel state information, the present invention also combines water flow data. The water flow data is usually obtained through the flow sensor 103 and represents the flow rate of the water flowing through the system. These two data sources are used for comprehensive analysis. For example, when a person is moving in the bathroom, the propagation mode of the Wi-Fi signals will change due to the presence and movement of the human body. At the same time, the change in the water flow rate also reflects the impact of human activities, and the change in the water flow rate is related to the start, duration, and end of the shower activity. These associations can be identified through a convolutional neural network, which is used to process and analyze the Wi-Fi channel state information received from the Wi-Fi receiving antenna 204 and the water flow data. Through training, it can learn to identify different signal patterns, thereby capturing the signal changes caused by human movement, and then the activities of the user in the bathroom can be obtained, such as undressing, drying the body, standing still, washing the head and face in an excited state, washing the body and legs in an excited state, washing the head and face in a slow state, washing the body and legs in a slow state, falling, etc. These identifications are particularly important for the elderly. For example, if the elderly have continuous slow activities, it may indicate the exacerbation of a certain disease.

[0045] In one embodiment, the controller 202 is further configured to, based on the environmental sensor 206 disposed in the bathroom, use the environmental temperature and environmental humidity measured by the environmental sensor 206 as the input variables of the fuzzy control algorithm, cooperate with the data of the temperature sensor 102 and the data of the flow sensor 103, and obtain the second control value of the solenoid valve 101 and the first control value of the heater 201 based on a preset fuzzy inference rule.

[0046] Supplementary, obtaining the first control value and the second control value based on preset fuzzy inference rules includes: dividing the data of the temperature sensor 102, the data of the flow sensor 103, the ambient temperature, and the ambient humidity into multiple fuzzy sets with different gradients; fuzzifying the data of the temperature sensor 102, the data of the flow sensor 103, the ambient temperature, and the ambient humidity obtained in real time into the corresponding fuzzy sets, and based on the fuzzy inference rules, performing logical inference and defuzzification on the fuzzified data of the temperature sensor 102, the data of the flow sensor 103, the ambient temperature, and the ambient humidity to obtain the first control value and the second control value. The fuzzy inference rules at least include: when the ambient humidity is fuzzified into a high-gradient fuzzy set, the output second control value is a low-gradient value.

[0047] Among them, the purpose of the fuzzy control algorithm is to reasonably adjust the working states of the heater and the solenoid valve according to the environmental and water flow conditions. The core of fuzzy control is to generate control values through the processes of fuzzification and defuzzification based on fuzzy inference rules. In defining the fuzzy sets, the input variables (ambient temperature, ambient humidity, water temperature, and flow rate) are divided into multiple fuzzy sets with different gradients. For example, the temperature Tenv can be: low temperature (L), medium temperature (M), high temperature (H), the humidity Henv can be: low humidity (L), moderate humidity (M), high humidity (H), and the water temperature Twater and the water flow rate Fflow will also be divided into different fuzzy sets respectively. The sensor data obtained in real time will be mapped into these fuzzy sets. Assume:

[0048] Tenv = 25°C;

[0049] Henv = 60%;

[0050] Twater = 38°C;

[0051] Fflow = 5 L / min;

[0052] According to the predefined fuzzy sets, the system will convert these real-time data into fuzzy values. For example, for an ambient temperature of 25°C, it can be mapped into the "medium temperature" set, and for a humidity of 60%, it is mapped into the "moderate humidity" set, and then the fuzzy inference rules are used to process the fuzzified input variables; for example:

[0053] Rule 1: If Twater is "low temperature" and Fflow is "large flow rate", then the control value of the heater should be increased;

[0054] Rule 2: If Tenv is "low temperature" and Henv is "low humidity", then the control value of the solenoid valve is high.

[0055] Then, defuzzification is performed. The result generated by fuzzy inference needs to be converted into a specific control output. For example: the second control value of the solenoid valve 101: controls the water flow rate to ensure the adequacy of the water flow; the first control value of the heater 201: controls the water temperature. The water temperature is adjusted by the output voltage or duty cycle of the heater 201.

[0056] Additionally, the fuzzy membership function range of the temperature input is selected as 10 - 44 °C. This input temperature range covers the expected temperature signal values of the shower water that the shower head can receive from the moment the faucet is turned on to the moment when hot water is supplied from the central heater; Flow rate input variable: the fuzzy membership function range of the flow rate input is adjusted to be between a minimum of 0.87 gpm and a maximum of 2 gpm; First control value output variable: The heater can be an in-line heater, single-phase, 220V. Therefore, the output membership function range is selected to be between a minimum of 0 volts and a maximum of 220 volts.

[0057] In one embodiment, when the controller 202 identifies the activities of a human body in the bathroom, it includes: preprocessing the Wi-Fi channel state information, reconstructing it into the input matrix shape of the convolutional neural network, performing denoising processing on the reconstructed Wi-Fi channel state information based on the wavelet denoising method, and storing it as a set of images. Each pixel point in the image represents the absolute value of a specified subcarrier at a specified antenna and time point; based on the convolutional neural network, classifying the type of human activity for the input image, and recording the duration of the activity. When the duration of a specified activity exceeds the threshold, an alarm is issued.

[0058] Specifically, the 4D Wi-Fi channel state information is processed into a 3D input matrix. The Wi-Fi channel state information is represented as n*m*f*t, where n is the Wi-Fi receiving antenna, m is the Wi-Fi transmitting antenna, f is the subcarrier of different frequencies for data transmission between the Wi-Fi receiving antenna and the Wi-Fi transmitting antenna, and t is the time step. The 3D input matrix is represented as H*W*D, where H and W are the height and width of the matrix respectively, and D is the depth. During reconstruction, H = n*m*f, W = t, and D = 1, such that the subcarrier data of each pair of the Wi-Fi receiving antenna and the Wi-Fi transmitting antenna is stacked in sequence.

[0059] Among them, in the preprocessing stage, the first step is to reconstruct the channel state information. The channel state information is initially a four-dimensional matrix (3*3*30*t), which is the performance of a 3*3 Wi-Fi antenna and its 30 subcarriers at time t. This four-dimensional matrix is transformed into a three-dimensional matrix required for the input layer of the convolutional neural network, with the shape of a 3D matrix (H*W*D). The data is flattened into a 3D shape with a depth of 1 (D = 1), that is, 270*t*1, where 3*3*30 = 270. This means that for each pair of antennas, the subcarriers are stacked together (for example, the first 30 rows represent receiving antenna 1 and transmitting antenna 1, the next 30 rows represent receiving antenna 1 and transmitting antenna 2, and so on). Next, the wavelet denoising method is used to remove the noise in the signal, and the denoised data is stored as a set of images, where each pixel in the image represents the absolute value (H) of a specific subcarrier at a specified antenna pair and time point. The input images can be classified using a convolutional neural network, which can retain the spatial and structural information of the channel state information. The convolutional neural network consists of three two-dimensional convolutional layers, with a dropout layer of 0.6 after the first and third layers. A max pooling, batch normalization, and activation layer with leaky ReLU are applied after each layer. Finally, the output is flattened and passed through a dense layer with 160 neurons (using the sigmoid activation function), and then reaches the final dense layer for classification (softmax classification layer).

[0060] In one embodiment, when obtaining the classification of the human activity type, the water flow rate is binarized, and based on the binarized water flow rate, the classification probability of each human activity type output by the convolutional neural network is weighted, and the human activity type with the highest probability is used as the final human activity type.

[0061] Among them, to simplify the processing, the water flow rate data can be transformed into a binary form, that is, binarized. Generally, the water flow rate can be divided into two categories: with water flow rate and without water flow rate. In the case of with water flow rate, it corresponds to certain specific activities, such as taking a bath; in the case of without water flow rate, it corresponds to no bathing activity.

[0062] The convolutional neural network outputs the classification probability of each activity type, indicating the possibility that the network considers a certain activity type. Assume that the CNN has multiple activity categories, and each activity type has a probability, indicating the prediction confidence of the network for that activity type. Next, based on the binarized water flow rate, the system weights the classification probability of each activity type. The purpose of weighting is that when there is water flow rate or no water flow rate, the corresponding activity type probability is strengthened or weakened. For example, if there is water flow rate, the probability of the activity type related to taking a bath or other water-using activities is enhanced.

[0063] Beneficial effects:

[0064] Based on the settings of the pump body 203, the heater 201, and the temperature sensor 102, the cold water first ejected by the shower head 100 is pumped back into the water heater. On the one hand, it avoids wasting water, and on the other hand, it avoids the discomfort caused by cold water to the human body. Based on the fuzzy control algorithm, the heating of the heater 201 is controlled to make the outlet water temperature of the shower head 100 more in line with the actual set temperature, providing a more comfortable experience.

[0065] Based on the Wi-Fi receiving antenna 204, the Wi-Fi transmitting antenna 205, and the convolutional neural network, using the strong penetration and anti-interference of the Wi-Fi signal, the types of human activities are classified to obtain the activity situation of the elderly in the bathroom to prevent risks. Wi-Fi devices already widely exist in daily life. The detection can be completed using an existing router and receiver without the need for additional hardware deployment. Moreover, the Wi-Fi device components can detect human activities and also perform data communication, enabling the controller 202 to communicate with the outside world for alarms and other data analysis.

[0066] Finally, 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention. In addition, those skilled in the art can understand that although some embodiments described herein include some features included in other embodiments but not other features, the combination of the features of different embodiments means that it is within the scope of the present invention and forms different embodiments. For example, in the above claims, any one of the claimed embodiments can be used in any combination. The information disclosed in this background art section is only intended to deepen the understanding of the overall background art of the present invention, and should not be regarded as an admission or any form of suggestion that this information constitutes prior art known to those skilled in the art.

Claims

1. A water-saving intelligent shower system, characterized in that , including a shower body and a control box, wherein the control box is connected between the shower and the water heater, wherein: The shower head is provided with a solenoid valve, a temperature sensor, and a flow sensor. The temperature sensor is used to sense the water temperature inside the shower head. The solenoid valve is used to control the water flow switch of the shower head. The flow sensor is used to measure the water flow to the shower head. The control box is provided with a heater, a controller, a pump body, and a WiFi receiving antenna. The pump body is connected to the water inlet pipe and the water outlet pipe of the water heater and the shower head. The heater is provided between the connecting pipes of the water heater and the shower head. The pump body is used to pump water back to the water heater when the temperature of the water transmitted from the water heater to the shower head is lower than a preset value. The heater is used to heat the water flowing from the water heater to the shower head. The WiFi receiving antenna is used to receive signals from one or more WiFi transmitting antennas at fixed positions. The controller uses the data of the temperature sensor and the data of the flow sensor as input variables in real time based on a fuzzy control algorithm, and uses the first control value of the heater as an output variable, so that the outlet water temperature of the shower head tends to a set value, and the first control value is a voltage or a duty cycle. The controller is also used to capture the multi-path signal propagation changes caused by human motion based on the WiFi channel status information and the water flow received by the WiFi receiving antenna, combined with a convolutional neural network, so as to identify human activities in the bathroom.

2. The water-saving intelligent shower system according to claim 1, characterized in that: The controller is also used to, based on an environmental sensor arranged in the bathroom, use the ambient temperature and ambient humidity measured by the environmental sensor as the input variables of the fuzzy control algorithm, coordinate the data of the temperature sensor and the data of the flow sensor, and obtain the second control value of the solenoid valve and the first control value of the heater based on preset fuzzy inference rules.

3. The water-saving intelligent shower system according to claim 2, characterized in that: Obtaining the first control value and the second control value based on a preset fuzzy inference rule includes: Presetting the data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity as a plurality of fuzzy sets with different gradients; The data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity obtained in real time are fuzzified into the corresponding fuzzy sets. Based on the fuzzy inference rules, logical reasoning and defuzzification are performed on the fuzzified data of the temperature sensor, the data of the flow sensor, the ambient temperature, and the ambient humidity to obtain the first control value and the second control value. The fuzzy inference rules at least include: when the ambient humidity is fuzzified into a high-gradient fuzzy set, the output second control value is a low-gradient value.

4. The water-saving intelligent shower system according to claim 1, characterized in that: When the controller identifies the activities of a human body in a bathroom, it includes: Preprocessing the WiFi channel status information, reconstructing it into an input matrix shape of the convolutional neural network, denoising the reconstructed WiFi channel status information based on a wavelet denoising method, and storing it as a set of images, wherein each pixel in the image represents an absolute value of a specified subcarrier at a specified antenna and time point; Based on the convolutional neural network, the input image is classified into human activity types, and the duration of the activity is recorded. When the duration of a specified activity exceeds a threshold, an alarm is issued.

5. The water-saving intelligent shower system according to claim 4, characterized in that: The convolutional neural network comprises: Three 2D convolutional layers; The 2D convolutional layers of the first and third layers are followed by a Dropout layer; After the 2D convolutional layer of each layer performs convolution operation, the maximum pooling, batch normalization and LeakyReLU activation functions are used; The final output layer is a fully connected layer with 160 neurons, the activation function is Sigmoid, and the human activity type classification is performed by the Softmax classification layer.

6. The water-saving intelligent shower system according to claim 4, characterized in that: When obtaining the classification of the human activity type, the water flow is binarized, and based on the binarized water flow, the classification probability of each human activity type output by the convolutional neural network is weighted, and the human activity type with the highest probability is taken as the final human activity type.

7. The water-saving intelligent shower system according to claim 4, characterized in that: The human activity types include at least: undress; Dry yourself off; still; Washing hair and face in an excited state; Washing the body and legs in an excited state; Wash your hair and face slowly; Wash the body and legs slowly; fall.

8. The water-saving intelligent shower system according to claim 4, characterized in that: When reconstructing the WiFi channel state information into an input matrix shape, it includes: The 4D WiFi channel status information is processed into a 3D input matrix, and the WiFi channel status information is expressed as n*m*f*t, where n is a WiFi receiving antenna, m is a WiFi transmitting antenna, f is a subcarrier of different frequencies for data transmission between the WiFi receiving antenna and the WiFi transmitting antenna, and t is the number of time steps. The 3D input matrix is ​​expressed as H*W*D, where H and W are the height and width of the matrix, respectively, and D is the depth. During reconstruction, H=n*m*f, W=t, and D=1, so that the subcarrier data of the WiFi receiving antenna and the WiFi transmitting antenna are stacked in sequence.

Images