TITLE .
A method for cleaning of washing/dishwashing articles in a washing/dishwashing machine and a device for performing the method
TECHNICAL FIELD
The present invention relates to a method for cleaning of washing/dishing material in a number of steps in a washing machine/dishwasher, into which the material is placed in a cleaning space that may be closed, water is supplied in one or more cycles to the cleaning space with or without chemical detergents, the material is cleaned during a certain time and at a certain temperature during one or more cycles in a cleaning process, the water with cleaning remains is discharged in one or more cycles from the cleaning space where supply of water, the cleaning process and discharge of water is controlled in dependence of, for instance, detection of the presence of cleaning remains and detergent in the water.
The present invention also relates to a device for cleaning of washing/dishing materialϊn a washing machine, with a cleaning space that may be closed into which the material is placed, where water is supplied in one or more cycles to the cleaning space with or without chemical detergents, the material is cleaned during a certain time and at a certain temperature during one or more cycles in a cleaning process, the water with cleaning remains is discharged in one or more cycles from the cleaning space where supply of water, the cleaning process and discharge of water is controlled in dependence of, for instance, detection of the presence of cleaning remains and detergent in the water.
BACKGROUND ART
The development of washing machines/dishwashers has gradually advanced towards a higher degree of automatization in order to ease the handling of in particular machines for household washing/dishing, enhancing running economics and sparing the environment.
Previously, different sensors have been proposed for detection of different parameters during a cleaning process with the intention to make it more efficient. For example, US patents 5291 626 and 5 446 531 disclose sensors that are arranged to detect if the cleaning water is muddled in order to determine the number of rinses. However, up to now the user has still chosen different washing programs for different types of material and different degrees of dirtiness by settings of manual controls. Further, one has up to now been able to take different water qualities into consideration only in a limited degree, for example one has taken the water hardness into consideration in order to determine the detergent dosage.
DISCLOSURE OF INVENTION
The purpose of the present invention is to further automatize and to increase the capacity of the cleaning process in washing machines/dishwashers.
Said purpose is achieved by means of the method according to the present invention, which is characterized in that the steps comprise:
• electrochemical detection of the water using voltammetry by means of application of a varying electric potential over at least two electrodes, which are in contact with the water in order to form an electric circuit and measurement of the present electric current's magnitude in the electric circuit,
• analysis of the current's magnitude as a response to supplied potential using multivariate data analysis (MVDA) to acquire information in dependence of the contents of the water,
• comparison with reference data for determining the contents of the water during the cleaning process,
• control of the cleaning process based on the determined contents of the water with respect to predetermined parameters, at least processing time and processing temperature.
Said purpose is also achieved using the device according to the present invention, which is characterized in that the device comprises sensor instruments for electrochemical detection of the water using voltammetry by means of application of a varying electric potential over at least two electrodes, which are in contact with the water in order to form an electric circuit and measurement of the present electric current's magnitude in the electric circuit, by instruments for analysis of the current's magnitude as a response to supplied potential using multivariate data analysis (MVDA) and instruments for comparison with reference data for determining the contents of the water and control of the cleaning process based on the determined contents of the water with respect to predetermined parameters, at least processing time and processing temperature
BRIEF DESCRIPTION OF DRAWINGS
The invention shall in the following be described using a preferred embodiment with reference to the enclosed drawings, on which fig. 1 schematically shows the method and the device according to the invention, fig. 2-4 show examples of recognition patterns that are used for the control of the cleaning process, and fig 5 and 6 show examples of selected cleaning processes in a machine as a result of detected water contents.
MODE FOR CARRYING OUT THE INVENTION
The invention is partly based on the application of a previously known technology in a new context, namely the use of an "electronic tongue" for controlling machines for, in particular, household washing/dishing. The principle behind the electronic tongue that is used for this application is based on electrochemical measurement of the type voltammetry. This measuring technique uses electrodes that serve as working electrodes and preferably are made of platinum, rhodium or stainless, which together with an opposing electrode are in contact with an electrically conductive fluid, in the current application the water that is supplied to a washing machine/dishwasher. The term water here refers to the fluid that is supplied to the machine and is used during the washing/dishing process and is discharged from the machine. This water is thus not chemically pure water, but contains in the different stages different soluble or non-soluble substances such as minerals and other impurities, chemical washing/washing-up detergent and/or impurities from the washing/dishing material.
The measuring technique voltammetry is based on that a varying potential is supplied to the electrodes, where an exchange takes place between the working electrodes and the current response is measured in the formed closed electric circuit. The current response, including transients, is detected and evaluated at different potentials and for each connected working electrode, when a recognition pattern is created, providing information regarding the contents and character of the fluid. The voltammetry measurement as such is not an object of the present invention, but is described in detail in the published international patent application with publication number WO99/13325, which is referred to and hereby is included to this application by reference.
By using the recognition pattern that is created at the voltammetry measurement, information may be created that is representative for the composition of the water before a cleaning process as well as during and after the process. According to the invention, this information is used for controlling the cleaning process, where parts of the process or the whole process may be controlled regarding different parameters. Examples of parameters are times and temperatures for different cycles, dosage of detergent, water amount, choosing type and number of cycles, such as pre- washing/dishing, main washing/dishing, rinsing and (when washing) spin- drying.
The system scheme according to fig. 1 illustrates the method and device according to the invention for cleaning washing/dishing material in a washing machine/dishwasher. Blocks made up by a dashed and dotted line illustrates the electrochemical sensor unit 1 that uses volammetry. For this purpose, a washing machine/dishwasher is equipped with an electrode unit 2 with one or more working electrodes 3, 4, consisting of a stable metal, such as platinum, rhodium, or stainless, enclosed in an electrically insulating material, preferably with an end surface in contact with the washing/dishing water. When two or more working electrodes 3, 4 are used, different metals are chosen for each electrode, which results in different electrochemical courses of events, and thus supplementary information regarding the contents of the water. At a suitable distance in the vicinity of the working electrodes an opposing electrode 5 made of, for example, stainless steel, is arranged. The electrode unit 2 and the opposing electrode 5 are arranged in a suitable manner in the washing machine/dishwasher, either directly in the cleaning space of the machine, or in a special container, communicating with the space, by way of example space for a circulation pump. Possibly further sets of electrodes may be may be arranged, for example a set at the water intake and outlet in order to selectively detect intake water, process water and outlet water. The electrochemical sensor unit 1 comprises a control unit 6 that is arranged to apply a varying potential, see e.g. the curve E(t), to one working
electrode 3, 4 at a time and the opposing electrode 5, according to a predetermined pattern. The potential may vary continuously, e.g. linearly or be pulsed. At least two types of pulsed voltammetry may be used in this context, which provide different current responses. These are generally called LAPV (Large Amplitude Pulse Voltammetry) and SAPV (Small Amplitude Pulse Voltammetry). Further, the electrodes and the control unit 6 are connected in such a way that a closed electric circuit 7 is formed, through which an electric current flows, see for example the curve E(t), as a response to the applied potential. The control unit 6 comprises a potentiostat which supervises that the predetermined potential at a certain moment is maintained and which measures the created current response. Further, the control unit 6 sees to that changing takes place between the working electrodes 3, 4 in combination with variation of the potential, whereby the current response, including its transients, provides information regarding the electrochemical course of events in the cleaning water. The technology described in the international patent application stated above may be used for the current electrochemical sensor 1 , but the technology may also be modified for adaptation to the current application.
The measurement of the current at different potentials and changes of potential at certain determined points of time and using a certain working electrode at different moments during the cleaning process results in an amount of data, which shall be reduced and sorted in order to be evaluated and used for control of the cleaning process. For this purpose, an analysis step'8 is included, receiving the great amount of data from the sensor unit 1. The function of the analysis step is based on a principally previously known analysis method, called multivariate data analysis MVDA. MVDA has two main purposes, one is to obtain structure and correlation for data, the other is to achieve calibration models which may predict groupings of data. MVDA may use different methods for treating data from the sensor unit.
A multivariate data analysis method which may be used for the present invention is PCA (Principal Component Analysis). PCA creates an overview for the great amount of data received from the sensor unit 1 , in this case the voltammogram. Information regarding the following may be acquired:
• Important/less important variables
• Correlation between variables
• Systematic variation separated from noise
• Deviating object • Groups of objects
PCA is a mathematical transformation that is used to explain the variance of a matrix (called the X matrix) with the number N of objects (i.e. measurements) and the number K of variables (i.e. output signals from the sensor unit 1 ). This creates a multidimensional space of K dimensions, containing N points.
By using PCA a vector is computed, describing the largest variance, i.e. the direction which describes the largest difference between the observations. This is the first main component PC1. The second main component PC2 is orthogonal to PC1. PC 2 describes as much as possible of the remaining information. This proceeds until all information has been accounted for.
The dimensions K are thus reduced to a lesser number of dimensions, which are defined by the main components. Not only he dimension have been reduced, but the latent structure of raw data, such as chemical or physical changes, may also be visualized. The main components define a plane maximizing the variation of raw data, where data are projected on this plane. This results in a resulting image, a so-called score-plot.
Another MVDA method is constituted by PLS (Partial Least Square), which also is called "Projection to Latent Structures". PLS does not only require process or sensor data in a matrix, but also data for a Y-matrix, which for example may consist of the results, known concentrations or biologic activity. A model is created to:
• Find a relationship between X and Y
• Predict new Y
• Create a model for one or more y variables
A PLS is created by performing a PCA on the X matrix and the Y matrix, after which a linear regression is carried out for each PC between the results for the X and Y matrixes. The algorithms try to maximize co-variances between X and Y. The purpose is to acquire a regression model between those X and Y matrixes that may be used for, by way of example, predicting unknown substances. PLS may handle "missing data", i.e. non-complete matrixes, in conformity with PCA.
A further alternative MVDA method that may be used for the present invention is ANN (Artificial Neural Networks). Using ANN, a model that may assume almost any mathematical transformation may be created. ANN is also tolerant regarding noise and errors. The disadvantages is that the network requires more measurements and a longer learning time compared to PLS. ANN can not handle many variables, since the number of required observations thus will increase rapidly. After learning and optimization the model may:
• Classify complicated (also non-linear) patterns
• Predict new objects
ANN is constructed to emulate the way in which the human brain works. In the brain, the function is based on signal transmissions between the neurones in a complicated network. The neurones are connected via synapses.
In the MVDA step 8 (consisting of a microcomputer), the neurones have been replaced by nodes. A node receives information from many other nodes, executes a simple computation of the information, and forwards it to the other nodes. The information strength is determined by the coupling constant of the node, which is multiplied with the strength of the signal. Such a net has the capability of learning and memory storage.
ANN is a layer structure, there is an input layer, hidden layer(s) and an output layer. The hidden layer(s) and the output layer are those that are active and treat information. The number of nodes in each layer and the number of hidden layers is determined by the user and depends on the problem. The properties and the knowledge of the network is determined by the characteristics of the single node (the weight factor) and the internal arrangement of the nodes (topology).
There are many learning algorithms available that are useful for learning the net. During learning, initial values from ANN are compared with real values, thus adjusting the coupling constant to provide minimal differences by minimizing the sum of the squared error.
Examples of other MVDA methods that may be used in this context are:
• MLR (Multiple Linear Regression), a simple linear method for prediction and calibration • PCR (Principal Component Regression), a linear method for prediction and calibration
• PPR (Projection Pursuit Regression), a non-linear method for prediction and calibration
After the reduction and structuring of data from the sensor unit 1 described above using the analysis step 8 MVDA, see fig. 1 , data indicating the contents of the cleaning water may thus be acquired, as will be explained in greater detail below.
The contents in the cleaning water is advantageously detected or measured at different stages of the cleaning process, either at different time intervals or continuously. It is also conceivable that two or more sensor units are placed in the machine to detect the cleaning water either parallel or in sequence.
An advantageous measurement or detection may be made for incoming water either in the form of a special sensor unit placed at the inlet or measuring during a short time when filling-up, before the water is affected by the cleaning material. The measure may take place before and after chemical detergents are supplied. An important parameter regarding water quality is water hardness, which depends on the contents of calcium and magnesium. This measurement may be used to supervise water softeners in for example a dishwasher, by way of example by controlling a supply of salt. There are, however, other parameters in the water that are important for the general water quality, such as copper and iron.
A second measuring step is to measure the water before the cleaning process is started. According to the invention, it may at this stage be determined which kind and what amount of dirt and impurities that are present in the machine by letting the washing material or the dishing material get into contact with the cleaning water, thus mixing or solving the contents in the water and enabling detection by the sensor unit 1. Based on the performed measurements, it is possible to choose a cleaning program that is optimal for cleaning while using as little water and power as possible.
A third measuring step may be executed during the cleaning process itself, where the process may be followed and controlled so that the process runs according to plan. If something unexpected occurs, for example if the user opens the door to the dishwasher and inserts more dirty dishing material, the machine may react to this and adapt the process by changing certain parameters such that the method secures a desired dishing result, usually that the dishing material is clean.
A fourth measuring step may be performed during the rinsing process, where the outlet water may be checked for being clean enough before the materiel is clean. Except dirt and impurities, it is essential to secure that all the detergent has been removed by rinsing.
Fig. 2 shows a concrete example of the score plot described above that is acquired during measurements of different types of incoming water qualities and after multivariable data analysis MVDA of the type PCA. The score plot shows results from a number of measurements, on one hand before chemical detergent is supplied, on the other hand after supply of chemical detergent. The result is presented in a system of coordinates where the measurement results are grouped in different positions in the system of coordinates. In the current example, position 9 and 10 thus represent distilled water before and after supply of detergent, respectively, position 11 and 12 soft seawater, position 13, 14 relatively hard tap water, position 15 and 16 softened hard tap water and position 17 and 18 not softened hard water, all in each group representing water before and after supply of detergent. In this way, by the positions of the different groups, one acquires a detection of the quality of the water, which consequently is given a new position after supply of detergent. As the results are recurring with limited deviations, the positions may be used to provide control instruction to the machine depending on the water quality.
Fig. 3 shows an example of a score plot or resulting image that has been acquired from measurements of four typical cases in a washing machine at the beginning of the washing process, when the washing material has been soaked. Each point originates from a measurement, where the washing material in the machine was dirtied with the current dirt. The score plot has in this case been acquired after using multivariate computer analysis of the type principal component analysis (PCA) too. After this analysis too, it may be observed that dirt of different characters has obtained clearly separated positions in the score plot. Here, the washing material has been dirtied before each measurement with one kind at a time of the dirt types that are used in the occurring standard tests, namely wine that appears at point or position 19, blood in position 20, oil in position 21 and chocolate in position 22. It may be added that in this case, the measurement has been performed after supply of detergent, and thus with a dirt type at a time for each test run.
Fig. 4 shows a corresponding measurement in a dishwasher, in this case, however, before supply of detergent. Here, a number of test runs was also performed, where one single dirt type occurred at each run. It could be observed, that the different substance types had a relatively good separation in the score plot. Thus margarine has obtained position 23, egg position 24, minced meat position 25, spinach position 26, tea position 27, milk position 28 and porridge position 29. It may be observed that egg and minced meat lie relatively close to each other, depending on that the minced meat contains egg. Spinach and tea also lie relatively close to each other, depending on similarities in their character. It may also be observed that by this reason, the substances that lie close to each other in position thus need similar a cleaning process.
Irrespective of how many types of dirt the washing/dishing material contains, each measurement results in a single measurement point after treatment in the MVDA step 8. The position of the measurement point is affected by the
particular types of dirt and the extent of their presence and thus provides information about this.
The measurement points from the particular types of dirt and mixtures thereof will thus end up somewhere in the space that is described by the score plot. By learning the machine control step 30, see fig. 1 , which area that corresponds to which dirt mixture (and which washing/dishing program that is necessary for each area) it is possible to divide the score plot into many specified areas. In Fig. 5 a schematical drawing of how such a division may look like is shown in a two-dimensional score plot. The learning step demands many measurements of many different dirt mixtures. Each area 36 thus corresponds to a certain dirt mixture.
When measurements take place continuously, the measurement points will move from area to area, and the program will thus adapt in order for the process to follow an optimal path to the desired end result. This is illustrated in fig. 6, where a dishwasher process is followed (from right to left) by quick, continuous measurements.
Fig. 7 and 8 show examples of changes to the cleaning process in a normal program. Fig. 7 shows an example of a normal dishing program that is visualized with a broken-line curve 32, where temperature changes over time during the run of the dishing process are visualized. For the sake of clarity, a case where the dirt solely consists of margarine has been chosen. Thus a program that is visualized with the curve 33 was chosen in the machine control step 30. The total change as a result of the invention, according to which the measurement has been performed, is
Time: -20% Washing-up detergent: -20% Supplied power: no change Water: -40%
By measuring the water contents with the sensor unit 1 , computer analysis of the measurement information in the analysis step 8 and choice of dishing process in the machine control step 30, a temperature curve over time was acquired, shown with a solid line 33 and thus clearly deviates from the normal program.
In a corresponding way, fig. 8 shows a simple situation in a washing machine in order to show the principle according to the invention. A normal program is shown with a broken line 34, which implies relatively high temperatures during the run of the washing process. Except power in the form of heat, mechanical treatment of the washing material is a parameter that may be controlled in dependence of detected dirt or impurities in the cleaning water, i.e. running time or possibly spinning velocity for the movement of the drum. When only blood is detected in the laundry, the machine control step 30 chose the following change, that also resulted in a curve 35 according to the solid line.
Time: -15% Washing-up detergent: -20% Supplied power: -75% Water: no change
The points 37 on the score plot in fig. 6 thus represent current status of the cleaning water at each measurement, where an indication of the final washing/dishing result 38 may be put in relation to the water quality (possibly after softening) in the process water or in the outlet water. However, in practice this does not mean that the outlet water shall have the same quality as the inlet water. In certain cases it is desirable to end the process, without obtaining final cleanness for the material, e.g. if one only wishes to the rinse the dishing material, or if one for any reason is satisfied with a lesser clean laundry. Thus one chooses an ending point 40, see fig. 6, i.e. ends the
process at a point which deviates from a terminal point that normally corresponds to the final washing/dishing result.
The invention is not limited to what had been described above and the examples shown in the drawings, but may be varied within the scope of the appended claims. Dosage of washing/washing-up detergent may, for example, be a part of the process, i.e. the dosage is controlled in dependence of partly the detected water quality, partly the continuously measured result. Choosing between different washing/washing-up detergent types may also be a part of the dosage control. It shall be added, that the tongue control step 6 as well as the analysis step 8 and the machine control step 30 are realized in the form of microcomputers (microchips) programmed with computer programs to execute the functions described above. By way of example, the tongue control 6 is realized as a microchip that also contains the potentiostat mentioned above and an A/D converter, while the analysis step 8 and the machine control step 30 are included in a second microchip.